Newspaper Industry Environmental Technology Initiative

Understanding Recycling
Behaviour Volume
Technical Monograph 2

Peter Tucker

Cheshire Recycling 
Understanding Recycling Behaviour
Volume 2

A Technical Monograph

PETER TUCKER

Copyright © The University of Paisley, 2003

All rights reserved. No part of this work may be reproduced by any person for any
purpose in any medium without written permission from the University of Paisley.

ISBN 1-903978-16-5
 About the Author
Professor Peter Tucker holds the Newspaper Industry Foundation Chair in Environmental
Technology at the University of Paisley. He is a physicist by training and has a doctorate in
geophysics. He spent the early part of his career working as a scientist in the UK Department
of Trade and Industry's Warren Spring Environmental Technology Laboratory, where he rose
to become head of Environmental Modelling and Statistics. Environmental modelling has
always been at the fore of his research interests, and he has developed a strong interest in
modelling uncertainty in complex environmental systems. During his career, he has
researched, developed and deployed modelling across a wide range of applications including
air and marine pollution, mineral processing and reclamation, land contamination and
environmental and waste management.

He has been involved in a number of Government waste research programmes to gather

national waste statistics and managed the database development for the National Household
Waste Analysis Programme. Since joining the University of Paisley in 1995, he has
strengthened and developed the University's thriving research initiatives across environmental
science, though has maintained a personal research focus on the environmental management
of solid and wet particulate wastes, including recycling, composting, and land application of
wastes.

Professor Tucker is a strong advocate of interdisciplinary and multi-disciplinary approaches to

problem solving, and is quite delighted to mix psychology with process engineering in order to
help facilitate understanding. He is also highly committed to building in social and
environmental factors into the decision making process.

Professor Tucker is the author of over 50 publications on solid waste management.

ii
 Foreword from the Author

Understanding Recycling Behaviour Volume 1 was well received by the waste management
community. The only criticism, perhaps, was that some found it too technical and too
academic in its content. I make no apologies for that. It was deliberately designed as a
technical document, providing rigorous scientific analyses that could underpin our
understanding of recycling behaviour. Plenty of other research has been carried out, and
plenty of popular reports and surveys have been published without that scientific rigour. Whilst
they provide good „coffee table‟ statistics they rarely further the scientific understanding. The
sponsors of this research were keen to inject more fundamental „hard‟ science into the subject
to help drive our understanding forward. So why a technical monograph? The answer is
simple. Most scientific publications by their very nature are normally restricted to the reporting
of single issues. Many are also limited by word counts. They can only provide a window on
the whole. The monograph provides a vehicle to link different issues together and to develop
the full holistic picture. Understanding recycling behaviour is not just about understanding
peoples‟ attitudes, or about understanding the impacts of programme design, or
understanding why or why not recyclers might compost. It is about developing the whole
picture.

Understanding Recycling Behaviour Volume 2 follows the tradition of being a technical

treatise that aims to further the science and our understanding. It does not provide all the
answers, though it does provide many new insights and pointers. Volume 1 set out the
scientific framework. Volume 2 now furnishes more of the details.

The research presented in this monograph focuses on the people. Minds may be fixed on
Government targets, but it is the people who will determine whether those targets will be met.
Understanding the people is crucial. Volume 1 of Understanding Recycling Behaviour finished
with the observation that “Everyone is different, but some are more explainable than others”
Volume 2 starts from the premise that “Everyone is different, but how significant are those
differences?” It also tackles the questions of “What causes people‟s behaviours to change?”
and “Are their new behaviours meeting expectations?” and “If not, why not?”

Understand Recycling Behaviour Volume 2 draws the results of the past 3 years of research
at the University of Paisley together into a single volume.

iii
 Acknowledgements

First and foremost I would like to thank the past and present sponsors and other supporters of
the Newspaper Industry Environmental Technology Initiative for their vision and deep interest
in the research. Those supporters include Associate Newspapers Ltd., Daishowa Forest
Products Ltd., Donohue Inc., Flint Ink Europe, Holmen Paper AB, Norske Skog., Shotton
Paper Co., The Newspaper Society, The Newspaper Publishers Association, and SMG.
Special thanks are given to the principal sponsors of the Initiative: Bridgewater Paper Co.
Ltd and Cheshire Recycling Ltd., Stora Enso, and Sun Chemical.

Special Thanks are also given to the co-funders of the research: The Cabinet Office Strategy
Unit, Lancashire Waste Services Lancashire Environmental Fund, Waste Recycling
Environmental Group (WREN), and North Warwickshire Borough Council.

The work would not have been possible without the support of the following local authorities,
recycling organisations and, of course, their local residents. All are gratefully thanked.

Blackburn with Darwen Borough Council, Blackpool Borough Council, Burnley Borough
Council, Chorley Borough Council, Fylde Borough Council, Hyndburn Borough Council,
Lancaster City Council, Pendle Borough Council, Preston City Council, Ribble Valley Borough
Council, Rossendale Borough Council, South Ribble Borough Council, West Lancashire
District Council, Wyre Borough Council, Clackmannanshire Council, Renfrewshire Council,
South Ayrshire Council, East Ayrshire Council, South Lanarkshire Council, Inverclyde Council,
North Warwickshire Borough Council, North East Derbyshire District Council, Chesterfield
Borough Council, Lancashire County Council, Hampshire County Council, and Alloa
Community Enterprises.

Finally, I wish to pay tribute to all the hard work, commitment and enthusiasm shown by the
Environmental Technology research team at the University of Paisley: David Speirs, Pat
Douglas and Jackie Polson. This monograph would not have been possible without them.

The monograph comprises a mixture of published and not previously published data. Special
acknowledgement is given the UK Cabinet Office Strategy Unit and to the Chartered
Institution of Wastes Management for permission to include the following material from their
original publishers:

Model Forecasts of Recycling Participation Rates and Material Capture Rates for Possible
Future Recycling Scenarios (2002). University of Paisley Report to the Strategy Unit.

Tucker, P., Speirs, D. (2003). Relationships Amongst Waste Management Behaviours. CIWM
Scientific and Technical Review, 4(1), 2-8.

Tucker, P., Speirs, D. “Count Me As Doing My Bit”: An Appraisal of the Accuracy of Waste
Attitude/ Behaviour Surveys. CIWM Scientific and Technical Review, 4(2)

iv
The Newspaper Industry Environmental
Technology Initiative
The Initiative was born in 1995, founded by a consortium of multi-national companies,
amongst the world leaders in the paper and inks industries. The major sponsors are
Bridgewater Paper Co. and Cheshire Recycling, Stora Enso, and Sun Chemical. Those
companies share a strong commitment to the environment, which brought them together to
plan strategically for a sustainable and healthy future for the industry and for society.

The NEWSPAPER INDUSTRY ENVIRONMENTAL TECHNOLOGY INITIATIVE

seeks to carry out environmental research into the TOTAL ENVIRONMENT OF
THE NEWSPAPER RECYCLING LOOP, to gain new or improved data sets and to
place these data sets in the public domain, and to disseminate the findings
widely amongst industry, Government and the general public.

It aims to research, develop and exploit new methods, processes and

procedures to promote continual environmental improvement and enhance
environmental decision making.
Research carried out under the Initiative comprises independent, objective
investigations, of high scientific integrity with output being subjected to
stringent, external, scientific peer review before publication.

"The Newspaper Industry Environmental Technology Initiative is not about one-off

projects, it seeks to provide the whole picture"

The themed research covers:

 Consumer Recycling Behaviours

 Recycling Performance Modelling, Prediction and Diagnostics

 Environmental Monitoring and Impact Assessment of Newspapers and their Recycling

Residues

 Materials Flow Analyses of the Newspaper Sector

 Markets and New Technologies for Old Newspapers and their De-inking Residues

Understanding Recycling Behaviour Volume 2 presents the latest research results from
the first two themes. It is the fourth technical monograph to be produced under the Newspaper
Industry Environmental Technology Initiative. The previous titles were:

 Understanding Recycling Behaviour

 Newspapers, Metals and the Environment
 Understanding Home Composting Behaviour

Summaries of all that previous research are available on the Newspaper Industry
Environmental Technology Initiative website <http://www.paisley.ac.uk/environment>

v
 Contents
About the Author ii
Foreword from the Author iii
Acknowledgements iv
The Newspaper Industry Environmental Technology Initiative v
Contents vii

Executive Summary xi
1. Introduction and Review 1
1.1 Research Goals 1
1.2 Structure of the Monograph 2
1.3 Factors Affecting Behaviour 3
1.3.1 Programme Design 3
1.3.2 Education and Promotion 5
1.3.3 Demographic Factors 7
1.3.4 Past Experience and History 7

1.4 Case Study Review 8

1.4.1 Introduction 8
1.4.2 Millennium Recycling Scheme 9
1.4.3 Milton Keynes 10
1.4.4 Wealden 11
1.4.5 Babergh 11
1.4.6 Eastleigh 12
1.4.7 Kerbside Glass Recovery 12
1.4.8 Hertsmere 13
1.4.9 Kensington and Chelsea 14
1.4.10 Other Schemes 14

1.5 Cross Authority Comparisons 15

1.5.1 UK Schemes 15
1.5.2 US Schemes 16

1.6 Intra-authority Comparisons 17

1.7 Discussions and Conclusions from Previous Research 17

2. Where are we Going 19

2.1 Introduction 19
2.1.1 Aims 19
2.1.2 Methodology 19
2.1.3 Model Calibration 20
2.1.4 Scenarios Modelled 20
2.1.5 Definitions 20

2.2 North West England 21

2.2.1 Intensification of Bring Facilities 21
2.2.2 Full Kerbside Coverage Regionwide 22
2.2.3 Organic Waste Recoveries 25

2.3 South West England 26

2.4 Comparability between Regions 28

vi
2.5 Discussion and Conclusions 29

3. Does the Programme Design Matter? 31

3.1 Introduction 31
3.2 Results 31
3.2.1 Total Kerbside Yield 31
3.2.2 Paper and Card Yields 32

3.3 Conclusions and Further Evidence 36

4. Implementing Change: A Case study in 39

Chesterfield and North East Derbyshire
4.1 Background 39
4.1.1 Recycling Provision 39
4.1.2 Data Considerations 39

4.2 Dry Recyclate Yields 40

4.3 Effect of Alternate Green/ Residual waste Collection 43
4.4 Effect on Bring Site Recoveries 44
4.4.1 Methodology 44
4.4.2 Results 44

4.5 Demographic Dependences 47

4.5.1 Methodology 47
4.5.2 Results 47
4.5.3 Alternate Residual/ Green Waste Collections 50
4.5.4 Similarities Between Rounds 51

4.6 Model Predictions 52

4.6.1 Current Performance 52
4.6.2 Future Scope 55

4.7 Summary and Conclusions 56

5. Results from other Communities 59

5.1 Lancashire and North West England 59
5.1.1 Introduction 59
5.1.2 Whole district Statistics 59
5.1.3 The Influence of Demographic Factors On Recycling Yields at 62
Neighbourhood scale
5.1.4 Conclusions 64

5.2 North Warwickshire 64

5.2.1 Introduction 64
5.2.2 Dry Recyclate Yields 64
5.2.3 Demographic Dependences 66
5.2.4 Similarities between Rounds 67

5.3 Discussion and Conclusions 68

vii
6. Relationships Amongst Waste Management 69
Behaviours
6.1 Introduction 69
6.1.1 Review 69
6.1.2 New Analyses 70

6.2 Basic Correlations 70

6.2.1 Levels of Activity 70
6.2.2 Correlations between Behaviours 71

6.3 Factor Analysis 74

6.3.1 Methodology 74
6.3.2 Results 75

6.4 Correlations and Length of Experience 77

6.5 Behaviours and Demographics 78
6.5.1 Significant Relationships 78
6.5.2 Relationships between Materials 79
6.5.3 Recycling, Reduce / Reuse and Demographics 81

6.6 Discussion and Conclusions 83

7. Time series Analyses 85

7.1 Introduction 85
7.2 Between District Comparisons 85
7.3 Intra-district Variations 90
7.4 Summary and Conclusions 93

8. Attitudes, Demographics and Behaviour 95

8.1 Literature Review 95
8.1.1 Attitudes 95
8.1.2 Attitude and Behavioural Models 95
8.1.3 Experience and Habit 97
8.1.4 Knowledge 97

8.2 Introduction to the Experimental Study 98

8.2.1 Pointers from Home Composting Behaviour 98
8.2.2 Focus of the Current Study 99

8.3 Attitude/ Behaviour Research in Renfrewshire 100

8.3.1 Introduction 100
8.3.2 Demographics and Participation 100
8.3.3 Attitudes 101

8.4 Attitude/ Behaviour Research in Clackmannanshire: 102

Part 1 Before Change
8.4.1 Introduction 102
8.4.2 Demographics and Behaviour 103
8.4.3 Material Capture Rates 104
8.4.4 Knowledge and Information 106
8.4.5 Attitudes 108
8.4.6 Attitudes and Demographics 112
8.4.7 Further Considerations of Attitudes 113
8.4.8 Performance Indicators 114

viii
 8.4.9 Discussion and Conclusions 116

8.5 Attitude/ Behaviour Research in Clackmannanshire: 117

Part 2 After Change
8.5.1 Introduction 117
8.5.2 Whole Sample Analyses 117
8.5.3 Prior Recyclers and New Recruits 119
8.5.4 General Comments and Conclusions 120

8.6 Normative Influences 121

8.6.1 Literature Review 121
8.6.2 Previous Research at the University of Paisley 121
8.6.3 New Results 122

8.7 Summary and Conclusions 123

8.7.1 General 123
8.7.2 Recyclers versus Non-recyclers 123

9. How Accurate are Waste Behaviour Surveys? 125

9.1 Introduction 125
9.1.1 Review 125
9.1.2 New Evaluations 126

9.2 Participation Behaviours 127

9.2.1 Results 127
9.2.2 Determinants of Participation Behaviour 129

9.3 Exaggerated Behaviours 130

9.3.1 Results 130
9.3.2 Determinants of Exaggerated Behaviours 132

9.4 Bias in Questionnaire Returns 133

9.5 Summary, Discussion and Conclusions 134

10. Waste Arisings 137

10.1 Review 137
10.1.1 Context 137
10.1.2 Methodologies 137
10.1.3 Early Research 138
10.1.4 Subsequent Research 138

10.2 New Results 139

10.2.1 Items Consumed 139
10.2.2 Consumptions and Recycling Behaviour 142

10.3 Conclusions and Implications 144

11. Discussion 147

11.1 Government Targets and Human Behaviour: Bridging the Gap 147
11.1.1 What are the Endpoints? 147
11.1.2 Designing the Programme 147
11.1.3 Making the Changes 148

ix
11.2 Performance Monitoring and Evaluation 150
11.2.1 Introduction 150
11.2.2 Methodological Limitations 151
11.2.3 Some Current Data Limitations 152
11.2.4 Household Surveys as Monitoring Instruments 153
11.2.5 Longitudinal Monitoring 154
11.2.6 Conclusions 155

11.3 Do Demographics Really Matter? 156

11.3.1 The Underlying Premise 156
11.3.2 Who is Now the Recycler 156
11.3.3 Who will Become the Recycler 157

11.4 Attitudes and Behavioural Change 157

11.4.1 Fundamental Attitudes 157
11.4.2 Initiation and Persistence 158
11.4.3 The Role of Knowledge 160
11.4.4 „Natural‟ Change 160

11.5 A Conceptual Model of Recycling Behaviour 161

11.5.1 Basic Concepts 161
11.5.2 Visualising Behavioural Change 163

11.6 Promotional Campaigns 166

11.6.1 Visualisation of Promotional Campaigns 166
11.6.2 Salience and Susceptibility 167
11.6.3 Message Content 167

12. Concluding Remarks 169

12.1 Research Direction 169

12.2 Barriers and Opportunities 170

12.2.1 Getting Sufficient and Representative Data 170
12.2.2 Opportunities 170

12.3 The Final Words 171

References 173

Appendix A 179
Market Research Indicators 179
Social Classes 180

x
 Executive Summary
Aims and Scope

This monograph details the research output achieved between January 2001 and June 2003
at the University of Paisley, under the Newspaper Industry Environmental Technology
Initiative sponsored by Bridgewater Paper Co. Ltd., Cheshire Recycling, Stora Enso, and
Sun Chemical. The new research presented here extends and complements that previously
published in Volume 1 of Understanding Recycling Behaviour. The latest research includes
two new household waste behaviour and attitude surveys carried out in Scotland, in the
districts of Renfrewshire and Clackmannanshire respectively. It also includes detailed
statistical analyses of recent kerbside recycling data from Chesterfield and North East
Derbyshire, from North Warwickshire, and from the fourteen districts of the county of
Lancashire. In the course of the research, some 500 new household surveys have been
carried out, and the performances of over separate 130 kerbside collection rounds have been
analysed in detail.

The central research thrusts have been: (i) to diagnose and explain monitored recycling
performances – are they above or below expectations, and if so why? and (ii) to develop a
better understanding of what causes behavioural change.

What are the Endpoints?

The monograph starts out by asking the question “What recycling levels could we ultimately
achieve?” To most waste management professionals that means hitting the targets. However
this research is not about targets. It is about people. Government targets represent quite
arbitrary levels of household behaviour. So given our imperfect society, given its history, and
given its current recycling ethos, what is reasonably possible? The research broke down the
answer into: (i) what could be achieved by making recycling more convenient, providing
everybody with kerbside collections, and collecting more materials, and (ii) what must rely on
fundamental attitude changes and mass education. The study concentrated on two regions of
England: the north west and the south west, which typified the poorer and better recycling
regions of the country.

The conclusions of the study were that, in both regions, region-wide recycling rates of 40%
should be achievable through a combination of (1) multi-material kerbside collections, (2)
kerbside green waste collections potentially coupled with (3) kitchen waste collections, backed
up by (4) strong promotional education and awareness campaigns. However, it is unlikely that
a 40% national recycling rate could be achievable without every one of those elements being
in place.

The introduction of enhanced recycling infrastructures was predicted to lead to comparable

step changes in material capture across all districts. A relatively well-performing area would
continue to be well-performing with better recycling provision, whilst a relatively poorly-
performing area would continue to be relatively poorly-performing despite the enhanced
recycling provision. This effect was fully borne out in case study investigations undertaken in
Derbyshire and Warwickshire. Inter-district differences will only be fully eroded through
additional promotional and education campaigns. With comparable infrastructures and strong
successful promotions, all districts should reach much the same endpoints. Those endpoints
were predicted to be:

 80 - 82% capture of newspapers, pamphlets and magazines;

 80 - 82% capture of glass bottles;
 75 - 78% capture of glass jars, aluminium cans, steel cans;
 71 - 73% capture of cardboard;
 71 - 73% capture of plastic beverage bottles.

xi
In those scenarios 85-87% of households would actively carry out some kind of recycling
activity, with 75-80% of them using the kerbside schemes.

Even with high kerbside participations, bring sites cannot be dispensed with. Significant
recycling will continue at the sites in parallel with the kerbside recoveries. The national model
showed that bring site recoveries might drop by around 35%. In Chesterfield and North East
Derbyshire the measured drops for paper and glass were around 40%. However, the falls in
bring site yields of cans and textiles were marginal in both communities.

Achievements in Practice

Specific details of recycling programme design should not affect the predicted endpoints,
though certain features of the designs, e.g. co-mingled as opposed to source separated, bins
rather than boxes rather than bags do appear to give the edge in poorer-performing areas.
Collection frequency appears to be unimportant to success.

The analyses concentrated only on the different options for the recovery of the dry recyclable
components of domestic waste. They did not specifically address how the residual wastes
might be collected and how those collections might impact on recyclate recoveries. However,
in the case study in Chesterfield it was clear that recyclate yields were higher, sometimes
more than 50% higher, in the areas of the borough where the residual waste was being
collected fortnightly rather than weekly. However, the size of effect varied. The highest yields
were all realised in areas where the recyclate collections and the residual waste collections
were changed together. In the areas where the fortnightly residual waste collections were
introduced prior to the introduction of multi-material recyclate collections, the effects on those
recyclate collections were quite modest. Timing was a critical factor.

The case studies in Chesterfield, North East Derbyshire and North Warwickshire tracked
those districts through the conversions of their paper-only kerbside collections to multi-
material, collecting paper, glass, cans and textiles. Before the conversions, the yield from the
paper-only collection in North East Derbyshire was 1.31 kg per serviced household per week.
The yields in Chesterfield and North Warwickshire were lower at 0.68 and 0.63
kg/household/week respectively. The conversions to multi-material collections resulted in
significant improvements in dry recyclate yield in all three districts, recording yields of 2.38,
1.79, and 2.31 kg/serviced household/week in North East Derbyshire, Chesterfield and North
Warwickshire respectively. The yields from the areas in Chesterfield with the fortnightly
collections of residual waste were 2.9 kg per serviced household per week. The increases in
yield were not only attributable to the new materials being collected. The paper yields
increased substantially as well. Those increases were by factors of 1.1 to 1.4 in North East
Derbyshire, and by factors of 1.5 or more in Chesterfield. In North Warwickshire the paper
yields doubled.

The recovered material comprised around 63-67% paper, 28-29% glass, and 4% and 3%
cans and textiles respectively. Can and textile recoveries were somewhat lower than
anticipated. It was concluded that in the case of textiles it was due to a reluctance to transfer
allegiance away from the traditional outlets. Can recoveries were simply low at all outlets. It is
noted that poor can capture rates have been seen in many other schemes as well. Too high a
personal cost [to wash and squash], discard in non-household waste streams [at work or at
school], and uncertainties about whether they can be recycled may all be contributory reasons
why.

Mirroring the national predictions, the poorer paper-recycling neighbourhoods in all three case
study areas still remained the poorest producers in the new multi-material collections, all
neighbourhoods showing more or less comparable step changes in production. Thus, given a
comparable intervention, most people appear to have responded in the same manner. The
question then to be asked is “Why were they poor producers in the first place?” Is there some
attributable factor or attribute of those people that equates with low recyclate production? Are
the differences expected? [It should, of course, be borne in mind that low recyclate yields do
not necessarily mean low recycling participations. They could alternatively indicate low waste
generations].

xii
What about Demographics?

The notion that the differences in waste generation (and in recycling participation) may be
linked to socio-demographic differences has been around for twenty years or more, and is still
generally well accepted. Based on that premise, different demographic groups might be
expected to contribute different recycling yields. So if we filter out the demographic influence,
the remaining variation should [in theory] provide the picture of who is recycling better or
worse than expected. The problem is that we do not know a-priori what the actual
demographic influences might be. The influences can only be estimated through fitting models
to the measured performance data. What fits tells us what the local influences might be. What
doesn‟t fit tells us where other explanations are needed.

This methodology was applied to the routine performance monitoring data (collection round
tonnages) collected in Chesterfield, North East Derbyshire, North Warwickshire, and across
five districts of Lancashire. The results were highly equivocal with no explanatory factors
found in common across each of the districts. The traditional demographic indicators of „good‟
recyclers (more affluent housing stock, older residents, two person households, car owners
etc.) generally tended to show the largest positive influences on recyclate yields, whilst the
traditionally poorer indicators (poorer housing stock, private sector rentals, families with young
children, young adults, larger households) generally tended to show the smallest or more
negative influences, though the effects were weak and there were many exceptions. Most of
the demographic factors tested were not significant at all in explaining the behaviours. Those
that were significant took on different importances in different localities. Between-district
factors were found to be at least as important as local demographic variations within districts.

After fitting the models, very few rounds could be unambiguously identified as performing
better or worse than expectations. The rounds with the fortnightly residual waste collections in
Chesterfield stood out as performing much higher than expectations. The other outliers were
often areas with the most extreme demographic profiles. Perhaps the main point is that
around 80% of all areas all performed quite similarly, irrespective of their demographic
profiles. Put another way, 80% of the community were all performing reasonably similarly at
some „intermediate‟ performance level. The key to understanding lies in the understanding
why the other 20% were not.

The correlations between demographics and recycling participations were explored across
other districts as well, and much the same pattern emerged there also. The groups
traditionally considered to be „good‟ recyclers did appear to participate more, though not
necessarily significantly more, than the traditional poorer recycling groups, but results were
quite inconsistent between districts. There was also evidence that the more recent recruits to
recycling are now being drawn preferentially from the poorer recycling groups. However this
may simply be an artefact of the statistics – there may just be more of them to recruit from.
There was also some evidence from Chesterfield and North East Derbyshire that paper and
glass recyclers might have a different demographic make-up to can and textile recyclers
respectively, however evidence from elsewhere was much more equivocal about this.

Does one Behaviour begat Another?

Paper and glass recycling were normally highly correlated across all areas and also week by
week within those areas. That is, a paper recycler was highly likely to be a glass recycler as
well, and on weeks when paper recovery was high, glass recovery was likely to be high also.
The correlations between paper and glass yields and can and textile yields were much
weaker, but were still positive. Weaker still were any correlations between recycling
behaviours and reduce and reuse behaviours, though again the correlations generally
remained positive. Home composting was poorly correlated with recycling behaviours and
with other reduce/ reuse behaviours as well, but the other reduce/ reuse behaviours were all
mutually correlated to a certain extent. These results indicate that recycling behaviours,
reduce and reuse behaviours and home composting behaviours may all have fundamentally
different roots. However, the difference in roots between recycling and reduce/reuse was not
founded on any demographic difference between the practitioners. [Home composting, of

xiii
course, is highly dependent on having a garden, and the size of that garden]. Generally, the
more dissimilar the activities, the weaker the correlations were between them, though there
was evidence that correlations could increase the longer the behaviours were practiced.
Overall the results suggest that there may be a tiered structure of attitudes in operation, from
the general to the highly activity specific. For example, a pro-environmental attitude may be
held in common by recyclers and reducers, though the two groups could have different
attitudes about the more specific facets of recycling and waste reduction.

Attitudes and Behaviour

When people are asked why they recycle or why they do not recycle, or what would make
them recycle, most people come up with the same stock responses. Doubt has been raised in
the research literature as to whether the reasons voiced in those surveys represent true or
complete justifications of people‟s behaviours. Psychological research suggests that more
fundamental attitudes, values and beliefs are instrumental in determining their behaviours.
Different attitudes might be important for the initiation and persistence of the behaviours. The
research reported in the monograph tried to look beyond the customary excuses and other
reasons for recycling [or for not recycling] and explored the underlying attitudes.

The attitudes tested in the research were:

 An acceptance of personal responsibility for recycling;

 Awareness of the consequences of [or of not] recycling;
 Self-efficacy, i.e. the self confidence that one is able to carry out the behaviour
satisfactorily;
 [Absence of] negative perceptions about recycling;
 Perceptions of local recycling facilitation.

Most of the attitudes tested were found to be stronger amongst recyclers than amongst non-
recyclers, however not all the differences were significant, and there were always
considerable overlaps between the two groups. A full separation of recyclers from non-
recyclers was not possible on the basis of those attitudes. However 73% of behaviours could
be predicted correctly from the attitudes held.

Many attitudes appeared to be reasonably long-lived and persistent. There were indications
that some of those attitudes might act as pre-determinants of whether behaviours would be
taken-up, including strongly-held negative perceptions about home composting, and possibly
the acceptance of a personal responsibility for recycling, though the latter was less clear.

Other attitudes appeared to become shaped quickly upon gaining experience. Notions of
personal cost (time, effort, storage demands etc.) tended to be set mainly through early
experiences. Other attitudes can also evolve as circumstances change. A weakening of some
attitudes can act as a pre-cursor to drop-out. This generally happens either through the
realisation of untenable personal costs (like being too busy) or through bad experiences (such
as non-collection of the recycling container). For example, recycling drop-outs showed weaker
levels of acceptance of responsibility than continuing users. However, it was impossible to
ascertain whether those attitudes weakened because of a bad experience or whether those
attitudes were already weak rendering the householder more susceptible to dropping out
when problems occurred.

In Clackmannanshire there were distinct attitude differences between the long-time recyclers
in the community, and those who were just about to become recyclers (through the launch of
a new kerbside scheme). All the attitudes were stronger amongst the longer-time recyclers.
After the kerbside scheme was introduced, the attitudes of both groups increased, and the
differentials between their attitudes were more or less maintained. The biggest change was a
sharp rise in self-efficacy (i.e. procedural knowledge). Other attitude components changed as
well, but to a lesser degree. The more pronounced of those other changes appeared to be
related to the new visibility of their behaviours, with behaviours becoming more open to
scrutiny by neighbours. In Elderslie, as well, the recent recycling recruits also held weaker

xiv
attitudes than their longer-recycling neighbours, and had much stronger attitudes towards
recycling than did their non-recycling neighbours.

Whilst the research has cast much new light on attitude/ behaviour dynamics much still
remains to be understood. However, our understanding can be complicated by the
exaggerated and erroneous claims made by some people about their recycling activities. The
research looked at the levels of false reporting that might occur. Based on a combination of
monitoring actual behaviours against the self reported claims and by setting „traps‟ in the
questionnaires, it was suspected that 15% of respondents might be falsely claiming to recycle.
A further 15% were suspected of exaggerating the numbers and amounts of materials that
they did recycle. Similar, or higher, figures have been noted in a number of other surveys as
well.

The attitudes of the false claimants were higher than the attitudes of the self-reported non-
recyclers, through lower than the confirmed recyclers. It was also noted that that the suspect
groups only claimed to participate in modest levels of recycling activity. They did not claim to
recycle a lot of materials, nor did they claim to recycle many of those materials, nor to recycle
them often. It was as if they were claiming just enough to be registered as recycling. Stronger
attitudes towards recycling were also held by those suspected of exaggerating their recycling
behaviour. Self-efficacy was particularly elevated amongst the exaggerators. It was
considered that both the false claimants and the exaggerators are expressing what they felt
they should be doing rather than what they actually were doing. They were expressing
attitudes rather than behaviours. With those groups, the necessary attitudes to support
increased recycling performances could already be in place. Measures (or interventions) to
increase performance may simply need to find ways to trigger those attitudes into action.

The Importance of Monitoring

The evidence suggests that understanding attitudes and attitude changes should provide the
key to explaining why and how behavioural changes occur, and how those changes might be
triggered. That knowledge will be crucial for developing more successful promotional
campaigns and educational strategies. To date, even the most intensive campaigns have only
managed to achieve a few percentage points increase in recycling rate. Step changes in
recycling behaviour have almost invariably been linked to step changes in recycling provision.
The latter are reasonably predictable. The impacts of campaigns are less well understood.
The next few years will probably see much campaigning being carried out across the country.
That will present us with a unique opportunity to learn – provided, of course, that the
campaigns are monitored effectively, reliably, and scientifically. An appropriate monitoring
protocol encompassing specific performance indicators of behavioural change has been
developed and applied in the course of the Newspaper Industry research.

Effective monitoring is essential, not only for campaigns but also in the day-to-day
management of recycling by district authorities. In the course of the research, it came as
quite a surprise that many district authorities do not keep records of individual round to round
performances, collection by collection. The research has demonstrated how long-term trends
in recycling performances can be much stronger at this level of resolution compared with
district-wide statistics. Recyclate yields can decrease in some parts of a borough at the same
time as they are increasing elsewhere within the borough. The whole is the sum of the parts.
Managing the whole needs knowledge of each of the parts. The monograph provides much
discussion on the problems and limitations of extracting key behavioural performance
indicators from monitored performance data.

Overall, the research of the Newspaper Industry Environmental Technology Initiative is

progressively developing and refining an explanatory model of behavioural change that takes
account of people‟s attitudes, the messages they receive and how those messages are
received. In the past, that model has proved successful in helping strategic planning. In the
future the model will take on more of a role in providing decision support for performance
evaluation and diagnostics.

xv
Final remarks
The future still needs to see much more research. The major barriers to our developing our
understanding lie in the sparseness, mixed quality, and limited availability of the current data.
The „industry‟ needs to carry out more objective, scientific investigations, gain new and
improved data sets, place those data sets in the public domain, and disseminate the findings
widely. The sponsors of the Newspaper Industry Environmental Technology Initiative,
Bridgewater Paper Co. Ltd., Cheshire Recycling, Stora Enso and Sun Chemical realised
just that when they set up the Initiative eight years ago. Their mission has guided the
research, enabled the science, and delivered the output. This monograph is part of that
mission.

xvi

1. Introduction and Review
1.1 Research Goals
Traditionally, public engagement in household
waste minimisation behaviours has been driven
by strong altruistic motivations, environmental
concerns and the dislike of waste. However
that is no longer enough. Stringent waste
minimisation targets have now been set for the
country. Bridging the gap between current
household waste management behaviours and
Government targets presents us with a
considerable challenge.
Those charged with meeting the challenge are
likely to adopt a combination of the following
three strategies:
Make the required behaviours more convenient
to perform;
Educate and instil greater awareness amongst
the public;
Make it more financially attractive to behave in
the required manner.
The goal is, of course, to achieve substantial
and lasting behavioural change amongst
residents. The problem lies in the decisions that
need to be taken in order to realise that
aspiration. The research presented here sets
out to examine some of the key issues that
need to be addressed within those decisions.
Recycling behaviour depends essentially on
three main factors:
Programme Past
design history
The
people
For a resident to participate in a local recycling
programme, the resident must first judge that
the programme is suitable, and then must be
sufficiently motivated to use it. That motivation
is inevitably linked to individual personal
characteristics. Those characteristics are likely
to have been shaped by a myriad of past
events and experiences. „Suitability‟ is often
taken to be synonymous with convenience.
1
Understanding Recycling Behaviour 2
Personal characteristics are effectively the
attitudes and values held by the individual.
The first goal of the research was to determine
how much improvement could be achieved by
increasing convenience alone, and how much
must depend on changing fundamental
attitudes.
The second goal of the research was to
determine how programme design variables
might affect the ultimate endpoints to recycling
behaviour.
The third goal of the research was to delineate
the specific attitudes and attitude changes that
are instrumental to achieving lasting
behavioural change. Here, attitudes that control
the initiation of behaviours and those that
control the persistence of those behaviours
need to be separately distinguished.
Household waste management is not a single,
homogeneous behaviour. It encompasses
waste reduction and reuse practices as well as
recycling. It involves the parallel management
of many different materials. Does the
involvement in one waste management practice
then imply any involvement in the others? The
fourth goal of the study was to ascertain the
relationships between different household
waste management behaviours. Knowing those
relationships will be important to designing
interventions that can increase all behaviours
across the board.
It is well known that household recycling
behaviours vary substantially from district to
district, from neighbourhood to neighbourhood,
and from street to street. What is not known is
“Why?” It is commonly perceived that spatial
variations in behaviour may in some way be
linked to differences in household
demographics. The fifth goal of the study was
to investigate how much demographic factors
actually matter. Are they relevant to today‟s
recycling programme design and its
management?
Overall, household waste management
behaviours tend to be quite stable though they
might fluctuate quite dramatically with time. The
sixth goal of the study was to ascertain how far

those temporal fluctuations are normal and
predictable and how much may result from real
behavioural changes. Identifying and acting on
those real changes will be essential for
programme sustainability.
Overall the study emphasises what
improvements in recycling performance might
be expected from given actions taken by local
authorities. Those authorities are already
progressively improving their recycling
programmes. But do the outcomes from those
programmes meet expectations? Here we need
to distinguish aspirational goals [to meet
targets] from the expected „behavioural norm‟
that is appropriate to the local programme, its
people, and their history. Monitoring the
programme performance is crucial. It provides
the feedback essential for managing ongoing
improvements.
1.2 Structure of the Monograph
The remaining part of chapter 1 reviews
previous research into household recycling
performances. The key findings from volume 1
of Understanding Recycling Behaviour are
summarised in section 1.3. The final part of the
chapter then reviews some recently published
case studies on specific recycling schemes,
and the available data on cross-scheme
comparisons.
Chapter 2 provides a model forecast of „where
we are going‟. It establishes what the practical
limits to recycling might be, and demonstrates
the staged improvements that are needed to
reach those end points. It provides the first
estimate of how much improvement can be
expected from enhanced programme design,
and how much may need to be realised through
education.
The details of programme design and how they
affect performances are discussed in chapter 3.
The real effects of enhancing local recycling
provision are reported in chapter 4. The
reported research is based on a case study in
Chesterfield and North East Derbyshire. The
chapter also starts the investigation into
demographics and behaviour and highlights
some key issues and uncertainties for effective
programme evaluation.
Chapter 5 extends and cross checks the
Chesterfield and North East Derbyshire findings
with parallel case studies in Lancashire and in
North Warwickshire.
Understanding Recycling Behaviour 2
The seventh goal of the study was to research
objective methods for more effective
performance evaluation and diagnostics.
Data quality is a key issue. The eighth goal of
the study asked the question “Do your
measurements really tell you what you want to
know?”
The objective that cut across the whole study
was to isolate the separate effects of the
programme from the people and the local
history. It is only by understanding those effects
separately, and their interactions together, that
we will understand how to manage them and
achieve best practice.
The relationships between different recycling
behaviours and between recycling, reduce, and
reuse behaviours are explored in chapter 6.
The demographic bases of the different
behaviours are also examined.
Chapter 7 continues the theme of programme
evaluation through exploring the information
that can be extracted from performance
monitoring time series data.
Recycling attitudes are then addressed in detail
in chapter 8, which combines a literature review
with two new attitude/ behaviour research
investigations. The investigations focus on
episodes of behavioural change around the
enhancements of local recycling provision. Both
„before‟ and „after‟ data are reported. A
performance evaluation protocol for quantifying
attitude and behavioural changes is presented.
The problem of gaining accurate information
from household questionnaires is addressed as
a separate issue in chapter .
Chapter 10 considers another key issue
impinging on effective recycling performance
evaluation - quantifying the waste arisings.
Chapters 11 and 12 conclude the discussions
focussing on the central issues of stimulating
behavioural change, and the evaluation and
diagnosis of performances.
2

1.3 Factors Affecting Behaviour
The contents of this section are a distillation of
the findings from volume 1 of Understanding
Recycling Behaviour. The full technical details
and supporting arguments are available from
that reference source (Tucker, 2001).
It is considered that the major factors that can
affect community recycling performances are:
 Programme design;
 Information, education, and promotion;
 Demographic factors;
 Past experiences and history;
 Other factors.
These factors are not necessarily independent.
Information and education, in particular, cut
across all categories. Past experience and
history can include several sub-factors, the
most important of which are: (i) poor service
from the operators, or other experienced
problems, (ii) local social influences
(behavioural norms), and (iii) previous
informational messages received. Social norms
can become strengthened in specific local
geographies, for example in cul-de-sacs as
opposed to main roads, close to schools, and in
areas with strong community identity. „Other
factors‟ comprise all the factors that we cannot
explain at present, and perhaps will never
explain. Effectively they contribute to a
pervasive randomness in behaviours.
Whilst each of the above factors can take on
different importances in different communities,
there are certain key features within those
factors that appear to act commonly towards an
overall good practice.
1.3.1 Programme Design
Many variables must be addressed when
designing a household recycling programme.
These impact not only on the effectiveness of
the programme (in diverting waste from landfill)
but also affect its cost. Previous research has
highlighted the following design variables as
being potentially important for performance:
1. Programme type (mandatory or
voluntary);
2. Specific materials collected and
number of materials collected;
3. Number of separations required;
3
Understanding Recycling Behaviour 2
4. For a co-mingled collection – separate
collection versus co-collection with
residuals;
5. Same day collection as residuals or on
a separate day;
6. Collection frequency;
7. Provision of a free container;
8. Type of container (one-trip bag, multi-
trip bag, box or bin);
9. Container size;
10. Collection vehicle type;
11. Education programme (provided or not
provided, and if provided, type of
programme);
12. Economic incentives (fees, fines,
rewards, etc.).
In general, research has been quite equivocal
as to whether each or any of these factors is
actually significant to programme success.
Mandatory schemes tend to produce higher
levels of participation than voluntary schemes,
however the best voluntary schemes still match
the performances of the best mandatory
schemes. Increasing the number of materials
collected tends to provide higher participations,
so a multi-material collection will generally
stimulate higher levels of participation than a
paper-only collection. However, if more
separations are demanded of the householder
(i.e. more separate containers) then there is
some evidence that participation and recovery
levels could drop. This can be due to: (i)
greater inconvenience (or higher personal
cost), and (ii) increased, and presumably
unacceptable, storage demands.
The specific materials collected can also be
important. Generally paper (newspaper,
pamphlets and magazines) will always be
included in a kerbside collection. Recovery
rates of available newspapers from recyclers
are generally very high. Schemes that also
collect glass are also looking successful, with
high recoveries of glass also being achieved.
Recoveries of the „minor‟ materials: cans, foil,
plastic bottles, film, and textiles tend to be
lower than those for paper and glass. The
possible explanations are that: (i) when small
quantities are generated there can be a
perception that “there is not enough to make
recycling worthwhile”, or (ii) when space in the
kerbside container is at a premium the bulky,
light items are the most vulnerable to being left
out, or (iii) hygiene, safety or distaste
considerations mitigate against those items
being prepared (i.e. cleaned out) for recycling,

or (iv) there may also be some ignorance that
such materials can be recycled through the
programme. Surveys also show that almost
nobody recycles „minor‟ materials like cans and
plastic bottles if they do not recycle their
newspaper or glass as well. The more types of
materials that are collected together can
convey a stronger message about the
seriousness of recycling.
There is also some evidence, though
somewhat equivocal, that same day collection
of recyclables with the residual waste may
increase participation in some instances,
though usually no effect is found.
Generally, the more frequent the collection, the
higher the material capture and participation
achieved. Too long between collections
increases the possibilities of information loss
and decreases the chance of habit formation.
Research has shown that changing established
paper-only collections from 2-weekly to 4-
weekly collections only had marginal impacts
on capture and participation, provided that
adequate and convenient paper bring sites
were also available in the locality. The „lower‟
convenience of the kerbside scheme switched
more recyclers to bring recycling [or to dual
use] rather than switching them off from
recycling altogether. Collection frequency,
however, must also be considered alongside
collection container size and type, as the
effects are inter-related.
Normally, the provision of a free container
produces higher participation levels compared
to where householders pay for their recycling
container. Container size is important. Too
large a container can inhibit some low waste
producers from recycling - enforcing the
perception that “they do not have enough waste
to make recycling worthwhile”. Here, the
container size can convey a perception of the
expectations of the collector in terms of waste
quantities required. Not every low waste
producer will save and accumulate their waste
until they have a respectable quantity to set out
for recycling. Concerns about long-term storage
(e.g. linked to smells, hygiene, tidiness, storage
space considerations, and so on) can provoke
a “get it out the door as fast as possible”
attitude.
Too small a container will lead to some
households filling it completely and having
additional waste as well. Some households
may take that excess to a drop-off site for
recycling. Others may simply commit it to the
dustbin – “I have done my expected bit”. The
Understanding Recycling Behaviour 2
minor components of a multi-material collection
are particularly vulnerable to loss in this way –
e.g. aluminium cans and especially plastic
drinks and milk bottles as they are so bulky.
Ideally optimally run schemes should aim at
householders „nearly‟ filling the provided
recycling container during one collection period.
The gross weight of the filled container is also
important. Box or bag collections of co-mingled
paper, glass, cans and textiles can be 10 – 12
kg or more for high waste producers on a
fortnightly collection. This can lead to serious
handling problems, both for the recycler and for
the collection crew. Providing two containers,
one for the paper and one for the other dry
recyclables can help solve this, and may be
essential anyway to alleviating concerns about
recovered paper qualities. However, it still
needs to be borne in mind that too large a
container or too many containers raises storage
issues. Box collections may benefit from
allowing householders a choice of container
size to suit their needs. Bag collections can
issue single or multiple bags on demand.
In general, participation and recovery increases
through the hierarchy: Own bag (e.g. old carrier
bag)->Plastic sack->Rigid box/ bin. Specific
issues raised in boxed collections are whether
to provide a lid or not. Some research has
shown that significantly higher recoveries may
be achieved with lidded boxes compared to
unlidded boxes. Some major considerations are
that the lidded boxes are weatherproof and
may be stored outside (thereby alleviating in-
house storage constraints). Unlidded boxes are
more vulnerable to rain infiltration when set out,
which can lead to rejections of some collected
loads. However, this must be balanced by the
greater operational burden in emptying the
lidded boxes, and the potential issues of the
lids „blowing away‟.
Whilst analyses demonstrate on average that
there could be positive benefits associated with
specific scheme variables, very different
responses can be achieved in different
localities. Research in the US (see section
1.5.2) is showing that good-performances do
not depend uniquely on any programme design
variable, except perhaps for container
provision. Put another way, no programme
design variable is identified as a necessary
condition for programme success.
Finally, it must be remembered that the
recycling collection cannot be decoupled from
the residual waste collection. Recent research
is now showing that reducing the capacity of
4

residual waste collections can stimulate
significant recovery increases in the recycling
stream. This can be achieved either by
reducing the size of the residual waste
container, or by reducing its frequency of
emptying.
1.3.2 Education and Promotion
There is relatively little well-documented
information on the quantitative effects that
information and promotional campaigns
actually have on recycling behaviours.
Generally, most past campaigns were not
monitored adequately, if they were monitored at
all. The small amount of monitoring that was
undertaken was never continued into the longer
term so we cannot assess the sustainability of
any of the performances that may have been
induced.
Intervention promotion and education)
strategies comprise both antecedent strategies
and consequent strategies. The major
antecedent strategies that have been used to
try to stimulate recycling are: prompting,
commitment, goal-setting, normative influence,
and the removal of barriers. The consequent
strategies that have been used have included
feedback, reward, and punishment. An
important part of most strategies is the
message delivery medium, and this needs to
be considered in detail alongside the strategy
itself. The potential and limitations of each type
of strategy are now considered in turn.
Reward: The general outcome of providing
economic incentives, such as lottery tickets or
vouchers, as rewards for specific recycling
behaviours were that: (i) people‟s behaviours
were normally enhanced during the
intervention, but (ii) those enhancements were
seldom maintained once the incentives were
withdrawn.
Commitment: Both written and verbal
commitments, normally elicited as pledges to
recycle, were generally found to have positive
effects on recycling behaviours. There was also
some evidence that those enhanced
behaviours might be sustained into the longer-
term, i.e. when the participant was no longer
bound by the terms of their pledge. Few
interventions of this nature have been
undertaken in the UK. The University of Paisley
tried it in the Borough of Fylde without success.
Prompting: These include verbal or written
persuasive messages. Impact can depend on
the credibility of the source – association with a
5
Understanding Recycling Behaviour 2
well-known environmentalist or personality can
help. Council endorsement can emphasise the
authority of the message though some
residents can react against Councils‟ prompts.
Prompts can utilise the fear factor but if they
do, they also need to prescribe specific actions
to be successful. Negative messages or
messages that emphasise personal relevance
tend to work best. For most effect, prompting
needs to be repeated regularly, and results can
depend critically on the message delivery
medium. It is also important to tailor the
message to the audience, i.e. to segment the
market.
It is generally found that the most effective
delivery strategies are those which achieve a
high contact with the population but are of low
cost to the system provider. Personal contacts
usually provide the highest success rates
overall but can be very cost-intensive. Press
articles (if they are read) can be effective
through publicising a norm that would make
others more likely to participate. Government
educational campaigns via the media can be
successful but tend only to produce small
increases in recycling behaviour (at least in US
studies – we await analyses of the recent UK
campaigns). Generally, more educated people
are most likely to be influenced by newspaper
coverage, whilst TV advertisements may be the
best medium to reach those not specifically
looking for recycling information. Leaflets and
billboard advertising may be contentious media
for promoting pro-environmental information.
They could be perceived by some as
environmental blights, and incongruous with the
message being conveyed.
Feedback and Goal Setting: As with
promotions, it has been found that feedback
works best if repeated over extended periods.
This can become expensive. To be successful,
it needs to demonstrate a relevant and tangible
connection to the individual‟s behaviour. Goal
setting is often linked to feedback. It works best
for charity collections where the saliency of the
goal is easily understood. Group identity can be
important for success.
Normative Influence: Normative influence
interventions are a special case of local
promoting using indigenous, committed
individuals (e.g. block leaders). Effects can be
highly dependent on local social factors, the
visibility of scheme, and on how high
participation rates were originally. As far as we
are aware, such interventions have not been
tried in the UK.

Whilst research has indicated that most
reported interventions have had at least some
success in increasing recycling behaviour, the
effects have generally been quite modest and
their durability has been questionable. The
weight of evidence shows that a combination of
strategies might be more successful than the
use of a single strategy alone. Different
strategies may achieve very different results in
different localities and across different cultures,
though the personal touch appears fairly
ubiquitous to any success.
Apart from lack of monitoring, another major
problem that has hindered successful
campaign development in the past is the lack of
adequate market research. Many campaigns
have been based on the perceptions of the
designer rather than on any identified needs of
the recipient. Increased use of focus groups is
now beginning to address this problem. Letting
the focus groups design the interventions for
their peers looks promising.
Research is showing that four classes of
educational material may be needed to cover
most educational needs:
1. Addressing general pro-recycling
attitudes.
2. Specific [procedural] information and
awareness.
3. Addressing perceived consumer
effectiveness – that every little bit of
material is important and that every
individual‟s action counts (This is
covered broadly by „Are you doing your
bit‟ but needs to be reinforced locally
specifically for the perception that
recycling is not worthwhile unless you
have a lot of waste).
4. Actions that may involve more effort or
personal cost to the recycler. These
include addressing forgetfulness (often
rooted in having a busy lifestyle),
stimulating any extra effort needed to
properly prepare materials for recycling
– e.g. washing and squashing tins,
what to do on experiencing personal
difficulty or encountering problems with
system provision, organisation of
household storage etc.
We are not yet able to unequivocally say what
specific pro-recycling „attitudes‟ need to be
tackled, though drawing on the small amount of
research that exists, it would appear that the
following factors may be the most important to
address:
Understanding Recycling Behaviour 2
 acceptance of a personal responsibility
 negative perceptions
Also important may be:
 conveying the saliency of the issue to the
individual – i.e. can it be linked to some
local aspiration such as saving local landfill
space
 raising an awareness of the consequences
of action or non-action
 instilling a „locus of control‟. That is, the
individual must feel comfortable that he/she
has the ability to perform the action
satisfactorily and that the action is carried
through to the desired outcome by all other
individuals downstream in the recycling
chain.
Addressing general environmental awareness
would not appear to be particularly relevant to
stimulating specific environmental actions such
as recycling. Addressing economic benefits or
disbenefits could also be of little relevance to
many people as they may be driven more by
altruistic motivations.
It is also becoming apparent that fundamentally
different messages may be needed for
addressing different pro-environmental waste
management behaviours. Reduce/reuse
behaviours may have fundamentally different
roots than recycling behaviours, and
composting behaviours may have different
roots to both.
Timing and frequency of campaigns can be
important. The effects of most promotions will
decay with time (which could be extremely
rapidly with reward-based promotions). There is
little research on the longevity of effects of
other types of promotion, though indications are
that significant decays in effect can occur within
six months or less. Repeat or reinforcing
messages may then need to be deployed at
shorter intervals [at least initially] to sustain the
levels of increased behaviour. Once that
behaviour becomes more habitual, the
reinforcing interventions will not need to be as
frequent. However, evidence shows that it may
take some time before stable new behaviours
become internalised.
Finally, it must be borne in mind that the
scheduling of interventions should generally
follow or be concurrent with the provision of the
necessary recycling infrastructure, i.e. you
need to have a product before you can market
it. Promotions should not raise expectations
without being able to deliver.
6

1.3.3 Demographic Factors
Recycling Participation:
In general, studies have found that older,
better-off and more educated people are the
most likely to recycle. Ethnic differences have
also been noted with non-white ethnic groups
tending to participate less than whites.
Residents of small detached and semi-
detached housing have been identified as
those most likely to recycle. Larger detached
households in addition to those from poorer
housing tend to perform less well. Those living
in a single-family dwelling are generally found
to be more likely to recycle than those from
multi-household dwellings. Mixed findings are
reported with respect to household size and
tenure. The presence of young children in the
household appears to have a negative
correlation with recycling. This has been put
down to households being too busy when the
children are young. Busyness of lifestyle, more
generally, is also found to correlate negatively
with recycling activity.
Overall, for most demographic factors, most of
the identified relationships vary significantly
amongst different studies. Furthermore,
different factors may be associated with the
recycling of different materials.
Overall, the strengths of the relationships
between socio-demographic factors and
recycling behaviour are weak. It is now
considered that any strong differences that
existed in the past may have been eroded with
time. Now that recycling and environmental
concern has become more diffused throughout
the community, demographic differences in
recycling behaviour are progressively
disappearing. Also, as recycling levels
increase, statistically more new recruits must
come from the less traditional recycling sectors.
Research into demographic factors has tended
to concentrate on small samples of the
population in restricted localities. Limited
studies across larger samples (whole cities)
undertaken ten years ago in the US, showed
that the cities‟ household size, tenure,
household income and occupational status
profiles were not significantly correlated with
any aspect of recycling behaviour. That is, at
city scale, demographic characteristics may be
relatively unimportant in explaining recycling
success.
7
Understanding Recycling Behaviour 2
Waste Arisings:
The above discussion refers to recycling
participations and material capture rather than
recycling yields. Demographic factors could
also affect the yields if there are differences in
material consumptions amongst different
demographic factions. The argument is that
those sharing a given demographics may be
more likely to share similar lifestyles and
purchasing habits as well. However, there is
little published data to quantify any such
dependences. Limited inferences can be drawn
from the various historical data compiled under
the National Household Waste Analysis
Programme, and from independent market
research, e.g. under Project Integra in
Hampshire, and previous University of Paisley
research. However the data is quite
fragmentary and generalisations are probably
not yet possible. Newspaper yields are better
understood than most materials, with better
yields normally emanating from the higher
socio-economic groups. This tends to reflect
the relative amounts of broadsheets and
tabloids taken rather than revealing any
differences in numbers of newspapers read.
1.3.4 Past Experience and History
Past experience and history tend to act on
individual behaviours. The effects can be both
positive and negative. Negative effects on
performance levels are essentially the
aggregated result of all the catastrophic and
other adverse events that have affected
individual households. Common problems
affecting kerbside collections are:
 News or observation that contributed
material is not actually being recycled;
 Poor service provision (this includes
missed kerbside pick-ups, non-return of
recycling containers, non-advertised
changes in kerbside collection timings,
etc.);
 Not having a recycling container, e.g. new
residents moving in;
 Insufficiency or loss of information (e.g. of
the day of collection of less frequent
schemes, procedural information on how to
participate and what to participate with);
 Design of recycling container – bags or box
lids blowing away, container too heavy etc.
Generally, attitude surveys of non-recyclers
tend to find that around half the problems are
related to information deficiency or are
container-related. Regular calendaring and

procedural information reminders are important,
also the provision of an information pack to new
residents could be worthwhile. Getting the
container right at the outset is also important.
The container is the prime interface between
the household and the scheme and many
householders tends to polarise their thoughts
around that container. More flexibility in
container choice could be important.
The service personnel picking up at kerbside or
supervising civic amenity sites are effectively
the ambassadors of the recycling scheme.
Getting the right personnel that interface well
with the public, answer questions, give advice
and help etc. can be very beneficial in
alleviating any developing problems.
Spotting the onset of any developing problems
can be difficult in practice. The normal temporal
fluctuations in monitored performances can
easily mask the outcomes of individual
problems. Continuing effects may only be
identified when longer-term statistical analyses
are carried out on the performances. Then it
may already be too late to apply the necessary
corrective measures. Nevertheless regular
performance monitoring is essential to sustain
optimum performance levels. That monitoring
needs to be carried out at collection round level
for kerbside schemes. Experienced and
developing problems tend to be localised. They
will not be diagnosed unless the monitoring is
undertaken at the necessary resolution.
Positive „natural‟ enhancements to recycling
levels might act primarily through local
normative influences, although pervasive
background messages (e.g. from television
campaigns) may also contribute – however this
is not yet proven or quantified. Normative
influences can act through social dialogue or
can act remotely through visual stimuli if the
visibility of the behaviour is high, as it is with
kerbside schemes. It is considered that once a
1.4 Case Study Review
1.4.1 Introduction
The preceding section illustrated some of the
research results into the determinants of
recycling behaviour. In theory, by knowing
which specific factors support the higher
recycling performances, one should then be
able to assemble a recycling scheme to provide
Understanding Recycling Behaviour 2
certain number of residents in a locality are
setting out their recyclables at kerbside, others
will be stimulated to do so as well. This effect is
often most noticeable in cul-de-sacs.
By and large, however, recycling behaviours
are quite stable. A recycler will generally
continue to recycle into the future whilst a non-
recycler will not recycle, even through periods
of applied impetus for behavioural change.
Several psychological studies that attempted to
determine the antecedents of recycling
behaviour have found that past behaviour can
be the dominant, and sometimes sole predictor
of current recycling behaviour. It is often found
to have much more influence than any
attitudinal or demographic factor. Put simply
this implies that recycling may be dominated
mainly by habit.
In practical terms, this means that many
campaigns will inevitably fail to stimulate any
significant behavioural change. On the other
hand, going through bad experiences may not
switch off recycling behaviour per se, unless
that experience is perceived to be of
catastrophic proportions (e.g. that learning the
council is sending all the collected recyclables
to landfill). For lesser irritations, the committed
recyclers may simply adapt their behaviours in
order to cope, provided it is practicable for them
to do so. They may, for example, turn to their
local bring-schemes when they experience
problems with their kerbside collection.
Kerbside collections will always be more
convenient than bring systems (for most
people) though some would consider bring
sites to be more convenient for their own
particular circumstances. A crucial point,
however, is that you cannot dispense with bring
collections even if the kerbside scheme is
operating well. Bring site back-up is essential in
mopping up recoveries when the kerbside
collection has problems or is not practical.
the best practice. However, many of the
research data on which such ideas could be
founded were derived from quite limited
experimentation based on fairly small samples.
The general applicability remains quite poorly
tested. Complementary information must now
be derived from looking at specific schemes
(with their ready-built assemblages of design
variables) to see how well they live up to
8

expectations, and to see how well they support
the hypotheses of the smaller-scale research.
This section now looks at a limited number of
UK case studies. Most of the studies are drawn
from published sources. It is noticeable, in this
respect, that few comparative studies have
been published, and the few that have been
published have only provided rather superficial
analyses.
It should be noted that the studies singled out
for presentation here do not necessarily
represent „Best Practice‟, though they do try to
illustrate some of key points. It should also be
borne in mind that the most in-depth data on
schemes tend to be collected during their initial
pilot stages. Those pilots may have been
„cherry-picked‟ and nurtured, so the reported
performances may not necessarily translate to
achievable borough-wide performances.
1.4.2 Millennium Recycling
Scheme: Pilot in Bradford
Reference: Barton, Perrin, and Barton. (2001).
This was a rather artificial experiment, carried
out in Bradford, to establish whether high
participations and high diversions could be
achieved by a scheme that had been made as
simple and as convenient as possible.
Elsewhere in Bradford a kerbside collection of
paper, card, plastics and cans, run as an opt-in
scheme using 240 litre wheeled bins, was only
achieving 30% household participation. The
Millennium scheme was designed as an opt-out
scheme using a 140 litre wheeled bin to collect
all recyclables (including plastic containers and
glass bottles). The residents were not told
specifically what recyclables were allowed,
though were specifically prompted that food
and garden waste were not wanted.
The pilot trial covered just 143 households,
comprising bungalows, detached and semi-
detached housing of „mature established home
owners‟ (All Acorn D type 26, see Appendix A
for explanation). Twelve weeks into the 6-
month trial, feedback was provided giving
information on recovery performance and
clarification of what could go in the bin.
Set out rates averaged 70-80% with the overall
participation rate being around 90%. Around 5
kg/household/week recyclables were collected,
rising to around 6.3 kg/household/week after
the feedback was given. Diversion rates were
37% (before), 44% after. The feedback had a
9
Understanding Recycling Behaviour 2
relatively small but positive effect on the paper
and glass recoveries (rising from 74% to 79%,
and 69% to 73% respectively) but increased
card recoveries from 54% to 73%, plastics from
22% to 41%, and metals from 29% to 48%.
Part of these changes may have been induced
by the feedback message: “I don‟t have to be
clean to be in the recycling scheme”
The total paper and glass in the combined
recyclate and residuals increased across the
scheme introduction, perhaps suggesting that
some of the capture of those materials was due
to a transfer away from bring outlets. Total
newspaper and pams arising before scheme
introduction were 2.17 Kg/household/week
compared to 3.31 Kg/household/week after
(around 2.53 Kg being recycled). So by
inference, up to 45% of the Millennium
newspaper and pams recyclate had been
transferred away from the bring sites.
Alternatively the low arisings prior to the
introduction may simply have been an artefact
of people saving up their material in
anticipation.
Contamination levels in the recyclate were
around 4% by weight, made up of 2%
putrescibles and 2% composites.
Perrin and Barton (2001) compared the
Millennium recycling scheme with the „Paper
Chain‟ scheme running in Leeds. The paper
chain scheme introduced kerbside recycling to
20,000 households in two suburban areas. It
collected newspapers and pams in a reusable
plastic sack on a fortnightly basis. At the start of
the scheme, residents had been given an
information leaflet and instructions were also
printed on the sack. For this research, the
scheme was monitored for a six-month period
in 1999.
The Paper Chain scheme only achieved 59%
recovery (1.24 Kg/hh/w) compared with the
83% diversion of newspapers and pams (2.5
Kg/hh/w) in the Millennium scheme, and the
scheme only attracted 49% of residents to
participate compared to the 90% participating in
the Millennium scheme. Ninety eight percent of
residents of both schemes had expressed the
positive intention to use their offered scheme
when asked prior to scheme commencement.
Those recycling (through bring sites) prior to
scheme commencement expressed similar
reasons for doing so in both schemes (table
1.1). Similarly the reasons expressed by non-
recyclers for not recycling were also
comparable between the schemes (table 1.2).

Table 1.1 Reasons for Recycling
Paper Chain N=183
Good facilities
Saves waste/landfill space
Saves dustbin space
For environment/future generations
Personal satisfaction/habit
Peer pressure/duty
Source: Perrin and Barton (2001)
Table 1.2 Reasons for Not Recycling
Paper Chain n=79
Inconvenience/ time 50.0%
Effort 16.3%
Facilities too far/ inadequate 28.8%
Lack of information 48.8%
Storage/handling problems 35.0%
Never thought about it 22.5%
Not enough to recycle 21.3%
Other 8.8%
Source: Perrin and Barton (2001)
Perrin and Barton drew attention to the fact that
the reasons were quite similar to reasons given
in many other surveys as well. Overall, they
argued that the behavioural differences
between the Paper Chain and Millennium
schemes may have little to do with attitudes,
and may be more connected with the
differences in programme design. They also
noted that fewer of those served by the Paper
Chain scheme claimed to recycle their glass
and cans after the paper collection was
introduced. The implication was that there was
a drop off in bring site usage for all materials,
not just paper.
The expressed reasons for recycling did not
change significantly as a result of the
introduction of either kerbside scheme, though
the numbers citing peer pressure or duty
decreased in both areas after the respective
scheme introductions, despite 72.4% of
residents admitting that they noticed their
neighbours‟ behaviours.
Whilst the Millennium Recycling Scheme did
demonstrate that high participations and high
diversions could be achieved in practice, the
exercise was artificial in that some of the
collected materials had no local market (e.g.
plastic containers, non-bottle glass). However
the researchers did demonstrate that MRF
recovery would be feasible on the Millennium
scheme recyclate and that acceptable picking
rates could be achieved.
Understanding Recycling Behaviour 2
Millennium N=48
50.3% 58.3%
59.6% 52.1%
42.1% 33.3%
73.2% 81.3%
51.4% 52.1%
7.7% 6.3%
Millennium n=14
64.3%
7.1%
28.6%
42.9%
50.0%
7.1%
14.3%
0.0%
1.4.3 Milton Keynes
Reference: Thomas (2001)
Milton Keynes operates a twin box kerbside
collection throughout its area of 80,000
households, on an opt-in basis. Around 75% of
properties have requested recycling boxes.
Current participation rates are unknown but are
likely to be a little less than the self-reported
rate of 71% returned in a 1995 survey. The
scheme now targets a high potential diversion
rate by targeting a large number of materials for
recycling, though it only achieves a low
recovery rate, and capture rate of 52%. Low
capture with relatively high participation points
to residents having a relatively poor
understanding on what and how to recycle.
In 1995, twenty-two items were listed as
acceptable for recycling through the scheme,
and 12 items were listed as being not
acceptable. Pre-1994 just 5 items were listed
as acceptable and 4 as not acceptable. Not
surprisingly, when questioned nobody was able
to accurately recall the later list although some
61% correctly identified the full range of items
on the previous list. Nearly 95% of respondents
correctly identified newspapers as being
wanted, with around 90% correctly identifying
plastic bottles, cans, glass bottles and
magazines. Plastic cartons, trays and foil were
less well recalled and under 20% considered
that textiles, shoes and handbags were also
wanted. Less than 40% of respondents
10

correctly knew that margarine tubs, drinking
glasses and envelopes were not wanted.
In 1995, the scheme achieved a diversion rate
of 17.5%. In 1993, collecting fewer materials,
the diversion rate had been 19%.
Thomas (2001) also undertook a parallel study
of public understanding across 11 kerbside
schemes in Hampshire. The results showed no
strong difference between the levels of
understanding and the recycling container type,
though it appeared that the grasp of
requirements was generally higher in districts
using twin wheeled bins and lower in districts
providing no recycling container. There were no
significant differences between weekly and
fortnightly collection regimes and no
correspondence between levels of
understanding and social class. However, the
numbers of participants correctly identifying
what could be recycled were higher (80%) in
schemes providing information on what
materials were acceptable than in schemes
where information was not provided (~60%).
1.4.4 Wealden
Reference: Woodard et al. (2001).
Wealden had previously operated a kerbside
recycling scheme based on a weekly 55 litre
box collection of paper (newspapers, pams and
white directories) and metals (food and drink
cans and clean foil) with the residual waste
collected via wheeled bins. In 1998, a pilot
programme „CROWN‟ was commenced across
1000 Wealden households in which an
additional green waste bin was introduced and
the green and residual waste bins collected on
alternate weeks. CROWN stands for Compost
and Recycle Our Waste Now. The scheme was
extended in 1999 to cover a further 4300
households with further extensions planned.
The Council implemented a carefully planned
publicity and information campaign before,
during and after each implementation.
Two hundred and forty seven households
serviced by the second phase of the CROWN
scheme were monitored for participation and
for the levels of recyclate that they set out.
They represented a mix of 45% better off
retirement areas, 45% modern family housing
and 10% high status non-family areas. A
detailed waste compositional analysis was
carried out on a smaller sub-set of 32
households, 3 weeks prior and 5 weeks after
the introduction.
11
Understanding Recycling Behaviour 2
Prior to the introductions, the residual waste
arising of the 32 households averaged 18.1
Kg/household/week. After the introduction this
had fallen to 8.2 Kg/hh/w, which represented a
55% reduction. The volume of dry recyclate
produced increased from 5.5 litres per
household (0.5 Kg/hh/w) to 17.5 litres (1.5
Kg/hh/w). Paper diversion increased from 0.44
Kg/hh/w to 1.02 Kg/hh/w and the diversion of
metals from 0.1 Kg/hh/w to 0.46 Kg/hh/w. The
percentage increases in both categories were
higher than the increase in scheme
participation (40% to 78%) suggesting that the
capture from participating households also
increased. Contamination levels increased
marginally from 0.0 Kg/hh/w to 0.02 Kg/hh/w.
The amount of garden waste in the residuals
decreased from 5.8 Kg/hh/w to 0.1 Kg/hh/w and
other putrescibles from 3.4 to 2.0 Kg/hh/w.
However it must be borne in mind that the prior
analysis was conducted in September and the
after analysis in November so seasonal effects
may have contributed to those figures.
It is also noted that total waste recovered at
kerbside (recyclate plus residuals plus
organics) decreased by 1.3 Kg/hh/w across the
introduction. Whether this was a seasonal
effect, a real decrease in waste consumption,
or a knock-on effect stimulating more bring
recycling is unclear.
The sustainability of the CROWN recoveries
was investigated for the year following the 1998
trial of 1000 households. Woodard et al. (2001)
argue that enhanced recyclate recoveries have
continued over the year though inspection of
their charted data indicates that there may have
been a drop from around 7.5 tonnes/month to
around 5.0 tonnes/month over the year.
1.4.5 Babergh
Reference: Waste Research Ltd., AEA
Technology. (2001).
In 2000, Babergh introduced a pink sack
(survival bag) for dry recyclables. The borough
was already operating a wheeled bin collection
for refuse. The new scheme collected
newspapers, pamphlets and magazines, other
paper and card, plastic bags and plastic film,
steel and aluminium cans, plastic bottles and
plastic containers. The pink sacks were
collected on the same day as the residuals, the
sacks being placed next to the wheeled bin.
Sacks were transported co-mingled with the

refuse and recovered at the MRF by hand
picking.
Babergh is generally more affluent that the UK
average (36% Acorn A, 20% B, 1% C, 42% D,
22% E, 5% F compared with 20% A, 12% B,
8% C, 24% D, 14% E, 22% F – see Appendix A
for a key to Acorn profiles). Two separate
neighbourhoods were sampled: a very affluent
neighbourhood in November (92% Acorn A+B)
and a less affluent neighbourhood (though still
more affluent than the UK average) in August.
Set out rates for both samples were
comparable at 68-69%. The weights of
recyclate recovered were 1.8 Kg/household per
week (August) and 2.4 Kg/household/week
(November) from a total arisings (recyclate +
residual) of 24.0 and 21.6 Kg per household
respectively. It was noted that the lower
recovery in August may be due to the samples
for analysis being recovered at the landfill site
(where there would be losses from split and
unrecovered bags) whereas the samples were
diverted directly from the kerbside in
November. The overall scheme diversion
ranged from 8% to 11%.
The recovery of newspapers and pams was
0.81 Kg/household per week (August) and 1.5
Kg/household/week (November), which
equated to 26% and 50% recoveries
respectively. Recoveries of other materials
were much lower: Cans (8-15%), plastic film
(4%), dense plastic (19%), and other paper
(15%). It was considered that the higher
newspaper recoveries in November were
attributable to the higher socio-economic profile
of that sample, though it must be noted that the
total arisings of news and pams were very
similar across the two samples (3.08 and 3.02
Kg/hh/w), so if there was a demographic
dependence is was in recovery rather than
consumption.
Contaminants made up around 4.2 and 5.9% of
the recyclate, split roughly evenly between
textiles, glass and other materials.
1.4.6 Eastleigh
Reference: Waste Research Ltd., AEA
Technology. (2001).
The scheme at Eastleigh was based on twin
140 litre wheeled bins for dry recyclables and
residual refuse (though larger families could opt
for 240 litre bins). Bins were collected on
alternate weeks at kerbside. The dry
Understanding Recycling Behaviour 2
recyclables collected were newspapers,
pamphlets and magazines, other paper, steel
and aluminium cans, plastic bottles and plastic
containers.
Like Babergh, Eastleigh is generally more
affluent that the UK average (22% acorn A,
38% B, 2% C, 26% D, 7% E, 5% F – see
Appendix A for key). Two separate
neighbourhoods were sampled in September
and April respectively, a very affluent
neighbourhood in September (84% Acorn A+B)
and a less affluent neighbourhood in April.
Set out rates were comparable and very high
across the two samples (94% and 95%
respectively). The weights of recyclate
recovered were 5.4 Kg/household per week
(September) and 4.5 Kg/household/week
(April). The diversion rates for the two sampled
areas were 36% and 31%, which were
comparable with the borough wide scheme
performance of 30%.
The recovery of newspapers and pams was 3.3
Kg/household per week (September) and 2.7
Kg/household/week (April), which equated to
79% and 82% recoveries respectively.
Recoveries of other materials were lower: Cans
(80% and 45%), and plastic (75% and 63%).
There was much higher arisings of paper in the
samples than the national average and much
lower levels of glass and putrescibles in the
residuals than nationally. It was conjectured
that this might imply a high local use of bottle
banks and composting also occurring.
Contamination of the recyclate was relatively
high at 6.1% (September) and 8.5% (April).
Glass and textiles formed a relatively minor part
of this contrary material.
1.4.7 Kerbside Glass Recovery
Reference: WRAP (2002).
Authorities who now operated kerbside glass
collections were found to have doubled their
yield of glass recyclate (on average) since 1993
compared with a more modest 1.5 times
increase (on average) from authorities not
collecting glass at kerbside. The study
considered five of those schemes in more
detail. Together those schemes serviced
230,000 households and yielded 6,663 tonnes
of glass per annum (0.56 Kg/household/week)
with 2,480 further tonnes recovered via bottle
banks (not including civic amenity sites).
12

Cotswold collected glass through a fortnightly
44 litre box scheme collecting mixed dry
recyclables. It was an opt-out scheme which
was widely publicised amongst the residents.
35,000 households were serviced by the
scheme. 95% of collections were on the same
day as the normal refuse collection. Materials
were separated at the vehicle by the collection
staff. Participation rates were unknown. Yields
were 0.66 Kg/hh/w (glass), 1.34 Kg/hh/w
(paper) and 0.06 Kg/hh/w (cans).
Lambeth collected glass through a weekly co-
collection of glass, paper, cans and textiles in a
55 litre box. Again the scheme was widely
publicised at the start, backed up by quarterly
newsletters and reminder cards. 74,000
households were served. Yields were 0.51
Kg/hh/w (glass), 1.07 Kg/hh/w (paper) and 0.03
Kg/hh/w (cans).
An anonymous district collected their glass
together with cans in a 55 litre unlidded box,
the collection was weekly (due to residents
concerns over longer-term storage) but not
necessarily on the same day as the refuse
collection. It was noted that theft of boxes was
considered to be a problem. Paper and textiles
were collected alongside in a separate bag,
with 37,500 households being served.
Participation in the scheme was estimated to
be 65%. Yields were 0.54 Kg/hh/w (glass), 0.99
Kg/hh/w (paper) and 0.04 Kg/hh/w (cans).
Another anonymous district collected glass
along with other dry recyclables in 35 litre
baskets, collected the same day as the refuse.
The collection staff carried out a kerbside sort.
Advertising and public awareness were thought
to be very important to the scheme where
58,247 households were served. Participation
in the scheme was reckoned to be 55% (1997
figures) in both the urban and rural areas of the
district. Yields were 0.46 Kg/hh/w (glass), 0.93
Kg/hh/w (paper) and 0.04 Kg/hh/w (cans).
Finally, Durham collected glass along with
other dry recyclables in 55 litre boxes supplied
without lids, though additional boxes or smaller
35 litre boxes were available on request. The
boxes were collected the same day as the
refuse and the collection staff carried out a
kerbside sort. 25,000 households were served.
Set out rates for the scheme were 70% and
80% in the urban and rural areas respectively.
Yields were 0.8 Kg/hh/w (glass), 1.56 Kg/hh/w
(paper) and 0.06 Kg/hh/w (cans).
13
Understanding Recycling Behaviour 2
1.4.8 Hertsmere
Reference: Clamp (2000)
Herstmere Borough contains 38,000 dwellings,
33,000 of which were serviced by a kerbside
collection of paper. Flats were excluded. The
paper categories collected were newspaper,
pamphlets, magazines, catalogues, directories
and white paper. Collections started in 1996
through a plastic sack collection. Recoveries
were then around 95 tonnes per month (0.66
Kg/household per week) but have risen to 139
tonnes per month (0.97 Kg/hh/w) on average in
2000. Based on a waste audit under taken in
1998, 6,700 tonnes of paper per annum (3.90
Kg/hh/w) are available for recycling, so there is
a 25% diversion rate. Another 8% diversion is
achieved via the borough‟s bring sites.
A trial in Potters bar started in September 1999
in which the collection bags were replaced by
boxes. The trial area covered 3,027
households. 1,445 households were issued
boxes with lids and 1,647 were issued boxes
without lids. The boxes were split between 53
and 35 litre capacities distributed randomly
across both areas. The new scheme was
promoted through a flier backed up by press
articles.
The set out rate for the trial areas rose from
29% (before) to 50% (after) in the area with
lids, and from 28% (before) to 40% (after) in the
area without lids. Overall yields rose from just
over 1.0 Kg/hh/w to 1.65 Kg/hh/w showing a
61% increase on introducing the scheme.
Substantially higher yields were achieved from
the lidded boxes than from the unlidded boxes
(2.46 versus 1.25 Kg/hh/w) however this may
not all be attributable to the lids. The unlidded
trial area also contained greater proportions of
local authority housing than did the lidded area,
and greater proportions of the smaller boxes
were distributed in the unlidded area. No
investigations of the effect of box size were
made.
A questionnaire survey administered 5 months
into the new scheme explored the reasons why
the new recruits had been stimulated to join the
scheme. Those reasons covered a previous
lack of awareness, experience of erratic
service, poor supply of bags, and bags blowing
away. Interestingly those answers implied that
the new recruits might have been drawn
primarily from drop-outs from the previous bag
collection scheme rather than from „fresh‟ non-
recyclers. It may have simply drawn back those
willing to give the scheme another shot now

that the scheme had been changed and
perceived to have improved. On the down side,
some residents (mainly the elderly) complained
that the new boxes were too heavy to handle.
In conclusion, whilst the results might indicate
that lidded boxes may yield more material and
stimulate higher participations than unlidded
boxes which, in turn, attract higher yields than
bags, the experimental design was insufficient
to deconvolute the effects of container type
from possible contributory effects of
demographics, box size, and from the
associated promotional messages.
It was considered that the reason why lidded
boxes were apparently more successful lay in
their convenience in as much as they could be
stored outside rather than needing inside
storage. However this must be set against the
operational considerations that it is easier for
the collection staff to empty and return unlidded
boxes.
1.4.9 Kensington and Chelsea
References: Read (1998, 1999).
Kensington and Chelsea has offered kerbside
recycling to all residents since 1993.
Collections were accomplished using used
carrier bags, collected twice a week. Despite
extensive promotion of the scheme via the
traditional methods of posters, leaflets, and
newspaper adverts, many households still
claimed to be unaware of the service and did
not recall any publicity. A new „roadshow‟
approach was adopted in 1996. It aimed to
make personal contacts with all residents by
talking to those residents at home, delivering
persuasive messages and distributing
supporting leaflets, badges and stickers.
In the first 18 months, 8% of the borough‟s
78,600 households were contacted. The
interviewers found that some 31% of residents
appeared unaware that a kerbside recycling
service was offered, and this lack of knowledge
was cited by 73% of households as their main
reason for non-participation. Across collection
rounds, there were clear associations between
the level of awareness and the tonnage
collected. The tonnages collected post
roadshow increased by 19% though there was
much variability across rounds (35% down to a
negative impact in one round). Figures were
based on a four-week average. Monitoring was
Understanding Recycling Behaviour 2
not extended beyond that period. Overall, it
was claimed that the roadshow helped to raise
the doorstep recycling schemes recycling rate
from 9% in April 1995 to over 11% in October
1996, though other factors may have
contributed as well.
1.4.10 Other Schemes
West Lothian moved from a plastic sack
collection to a wheeled bin collection for paper
in 2001. Three months into the scheme, plastic
bottles and carriers, card and food and drink
cans were also added to the list of accepted
items. A third wheeled bin was additionally
provided for green waste. The pilot scheme
covered 7,200 houses. Recoveries in the first
year of the pilot scheme averaged 1.97
Kg/hh/w (mixed paper and card), 0.13 Kg/hh/w
(mixed plastics) and 0.03 Kg/hh/w (mixed
cans). Before figures for the pilot area were not
available, but as a comparison the 59,000
households still on the sack collection for paper
only recovered an average of 0.54 Kg/hh/w
over the same period. Green waste recoveries
in the new scheme averaged 2.98 Kg/hh/w.
Most of the households in the trial area
participated with all 3 bins, though some took
just a dry recyclate bin or organic bin rather
than both.
West Farnborough introduced a trial scheme in
1999 for a blue bin collection of mixed dry
recyclables. Previously the same materials had
been collected through a „bag and hook‟
method, whereby recycling bags were attached
onto hooks on the side of the residual waste
wheelie bin. In the trial area of 1,172
households, participation increased from 62%
to 74% with the introduction of the new
scheme, with yields rising from 1.31 kg/hh/w to
2.04 Kg/hh/w. There was noticeably more cans
and plastics in the recyclate compared with
previously. The bag and hook method had
engendered much more selective recycling.
The new scheme was also considered to offer
the added advantages of more storage,
possibility of outside storage, and greater ease
in setting out.
Finally, Smith et al. (1999) studied plastic bottle
recycling. They noted a difference in kerbside
recovery rates depending on the type of
container used for collection. Averaging over a
number of schemes diversions ranged from
27% (box), 21% (supplied bag), 20% (wheeled
bin) down to 11% (own bag).
14
 Understanding Recycling Behaviour 2

1.5 Cross Authority Comparisons

1.5.1 UK schemes
The case studies listed above refer mainly to
single schemes without setting those schemes
in the broader context of kerbside collections.
As such, they can only provide hypotheses
rather than explanations of why those schemes
are considered to be successful (or otherwise).
Relatively few studies have attempted to tackle
the broader issue, in any greater depth than
producing league tables of authorities‟ recycling
rates.
DEFRA statistics classify recycling rates by
Government Office Region. The statistics show
clear differences between the poorer-
performing north English regions to those
generally performing much better in the south.
Parfitt et al. (2001) analysed and compared
some alternative classification regimes. He
showed that the Office for National Statistics
ONS) classification also produced significant
differences in the household waste collected for
recycling (table 1.3). Parfitt et al. (2001) then
went on to undertake a cluster analysis of
authorities based on a [limited] number of
programme design variables. The derived
classification shown in table 1.4 provided eight
primary clusters.
Residual waste arisings, percentages recycled
and civic amenity collected weights were all
found to be significantly different across
clusters, with averages ranging from <5%
recycled (Wheeled bin I, urban), 5-10% (other
wheeled bin clusters, no method I, plastic sack
I), 10-15% (no method II, plastic sack II and III).
Rural authorities in any cluster generally
performed better than their urban counterparts.
There was, however, considerable overlap
between clusters, with considerable variation
amongst authorities in any given cluster.
The results do show how both residual waste
containment and kerbside provision may be
important to the total recoveries of recyclate
achieved. They do not, however, explicitly
consider the different design options for
recovering that recyclate. We could not find any
published information of such studies being
undertaken in the UK, though there were a
handful of publications relating to the US.
Those are reviewed in the next section.

Table 1.3 Yield Kg/household/week (ONS Categories)

Type of authority Rural Prosp- Mature Urban Mining/ Inner
erous centres Industrial London
Household waste 2.7 2.9 2.7 2.0 0.9 1.1
collected for recycling
Source Parfitt et al. (2001)

Table 1.4 Clusters Based on Programme Design Parameters

Residual W aste Description Average % Bring sites per
Containment households on 1000 households
kerbside schemes

Wheeled bin I Low bring/ca & kerbside 10.6 0.59

Wheeled bin II Low bring/ca high kerbside 88.4 0.60
Wheeled bin III High bring ca & high kerbside 60.3 1.63
No method I Moderate bring/ca & low kerbside 10.9 0.85
No method II Moderate bring/ca & high kerbside 84.4 0.74
Plastic sack I Low bring/ca & low kerbside 13.4 0.84
Plastic sack II Moderate bring/ca high kerbside 87.1 0.88
Plastic sack III Very high bring/ca moderate 57.8 3.25
kerbside
Source Parfitt et al. (2001)

15

Other variables that have been considered as
affecting community recycling performances
include the community‟s deprivation index
Hogg and Mansell).
1.5.2 US Schemes
Folz (1991) analysed 264 recycling schemes
across the US in 1989, 109 of which provided
voluntary kerbside recycling. Overall, recycling
participations (estimated by the municipalities)
showed significant correlations with programme
type (voluntary versus mandatory) and whether
or not kerbside collections were provided.
Statistically significant relationships occurred
for several other policies as well. Cities that
imposed sanctions or issued reminders were
more likely to have higher participations.
Establishing a specific goal (diversion target)
was also correlated to higher participation.
Schemes using private sector contractors to
collect the recyclables performed significantly
better than those using public services. For
voluntary kerbside schemes, the provision of a
free container was important, though the effects
of same day collection as the refuse and co-
mingled versus segregated collections were not
significant. The most significant factor,
however, was the involvement of citizen
participation in the programme design (through
citizen surveys and meetings with community
groups).
In looking at socio-economic differences
amongst municipalities, Folz and Hazlett (1991)
showed that per-capita income was positively
correlated with kerbside participation but not
drop-off participation. Being female, or older, or
more educated correlated with drop-off
participation but not kerbside participation.
Other characteristics tested (household size,
tenure, household income and occupational
status) were not significantly correlated with
any aspect of recycling behaviour. Overall,
demographic characteristics appeared to be
relatively unimportant in explaining recycling
success.
Folz carried out a repeat analysis in 1996 (Folz,
1999). Participation and diversion rates in most
schemes had increased substantially since
1989 (from 49.8% to 68.8% participation for
voluntary kerbside schemes). Amongst
voluntary programs, participation gains were
highest in cities that had established a near-
future recycling goal, in cities that provided free
containers, and in cities that had established
„block leaders‟ (indigenous local champions
who reminded or encouraged citizens to
Understanding Recycling Behaviour 2
recycle). Same-day pick up and separation
requirements were not significantly related to
participation gains in voluntary schemes, nor
were speeches of city officials. Cities with
variable fee or volume pricing for residual
waste collections did not return significant gains
in participation, though they had much higher
mean rates of participation to begin with. It was
concluded that such financial policies, at the
very least, had an enduring positive effect on
recycling behaviour. Diversion levels rose
alongside participation, though it was noted that
generally more materials were being targeted.
Another longitudinal study by Feiock and Kalan
(2000) looked at changes in diversion rates
from 1989 to 1995 for 67 counties in Florida.
Programme design variables were not found to
be strong predictors of any changes in
recycling rates over time. However, county
education levels and per-capita income were
significantly correlated with the temporal
increases in diversion, though levels of support
for environmental protection were not.
The above studies considered averages
without explicitly considering variances. The
results of a separate study by Noehammer and
Byer (1997) compiled survey data for 104
kerbside schemes in the US from surveys
undertaken between 1987 and 1991,
concentrating on the ranges of performances
being achieved. They found that the best
performing schemes all tended to be
comparable in performance, whether they
collected weekly or monthly, whether they were
based on a single separation or on 3 or more
separations, whether they are mandatory or
voluntary and so on. However amongst poorer
performing schemes, the poorest performances
were returned by those with monthly
collections, high numbers of separations, and
voluntary participation. It follows that there may
be key factors other than programme design
that will determine a successful scheme.
However, for less successful schemes,
inappropriate design parameters can further
reduce the quality of performance. This
hypothesis raises a crucial issue as to what
constitutes best practice: If you are managing a
kerbside scheme well, in a population that is
sympathetic to the recycling ideal, then it does
not matter how you collect the recyclate!
However, when circumstances are less good,
you can minimise the resultant under-
performance through judicious programme
design.
16

1.6 Intra-authority Comparisons
Most analyses of recycling performance are
undertaken at municipality level or at even
coarser resolution. Very few published
analyses have looked deeper inside the
municipalities, and at the spatial variations that
occur within those municipalities. However
those variations can be quite significant.
Achieving best practice in a municipality means
achieving best practice in all constituent parts
of that community, and that involves identifying
and understanding neighbourhood differences.
For example the promotional roadshow
discussed by Read (1998,1999) had a negative
effect on the diversions from one
neighbourhood, null effects on two others, but
generally larger, positive effects on the rest.
However, the reasons for this were not
answered.
Tucker (2001) analysed neighbourhood
variations in recycling participation at even finer
levels: (i) at enumeration district level – around
250 households, and (ii) at the level of
individual streets. At street level, simple
randomness was found to be a dominant factor,
whilst at enumeration district level, more of the
1.7 Discussion and Conclusions from Previous Research
The above research has clearly demonstrated
the wide variability that can occur in household
waste recycling behaviours. Such variabilities
occur amongst individual households within the
community, amongst whole neighbourhoods
within a municipality, and amongst
municipalities. Much research has been
undertaken to establish the determinants of
these behavioural differences, but by and large
the results have been equivocal. How well a
municipality performs can be influenced by
many factors. Basically, individual
householders‟ attitudes form the principal
drivers of their behaviour, and collectively for
their collective behaviour. For best practice, a
kerbside recycling programme must be
designed to mesh with those attitudes, and to
shape and improve any weaker or more
negative attitudes that may be held about the
scheme. Unfortunately, we do not yet know the
precise attitudes that fundamentally underpin
the recycling ethos. We do however know more
about the negative factors that are cited for not
17
Understanding Recycling Behaviour 2
variability could be explained in terms of
identified differences in „features‟ of the
individual neighbourhoods. Tucker (2001)
carried out a large-scale study of participations
in a kerbside paper collection across 20,000
households in Fylde Borough. The study found
that demographic factors (housing type and the
proportion of retired residents) could explain
around 50% of the observed variability,
however some 20% of neighbourhoods did not
conform at all to this behavioural model. In
investigating possible reasons, Tucker found
that neighbourhoods with the more
heterogeneous mixes of housing types, and
those adjacent to particularly high performing
areas often performed better than expected. It
was hypothesised that such effects could be
attributable to strong normative influences. On
the other hand, residential households in the
urban centres performed worse than expected.
Another class of poor performers identified
neighbourhoods in which half the area was
performing badly whilst the other half
performed as expected. It was hypothesised
that local incidents of poor service provision
might have triggered these polarisations.
recycling. Not having a kerbside scheme is a
major reason if there is not a kerbside scheme,
whilst problems with the kerbside scheme is a
major factor if there is. The other major factors
are not having enough information or not
having enough waste to make it worthwhile.
Most kerbside schemes are intrinsically stable.
What they collect and divert next year will not
be too different to what they collect and divert
this year, unless there is a step change in
program design and management, or unless
serious incidents occur. The participation levels
achieved at scheme start up would appear to
be crucial. High levels of public participation or
expert consultation at the programme design
stage can be important, with awareness raising
and education being important at that stage
too.
The actual design parameters of the
programme may not be too important in a
global sense, but may be important locally.

Collection by bag or box or bin, weekly or
fortnightly, number of segregations required,
day of collection and so on can aid the
convenience of using the scheme but appear
not to be essential requirements for a well-
performing scheme. Good and mediocre
performances can occur with all systems.
Sufficiency of information may be a more
necessary requirement towards achieving good
performance, but may not be sufficient on its
own to ensure a good performance.
Outside scheme start up, significant positive
increases in performance over time appear
almost entirely linked to implementing step
changes in programme design. That is, when
an existing kerbside scheme is upgraded, from
a paper only to multi-material collection, from a
bag collection to a box or bin and so on.
Changing an existing scheme to become „more
convenient‟ can increase participation through
several routes. It is easiest to visualise the
possible contributory factors through a concrete
example. Let us consider, as that example, the
change from a paper-only bag collection to a
multi-material collection in a lidded box:
(i) Those previously not using the
scheme because they “did not have enough
paper” may find that their bottles and cans do
now make up reasonable amounts of material
that are worthwhile to recycle.
(ii) Those previously unaware of the
scheme, or of its collection days, or not feeling
comfortable that they know how to use the
scheme will receive the necessary „education.‟
(iii) Those not using the old scheme
because of various „convenience‟ concerns
about the scheme (e.g. those not wanting to
store the bag inside, may now see definite
advantages in a weatherproof storage box).
Understanding Recycling Behaviour 2
(iv) Those who had dropped out of the old
scheme because of problems and incidents
may write off those incidents as history, and be
willing to try a fresh new start with the fresh
new scheme.
v) Committed recyclers will simply
continue.
Almost universally, such changes are
accompanied by attendant publicity, promotion
and education campaigns, so it is hard to
isolate the direct effects from the physical
change from any effects caused by the
promotion. Perhaps what is most important is
that they occur together.
Even in the most intensive promotional
campaigns applied in the UK the promotions
actually achieved very little behavioural change
when there was not an accompanying system
change. The Kensington and Chelsea
experience (section 3.9), for example, only
realised around 2% increase in recycling rate
after 18 months intensive door-to-door
campaigning.
There may be two main issues here. Firstly,
generic campaign messages may do little to
reassure the disillusioned that their grievances
will be rectified. Secondly, the message may
not be received at a time when the issue is
salient to the recipient. In that respect, the
message needs to be both congruent with the
point of action (e.g. received in the household,
not [say] at a bus stop) and needs to be
provided when waste management is high on
the agenda of the recipient. Effectively that is
when recycling issues are impacting on normal
household routine and habit. Imposing/ offering
a new recycling scheme provides just that
timing and trigger.
18

2. Where Are We Going?
The research reported in this section was commissioned by the UK Cabinet Office Strategy Unit to
inform the development of the SU „Waste Not Want Not‟ report. This monograph presents a slightly
abridged version of the study.
2.1 Introduction
2.1.1 Aims
The study aimed to provide a general forecast
of the possible levels of recycling
performances, in English regions, for a number
of alternative management scenarios. Those
scenarios explored increased provisions of
recycling infrastructure, with or without parallel
promotional and educational campaigns. It
aimed to provide the first estimates of how
much might be achieved simply by making
recycling more convenient and how much must
additionally rely on changing people‟s attitudes.
Separate model predictions were developed for
a currently relatively poorly performing region of
England, the North West, and for a better
performing region, the South West.
2.1.2 Methodology
The predictions were made using the University
of Paisley Integrated Household Waste
Management Model, and are based on the
theories and heuristics contained within that
model, relying on the assumptions within the
model, together with assumptions made about
the likely waste arisings in the English regions.
A full technical description of the model and its
assumptions can be found in technical
monographs Understanding Recycling
Behaviour (Tucker, 2001) and Understanding
Home Composting Behaviour (Tucker and
Speirs, 2001).
The Integrated Household Waste Management
Model is essentially a model of individual
household behaviours that models how
individual household attitudes and behaviours
might contribute to community-wide household
recycling performances. Basically, the
behavioural model forecasts how many
households might participate in available
recycling schemes and how much of their
available waste they might contribute to those
schemes.
19
Understanding Recycling Behaviour 2
The forecasts are made by firstly calibrating the
household attitude profiles to produce model
output that matches the community
performance indicators for the current levels of
recycling provision. After calibration, the effects
of change can be forecast through running
„what-if‟ scenarios. Model sensitivity analyses
can then be used to explore the likely ranges of
possible outcomes.
Basic Assumptions
The Integrated Household Waste Management
Model comprises a number of inter-linked sub-
models describing waste arisings and waste
diversions, principally kerbside and bring
recycling of dry recyclables and green wastes,
and home composting. The model takes as
input: (i) an estimated average waste
compositional analysis for the community being
studied, (ii) socio-demographic profiles for each
neighbourhood within that community, (iii)
specifications of the waste management
provision provided to each neighbourhood. The
model was initially developed with the district
authority being the community, with the
neighbourhoods being individual kerbside
collection rounds or enumeration districts within
that authority. The current application now
takes the community to be a whole region of
England with its constituent neighbourhoods
being the local authority boroughs contained
within that region. In adopting that definition, it
was no longer practical to simulate every
household in the community, nor to specify the
precise details of the variety of waste recovery
schemes operating in different
„neighbourhoods‟. The expedient adopted in the
analyses was to simulate a demographically
representative sample of 1% of the households
of the region, and to undertake the modelling
simply for „generic‟ bring sites and „generic‟
civic amenity (ca) sites. Individual kerbside
collection schemes for dry recyclables were
specified as being one of three generic classes:
paper only, multi-material 1 (paper, glass, cans
and textiles), or multi-material 2 (paper, glass,

cans, textiles, plastic beverage bottles and
cardboard).
Socio-demographic factors were derived from
1991 census data, augmented by a Mosaic
classification of household types at postcode
sector level.
2.1.3 Model Calibration
The model was calibrated to give a reasonable
fit to the reported 2000-01 recycling factors for
each local authority within the modelled region
and the aggregated values checked against
DEFRA 2000-01 waste management statistics.
Material specific recoveries were matched to
the regional level statistics reported in the
DEFRA 1999-00 waste management survey
(2000-01 compositional data were not available
at the time of modelling).
2.1.4 Scenarios Modelled
The scenarios modelled were:
 Intensification of bring sites
 Introduction of kerbside paper collection to
all areas currently without kerbside
provision
 Multi-material kerbside collection (paper,
glass, cans, and textiles) introduced region-
wide
 Extended multi-material collection (paper,
glass, cans, textiles, plastic bottles and
card) introduced region-wide
 Green waste [and kitchen waste] kerbside
collections introduced region-wide
 Intensified home composting promotional
bin offers
The time and resource constraints of the study
did not allow for all different operational
alternatives of the schemes to be comparatively
assessed (e.g. full source separation versus
co-mingled, weekly, fortnightly or monthly
collections, collection container size and type,
etc.). Instead the analyses simply sought to
generate the typical outcomes for each generic
collection type.
Understanding Recycling Behaviour 2
In addition four types of accompanying
intervention were studied:
 Enhancement of pro-recycling attitudes
 Procedural information and awareness
 Addressing the perception that “It is not
worth recycling if you do not have much to
recycle” (Doing your bit, Do a little change
a lot)
 Addressing actions that require „extra effort‟
from the recycler – e.g. distaste in washing
out cans, coping with „busy‟ lifestyles, what
to do if things go wrong etc.
Again, it was not possible to consider every
type of possible intervention and every means
of delivery. Those aspects are more within the
province of the national waste awareness
campaigns. It was also recognised that more
coercive interventions such as waste charging
or reduced residual waste servicing may be just
as appropriate as promotions or persuasive
messages in addressing some of the issues
involved. The study sought to typify results from
fairly strong interventions, of each type,
towards the top end of the range of what could
potentially be achieved in practice. Those
outcomes are contrasted with the
corresponding „worst case‟ scenarios
representing likely outcomes with little or no
promotion.
2.1.5 Definitions
The definitions used in this section are:
Recycling participation = those using the
scheme at least once in a four week period 
those with access to the scheme over that
period
Capture rate = weight of a given material
diverted from the waste stream through a given
scheme(s)  total amount of that material in the
waste stream from all households in the region.
Note, that to provide comparability across all
scenarios this definition differs from the
standard DETR definition of the „recognition
ratio‟ capture performance indicator.
20
 Understanding Recycling Behaviour 2

2.2 North West England

2.2.1 Intensification of Bring what might be practically achieved. The
Facilities promotion investigated for scenario B was:

The scenarios modelled were: Scenario B2:

Scenario A: Can collections at all bring
sites & doubling of number of bring sites in
districts currently without high density bring
Scenario B: Can collections at all sites &
high density bring introduced across the entire
region
Kerbside collections were maintained at their
current levels of provision.
The results are given in table 2.1 where they
are compared with baseline (or current) values
obtained from the model calibration.
These results provide likely outcomes for a
basic „minimal‟ promotion and awareness-
raising of the new schemes. The additional
changes that may be achieved with intensive
promotions are given below. It should be noted
here that quite strong promotions have been
modelled and, as such, represent the limits of
Addressing pro-recycling attitudes + Procedural
& awareness campaigns
Other scenarios were then investigated for
current bring densities but with can collections
introduced at all sites:
Scenario C2:
Pro-recycling attitudes + Procedural information
& awareness campaigns
Scenario C3:
Pro-recycling attitudes + Procedural information
& awareness campaigns + Addressing
perceptions of minimum amounts to make
recycling worthwhile
In scenario C3, participation at bring/ca sites
was predicted to increase to 39.5% and
participation in the currently operating kerbside
schemes would be expected to range between
41 and 84% (paper only) and to around 80%
(multi-material schemes) according to the
locality. The predicted effects on capture rates
are listed in table 2.2.

Table 2.1 Predicted Capture Rates (%) for an Intensification of Bring Facilities
Material Baseline Scenario A Scenario B
Bring All Bring All Bring All
+ca recycling +ca recycling +ca recycling

Paper 5.6 14.2 9.2 17.0 14.7 25.9

Glass 9.5 11.2 16.3 17.7 27.4 29.5
Cans/ foil 2.2 2.6 8.1 8.9 15.1 16.2
Plastic 1.2 1.2 3.8 3.8 5.5 5.5
Textiles 2.5 3.6 2.0 3.1 2.9 4.6

% Participation 14.2 32.9 25.3 41.0 44.5 52.2

Table 2.2. Effect of Interventions on % Capture Rates: Non-bracketed figures refer to all
recycling activities. Bracketed figures refer to captures at bring and ca sites only
Material Baseline Scenario B2 Scenario C2 Scenario C3

Paper 14.2 36.1 (22.9) 26.9 33.1 (19.2)

Glass 11.2 48.2 (45.1) 26.0 33.2 (31.3)
Cans/foil 2.6 34.5 (33.0) 27.8 23.8 (22.2)
Plastic 1.2 6.9 (6.9) 2.5 4.7 (4.7)
Textiles 3.6 7.9 (6.6) 7.8 12.0 (10.8)

Participation 32.9 72 53 67

21
 Understanding Recycling Behaviour 2

2.2.2 Full Kerbside Coverage different multi-material collection regimes were

Regionwide investigated:

The first scenario modelled was the
introduction of paper-only kerbside collections
to all areas currently without any kerbside
provision. Results are shown in table 2.3.
It is additionally noted that an introduction of
paper-only (news and PAMs) kerbside
recycling may only have a minor effect on drop-
off participations, though those participations
could be significantly increased again with the
generic recycling promotions. Without those
promotions, paper capture at the bring and ca
sites is predicted to decrease from 5.6% to
4.1% following the expansion of kerbside
collection to the whole region. However, around
20% of the total paper collected would still
come from the bring and ca sites. The kerbside
paper collection would not obviate the need for
providing parallel paper recycling facilities
through those bring and ca sites.
The second scenario modelled was a full
implementation of multi-material kerbside
collections Region-wide In the study two
Paper + glass + cans + textiles
Paper + glass + cans + textiles + plastics
drinks bottles + cardboard
100% coverage was assumed for the kerbside
collections, and it was assumed that bring/ ca
provision remained the same as that currently
operating.
The first set of simulations investigated the
minimum promotion (worst case) scenario for
introducing type1 multi-material collections with
the scenario being compared with paper only
collections and with the current baseline. As
any scheme introduction will almost certainly be
accompanied by some promotional activity, a
further simulation is included, with a typical
„normal‟ level of (scheme specific) promotion
being applied. It is considered that this will
represent the most likely outcome for a careful
and well-orchestrated, inaugural promotion. All
these initial simulations exclude doorstep
collections from flats. The results are shown in
table 2.4.

Table 2.3 Effect of Region-wide News and PAMs Collection: All recycling activities (Bring+ca
only)
Material Baseline Flats not included Flats included
Scenario P Scenario P3 Scenario P Scenario P3
(minimum (intensive (minimum (intensive
promotion) promotion)* promotion) promotion)*

All paper 14.2 (5.6) 19.4 (4.1) 39.0 (14.0) 20.5 40.4

Newspaper 27 (11) 39 (8) 71 41 74

Magazines 22 (8) 30 (6) 63 32 66
Junk mail 22 (9) 31 (6) 63 31 65
Catalogues 17 (6) 25 (5) 61 27 65
Other paper 2 (0.4) 2 (0.4) 0.4 2 0.4
Card 4 (4) 4 (4) 17 4 15

Participation 33 53 83 56 88
(all recycling)
Participation - 48 75 51 80
(kerbside)
Participation 14 13 36 13 35
(bring+ca)
* as per scenario C3

22
 Understanding Recycling Behaviour 2

Table 2.4 Effect of Multi-material Kerbside Collection on % Material Capture Rates: All
recycling (Bring + ca only)
Material Baseline Scenario P Scenario M Scenario MN
(paper-only) (worst case) (normal
promotion)

Paper 14.2 (5.6) 19.4 (4.1) 24.0 (4.2) 31.6 (4.3)

Glass 11.2 (9.5) 11.2 (9.5) 41.5 (6.2) 50.6 (7.5)
Cans/foil 2.6 (2.2) 2.6 (2.2) 19.8 (3.2) 24.5 (4.0)
Plastic 1.2 (1.2) 1.2 (1.2) 0.8 (0.8) 0.8 (0.8)
Textiles 3.6 (2.5) 3.6 (2.5) 25.0 (1.3) 30.1 (1.5)

Participation (all recycling) 33 53 70 76

Participation (kerbside) - 48 66 70
Participation (bring + ca) 14 13 8 9

The results show a significant increase in the associated increases in kerbside recoveries.
recoveries of all collected materials could be The casualties will be amongst those materials
achieved compared to paper-only collections, not included in the kerbside collection,
with increases of up to 50% (relative) in paper principally the plastics and cardboard.
recoveries compared to paper-only kerbside
collections. (Typically, in authorities where With an associated or independent
such changes have already been made, paper promotional activity aimed at enhancing
recoveries do actually rise by 40% or more – recycling behaviours more generally,
see sections 4.2, 5.2). There is however, a participations and capture rates could be
negative knock-on effect at the bring and ca increased further. Table 2.5 shows the results
sites with participations at those sites for a range of promotions operating with the
predicted to fall by some 5% absolute (35% worst case scenario, M. In these simulations
relative), however significant recycling will the strong promotion is the same as that used
continue at the sites in parallel with the in scenario C3 (addressing recycling attitudes,
kerbside recoveries. Capture of all materials minimum amounts, awareness and procedural
by bring recycling will decrease approximately information) whilst the maximum promotion
pro-rata with the fall in bring participation, scenario (M4) additionally addresses
though the overall recoveries of the region will forgetfulness, extra effort and what to do in
be more than compensated for by the large difficulties.

Table 2.5 Effect of Multi-material Kerbside Collection and Promotion on % Material Capture
Rates: All recycling (Bring + ca only)
Material Scenario M M3 (strong M4 (max
(worst case) promotion) promotion)

Paper 24.0 (4.2) 38.0 (8.8) 45.3 (14.6)

Glass 41.5 (6.2) 60.9 (14.6) 72.7 (27.7)
Cans/foil 19.8 (3.2) 41.3 (10.0) 55.0 (18.4)
Plastic 0.8 (0.8) 3.3 (3.3) 3.9 (3.2)
Textiles 25.0 (1.3) 30.3 (4.0) 37.0 (7.2)

Participation (all recycling) 70 85 87

Participation (kerbside) 66 77 80
Participation (all bring +ca) 8 29 26

23
 Understanding Recycling Behaviour 2

With flats included in the collections, the
capture rates for all collected materials would
rise by a further 2-3% points (approximately).
As with paper only collections, intensive
promotional activities would appear to be
crucial to the schemes achieving their full
potential. Whilst the basic recycling attitude
promotions, awareness campaigns and
provision of procedural information could
potentially achieve significant levels of
participation, they would on their own appear
insufficient to realising the full potential that is
ultimately available from the scheme. The
problem may be more with capture than with
participation, with many recyclers still not
recycling the maximum they are able to. Typical
problems include „being too busy to always
recycle‟, „perceptions that small amounts of
materials are not worth recycling‟, personal
difficulties and absences on recycling day
(when the resident holds a get rid as soon as
possible attitude), and having too much to
recycle (box full or too heavy – what to do with
the rest). All are potentially correctable, and if
that could be achieved there would be
significant further gains – as witnessed in the
maximum intervention scenario. Ultimately,
however, capture will be limited by other factors
such as safety considerations for broken glass
and sharp edges of cans, using newspaper for
wrapping dustbin waste or for various
contaminative uses (e.g. mopping spillages,
protective covering etc.) etc.
As the intensity of the promotions increase,
particularly with the maximum promotion
scenario, the relative amount of material
captured will increase at the drop-off sites. At
the ultimate limit, the final marginal increases in
capture are effectively all contributed by the
overflows and uncollected contributions from
the kerbside scheme being recycled at the
drop-off sites. Maintenance of satisfactory
levels of drop-off facilities will be essential in
the high capture scenarios.
The predicted capture rates of individual
component materials are shown in table 2.6 for
the modelled scenarios.
With the full range of promotional and
informational measures in association with a
100% multi-material kerbside coverage,
captures of around 80% of the newspapers,
PAMs and glass should be possible. Can and
textile recoveries would be expected to be
lower, but around 75% capture of those
materials should still be achievable ultimately.
Much lower capture rates would be achieved
for the materials not included in the kerbside
scheme, and of those materials, plastic bottles
and cardboard would appear the two materials
most needing to be addressed. The following
scenarios now extend the previous simulations
by including plastic drinks bottles (for soft
drinks, beer and milk) and cardboard into the
multi-material kerbside collections.

Table 2.6 % Material-specific Capture Rates: All recycling activities (Bring + ca only)
Material MN M3 (strong M4 (max
(normal promotion) promotion)
promotion)

Paper Newspaper 63 (11) 69 81

Magazines 49 (9) 62 79
Junk mail 45 (10) 61 82
Catalogues 42 (8) 57 77
Cardboard 5 (5) 14 16
Glass Green bottles 65 (12) 71 81
Clear/amber 64 (10) 70 80
Jars 34 (6) 58 77
Cans/foil Steel cans 33 (6) 56 75
Alu cans 44 (7) 58 77
Alu foil 3 (3) 13 15
Plastic Drinks bottles 2 (2) 12 16
Carrier bags 3 (3) 13 15
nd
Textiles 2 hand clothes 60 (3) 64 74

24

Table 2.7 % Capture Rates for kerbside collection of plastic drinks bottles and cardboard
(Over all recycling activities)
Material Baseline Scenario X2 XN
(worst case) (normal
promotion)
Cardboard 7 45
Plastic bottles 3 33
(All plastic) 2 7
The simulations show that including cardboard
and plastic bottles in the kerbside schemes
would not significantly affect the capture of the
materials previously collected at kerbside and
would only have a marginal effect at best on
kerbside participation rates. The region-wide
recovery of plastics and cardboard, however,
would increase significantly (table 2.7). The
additional interventions modelled in these
scenarios are the same as those considered
previously.
It is specially noted that the full range of
attitudes and perceptions need to be tackled
before plastic bottle recoveries rise above 50%.
In practice, plastic bottles are particularly
susceptible to not being recycled under
adverse circumstances (too few to be
worthwhile and too bulky to fit in the recycling
box if the recycling box becomes close to being
full).
2.2.3 Organic Waste Recoveries
So far, the discussion has concentrated on the
dry recyclable fractions of household waste.
This section now concentrates on the
putrescible fractions, principally the kitchen and
garden putrescibles. The baseline model
recoveries of those fractions from kerbside and
ca collections for the north west region were
estimated to be 27. Kg/household/year. It was
also estimated in the calibration that just under
30% of the population would be home
composting some or all of those wastes as well.
Whilst home composting diversions do not play
a part in recycling rate calculations, it is
nevertheless important to get a measure of how
much waste could be diverted into home
composting as this can impact directly on how
much waste is left for centralised composting
and how much of it could end up in the dustbin
or in the ca general waste skip.
The first scenario to be modelled was to instil a
positive change in recycling attitude similar to
the recycling attitude interventions for dry
25
Understanding Recycling Behaviour 2
X3 (strong X4 (max
promotion) promotion)
50 58 73
39 48 72
8 9 12
recyclables. The current provisions for green
waste collections at kerbside or at ca sites were
left untouched. The model scenario produced a
modest rise in organic waste capture across all
collections from 7.4% (current baseline) to
10.4%, with ca recoveries rising from 5.8% to
8.3% and participations there increasing from
8.3% to 11.1%.
The second scenario saw the introduction of
green waste collections extended across the
whole region (though not to the flatted
properties). Three cases were considered: (i)
the worst case scenario which could relate to a
minimum promotion or, alternatively, to a
scenario with separate charging for green
waste collections – e.g. tagged bags, (ii) the
most likely outcome with „normal‟ well-
orchestrated levels of promotion, and (iii) with a
particularly strong promotion of the scheme.
The results (figure 2.8) demonstrate that
enhancing kerbside recovery will have a
detrimental effect on the green waste
collections at ca sites, though as with dry
recyclables it does not obviate the need for
drop-off collections. „Larger‟ woody items, for
example, may still need be taken to the ca
sites. Overall, some 80% of green wastes
should be removable from the household
residual waste stream. This figure does not
include the potential for an absolute reduction
in the size of that waste stream through a
regime of intensified home composting. That
possibility is now considered. The scenario
modelled was a modest compost bin promotion
(for free or heavily subsidised bins through
special events) coupled with a green waste
collection region-wide.
The results predicted a marginal drop in
capture rate from the kerbside scheme, from
65.4% to 63.3%, when the parallel compost bin
promotion was mounted. In the scenario the
modest compost bin promotion increased home
composting participations from 28% to 38% (of
houses with gardens) diverting an extra 87,000
tonnes (24,000 tonnes kitchen waste and
 Understanding Recycling Behaviour 2

63,000 tonnes garden waste) across the
region.
A stronger compost bin promotion (through
doorstep offers of free bins) was predicted to
increase home composting rates to 52% and
divert a further 56,000 tonnes of organic waste.
More detailed model predictions of diversions
of garden and kitchen wastes through home
composting are given in the companion
monograph „Understanding Home Composting
Behaviour (Tucker and Speirs, 2001).
In the current development of the Integrated
Household Waste Management Model there is
no facility to model the collection of kitchen
putrescibles through kerbside collections, so it
was not possible to estimate their potential for
recovery. However, it is considered that the
capture rate for source separated kitchen
wastes or for kitchen wastes co-mingled with
garden wastes should at minimum match the
capture rates predicted above for green
wastes.

Table 2.8 % Material Capture Rates with an Extension of Kerbside Green Waste Collections
Region-wide (All recycling unless otherwise stated)
Fraction Current Worst case Most likely Strong
Baseline promotion

Kitchen compostable 0.8 1.9 2.7 3.7

Kitchen non-comp 0.7 1.8 2.4 3.0
Garden 15.2 37.8 65.4 85.0
(Total putrescible) (7.4) (18.0) (29.9) (38.4)

(Total at ca site) (5.8) (4.0) (3.5) (3.1)

Participation* ~50 30-50 65-70 70-90

(kerbside) 40-55
(urban areas)
Participation* 8.3 4.0 3.0 2.3
(ca site)
* of households with gardens

2.3 South West England

The same scenarios are now run for the South

East region. They are tabulated here without
additional comment.

Table 2.9 Effect of Intensification of Bring Facilities on % Capture Rates: All recycling (Bring
+ ca only)
Material Baseline Scenario B2 Scenario C3
(current)

Paper 17.5 (5.6) 35.0 (18.5) 30.5 (13.0)

Glass 25.4 (9.5) 54.0 (35.2) 40.8 (20.6)
Cans/foil 10.1 (1.2) 36.9 (25.0) 28.2 (14.7)
Plastic 1.3 (0.7) 2.6 (1.7) 3.3 (2.5)
Textiles 8.6 (1.2) 12.9 (3.3) 14.2 (5.2)

Participation 40.9 74.0 63.2

(all recycling)
Participation (bring/ ca) 10.6 40.0 23.0

26
 Understanding Recycling Behaviour 2

Table 2.10. Effect of Region-wide News and PAMS Collection: All recycling (Bring + ca only)
Material Baseline Scenario P Scenario P3
(current) (minimum (intensive
promotion) promotion)*

All paper 17.5 (5.6) 24.2 (5.0) 40.0 (15.5)

Newspaper 31.9 (10.6) 45 (10) 71 (34)
Magazines 27.2 (8.9) 39 (8) 63 (27)
Junk mail 28.1 (9.2) 40 (9) 63 (30)
Catalogues 25.5 (8.3) 36 (8) 61 (26)
Other paper 1.5 (0.3) 2 (0.1) 0.4 (0.1)
Card 7.5 (2.8) 8 (3) 19 (15)

Participation 41 53 81
(all recycling)
Participation - 41 49
(kerbside)
Participation 11 12 34
(bring +ca)
* as per scenario C3

Table 2.11 Effect of Multi-material Kerbside Collection on % Material Capture Rates: All
Recycling (Bring +ca only)
Material Baseline Scenario P Scenario M Scenario MN
(current) (paper-only) (worst case) (normal
promotion)

Paper 17.5 (5.6) 24.2 (5.6) 28.4 (4.8) 31.6 (4.9)

Glass 25.4 (9.5) 25.4 (9.0) 47.3 (8.4) 53.4 (8.2)
Cans/foil 10.1 (1.2) 10.1 (1.0) 24.6 (4.0) 33.3 (3.7)
Plastic 1.3 (0.7) 1.3 (0.6) 1.1 (0.7) 1.7 (0.8)
Textiles 8.6 (1.2) 8.6 (1.1) 17.8 (1.0) 21.9 (1.3)

Participation (all recycling) 41 53 69 73

Participation (kerbside) - 46 60 64
Participation (bring +ca) 11 12 9 12

Table 2.12 Effect of Multi-material Kerbside Collection and Promotion on % Material Capture
Rates: All Recycling (Bring + ca only)
Material Scenario M M3 (strong M4 (max
(worst case) promotion) promotion)

Paper 28.4 (4.8) 39.3 (10.8) 47.0 (16.1)

Glass 47.3 (8.4) 62.9 (16.4) 74.0 (30.5)
Cans/foil 24.6 (4.0) 40.9 (9.2) 54.9 (21.0)
Plastic 1.1 (0.7) 3.5 (3.5) 4.7 (3.9)
Textiles 17.8 (1.0) 25.0 (6.0) 29.4 (7.0)

Participation (all recycling) 69 78 85

Participation (kerbside) 60 65 75
Participation (bring + ca) 9 27 28

27
 Understanding Recycling Behaviour 2

Table 2.13 % Capture Rates for Kerbside Collection of Plastic Drinks bottles and Cardboard:
Overall Recycling Activities
Material Baseline Scenario M MN M3 (strong M4 (max
(current) (worst case) (normal promotion) promotion)
promotion)

Cardboard 8 43 51 58 72
Plastic bottles 6 31 39 52 71
(All plastic) 2 7 8 9 12

Table 2.14 % Material Capture Rates with an Extension of Kerbside Green Waste Collections
Region-wide: All Recycling unless otherwise stated
Fraction Baseline Worst case Most likely Strong
(current) promotion

Kitchen compostable 0.2 0.4 1.4 3.7

Kitchen non-comp 0.2 0.4 0.4 3.0
Garden 26.3 49.3 64.5 85.0
(Total putrescible) (12.7) (24.2) (31.7) (38.4)

(Total at ca site) (10) (6) (5) (7)

Participation* ~50 - 70 50 - 70 60 - 70 70 - 80
(kerbside)
Participation* 9.0 4.0 3.1 5.1
(ca)
* of households with gardens

2.4 Comparability Between Regions

These results show that, on the model adopted,
the relatively poorly current recycling region of
North West England and the better performing
region of South West England will both respond
in a broadly similarly fashion when improved
recycling infrastructure is provided and when
pro-recycling promotions are applied. The end
points for the two regions are expected to be
quite similar when comparable intensive
infrastructures and promotions are adopted.
At a finer scale of resolution, the intermediate
paths towards those end points will still echo
current differences in behaviour. Those
differences will be due in part to the current
differences in their respective current recycling
systems – e.g. currently more multi-material in
the south west and more paper-only collections
currently in the north west. They will also reflect
the current differences in recycling ethos
between the two regions, until those differences
are ultimately corrected through promotion and
education. Promotional and educational
campaigns will need to work relatively harder in
the poorer performing region to achieve the
ultimate endpoint.
The relative contributions of system
infrastructure and promotion and education to
the overall recycling performances are shown
in figure 2.1. In producing this figure, the
individual material results have been
aggregated to show the total material
contribution to the region-wide recycling rate for
each increment of scheme provision and
promotion.

28
 Understanding Recycling Behaviour 2

Figure 2.1 Predicted Relative Contributions of Infrastructure Provision and Auxiliary

Promotion to Overall Recycling Rate

50

45

40
Regional recycling rate

+max promotion
35
+ strong promotion
30
+100% ks green
25 100%multi+
20 +100% multi
+100% ks paper
15
Current
10

5
0
North West South West

multi = paper, glass,cans, textiles; multi+ additionally includes plastic bottles and cardboard
strong promotion assumes a successful targeting of „pro-recycling‟ attitudes, awareness and procedural
information and minimum weights
max promotion additionally successfully targets „personal cost‟ factors such as time, effort, distaste and
forgetfulness

Note: The contributions to recycling rates from white goods/ scrap metal, oil and other minor materials not
considered in the simulation have not been included in the figure and could add another 2 or more percentage
points to the overall recycling rates. Kitchen waste recoveries are also excluded. It is additionally considered that
carrier bag and aluminium foil collections may be underestimated in the simulations and there could be further
potential for increasing the capture of those materials at supermarkets through enhanced promotional activities
and scheme visibilities.

2.5 Discussion and Conclusions

Overall, the conclusions are that in both
regions, region-wide recycling rates of 40%
may be achievable through a combination of (1)
multi-material kerbside, (2) kerbside green
waste collections potentially coupled with (3)
kitchen waste collections, backed up by (4)
strong promotional education and awareness
campaigns. It would however appear to be
unlikely that a 40% recycling rate could be
achievable without every one of those elements
being in place.
Intensification of bring facilities across the
regions were found to be substantially less
productive than increasing multi-material
kerbside coverage with outcomes [possibly]
being tempered by past experiences and
prejudices about drop-off recycling and, in the
north west region in particular, still reflecting a
relatively poor recycling ethos across the region
as a whole. It was also evident in the model
results that the poor current capture rates had a
significant origin in the under-recovery of
available materials from active recyclers as well
as reflecting the relatively poor overall
participation rates. The studies predict that
interventions to decrease this under-recovery
may be essential if the highest capture rates
are ever to be achieved in the region.
Also of importance, not discussed previously, is
that the modelling studies showed that the
current „pro-recycling‟ attitudes held by the
better-performing districts in the north west

29

were identical to those held by the better
performing districts in the south west. That is,
given the same demographic profiles and the
same recycling infrastructures, there should be
no differences in recycling performance. Part of
the differences between the north west and the
south west lies in the number of district
authorities in each of the regions that do not
currently reach their highest potential.
Introduction of enhanced recycling
infrastructures was predicted to lead to
comparable step changes in material capture
across all districts, with the differentials in
performance between well-performing and
poorly-performing districts being largely
maintained. That is, a relatively well-performing
district will continue to be well-performing with
better recycling provision, whilst a relatively
poorly-performing district will continue to be
relatively poorly-performing despite enhanced
recycling provision. Inter-district differences will
only be fully eroded through additional
promotional and education campaigns. Those
campaigns are expected to produce relatively
higher effects on the poorly performing districts.
With comparable infrastructures and strong
successful promotions, all districts should reach
much the same endpoints. Whilst the study
concentrated solely on the north west and
south west regions of England, the parallels
between those regions would indicate that the
same trends might also be expected for other
regions of the UK as well.
The maximum captures possible based on the
most advantageous set of circumstances are
predicted to be of order:
80 - 82% (Newspapers and PAMs)
80 - 82% (Glass bottles)
75 - 78% (Glass jars, aluminium cans, steel
cans)
71 - 73% (Cardboard)
71 - 73% (Plastic beverage bottles)
These captures would be achieved by multi-
material kerbside collections of all the above
materials and the high levels of successful
promotional and education campaigns.
In these scenarios 85-87% of households
would be engaging in some kind of recycling
activity, with 75-80% of them using the kerbside
schemes.
Understanding Recycling Behaviour 2
All the above statistics exclude kerbside
collections from flats. With comparable
doorstep collections from flats, all statistics
would be expected to rise by a further 4 - 6%.
With free kerbside green waste collections,
garden waste recoveries are expected to be of
order 65% with normal levels of promotion but
could be increased to around 85% with strong
promotion.
The study has not yet discriminated how
different modes of kerbside recycling (bag or
box or bin, collection frequency, number of
household separations required, survival bags,
reducing dustbin size or the frequency of its
collection etc.) affect the capture rates or
participations. However the study had
delineated worst case scenarios for each type
of collection and maximum capture (best case)
scenarios assuming significant promotion. More
convenient collection regimes (e.g. comingled
collections, boxes versus bags, weekly versus
fortnightly or monthly collections, constraints on
residual waste volumes) may be hypothesised
to lead to starting points higher up the scale.
However as promotions are introduced and
intensified, regime-specific differences should
be eroded and all regimes should converge on
very similar endpoints. It is probably less
relevant then to ask what type of scheme to
deploy as to ask what type of promotion to
deploy. Irrespective of whatever type of
scheme is deployed, national goals are unlikely
to be met by the north west region unless
effective promotions are deployed as well. The
national waste awareness campaigns must be
charged with a high responsibility for achieving
those promotions. The study has shown that
the highest capture levels will not be met simply
by instilling awareness, increasing knowledge
and promoting a stronger pro-recycling ethos.
Ultimately, the limiting factor will be how much
extra effort and personal cost must be put into
ensuring that everything is recycled that can be
recycled, even under the most adverse
conditions.
The actual recycling programme designs and
their [staged] implementations must be
determined locally by each district authority.
The following sections of this report will now
examine the practical outcomes of some of
those decisions.
30

0
3. Does the Programme Design Matter?
3.1 Introduction
The US studies discussed in section 1.5.2
provided some insights into the relative effects
of specific design variables. As far as the
author is aware, comparable analyses have not
been reported previously for the UK. Those
analyses are now undertaken. The analyses
were based on the local authority statistics
compiled by CIPFA using the 1999/2000
actuals (CIPFA, 2001).
The performance indicator selected for the
comparative study was the weight of kerbside
recycled material per household serviced by a
kerbside collection. Firstly, all recyclate
recovered at kerbside is considered, though the
bulk of the statistical comparisons concentrate
on the paper and card fractions of that
recyclate. The specific programme variables
available from the CIPFA statistics are
recycling container type: {sack, one box,
multiple boxes, wheeled bin, split wheeled bin,
no-container}, collection frequency, and
recyclable materials separately identified, or co-
mingled. The study looked at the following
comparisons:
Total kerbside Yield
 Paper-only vs. Multi-material (separately
identified) vs. Multi-material (co-mingled).
3.2 Results
3.2.1 Total Kerbside Yield
Firstly it is noted that both the paper-only and
multi-material data sets were positively skewed,
that is, the modal value was substantially less
than the mean and there was a long tail at the
high weight end. The co-mingled weight
distribution, whilst also positively skewed, was
much closer to being normally distributed.
Table 3.1 provides summary descriptive
statistics where it can be seen that, as
expected, more weight is collected in a multi-
material collection than through a paper-only
collection, with the most weight being when the
31
Understanding Recycling Behaviour 2
Paper and Card Yield Only
 Percentage of borough covered by
scheme;
 Weekly (or more frequent) vs. fortnightly
(or less frequent) collections;
 Paper only vs. Multi-material;
 Bag vs. Box;
 Paper only (bag) vs. Paper only (box) vs.
Multi-material (bag) vs. Multi-material
(box).
Whilst the CIPFA 1999/00 statistics received
returns from over 340 authorities, several of
those returns did not operate kerbside
collections, or were incomplete in some or all of
the relevant columns, or appeared to be
potentially inaccurate in the data reported. In
the analyses, authorities returning >6 Kg
recyclate/hh/w or <0.1 Kg recyclate/hh/w were
considered to be extreme outliers and possibly
suspect, and were excluded from the analyses.
It was also noted that very few of the remaining
eligible schemes operated wheeled bin dry
recyclable collections or offered no containment
method for the recyclate. The sample bases for
such schemes would be too small for
meaningful statistical analyses, and those
schemes, too, were excluded from the
analyses.
multi-materials are co-mingled rather than
separately identified.
However, as per the US studies there were
substantial overlaps between groups with high
and low performing authorities spread across
all three groups (figure 3.1).
Statistical analyses showed that the paper-only
collection yields were significantly lower than
multi-material collection yields; however there
was no statistical evidence that co-mingled
yields were significantly greater than multi-
material yields.
 Understanding Recycling Behaviour 2

Table 3.1 Total Kerbside Yield (Kg/hh/w)

N Mean Median StdDev Min Max

Paper only 75 1.11 0.92 0.71 0.18 4.41

Multi-material 85 1.56 1.32 0.83 0.37 4.57
Co-mingled 24 1.67 1.44 0.81 0.54 3.43

Figure 3.1 Total Kerbside Yield (Kg/hh/w)

5

3
Kg/hh/w

0
Comingled Multi-material Paper-only
Scheme type

Note: The variations in yield are presented in this report in the form of as box and whisker plots. The box
delineates the inter-quartile range of the observations, with the horizontal line across the box showing the median
value. The whiskers extend out to the lowest and highest observations that are still inside the region defined by
the following limits:
Lower Limit: Q1 - 1.5 (Q3 - Q1)
Upper Limit: Q3 + 1.5 (Q3 - Q1)
where Q1 and Q3 are the first and third quartiles respectively.
Outlying points beyond those limits are then plotted individually.

3.2.2 Paper and Card Yields
Not too much more can be drawn from the
analyses of total yields, as there is probably too
much disparity in the types and ranges of
actual materials collected across the multi-
material and co-mingled categories for the
analyses to be extended meaningfully to any
finer detail. A more comparable yield statistic
may be given by the paper and card tonnages
recovered. As almost every scheme collects
paper and card, analysis of that fraction would
still retain a large sample base. The 24 co-
mingled collections must however be excluded
from the analyses as paper and card yields are
not separately identified in their statistics.
However, it must still be recognised that the
ranges of paper and card collected by different
schemes can be very different. Nearly all
schemes will collect newspapers, pamphlets
and magazines (and normally make up the bulk
of the recovered weight). Fewer schemes tend
to collect other white papers, or card
packaging. This needs to be borne in mind
when interpreting the data.
Firstly, a test was undertaken to see if the
proportion of households in the borough
serviced by the scheme was related to scheme
performance. The hypothesis was small pilot
programmes may behave significantly better
than borough-wide schemes. The results are
shown in figure 3.2. These results exclude five
extreme outliers (>5 Kg/hh/w) all of which were
associated with the low coverage fraction.
The results show that with the exception of a
small number of outliers, schemes that service
high proportions of a borough can yield just as
much paper and card per serviced household

32
 Understanding Recycling Behaviour 2

as that recovered in the more selective
schemes. The differences in yield between
levels of coverage were not statistically
significant.
The hypothesis that frequency of collection may
affect yield was tested in figures 3.3 and 3.4.
Figure 3.3 records the results for all schemes
whilst figure 3.4 records the result for multi-
material schemes only. Only five of the
seventy-nine paper-only collections were
carried out on a weekly basis and just three
multi-material schemes collected on a monthly
basis.
Frequency of collection was not a statistically
significant determinant of yield.

Figure 3.2 Paper and Card Yields by Scheme Coverage

2
Kg/hh/w

0
<25% 25-50% 50-75% 75-100%
Coverage

Figure 3.3 Paper and Card Yields by Collection Frequency – All Schemes

2.5

2.0

1.5
Kg/hh/w

1.0

0.5

0.0
>2 weekly fortnightly weekly
Frequency

33
 Understanding Recycling Behaviour 2

Figure 3.4 Paper and Card Yields by Collection Frequency – Multi-material Schemes
only.

2.5

2.0

1.5
Kg/hh/w

1.0

0.5

0.0
fortnightly weekly
Frequency

Whether the paper was collected on its own or
alongside other materials appeared to have
some effect on weights of material collected,
with around 15% more paper and card being
collected in a multi-material collection than in a
paper and card only collection (table 3.2, figure
3.5). These effects are statistically significant..
The type of container used appeared also to
affect the weights of paper and card collected.
Collection of kerbside recyclables through bins
remained relatively uncommon with only four
councils using this type of collection container,
although the average weight collected through
bins was higher than that for both bags and box
collections at 1.3 kg/hh/w. Because of the small
sample sizes for the other containment
methods they cannot be analysed here.
On average, box collections were yielding
some 15-20% more paper and card per
household served than were bag collections
(table 3.3, figure 3.6). The effect was
statistically significant.
Overall the results show that, in general, higher
yields of paper and card will be achieved by
those kerbside schemes that collect more than
just paper and card, and use boxes [or bins]
rather than bags. However there is also a high
correlation between these two factors, with
most multi-material schemes using boxes
(65%) and most paper only schemes using
bags (73%). Is it possible therefore that only
one of these factors is significant? To test this
hypothesis we then looked at the four groups
separately: i.e. multi & bag, multi & box, paper
& bag and paper & box. Results are shown in
figure 3.7.
Qualitatively, the two box collections appear to
follow similar patterns with relatively high
means and large standard deviations. Bag
schemes appear to have tighter distributions
with lower average weights. It could therefore
be hypothesised that the box may be the
dominant part of the combination contributing to
the highest yields. However, the only significant
statistical difference was between multi-
material box schemes and paper-only bag
schemes.

Table 3.2 Weights of Paper and Card Collected according to Collection Type (Kg/hh/w)
Scheme N Mean Median StDev Min Max

Paper-only 85 0.95 0.84 0.48 0.17 2.30

Multi-material 80 1.09 1.03 0.51 0.22 2.41

34
 Understanding Recycling Behaviour 2

Figure 3.5 Paper and Card Yields by Scheme Type

2.5

2.0

1.5
Kg/hh/w

1.0

0.5

0.0
multi-material(paper) paper only
Scheme type

Table 3.3 Weights of Paper and Card Collected according to Container Type (Kg/hh/w)
Container N Mean Median StDev Min Max

Bag 86 0.91 0.87 0.36 0.17 2.21

Bin 4 1.31 - - 1.05 1.69
Box 74 1.14 1.01 0.59 0.20 2.41

Figure 3.6 Paper and Card Yields by Container Type

2.5

2.0

1.5
Kg/hh/w

1.0

0.5

0.0
bag bin box
Container

35

Figure 3.7 Paper and Card Yields by Scheme Combination
2.5
2.0
1.5
Kg/hh/w
1.0
0.5
0.0
Multi & bag Multi & box
Combination
3.3 Conclusions and Further Evidence
The analyses of published English and Welsh
recycling statistics indicate that, on average,
kerbside yields of paper and card are higher in
(a) schemes that collect other materials
alongside that paper and card, and (b)
schemes that use collection boxes (or bins)
rather than bags or sacks. The average
difference in yield may be as much as 20%
between multi-material box collections and
paper-only bag collections. Collection
frequency does not significantly affect the yield.
However, there remains a strong caveat to
these conclusions. Paper and card yields may
also be strongly dependent on the range of
papers and cards targeted by individual
schemes. By their very nature, multi-material
collections may be more likely to target card
and mixed paper than would paper-only
collections. It could also be that box collections
are more likely to take mixed paper and card
compared to bag collections.
Subsequently, it proved possible to undertake
some further analyses using a data set from
Hampshire (MEL, 1999) supplied courtesy of
Project Integra and a data set elicited from
authorities in north west England. These data
provided a more detailed breakdown by
material category and, as such, allowed
individual paper fractions to be analysed
separately. The results shown in figures 3.8 a-d
Understanding Recycling Behaviour 2
Paper & bag Paper & box
all report newspaper yields. Because of the
limited number of samples in some categories,
the graphs should be viewed simply as being
indicative of possible trends.
The results generally follow the same kind of
trends that were revealed in the analyses of the
CIPFA statistics. No discernible trend was seen
for newspaper yield with respect to collection
frequency. Box or bin collections tended to
collect more newspaper per household than did
bag collections or collections where no
container was supplied. Acorn groups A
(principally A1) and D (all D9) gave better
yields of newspaper than did groups B and E/F,
though it was noted that the majority of B
households were in areas operating bag or „no-
container‟ collections. An expanded breakdown
by Acorn/ container combinations showed that
container type appeared to be dominant over
demographics in terms of newspaper yield.
Most design differences were impossible to test
for in the north west England sample. All but
one was a paper only scheme and only two
were not being served on a fortnightly basis.
However, it was possible to investigate
container type. Unlike the results from both the
CIPFA and Hampshire analyses, the type of
container appeared to have no statistically
significant effect for the north west of England
sample (figure 3.9). However, it is noted that
36
 Understanding Recycling Behaviour 2

the 16 bag schemes were giving much more

consistent yields than the 8 box schemes
operating there.

Figure 3.8 a-d Results from Hampshire

2.5
Kg/hh/w Newspaper

1.5

0.5

Bag Box None WB (Wheeled Bin)

Container

2.5
Kg/hh/w Newspaper

1.5

0.5

F W
Collection frequency

2.5
Kg/hh/w Newspaper

1.5

0.5

A B D EF
Acorn

37
 Understanding Recycling Behaviour 2

Kg/hh/w Newspaper 2.5

1.5

0.5

F/A F/B F/D F/EF R/A R/B R/D R/EF

Container/Acorn

Key: F=bag or no container, R=box or bin

Figure 3.9 Results from north west England

1.7

1.2
kg/hh/w

0.7

0.2
bag box
container

38

4. Implementing Change: A Case Study
Chesterfield and North East Derbyshire
4.1 Background
The national analysis presented in section 2
predicted the practical limits to recycling for the
UK, and provided some guidance on the staged
enhancements that will be needed to reach
those endpoints. However, it did not go into the
finer details of how individual local authorities
might best progress towards their own
endpoints, given their very different local
circumstances. Such solutions need to be
worked out locally. This section of the report
presents a comprehensive analysis of the
progress achieved in 2002 by two local
authorities: Chesterfield Borough Council and
North East Derbyshire District Council.
4.1.1 Recycling Provision
Kerbside recycling collections started in
Chesterfield in 1995 and in North East
Derbyshire in 1998 through the introduction of
paper-only (blue bag) collections servicing
around 41,000 households in Chesterfield and
around 9,000 households in North East
Derbyshire. A further 23,000 household were
subsequently added to the service in North
East Derbyshire in September 2001. From
November 2001, both authorities commenced a
rolling programme to convert those paper-only
rounds to multi-material, collecting paper,
glass, mixed cans, and textiles. Householders
were provided with blue boxes for their mixed
dry recyclables, with the paper being collected
in a separate bag. The first conversion in
November 2001 put 5,700 households in
Chesterfield and just over 5,000 households in
North East Derbyshire onto the new scheme. In
March 2002, another 5,700 households were
converted in Chesterfield and another 4,700
households in N.E. Derbyshire. The third major
conversion occurred in September 2002 adding
another 6,000 to the scheme in Chesterfield
and over 12,000 more in N.E. Derbyshire.
All the new multi-material rounds operated
fortnightly collections. The round designations
relating to the three staged expansions are 51 -
55, 56 - 60, 61 - 65 (in Chesterfield), and 101 -
105, 106 - 110, 111 - 120 respectively (in North
East Derbyshire).
39
Understanding Recycling Behaviour 2
in
In addition to converting its paper-only
collections to multi-material, Chesterfield
Borough Council is also now operating green
waste collections in some areas of the
Borough, using dual green and black bin
collections (for residual waste) The green and
black bins are collected on alternate weeks.
Those rounds map directly onto multi-material
rounds 56 - 61 and 63 - 65 respectively. The
green waste collections 61 - 65 commenced at
the same time as their corresponding multi-
material collections whereas rounds 56 - 60
were converted to multi-material approximately
6 months after the green bin collections had
been introduced.
4.1.2 Data Considerations
The performance indicator used in the
programme evaluation is the weight of material
collected per household served, normally
expressed in terms of Kg/household/week.
Computing these statistics relies on knowing:
(i) the weight recovered per collection round,
(ii) the number of houses serviced in that
collection round, and
(iii) the time period spanned by the recorded
information.
Some of those figures were not always certain.
The issues concerning the inherent
uncertainties in performance monitoring are
discussed in greater depth later, in section
11.2.3 of this report. The yield data was
collected on a round by round basis by the
scheme operator, with all figures broken down
by material category.
One important aspect of the research was to
quantify the effects of the change. This proved
to be difficult as the boundaries of the old
paper-only rounds rarely corresponded to the
boundaries of the new multi-material rounds.
The new multi-material rounds either comprise
a small fraction (typically 1/3 to 1/2) of a paper-
only round, or are constructed from smaller
parts of two or three of the old rounds. A
number of new properties have been added as
well. Whilst the contributing proportion of

households from each of the old rounds is
known, the contributing performance of the part
rounds cannot be distinguished separately.
This source of uncertainly must be borne in
4.2 Dry Recyclate Yields
Table 4.1 lists the total yield for the first two
staged conversions in Chesterfield and North
East Derbyshire. It is given as year 2002
averages. In the table results are compared
with the old paper-only yields across the two
districts for the previous year and with the
national averages given in the CIPFA statistics
(CIPFA, 2001).
The total material yielded nationally from multi-
material collections ranged from 0.37 kg per
household served per week up to 4.57 Kg/hh/w
(see section 3). The N.E. Derbyshire statistics
were well above the 1999-2000 national
average, and the Chesterfield performance also
sat comfortably amongst the top 50% of
authorities. (However, it must be recognised
that relatively few authorities in the CIPFA
compilation were collecting glass at kerbside at
the time those data were compiled). The weight
of the paper collected through the multi-
material collections in Chesterfield and N.E.
Derbyshire were also above average, as were
the yields from the previous paper-only
collections in N.E. Derbyshire. The previous
paper-only recoveries in Chesterfield, in
contrast, were somewhat below the national
average. The change from that paper-only
collection to multi-material appears to have
raised Chesterfield significantly up the paper-
recovery league table.
Tables 4.2 a-b provide estimates of the local
increases in yield across the conversions.
These results show that a general 20-40%
increase in the paper collected had occurred in
N.E. Derbyshire, and a significantly larger
increase (maybe a doubling in some areas) had
occurred in Chesterfield. The figures for the
third stage of conversion (rounds 61 - 65 and
111 - 120) are included here, though it should
be noted here that they are based on just two
months collection statistics. Actual figures may
be subject to revision once longer time series
are available. The data from the first two
collections of any of the new rounds have not
been used in the calculations, as those figures
can be untypical of later stable behaviours.
Overall, whilst some of the rounds appear to
have rather larger changes than others, part of
those differences are merely artefacts of
Understanding Recycling Behaviour 2
mind in the analyses when assessing the
effects of change at collection round levels.
matching old to new rounds (particularly where
the old rounds were highly [spatially]
heterogeneous in terms of performance). It is
particularly noted that rounds 56-65 on average
returned larger rises than rounds 51-55 in
Chesterfield. The rounds 56-65 were in areas
where alternate week green waste and residual
waste collections had been introduced. A more
detailed analysis of the impact of these
changes is given later in section 4.3.
The overall increases in yield from paper only
to all materials were by factors of 1.8 in N.E.
Derbyshire and 2.6 in Chesterfield (stages 1
and 2). However, the Chesterfield stage 3
conversions recorded a massive 4-fold
increase in total yield from 0.7 Kg/hh/w (paper
only) to 2.9 Kg/hh/w.
The measured outcomes show good qualitative
agreement with the national differences
expected between the respective types and
means of collection, noted previously in the
analyses of the CIPFA statistics (section 3).
The scales of the yield change from paper only
to multi-material also show good matches with
the regional model predictions for England
moving from a current scenario to a 100%
multi-material scenario (section 2, figure 2.1).
The change from paper-only collections to
multi-material collections still maintained the
historical differential in recycling performance
between the two districts except for the last
stage of conversions in Chesterfield where that
differential was eroded). That is, the historically
poorer-performing areas still remained the
poorer performers in the new scheme. These
results again confirm the features already
predicted to occur at regional level (Section 2,
figure 2.1).
Tables 4.3 a-b break down the multi-material
yields for Chesterfield and N.E.Derbyshire by
individual materials. Overall, the recovered
material in the new scheme was found to
comprise around 63-67% paper, 28-29% glass,
and 4% and 3% cans and textiles respectively
(figure 4.1).
40
 Understanding Recycling Behaviour 2

Table 4.1 Kerbside Yields (Kg/hh/w)

N Mean StdDev Min Max
Multi-material (Total)
National (CIPFA) 85 1.56 0.83 0.37 4.57
N.E. Derbyshire 10 2.38 0.28 1.87 2.74
Chesterfield 10 1.79 0.38 1.16 2.67

Multi-material (paper
fraction)
National (CIPFA) 80 1.09 0.51 0.22 2.41
N.E. Derbyshire 10 1.49 0.20 1.09 1.78
Chesterfield 10 1.14 0.21 0.79 1.63

Paper only
National (CIPFA 1999/00) 85 0.95 0.48 0.17 2.30
N.E. Derbyshire 4 1.31 0.15 0.99 1.54
Chesterfield 20 0.68 0.25 0.38 1.20

Table 4.2a Changes in Paper Recovery with Conversion to Multi-material (N.E. Derbyshire)
Round Kg/hh/w Paper-only Multiplier Kg/hh/ Paper-only Multiplier
equivalent Calander equivalent
month

101 1.31 1.38 0.95 5.37 6.06 0.89

102 1.66 1.40 1.18 6.90 6.16 1.12
103 1.42 1.30 1.09 5.97 5.64 1.06
104 1.58 1.25 1.26 6.62 5.42 1.22
105 1.17 1.01 1.16 4.93 4.35 1.13
106 1.55 1.43 1.08 6.60 6.21 1.06
107 1.60 1.51 1.06 6.83 6.56 1.04
108 1.58 1.17 1.35 6.49 5.09 1.28
109 1.78 7.59 3.54 2.14
110 1.24 5.56 2.90 1.92

101-110 1.48 1.31 1.14 6.29 5.19 1.29

111-120 5.45 3.89 1.41

Table 4.2b Changes in Paper Recovery with Conversion to Multi-material (Chesterfield)

Round Kg/hh/w Paper-only Multiplier
equivalent

51 1.13 1.09 1.03

52 1.11 0.73 1.52
53 0.79 0.61 1.31
54 1.13 0.51 2.21
55 1.23 0.49 2.02
56 1.63 1.15 1.42
57 1.09 0.59 1.84
58 0.95 0.60 1.58
59 1.14 0.63 1.80
60 1.25 0.66 1.90

51-60 1.14 0.71 1.67

61-65 1.82 0.75 2.54

41
 Understanding Recycling Behaviour 2

Table 4.3a North East Derbyshire Multi-material Yields (Kg/hh/w)

Round Paper Glass Cans Textiles All

101 1.31 0.66 0.108 0.054 2.13

102 1.66 0.71 0.090 0.048 2.50
103 1.42 0.57 0.089 0.043 2.12
104 1.58 0.89 0.107 0.077 2.66
105 1.17 0.65 0.088 0.024 1.93
106 1.55 0.61 0.092 0.047 2.31
107 1.60 0.68 0.123 0.110 2.02
108 1.58 0.67 0.105 0.073 2.42
109 1.78 0.70 0.144 0.118 2.75
110 1.24 0.54 0.090 0.083 1.95

M101-M110 1.48 0.67 0.103 0.069 2.33

M111-M120 1.39 0.54 0.150 0.070 2.14

Table 4.3b Chesterfield Multi-material Yields (Kg/hh/w)

Round Paper Glass Cans Textiles All

M051 1.13 0.46 0.11 0.02 1.72

M052 1.11 0.60 0.09 0.03 1.83
M053 0.79 0.29 0.06 0.02 1.16
M054 1.13 0.40 0.07 0.02 1.62
M055 1.23 0.41 0.08 0.03 1.76
M056 1.63 0.82 0.13 0.09 2.67
M057 1.09 0.50 0.14 0.06 1.78
M058 0.95 0.47 0.12 0.06 1.59
M059 1.14 0.45 0.10 0.06 1.75
M060 1.25 0.59 0.17 0.06 2.07

M051-M060 1.14 0.50 0.11 0.05 1.79

M061-M065 1.82 0.79 0.22 0.10 2.93

Figure 4.1 Average Kerbside Yields before and after Conversion

2.5
Kg/ household/ week

1.5 Textiles
Cans
1
Glass
0.5 Paper
0
Before After Before After
Chesterfield N.E.Derbyshire

42

4.3 Effect of Alternate Green Waste / Residual Waste Collection
The differences between the last two
conversions in Chesterfield were especially
significant. Both conversions were associated
with areas where residual waste management
had been put on a fortnightly collection cycle,
alternating with kerbside green waste
collections. The essential difference was that
the third multi-material conversion was timed to
coincide with the introduction of the alternating
green/ residual waste collections, whereas the
second conversion happened six months after
the residual waste collection was changed. The
differences in recyclate yield between the two
regimes are shown in figure 4.2. Whilst the
prior introduction of the alternate week
collection did end up with higher dry recyclate
yields compared with the areas with „normal‟
(weekly) residual collections, the improvement
was relatively small compared with that
achieved in the third conversion. Timing would
therefore appear to be the critical factor.
The explanation for the observed differences is
thought to be as follows: The imposition of a
reduced frequency of residual waste collection
effectively reduces the amount of „dustbin
space‟ available for that waste. The intended
outcome is that householders will have to seek
alternatives, preferably recycling, for disposing
of their excesses. For some householders, that
Figure 4.2 Average Kerbside Yields after the Chesterfield Conversions
3.5
Kg/ household/ week
3
2.5
2 Cans
1.5
1 Paper
0.5
0
Normal Alternate
(prior)
43
Understanding Recycling Behaviour 2
would necessitate a significant life-style
change. Such events will significantly raise the
saliency of the waste management issue to
householders (in terms of potential personal
costs). Providing the solution (i.e. the multi-
material recycling box) at the same time, leads
householders into preferentially adopting that
solution. In contrast, imposing the lifestyle
change prior to providing the solution effectively
renders the solution almost irrelevant when it
eventually comes. Lifestyle adjustments have
already been made.
However, on a note of caution, it must be noted
that whilst all the evidence strongly points to
„timing‟ being the critical factor, the evidence
presented so far does not prove it. The
difference could be due to some further
unconsidered factor. The high-performing
rounds in the third conversion were all in areas
of „favourable‟ demographics, i.e. containing
relatively high proportions of affluent housing
stock. So could demographics be the real key?
The answer is, whilst demographics does play
some part, the timing of the introduction of
alternate week collections still remains a
significant factor in their success. The analyses
and arguments that support this answer are
presented later in section 4.5.3 of the report.
Textiles
Glass
Alternate
(same
time)

4.4 Effect on Bring Site Recoveries
4.4.1 Methodology
It is difficult to get a reliable measure of the
impact of the kerbside scheme on bring site
yields. Firstly, the bring sites do not map
uniquely onto kerbside collection rounds. The
sites can have a much wider catchment than
the immediate area in which they are located.
Also residents may not choose to use their
nearest bring site and may choose to travel to a
more distant site if that site is more convenient
to their life style. Some residents may also
patronise sites belonging to other districts.
Secondly, bring site records relate only to when
the banks are emptied, which can be quite
infrequently and irregular in timing. To achieve
a reasonably consistent time series of bring site
yields, it becomes necessary to aggregate
those yields over a relatively long time period.
A 4-month aggregation period was chosen for
the analyses presented here.
The bring site tonnage data supplied by North
East Derbyshire was in the units of tonnes per
calendar month and covered all bring sites in
the district. The data supplied by Chesterfield
related to bring sites specifically located within
the kerbside collection areas. The yields from
those sites were recorded as the estimated
fullness of the recycling banks at their time of
emptying - on a scale of eighths full. Volume to
weight conversion factors were also provided.
4.4.2 Results
The yields from the two paper banks in
Dronfield (in N.E. Derbyshire) are shown in
figure 4.3. The time series plotted extends from
November 1997 to October 2002. The solid
vertical line shows when the first kerbside
paper collections were introduced in that town,
with the dotted vertical lines showing the
timings of the two phases of the conversion of
those rounds to multi-material. It is evident from
the diagram that the initial kerbside roll-out had
a very much more significant effect on paper
Table 4.5 Change in Bring Site Recoveries on Conversion to Multi-material Collections
Paper Glass
Dronfield -25% -50%
Chesterfield -30% -25%
Note: All these figures are subject to considerable uncertainties in estimation
Understanding Recycling Behaviour 2
bank tonnages than did its subsequent
conversion to multi-material.
The effects of the change to multi-material on
the can and textile bank yields are shown in
figures 4.4 and 4.5 respectively. For both of
these materials there appears to be only a very
minor effect, if any effect is discernible at all.
Figure 4.6 shows the impact on the glass bank
recoveries. Quite a significant effect is seen in
the Dronfield glass bank yields after multi-
material kerbside collections were introduced
there, with a significant drop in glass yields
being recorded. This can be contrasted with the
steady increase in glass yields at the banks
elsewhere within the district.
In Chesterfield, the paper banks and the glass
banks within the kerbside collection areas both
showed a marked loss of yield following the first
two multi-material conversions in November
1999 and March 2002 respectively (figure 4.7).
There was no associated discernable impact
evident on the can bank yields.
Overall, over both communities, the average
yields of the bring sites fell by around 40-44%
with the introduction of the kerbside paper only
collections (table 4.4). With the subsequent
change to multi-material kerbside collections
the bring site yields fell yet further by around
another 25 percentage points (paper), and by
25-50% (glass). The can and textile bank
recoveries, in contrast, remained fairly stable
across the change, falling by 10% at most
(table 4.5).
Table 4.4 Change in Bring Site Recoveries
on Introduction of Paper-only Kerbside
Collections
Location Change
Dronfield -40%
Other in N.E. -44%
Derbyshire
Cans Textiles
-10% +5%
-10% N/a
44
 Understanding Recycling Behaviour 2

Figure 4.3 Paper Bank Yields in Dronfield, N.E. Derbyshire

5
Tonnes/month

4
DRONFIELD, SPORTS
CENTRE
3
DRONFIELD, GREENDALE
SHOPS
2

0
JASO

JASO

JASO

JASO

JASO
NDJF

NDJF

NDJF

NDJF

NDJF
MAMJ

MAMJ

MAMJ

MAMJ

MAMJ

Note: The vertical lines show the timings of the system changes, the first being the introduction of paper-only
kerbside collections and the later two being the first and second phases of the multi-material conversions.

Figure 4.4 Can Bank Yields in N.E. Derbyshire

350

300

250
Kg/month

CLAY CROSS
200
DRONFIELD
150
KILLAMARSH
100

50

0
JASO

JASO

JASO
NDJF

NDJF

NDJF
MAMJ

MAMJ

MAMJ

Figure 4.5 Textile Bank Yields in N.E. Derbyshire - Council collections only

1600
Ashover
1400
1200 Clay Cross
Kg/month

1000
Dron Sports Centre
800
600 Greendale Shops
400
Eckington Baths
200
0 Killamarsh
JASO

JASO

JASO
NDJF

NDJF

NDJF
MAMJ

MAMJ

MAMJ

Parkside

45
 Understanding Recycling Behaviour 2

Figure 4.6 Glass Bank Yields in N.E. Derbyshire

5 HOLYMOORSIDE

4.5
ASHOVER
4

3.5 DRONFIELD, GREENDALE

SHOPS
Tonnes/month

3
ECKINGTON
2.5

2
TUPTON, Green Lane

1.5

1 CLAY CROSS

0.5
WINGERWORTH (opened Sept
0 1998) Allendale
JASO

JASO

JASO
NDJF

NDJF

NDJF
MAMJ

MAMJ

MAMJ

KILLAMARSH, PARKSIDE
(adopted 01/02/00)

Figure 4.7 Bring Site Recoveries in Chesterfield (Nov 1999-Oct 2002)

1000

800
Eighths of bank

600 paper
glass
400 cans

200

0
JASO

JASO

JASO
NDJF

NDJF

NDJF
MAMJ

MAMJ

MAMJ


Average weights of a full 1,100 litre bank: Paper 250kg, Glass 350 Kg, Cans 35Kg (source: manufacturer)

46

4.5 Demographic Dependences
The prime questions that the research set out
to answer were:
 “What were the [demographic] profiles of
the old paper-only recyclers? and “are the
new recyclers any different?”. Essentially
“Who continues to recycle?” and “What
sort of new people may have joined the
scheme?”
 “How far can the observed performance
differences be explained by demographic
differences?” and perhaps specifically
“Can demographic differences offer a
possible alternative explanation for the
high success of the third roll out in
Chesterfield?”
4.5.1 Methodology
Normally data of this nature are gathered
through questionnaire surveys of householders.
In the current study such analyses were not
resourced, so inferences had to be drawn from
data that was already available. The
demographic profiles used in this study were
constructed from census statistics and from the
Acorn market research indices see Appendix A
for details). Here, it must be noted that there
are considerable problems in mapping census
enumeration districts onto individual collection
rounds (as the boundaries do not coincide).
The necessary demographic assignments have
been made by assigning demographic profiles
pro-rata according to the numbers of
households shared between each unit. The
statistics used are all in the form: „percentage
of households serviced by a round that are
from a given demographic grouping‟. The
classes of demographic descriptor used for
those statistics were: housing type, family life
stage, household size, car ownership, tenure,
and Acorn index. As each collection round
provided a mix of groupings within each of
those classes, it was necessary to deconvolute
the individual effects of each group from the
round total. This was done using the statistical
technique of multi-linear regression.
The regression equations used for modelling
the round collection data were all of the form:
Kg/hh/w(a) = c1  x(a) + c2  y(a) + c3  z(a) +
where x, y, z etc. are percentages of
households in a given group and c1, c2, c3 are
the scaling coefficients obtained through the
47
Understanding Recycling Behaviour 2
regression of the equation onto the measured
round yield data, Kg/hh/w(n), for rounds n = a,
b, c, …..
It must be strongly stressed here that the
regression coefficients c1, c2, … should not be
interpreted in terms of the weights of material
expected from each group. They are simply
scaling factors that apply when all the other
terms in the equation are present.
4.5.2 Results
On including all demographic descriptors in the
regression equations, it was found that most of
those descriptors were not statistically
significant to the regression. In order to identify
and retain only statistically significant terms,
stepwise regression was performed. In the
current analyses a combined forward and
backward stepwise regression technique was
used. This technique starts off by identifying the
single most significant factor in the prediction,
then tries to add in the „best‟ second factor that
can improve the overall regression. The new
term is the most significant available given the
other terms already included in the model. If
that addition is successful, then the „best‟ third
factor is found, and so on. However, if any
previously included term(s) lose their
significance when new terms are added, then
those terms are dropped from the model.
The regressions were applied separately to
Chesterfield and N.E.Derbyshire and to the
combined rounds from both districts. The initial
combined results revealed that the regression
generally under-predicted the North East
Derbyshire yields and generally over-estimated
those in Chesterfield. In order to correct for this
apparent „District‟ factor, a formal district-
dependent term was then added into the
regression. It was arbitrarily defined as the
percentage of households within the district that
are located within Chesterfield. It assumes the
values: 100% (households in Chesterfield) and
0% (households in N.E. Derbyshire). The
individual regression equations for Chesterfield
and N.E. Derbyshire and the modified
regression equations for Chesterfield/ N.E.
Derbyshire combined are shown, for the paper-
only collections, in box 4.1.for the Acorn
classification and for all census variables
together. Regressions for individual class
descriptors (e.g. housing types) were also
undertaken but are not reproduced here.
 Understanding Recycling Behaviour 2

Box 4.1 Stepwise Regression of Paper Only Collection Yields against Demographics

N.E. DERBYSHIRE
Acorn
Kg/hh/w (paper) = 0.0145  A + 0.0141  F + 0.0120  B + 0.0110  D

Census
Kg/hh/w (paper) = 0.0430  More mature – 0.092  Private sector rentals

CHESTERFIELD
Acorn
Kg/hh/w (paper) = 0.0140  A + 0.0132  B + 0.0101  D

Census
Kg/hh/w (paper) = 0.0780  Families/older children + 0.0085  Detached

COMBINED
Acorn

Kg/hh/w (paper) = 0.0140  A + 0.0143  B + 0.0119  D + 0.0089  F

- 0.00249  In_Chesterfield
Census
Kg/hh/w (paper) = 0.0398  More mature – 0.0096  0_car households
- 0.00154  In_Chesterfield

Acorn A = Thriving, B = Expanding, C = Rising, D = Settling, E = Aspiring, F = Striving

The magnitudes and signs of the derived
regression coefficients, by and large conform to
traditional wisdom that rounds with more
affluent housing stock and the more older
residents were the more prolific recyclers; the
regression coefficients being larger and more
positive for these groups. However it is noted
that different factors take on different
significances in the two districts. Both the Acorn
and census-based formulations also reveal that
the systematic difference between districts (the
„in-Chesterfield‟ factor) to be statistically
significant.
However, in interpreting these equations, it
should be noted that the available census
variables are not necessarily independent of
each other. For example, flat dwellers are more
likely to be young adults or retired residents
rather than large families with children; car
ownership is likely to be higher amongst
detached households than amongst terraced
households or flats, and so on.
Boxes 4.2 and 4.3 show the corresponding
regressions for the multi-material collections.
The results for Chesterfield and North East
Derbyshire are shown separately for total yield
and for each individual material.
It is noticeable that the derived regression
equations differ for each commodity recovered,
and again for each district. There are
indications, particularly in the Acorn-based
equations, that the least affluent group F may
be having more influence on yields in the multi-
material collections than it had in the paper-
only collections. The influence of Acorn F
households appears strongest with respect to
the textile yields in both districts. Acorn E
households appear to consistently exert the
highest influence on can yields but have little
influence on the yields of the other
commodities. It may be tentatively concluded
that there could have been some erosion of
demographic bias with the conversion, and
perhaps more strongly stated that the more
prolific can and textile recyclers might belong to
different sectors of the population than the high
recyclers of paper and glass.
The developed models, listed above, were able
to explain around 50-60% of the observed
round-by-round variations in paper yields and
glass yields, but less than 40% of the observed
variations in the recovered weights of cans and
textiles. The remaining variations could not be
explained by any of the chosen demographic
factors. Other causal factors may need to be
invoked to explain those residual variations.

48
 Understanding Recycling Behaviour 2

Box 4.2 Stepwise Regression Equations for N.E. Derbyshire Multi-material Yields

Acorn

Kg/hh/w (all) = 0.0360  A + 0.0225  D + 0.0218  F + 0.0180  B + 0.0169  E

Kg/hh/w (paper) = 0.0243  A + 0.0151  D + 0.0126  F + 0.0108  B + 0.0108  E
Kg/hh/w (glass) = 0.0100  A + 0.0065  F + 0.00585  D + 0.00543  B + 0.0037  E
Kg/hh/w (cans) = 0.00166  E + 0.00143  A + 0.00132  F + 0.00118  B + 0.00092  D
Kg/hh/w (textiles) = 0.00236  F + 0.00091  A + 0.00075  D

All census variables

Kg/hh/w(all) = 0.0664  More mature + 0.0302  Retired

- 0.245  Private sector rentals
Kg/hh/w(paper) = 0.0404  More mature + 0.0136  Detached
- 0.051  Families/older children
Kg/hh/w(glass) = 0.02213  More mature - 0.034  Public sector rentals
Kg/hh/w(cans) = 0.00749  2-person_hh - 0.0018  1-person_hh –0.00  More mature
Kg/hh/w(textiles) = 0.00575  More mature + 0.0047  Public sector rental
– 0.00327  No_car – 0.0060  3-person_hh – 0.0096  Private sector rental

Box 4.3 Stepwise Regression Equations for Chesterfield Multi-material Yields

Acorn

Kg/hh/w (all) = 0.0362  A + 0.0312  B + 0.0176  F + 0.0147  E + 0.0143  D

Kg/hh/w (paper) = 0.0232  A + 0.0186  B + 0.0116  F + 0.0099  E + 0.0089  D
Kg/hh/w (glass) = 0.0099  B + 0.0091  A + 0.0051  E + 0.0039  D
Kg/hh/w (cans) = 0.00236  E + 0.00211  A + 0.0020  F + 0.00120  D
Kg/hh/w (textiles) = 0.00179  F + 0.00172  A + 0.00056  D

All census variables

Kg/hh/w(all) = 0.0472  Detached + 0.047  Flats + 0.0176  Terraced

Kg/hh/w(paper) = 0.0271  1_person_hh + 0.0229  Detached
Kg/hh/w(glass) = 0.0123  Detached + 0.0068  Public sector rentals
Kg/hh/w(cans) = 0.00574  2_cars
Kg/hh/w(textiles) = 0.00825  1_person_hh + 0.0061  Families/older children
–0.00142  Owner_occupation – 0.00243  No_car

49
 Understanding Recycling Behaviour 2

4.5.3 Alternate Residual / Green It is seen that with a single green/residual

Waste Collections
To test whether the two multi-material
conversions with associated alternate week
residual waste collections were significantly
different, further stepwise regressions were
performed, now including “green/residual”
descriptors in the pool of potential explanatory
variables. The new models took on the two
forms:
Kg/hh/w(n) = c1  x(a) + c2 y(a) + c3  z(a) +
… + cg  Green
Kg/hh/w(n) = c1  x(a) + c2 y(a) + c3  z(a) +
… + cog  Oldgreen + cng  Newgreen
where Green did not distinguish between the
two conversions and where Oldgreen and
Newgreen did differentiate the two sets of
rounds. The regressions were carried out over
all rounds in Chesterfield and N.E.Derbyshire
combined.
The results for a single green/residual factor
are shown in box 4.4 where they may be
compared with those for a twin green/residual
factor (box 4.5).
factor, the significant explanatory variables
remained with the Acorn descriptors. A
green/residual factor was only significant for
can recovery. The goodness of fit of the models
2
to the data, measured by the R statistic, were
substantially improved when the single
green/residual factor was replaced by twin
factors (table 4.6), The key features associated
with the twin descriptor models are:
1. The „Newgreen‟ factor (Rounds 61, 63 - 65)
has been identified as a significant factor in
fitting total yields, paper yields, glass yields
and can yields. Newgreen was always
positively correlated with increased yield.
2. „Oldgreen‟ was never identified as being
significant in any of the regressions.
3. A discriminatory factor between
Chesterfield and North East Derbyshire
was only found to be significant for paper
recoveries. There was no significant
discriminator of glass, can, or textile yields
between the two districts. The differences
in glass, can and textile yields appear to be
dominated by local demographic
differences. However, paper yields still
remain systematically lower in Chesterfield
than in N.E. Derbyshire (like they were
before the conversions).

Box 4.4 Stepwise Regression Results for a Single Green/residual Collection Factor
Acorn

Kg/hh/w (all) = 0.0366  A + 0.0231  B + 0.0197  F + 0.0185  D + 0.0166  E

Kg/hh/w (paper) = 0.0243  A + 0.0138  B + 0.0120  D + 0.0116  F + 0.0110  E
Kg/hh/w (glass) = 0.0097  A + 0.00723  B + 0.0049  F + 0.0485  D + 0.00425  E
Kg/hh/w (cans) = 0.00196  E + 0.00161  B + 0.00125  A + 0.00112  Green
+ 0.00094  D – 0.00059  In_Chesterfield
Kg/hh/w (textiles) = 0.00210  F + 0.00115  A + 0.00051  D + 0.00040  B

Box 4.5 Stepwise Regression Results for a Twin Green/residual Collection Factor

Acorn
Kg/hh/w (all) = 0.0319  A + 0.0229  B + 0.0207  D + 0.0191  F + 0.0186  E
+ 0.00890  Newgreen – 0.00280  In_Chesterfield
Kg/hh/w (paper) = 0.0219  A + 0.0139  B + 0.0134  D + 0.0122  E + 0.0113  F
+ 0.00460  Newgreen – 0.00181  In_Chesterfield
Kg/hh/w (glass) = 0.0084  A + 0.00713  B + 0.00540  D + 0.0480  F + 0.00474  E
+ 0.00238  Newgreen
Kg/hh/w (cans) = 0.00144  F + 0.00137  B + 0.00124  Newgreen + 0.00120  D
+ 0.00105  E – 0.00098  A
Kg/hh/w (textiles) = 0.00210  F + 0.00115  A + 0.00051  D + 0.00040  B

50
 Understanding Recycling Behaviour 2

Table 4.6 Goodness of Fit (R-squared) and Chesterfield rounds 51 - 65 (labelled 21 -

Single green Twin green 35).
factor factors
The figure shows there to be a number of
Paper 0.627 0.515 outliers, which appear to be quite unique in
Glass 0.603 0.512 behaviour. The round that is least similar to any
Cans 0.511 0.325 of the others is round 118 in N.E.Derbyshire,
Textiles 0.243 0.196 which was the highest yielding round across
the two districts, and probably the most affluent
All 0.699 0.538 area within those districts as well. The other
major outliers are the Chesterfield rounds 61,
and 63 - 65 (labelled 31, 33 - 35 on the figure).
These were precisely the rounds where the
4.5.4 Similarities Between Rounds multi-material dry recyclable collections and the
alternate week green/ residual waste
Whilst preceding analyses have investigated
collections had started at the same time. The
how individual round performances may be
analysis strengthens the case that there is
correlated with demographic factors, the
something „special‟ about those rounds.
analyses did not investigate how those
performances might be correlated with each
Of the rounds that did cluster, two major large
other. That is, do some rounds behave similarly
clusters formed at > 90% similarity. Analyses
to each other and, if so, can a common factor
were undertaken to ascertain whether there
be identified that might link those rounds. The
were any consistent demographic similarities or
statistical technique used to investigate these
dissimilarities between those two clusters. The
effects was cluster analysis. In this application,
results failed to reveal any consistent or
cluster analysis was used to investigate
significant differences between the two. The
similarities amongst the „patterns‟ in paper,
overall results show that most areas of the two
glass, can and textile yields across the two
communities are actually behaving quite
districts. Results are shown in the form of a
similarly, irrespective of their local fluctuations
dendrogram, which portrays how individual
in demographics. It follows that demographic
variables progressively cluster at lower and
factors might not have much influence on the
lower levels of similarity. The full dendrogram is
„gross‟ recycling behaviour of the two
shown in figure 4.8. The individual rounds
communities. They might only be important in
labelled on the x-axis refer to North East
moderating the finer details.
Derbyshire rounds 101 - 120 (labelled 1 - 20)

Figure 4.8 Cluster Analysis of Individual Round Performances

Similarity
76.32

84.21

92.11

100.00
18 9 15 13 16 28 3 4 26 2 6 7 8 12 19 1 10 20 30 5 22 14 17 25 11 27 21 29 24 32 23 33 35 31 34

Observations

51

4.6 Model Predictions
4.6.1 Current Performance
This part of the research describes a simulation
of the Chesterfield and North East Derbyshire
data using the University of Paisley Integrated
Household Waste Management Model.
The model calibration to the N.E. Derbyshire
data is shown in figure 4.9 and that for
Chesterfield in figure 4.10. The size and
direction of the residuals in those figures
provide model estimates of those rounds
performing better or worse than expected.
Larger residuals imply larger deviations from
the norm. It is noted has the highest „over-
performer‟ (circled on the figure) was round
118, which was also singled out in the cluster
analysis.
Similarly, in Chesterfield, the highest over-
performers were the now famous rounds 61,
and 63 - 65, though it is noted that round 62
was the fifth highest over-performer. This round
was part of the same roll-out as rounds 61 and
63 - 65 but did not have any change in its
residual waste collections. It could therefore be
conjectured that some hitherto unidentified
feature of the roll-out may also have
contributed something to its success.
Figure 4.9 Model Fit to N.E. Derbyshire Rounds
4
3.5
Predicted Kg/hh/w
3
2.5
2 118
1.5
1
0.5
0
0 0.5 1 1.5
Observed Kg/hh/w
Understanding Recycling Behaviour 2
It should also be noted here that the predicted
over and under-performers did not necessarily
map directly onto those predicted to over or
under perform in the regression model fits.
However some over or under performers (like
rounds 61 - 65 and 118) were identified
consistently as outliers across all model fits.
Overall, the Integrated Household Waste
Management Model simulations provided
reasonably close fits to the individual round
paper and glass recoveries in both districts, but
gave relatively poor predictions of the can and
textile recoveries (figures 4.11 and 4.12). Both
of these recoveries were over-predicted by the
model and much more uniform behaviours
were predicted to occur amongst the rounds
than were actually seen in reality. The reasons
for these results are unclear. The results could
point to inadequacies in the model.
Alternatively, they might also indicate that can
and textile yields might not have any „global‟
demographic predictors at all, and that their
recoveries are dominated by local
„peculiarities‟. It should be noted here that the
previous statistical analyses for the two
materials, also showed weaker demographic
associations than were found for paper or
glass, and that those associations (if
significant) were very different to the
associations found for paper and glass.
R2 = 0.3706
2 2.5 3 3.5 4
52
 Understanding Recycling Behaviour 2

Figure 4.10 Model Fit to Chesterfield Rounds

3.5
3
Predicted Kg/hh/w

2.5
2
1.5 61 & 63-65

1
62
0.5
0
0 0.5 1 1.5 2 2.5 3 3.5
Observed Kg/hh/w

Figure 4.11 Model Prediction of Can Recoveries (Chesterfield)

400
350
Predicted Kg/hh/w

300
250
200
150
100
50
0
0 50 100 150 200 250 300 350
Observed Kg/hh/w

Figure 4.12 Model Fit to Textile Recoveries (Chesterfield)

160
Predicted Kg/hh/w

120

80

40

0
0 40 80 120 160
Observed Kg/hh/w

53

The model predictions for each of the collected
materials (averaged across all rounds) are
compared with the average measured weights
in table 4.7 for North East Derbyshire. The
table also compares the model change from
paper-only to multi-material collections with
what actually occurred in reality.
The estimated changes in participation rates
following the change are shown in table 4.8.
These are defined as the percentage of
residents serviced by the kerbside scheme who
use the scheme at least once during a four
week period. The average set out rate (the
percentage of those serviced who use the
scheme in a given collection was estimated to
increase from 38% (paper only) to 53% (multi-
material).
Table 4.8 also provides an approximate
breakdown of participation by demographic
grouping. These results show that the largest
rises in participation may have occurred
amongst the terraced and, to a lesser extent,
the local authority (and ex-local authority)
housing stock. The responses by family life
stage were fairly uniform across all life stages,
but with slightly larger increases seen amongst
households where children were present.
Similar features (not shown here) were also
observed in the Chesterfield model analyses.
A further very significant factor emerging from
the modelling study was that the model
„attitude‟ change across the conversion was
exactly the same in the model Chesterfield as it
was in the model N.E. Derbyshire. That is, the
„pro-kerbside recycling attitudes‟ in both
communities increased by exactly the same
amount on conversion. This is a very important
finding as it means that the average attitude
change appears to be quite uniform and
predictable across the conversion event.
However, as always, some areas reacted
differently to the norm. Rounds 51 and 53
appeared to under-achieve, whilst rounds 61-
65 significantly over-achieved.
Overall, the differences in the final outcome of
the conversion in the two communities may
simply lie in the legacies of their past histories.
Chesterfield residents simply had lower pro-
kerbside recycling attitudes before the
conversion. It is not known why this should be
so, though it is noted that the paper-only
collections in Chesterfield had markedly
declining yields over their last year of operation
Understanding Recycling Behaviour 2
(see figure 7.9, section 7.4). This points
strongly to some negative event or events
having occurred during that year.
Model predictions of the change in bring site
recoveries are shown in table 4.9 for N.E.
Derbyshire. The data refer to the tonnages of
materials deposited by residents from the multi-
material collection areas before and after the
scheme introduction. The before figures all
assume that the areas were fully covered by
paper-only collections.
Similar predictions applied to Chesterfield, with
a predicted drop of around 40% overall in bring
site tonnages. That drop was predominantly
associated with a fall in glass bank recoveries.
Unfortunately it is difficult to provide any
meaningful comparison with the changes
recorded in the actual bring site tonnages (table
4.5) as the sample bases for the two
estimations was dissimilar. However, it is noted
that the predictions of the relative effect on
paper and glass bank recoveries are probably
in the right ball park, with the change in paper-
bank recoveries being a little under-predicted
and the change in glass recoveries being a little
over-predicted.
The predicted impacts on can and textile bank
recoveries, however, would seem to be at
some variance to the observed data, and the
results for those two materials are, in
themselves, quite different. The model has
overestimated not only the kerbside can
recoveries but has overestimated the bring site
can recoveries as well. In contrast, the model
has overestimated kerbside textile recoveries
but has underestimated the textile recoveries at
bring sites.
Basically this points to:
(i) lower can recoveries than expected across
all recycling points, and to
(ii) a seeming reluctance to transfer allegiance
for textile recycling away from the bring
sites onto the kerbside collection.
Put another way:
(i) can recycling is low across both districts for
all can-recycling outlets, and
(ii) people are not forsaking the traditional
routes for textile recovery even when new
[and potentially more convenient] outlets
are provided.
54
 Understanding Recycling Behaviour 2

Table 4.7 Whole District Kg/hh/w: Model Comparisons– N.E. Derbyshire

All Paper Glass Cans Textiles
Multi-material
Model 2.35 1.40 0.64 0.188 0.112
Actual 2.24 1.44 0.60 0.126 0.070
Paper-only
Model 1.24
Actual 1.31

Table 4.8 Model Estimations of Kerbside Participation before and after the Change to Multi-
material – N.E. Derbyshire
Family life stage* House type
Local
Semi & auth-
All YA FYC FOC MM RET Det ority Terrace

Before 50 37 41 51 57 53 65 53 35
After 68 55 61 72 74 68 78 73 63

% change 18 18 20 21 17 15 13 20 27
* YA=young adults, no children, FYC=families with young children, FOC= families with older children,
MM=more mature, RET=retired

Table 4.9 Predicted Change in Bring Site Tonnages (Tonnes/month) – N.E. Derbyshire
All Paper Glass Cans Textiles

Before 25.45 9.60 11.40 1.44 3.67

After 15.42 8.94 4.26 0.56 1.00

% change -39% -7% -63% -61% -73%

4.6.2 Future Scope Table 4.10a Effects of a Procedural

Information and Scheme Awareness
The final question to be answered is “Is there Reinforcement Campaign – N.E.Derbyshire
any scope for further improvement from follow- All Paper Glass Metal
up promotional and awareness campaigns?”
Current 2.32 1.36 0.64 0.193
One of the scenarios modelled was a After 2.44 1.40 0.68 0.217
procedural information and kerbside awareness campaign
campaign. It was assumed that this would be a
fairly strong and intensive campaign. Whilst it % change +5% +3% +6% +12%
may not represent the ultimate improvements
that could be achieved, it does show what
might be realistically achieved in [best] practice, Table 4.10b Effects of a Procedural
with carefully thought out messages and well Information and Scheme Awareness
targeted and orchestrated campaigning. At this Reinforcement Campaign – Chesterfield
stage, we do not yet consider what precisely All Paper Glass Metal
those campaign messages should be.
Current 2.17 1.33 0.60 0.147
Results are shown in tables 4.10 a-b. After 2.35 1.43 0.64 0.178
campaign

% change +8% +5% +7% +21%

55

The participation rate was predicted to increase
from 68% to 71% in N.E. Derbyshire and from
61% to 65% in Chesterfield.
Stronger promotions with regular reinforcement
messages could potentially increase
participation and yield to higher levels still. With
the full range of promotional and informational
measures in association with a 100% multi-
4.7 Summary and Conclusions
The conversions from paper-only to multi-
material kerbside collections in Chesterfield
and North East Derbyshire produced significant
step changes in material yields. Typically paper
yields increased by factors of order 1.1 to 1.4 in
N.E. Derbyshire, and by factors of 1.5 or more
in Chesterfield. Total kerbside yields in the new
scheme averaged in excess of 2.3 Kg per
household served per week in N.E. Derbyshire
and nearly 1.8 Kg/hh/w in Chesterfield
excluding the latest conversions where yields
of over 2.9 kg/hh/w were recorded). The
recovered material comprised around 63-67%
paper, 28-29% glass, and 4% and 3% cans and
textiles respectively.
The change from paper-only collections to
kerbside multi-material collections still
maintained the historical differential in recycling
performance between the two districts except
for the four latest conversions in Chesterfield
where that differential was eroded). That is the
historically poorer-performing areas remained
the poorer performers in the new scheme.
Model analysis indicated that identical step
changes in attitudes had occurred in both the
two districts. The disparity between the two
districts could not be accounted for by
demographic differences. It would appear to
have its roots somewhere in past history.
Model predictions showed that the introduction
of the multi-material kerbside collection should
reduce the districts‟ bring site recoveries by
around 40%. Such levels of reduction were
confirmed in the measured glass and paper
bank statistics. However, there was little if any
drop in the recoveries from the can and textile
banks.
Paper and glass recoveries were reasonably
predictable. However can and textile recoveries
were much less explainable with substantially
Understanding Recycling Behaviour 2
material kerbside coverage, captures of around
80% of the newspapers, pams and glass
should be possible. Can and textile recoveries
would be expected to be lower, but around 75%
capture of those materials should still be
achievable ultimately. These endpoints were
very similar to those predicted for a „perfect‟
England (section 2).
lower recoveries being achieved than might be
expected for those materials.
It would appear that: i) there was a relatively
poor ethos for can recycling per se across the
two districts, and ii) there was a reluctance to
transfer allegiance for textile recycling away
from its traditional outlets onto the kerbside
collection.
Statistical analyses were used to estimate the
possible influences of household demographics
on kerbside yields. By and large, the derived
rankings supported the traditional wisdom that
it was the more affluent and older residents that
tended to contribute most to the recycling
scheme. Whilst some demographic differences
were noted between Chesterfield and North
East Derbyshire recyclers, the major
differences appeared to be much more to do
with who recycles what. Can and textile
recyclers appeared to be drawn more from the
poorer, younger households compared with the
paper and glass recyclers.
The parallel alternate green waste and residual
waste collections had mixed effects on the dry
recyclable yields. Four of the targeted rounds
(the latest conversions identified above)
showed very high yields, whilst the yields of the
other five rounds in the green waste trial area
were close to those of the normal collection
regimes.
Overall, the statistical analyses showed that the
performances of most of the collection rounds
had a relatively high level of similarity
irrespective of their demographic make-up. Two
core behavioural clusters were identified. Major
outliers to those clusters were the latest green
waste conversions. Behaviours there proved to
be highly dissimilar to any of the other rounds.
Few other rounds could be singled out as
56

unambiguously performing better or worse than
expected from their demographic profiles.
The high performing rounds, yielding over 2.9
kg/hh/w all occurred in the areas where the
alternate week residual waste collections were
launched at the same time as the multi-material
kerbside recycling collections. The poorer-
performing dry-recyclate collections mapped
onto the areas where alternate week green and
residual waste collections had already been
running six months prior to the multi-material
conversion.
In psychological terms, the imposition of a
reduced frequency of residual waste collection:
(i) raises the saliency of the waste
management issue to the householder, and
(ii) may also necessitate a lifestyle change to
cope
Providing a ready-built solution (i.e. the multi-
material box) at the same time, leads
householders to adopting that solution.
Imposing the lifestyle change prior to providing
the solution can render that solution almost
irrelevant when it eventually comes.
However, whilst the simultaneous changes to
the residual waste collections appeared to be a
major factor in why those conversions were so
successful, other causal mechanisms for the
57
Understanding Recycling Behaviour 2
success cannot be ruled out completely. The
four rounds were all relatively affluent areas,
and a fifth conversion at that time, but without
any associated change to its residual waste
collection, also performed above expectations.
Overall, the current research has provided
reassurances that the new multi-material
collections in Chesterfield and North East
Derbyshire are performing at respectable levels
compared to multi-material kerbside collections
operating elsewhere in the UK.
The research centred on developing
performance predictions and undertaking
performance diagnostics. Calibrated
behavioural models have been built for both
communities, and those can now be utilised for
developing future performance predictions, for
different management scenarios and for
different event chains.
One question that still has to be posed is
whether the findings in Chesterfield and North
East Derbyshire can be transferred to other
localities. That question can only be answered
through undertaking parallel studies for other
districts. The research programme at the
University of Paisley has always sought to carry
out replicate studies for validation wherever
possible. The results from parallel studies in
Lancashire and in North Warwickshire will be
presented in the next section.

5. Results from Other Communities
5.1 Lancashire and North West England
5.1.1 Introduction
Effectively, in Chesterfield and North East
Derbyshire we were asking the question “How
much of the observed spatial variations in
recycling performance can ever be predicted?”,
or by implication “How much of the observed
variations must be considered irrational”
Gaining a full understanding of this can only be
achieved through analysing large data sets
from many municipalities. This represents a
major step change from the much more
restricted analyses normally undertaken in
recycling research. The necessary large-scale
research cannot be addressed (realistically) by
running out bigger and more extensive surveys
or through undertaking more extensive
household monitoring. The costs would be
prohibitive. The analyses must be based on
data that is relatively cheap to obtain. Suitable
data exists only in the form of the round-by-
round tonnages that are recovered from
kerbside collections. A comprehensive analysis
of those data could provide an invaluable
source of case study information from which
inferences about best practice and the basis of
recycling behaviour may be drawn. The
intention here was to simultaneously analyse
data sets from all fourteen district authorities in
Lancashire.
In the outturn, only 7 of the 14 Lancashire
authorities were able to supply data in a form
suitable for the detailed round-by-round
collection-by-collection analyses, though most
could supply aggregated totals across the
whole district. Only three of those councils
supplied long time series for three years or
more. Whilst this did compromise the scope of
analyses that could be undertaken, there was
nevertheless still sufficient data for a
meaningful study from which valid conclusions
could still be drawn.
The analyses that were undertaken could only
be based on the data actually received from the
authorities in suitable form. Many of the data
sets received were not as clean and as perfect
as they might have been. The wider
implications of that and other performance
monitoring issues are discussed later in the
report (section 11.2).
Understanding Recycling Behaviour 2
5.1.2 Whole District Statistics
Firstly an attempt was made to regress district-
wide yields against the demographic profiles of
the districts. Overall, at the inter-district level,
1
both Mosaic descriptors and raw census data
provided equally poor explanations of paper
yields.
Taking the whole of the North West Region, the
stepwise regression could only find one
significant term: Mosaic „A‟ (high income
households) in predicting paper yields. For
Lancashire on its own, stepwise regression
showed that Mosaic „K‟ (Country dwellers) and,
more marginally, Mosaic „I‟ (independent
elders) correlated positively with paper yield,
whilst Mosaic „B‟ (Suburban semis) and Mosaic
„F‟ (Victorian housing) had negative
correlations.
With Acorn groupings, the only [marginally]
significant predictor of yield for Lancashire
districts was the proportion of Acorn „E‟
housing, which was negatively correlated with
paper yield. (Acorn E = „Aspiring‟ – see
Appendix A for further details).
Overall, the analyses indicated that at district
level, the differences in district demographic
profiles might offer relatively little explanatory
power for the observed differences in their
kerbside paper yields. Few demographic
variables appeared to be significant and there
was no consistency between analyses at
county level or at regional level.
Similarity, or lack of similarity between
populations with multiple attributes can also be
tested through cluster analysis – similar to the
techniques used to derive the Mosaic and
Acorn clusters from the raw census data. In the
current study, cluster analysis was applied to
the identified demographic variables across 26
districts in the North West Region. The results
showed that only one central cluster emerged,
with the outliers each appearing to be „unique‟
and not similar at all to any other outlier (figure
5.1).
1
Mosaic is an alternative market research
classification system to Acorn.
59

In the figure, individual districts are represented
by a number from 1 to 26. They are ranked in
order of their kerbside paper yield – 1
representing the district with the highest yield
(Kg/hh/w) through to 26 - the district with the
lowest yield. It is noticeable that the major
outliers are districts 1, 21, 2, 7, 25, 22, 6, 26,
and 23, which are drawn most heavily from the
extreme performers (both good and bad).
Basically this shows that the majority of
authorities in the North West might be
reasonably similar demographically when taken
as districts as a whole, and may all perform to a
reasonably similar „intermediate‟ standard in
their kerbside recycling performance. This ties
in well with the results from section 2, which
show that there is little variance between the
schemes with borough-wide coverages or
between schemes using the fairly ubiquitous
bag collections. The „outliers‟ however may
behave as such for quite different reasons.
Some of those possible reasons are explored in
figures 5.2 and 5.3. These figures plot the
distance of the outliers from the centroid of the
core cluster. Positive distances for a given
demographic category indicate that that
category is more prevalent in the outlier than in
the core cluster, whilst negative distances
indicate that the demographic category is less
prevalent. Figure 5.2 shows the results for the
two outliers with better than average kerbside
yields, whilst figures 5.3 a&b show the results
for the four outliers with lower than average
yields.
Figure 5.1 Cluster Analysis of the North West Region of England
Similarity
40.35
60.23
80.12
100.00
1 21 2 3 5 9 24 17
60
Understanding Recycling Behaviour 2
The two „better-performing‟ districts (figure 5.2)
show a slightly older age pofile than the core
districs with more retired (ret) residents and
also more more-mature (mm) residents. They
also have higher owner occupation and higher
numbers of households with cars. The areas
additionally have more detached houses with
generally lower representations of all other
housing types. There may also be
proportionally fewer single person households
and fewer large households (five+) in the
areas.
Amongst the „poorer-performing‟ districts (figure
5.3), districts A and B show proportionally more
families with children than there are in the core
cluster, with significantly higher levels of young
adults with no children also present in district B.
More households are renting from the private
and public sectors (rent(o) and rent(p)
respectively), and car ownership levels are
depressed. The areas have greater proportions
of flats than the core cluster with more large
households (five+) in area A and high
proportions of single person households in area
B. Poor-performing areas C and D, in contrast,
show larger retired populations with higher than
average numbers living in converted flats.
There is also significant private sector rental in
area D. Areas C and D have smaller household
sizes than the core, with especially more single
person households.
4 10 11 19 16 8 18 20 14 15 13 12 7 25 22 6 26 23
Observations
 Understanding Recycling Behaviour 2

Overall, the results again support the traditional
wisdom that those from more affluent housing
and those from the older generations tend to be
the better recyclers, whilst those living in multi-
family dwellings, those renting from the private
sector and the younger residents tend to
recycle less. However, whilst the model has
revealed that those demographic features may
offer some explanation of the differences in
recycling performance amongst districts, no
definitive model emerged from the analyses.
Part of the problem could be due to the general
„smoothing out‟ of differences when
aggregating over large populations. As several
districts were quite similar demographically,
identifying models based on demographic
differences may lead to uncertain results. In
order to investigate the influence of possible
demographic dependencies any further, it
becomes necessary to concentrate on smaller
populations. Such analyses will be considered
in the next section.

Figure 5.2. Differences in Demographic Figure 5.3 a&b. Differences in Demographic

Profile of Two Well-performing Outliers Profile of Four Poorly-performing Outliers
Compared to Mean Performance Compared to Mean Performance

ret ret

mm mm

foc foc

fyc
fyc
yanc
yanc
Rent(o)
Rent(o)
Rent(p)
Rent(p)
Ow ner
Owner
Car
Car
Converted
Converted
Purpose
Purpose
Terraced
Terraced Semi
Semi Detached
Detached Five+

Five+ Four

Four Three

Three Tw o

Two One

One -4 -2 0 2 4

-2 -1 0 1 2 3
Distance from core cluster centroid

Distance from core cluster centroid A B

61
 Understanding Recycling Behaviour 2

each round. Also, it must again be noted that at

ret the time of the study the latest available census
mm data were quite dated. These possible sources
of error must always be borne in mind when
foc
interpreting the results.
fyc

yanc
Individual Authority Analyses
Rent(o)

Rent(p) As a prelude to the main inter-authority

Ow ner analysis, attempts were made to fit
Car
demographic models to the collection rounds of
individual authorities. The models were
Converted determined by a combined forward and
Purpose backward stepwise regression (see section
Terraced
4.5.2).
Semi
In Blackburn, the most significant predictor was
Detached the proportion of private sector rentals, which
Five+ was negatively correlated with the paper yields
amongst that district‟s 24 collection rounds.
Four
However, that term was eliminated in a
Three subsequent iteration of the stepwise
Tw o regression. The best fitting model contained
three significant terms. The proportion of
One
detached houses and the proportion of two-
-4 -2 0 2 4 6 person households both showed a positive
Distance from core cluster centroid effect on yield. Curiously, the third significant
term was having one car (as distinct from no
C D cars or two or more cars). One-car households
correlated negatively with yield.

For Blackpool (10 rounds) no significant

5.1.3 The Influence of Demographic
Factors On Recycling Yields
At Neighbourhood Scale
The demographic analyses described above
are now repeated at a finer resolution amongst
the individual daily collection rounds. Those
collection round sizes are typically 1000 to
3000 households, and represent the finest
resolution to which routine monitoring data is
normally collected. However, as discussed in
section 5.1.1, not all Lancashire councils
recorded those data, and even where records
were kept, recorded weights did not always
map one-to-one onto an individual daily
kerbside collection. Sometimes bring site
tonnages are subsumed into the figures, and
sometimes weighbridge recordings did not
correspond exactly to round beginnings and
round ends. Also, it must be borne in mind that
the small area census statistics enumeration
district boundaries used in the analyses do not
normally correspond with collection round
boundaries. Where an enumeration district
spanned a collection round boundary, the
household data were proportionally allocated to
predictor variables were identified.
For Burnley, like Blackburn, the most significant
predictor of neighbourhood performance was
the proportion of private sector rentals. This
predictor retained its significance throughout
the stepwise regression, correlating negatively
with yield. Of the different housing types,
terraced housing and converted flats had
negative influences whilst the proportion of
detached housing corresponded positively with
yield. Other significant factors were the
proportion of 3 person households (negative)
and the proportion of the more-mature but not
yet retired fraction (positive).
In Ribble Valley (6 rounds operating at the time
of data collection), the proportion of young
adults without children emerged as the most
significant predictor (negative effect), with lack
of car ownership (also negative) proving to be
significant as well.
In Wyre (10 collection rounds), only one
significant predictor emerged. That was the
proportion of detached houses, which had a
positive effect.

62

Overall, no general model emerged that could
show comparable explanatory powers across
all authorities. However, in the main, the terms
identified were still amongst those that were
expected according to traditional wisdom. In
general, the models were quite peculiar to the
specific ranges of local demographics and did
not extrapolate to areas of different
demographics, or even to the neighbouring
authorities in Lancashire. To derive the County-
wide picture, it is necessary to consider all
authorities together. Such an aggregation takes
advantage of the wider pool of demographic
profiles that are available, and should result in
a more general model.
County-wide Analyses
The „County-wide‟ analyses were based on an
aggregation of the five districts considered
individually above: Blackburn, Blackpool,
Burnley, Ribble Valley, and Wyre. The
aggregated data set covered 70 collection
rounds.
Stepwise regressions were carried out over all
70 rounds from the five districts. The
demographic terms included were housing
type, family life cycle stage, household size, car
ownership and tenure. Initially the functions
fitted were of the form:
Yield (Kg/hh/w) = Constant + C1 V1 + C2 V2
+ . . . + Cn  Vn
where V1,2,..n are the values of the significant
demographic variables and C1,2,…n are the
associated set of regression coefficients.
However, it was found that the residuals from
this model (i.e. the disparities between
individual observations and the model
predictions of those observations) were
significantly correlated with districts. Whilst the
residuals were normally distributed for three of
the districts, the model had consistently over-
predicted performances in each of the rounds
of the poorest performing district and had under
predicted the performances in all of the rounds
of the best performing district.
To try to overcome this disparity, a second
model was also fitted. This model took the
form:
Yield (Kg/hh/w) = Constant (k) + C1 V1 + C2
V2 + . . . + Cn  Vn
Understanding Recycling Behaviour 2
where a separate and independent constant
term was allowed for each district. (Analogous
to the „In Chesterfield‟ factor of section 4.5.2).
The hypothesis behind this was that there may
be a significant „Borough-specific‟ term as well
as demographic terms in the determining
equations. That is, some as yet unidentified
feature of the borough might be moderating any
more global demographic dependences.
The derived regression equation was:
Yield (Kg/hh/w) = 1.75 Detached + 2.46
Rent(public sector) - 3.00 Converted flats +
3.81 Retired - 2.53 Purpose-built flats - 1.31
Semis - 2.30 No car + 8.71 Families with older
children
On including the possibility of district-specific
constant factors, the method yielded a
somewhat simpler equation:
Yield (Kg/hh/w) = 2.26 Detached + 0.687
Retired - 0.680 Owner occupied + 2.76 2-
persons - 0.707 C2 + 0.608 C4
Where C2 and C4 are the constant terms for
Blackpool and Ribble Valley respectively.
The common descriptors across the two
equations were the proportion of detached
housing and the proportion of retired residents
respectively. It is also noted, however, that the
proportion of public sector renters actually
showed a positive influence and owner
occupiers producing a negative impact on yield.
It should be noted, however, that those terms
only retain that significance when all the other
listed terms are present. It does not necessarily
mean that public sector renters are actually
better recyclers than owner-occupiers. It is also
noted that only the constant terms for the worst
performing district and the best performing
district were actually significant.
Normally, in analyses such as these, the
outliers of concern are considered to be those
with standardized residuals (actual residual/
standard deviation) of greater than about 2. In
the analyses it was found that just 6 out of 70
rounds returned standardized residuals greater
than 1.9. The demographic profiles of those six
major outliers did not reveal anything other than
marginal differences in the household sizes and
family life-cycle stages of their residents
compared to the 70 rounds taken as a whole.
The major differences appeared to be in
housing type, tenure and car ownership. Two of
the outliers were quite similar. They had much
higher proportions of detached housing and
63

lower proportions of terraced housing than
average, and substantially higher levels of car
ownership and owner-occupation. Both
performed much better in reality than was
predicted by the models. One of the other
outliers, by contrast had very high proportions
of flats, council tenants, single occupancy, and
retired residents, with low levels of car
ownership. This district also performed much
better in reality than was predicted by the
model. The common factor between all three
was that they represented the extremes of the
demographics of the five districts. It is not
surprising, perhaps, that the model fit tails off at
those extremes. The poor fit may be as much
to do with the inadequacy of the model as it
may be to do with any physical reason.
The three other outlying rounds did not show
such extreme demographics, yet they, too,
were poorly fitted by the model. The model
significantly under-predicted the performances
of two of those rounds, whilst the third was
significantly over-predicted. The reasons for
such poor fits remain unclear.
5.2 North Warwickshire
5.2.1 Introduction
North Warwickshire, like Chesterfield and North
East Derbyshire, has recently converted its
kerbside paper-only collections into multi-
material collections recovering paper, glass,
cans and textiles. The new fortnightly collection
provides householders with red boxes plus
separate bags to hold the paper and the
textiles. Collections started in January 2003
with just under 21,000 households being
serviced by the new scheme.
The previous paper-only collections were split
into 8 separate collection rounds, one round
collecting fortnightly with the other seven
operating monthly collections. Eighteen multi-
material rounds now operate. In this report they
are designated as rounds 1 – 19, excluding a
round 10.
5.2.2 Dry Recyclate Yields
At the time of writing, the analyses from North
Warwickshire must be limited to just the first
seven collections of the new scheme. The data
from the first collections after conversion have
64
Understanding Recycling Behaviour 2
5.1.4 Conclusions
Overall, it can be concluded that whilst there
appears to be a general pattern between
observed performance variations and
demographic profiles, the performances of
individual neighbourhoods show much scatter
around that general trend. A general
demographic predictor cannot explain all of the
variance between neighbourhoods. Different
demographic factors may take on different
importances in different districts. However, like
the whole district analyses that were presented
in the previous section, the picture appears to
be one of reasonable conformity across the
majority of areas, but with a small number of
outliers; those outliers often having the most
extreme demographic profiles. As such, it might
be concluded that simple linear models may not
be the most appropriate models to use to
explain demographic effects. There was
insufficient data to test the efficacy of using
more complex models.
been excluded from the analyses. Such data
are rarely typical of longer-term performances.
Table 5.1 records the data for the North
Warwickshire multi-material collection rounds
together with historical figures for the old paper-
only rounds.
The results show, typically, that kerbside paper
yields have doubled. The total kerbside yields
in the new scheme are in excess of 2.3 Kg per
household served per week, which is a 3.6-fold
increase in dry recyclate recoveries at kerbside.
Of this total, 1.37 Kg/hh/w (59%) consists of
paper; 0.67 Kg/hh/w (27%) glass; 0.14 (6%)
cans; and 0.18 Kg/hh/w (8%) textiles.
The multi-material yields achieved in North
Warwickshire are very comparable to those
realised in North East Derbyshire (table 4.3a),
and are well in excess of those generally
achieved in Chesterfield (table 4.3b). The
striking feature about the North Warwickshire
results, however, is that the prior paper-only
performances matched those in Chesterfield,
well below those of North East Derbyshire. A
much greater step change in performance was
 Understanding Recycling Behaviour 2

achieved in North Warwickshire than in either
of those other two districts.
It is again very difficult to gain individual
measures of how individual rounds have
responded to the conversions changes, as the
new multi-material rounds comprise only a part
of an old paper-only round, or span parts of two
(and sometimes three) old rounds. Some
previously un-serviced properties have been
included as well. In the figures, presented in
table 5.2, estimations have been made to
reconstruct the equivalent paper-only
performances within the new round boundaries.
These are computed from the whole round
paper-only performances weighted by the
numbers of houses drawn from each old round.
Minority inclusions of new properties have
assumed the Borough average as their prior
performance.
The results show significant increases in total
yield by a factor of 3 to 4 in nearly all rounds,
with a fairly even increase of around 2 in the
paper yields of those rounds.
The comparable statistics for N.E. Derbyshire
were a range of paper fraction multipliers
ranging from 0.9 to 2.1 for individual rounds
(average 1.3) and from 1.0 to 2.2 in
Chesterfield (average 1.7).
There is a significant correlation between the
old and new performances (figure 5.4). If a
round was a relatively good performer before
the conversion, it remained a relatively good
performer after. However, it was a relatively
poor performer historically, it still tended to
remain so after the conversion. Effectively, a
legacy of past history persists through the
conversion. Very similar effects were seen in
Chesterfield and N.E. Derbyshire.

Table 5.1 North Warwickshire Kerbside Yields

Paper-only Kg/hh/w Multi-material Kg/hh/w
Round Round
Paper Glass Cans Textile Total

0 0.49 1 1.62 0.82 0.15 0.16 2.75

1 0.65 2 1.50 0.67 0.19 0.22 2.57
3 0.42 3 1.44 0.69 0.17 0.23 2.53
4 0.77 4 1.55 0.62 0.14 0.22 2.53
5 0.67 5 1.59 0.60 0.11 0.20 2.49
6 0.88 6 1.33 0.64 0.15 0.18 2.30
7 0.61 7 1.05 0.34 0.11 0.17 1.67
8 0.52 8 1.00 0.42 0.12 0.08 1.61
9 1.34 0.73 0.14 0.15 2.36
11 1.57 0.67 0.16 0.10 2.50
12 1.37 0.54 0.12 0.16 2.20
13 1.41 0.65 0.13 0.20 2.39
14 1.47 0.69 0.13 0.18 2.48
15 1.40 0.60 0.12 0.20 2.33
16 1.25 0.55 0.12 0.19 2.11
17 0.95 0.56 0.09 0.23 1.83
18 1.51 0.71 0.21 0.12 2.54
19 1.35 0.75 0.11 0.17 2.38

Average 0.63 Average 1.37 0.63 0.14 0.18 2.31

65
 Understanding Recycling Behaviour 2

Table 5.2 Changes in Paper Recovery with Conversion to Multi-material (North Warwickshire)
Round Kg/hh/w Paper-only Multiplier Multiplier
Equivalent (Overall) (Paper)

M0601 2.75 0.76 3.60 2.12

M0602* 2.57
M0603 2.53 0.77 3.29 1.87
M0604 2.53 0.67 3.75 2.30
M0605 2.49 0.88 2.84 1.81
M0606 2.30 0.51 4.49 2.59
M0607 1.67 0.52 3.19 2.00
M0608 1.61 0.54 3.01 1.86
M0609 2.36 0.66 3.59 2.04
M0611* 2.50
M0612 2.20 0.67 3.30 2.06
M0613 2.39 0.67 3.55 2.09
M0614 2.48 0.65 3.80 2.26
M0615 2.33 0.88 2.66 1.60
M0616 2.11 0.52 4.04 2.40
M0617 1.83 0.49 3.71 1.92
M0618 2.54 0.53 4.79 2.84
M0619 2.38 0.61 3.91 2.21

Average 2.30 0.64 3.60 2.12

* 100% newly-serviced properties

Figure 5.4 Changes in Yield for Individual 5.2.3 Demographic Dependences

Rounds
A stepwise regression of demographic factors
3 against yield data was undertaken in North
Warwickshire, as it was in Chesterfield and
2.5 North East Derbyshire (section 4.5) and
Kg/hh/w multi-material

Lancashire (section 5.1). See section 4.5.2 for

details of the methodology.
2
The identified models for North Warwickshire
1.5 are presented in box 2.1.

1 The main features that emerge from the results

are: (i) the overall high rankings achieved by
Acorn E households and families with young
0.5 children respectively, and (ii) the relatively low
rankings achieved by Acorn A households. The
0 regression results for these categories appear
0 0.5 1 contrary to traditional wisdom that it is the older
and more affluent households who recycle
Kg/hh/w paper-only more and the poorer and younger households
who recycle less.

The results for North Warwickshire are very

different to those found in Chesterfield or North
East Derbyshire, and indeed to those found
across Lancashire as well.

66
 Understanding Recycling Behaviour 2

Box 2.1 Stepwise Regression Equations for North Warwickshire Multi-material Yields
Acorn

Kg/hh/w (all) = 0.0325  E + 0.0253  B/C + 0.0227  D + 0.0218  A

Kg/hh/w (paper) = 0.0186  E + 0.0161  B/C + 0.0140  D + 0.0112  A
Kg/hh/w (glass) = 0.0091  E + 0.0078  A + 0.0067  B/C + 0.0055  D
Kg/hh/w (cans) = 0.0024  E + 0.00142  D + 0.00120  B/C + 0.00097  A
Kg/hh/w (textiles) = 0.0024  E + 0.0019  A + 0.0018  B/C + 0.00118  D

All census variables

Kg/hh/w(all) = 0.083  3_person_hh + 0.040  4_person_hh

Kg/hh/w(paper) = 0.0271  Families/young children + 0.0229  More mature
Kg/hh/w(glass) = 0.010  Car - 0.011  Young adults
Kg/hh/w(cans) = 0023  1_person_hh + 0.006 Families/young children
- 0.005  Young adults
Kg/hh/w(textiles) = 0.0096  3_person_hh

Quantitative regression models based on 5.2.4 Similarities between Rounds

demographics could only statistically account
for around 30-50% of the variability [in total and
individual materials yields] seen amongst
individual collection rounds in the Borough. As
such, demographic factors appear to be
relatively poor predictors of performance in
North Warwickshire. Previous performance in
the old paper-only collections was an equally
good predictor of current performance.
However, much of the observed variation
between rounds has not been explained.
No evidence was found that the demographic
base of the old paper-only recyclers was
significantly different from the new multi-
material recyclers.
The regression analyses identified three rounds
with high negative residuals – i.e. behaving
significantly worse than expected from the
demographic model. It also identified one round
behaving significantly better than expected.
The differences were confirmed through a
cluster analysis of round performances (figure
5.5). The results show that the three poor
performing rounds (7, 8, and 17) cluster loosely
together and are highly dissimilar to all the
other rounds. Fourteen of the eighteen rounds
form a strong cluster of relatively high similarity.
The best performing round, number 1, is a
distinct outlier to this cluster.

Figure 5.5 Similarities amongst Collection Rounds (Kerbside Yields)

Similarity

60.93

73.95

86.98

1 2 3 13 15 14 12 6 4 5 9 19 11 18 15 7 8 17

100.00
1 2 3 12 14 13 11 6 4 5 9 18 10 17 15 7 8 16

Observations
67

Analyses of the three poor performing rounds
failed to reveal any common demographic
factor or factors that might uniquely differentiate
those rounds from the rest of the Borough.
5.3 Discussion and Conclusions
The statistical analyses of kerbside paper
yields across Lancashire and the North West of
England provided some supporting evidence
that areas containing the greater proportions of
traditional recycling classes do tend to recycle
greater quantities of paper. The proportions of
detached households, car ownership, and
retired residents often correlated positively with
higher yields whereas the proportion of flats
(especially converted flats), private sector
rentals and younger adults often correlated
negatively with yield. However no one model
was found to apply to all districts, and no
individual demographic variable was found to
be equally significant within every district.
Demographic models fitted simultaneously
across the whole county also failed to provide
any unique solution. Several different models
all provided comparable fits. Furthermore, a
significant proportion of the observed variance
remained unexplained by every model. The fits
to around 10 - 20% of the collection rounds
were particularly poor. Some, but not all, of
those outliers were of relatively extreme
demographics (e.g. very affluent, or a high
proportion of younger residents, single
householders, and private sector rented flats).
The results from Lancashire echo the results
from Chesterfield and North East Derbyshire,
where different demographic models were also
identified for each district, and where most
collection rounds also returned similar
performances. The relative strength of the
traditional recycling base was apparent there
also. North Warwickshire appeared contrary to
both Lancashire and the Derbyshire districts
with the recycling yields correlating most
strongly with the proportions of local authority
housing and families with young children. The
proportion of affluent housing stock had much
weaker associations with yield. However, like
the other counties, most collection rounds in
North Warwickshire showed quite similar
performances overall, but with a small number
of outliers that did not appear to fit into the
general pattern.
68
Understanding Recycling Behaviour 2
Finally, it is again noted that the analyses
represent the first three months of the new
multi-material collection. The data should not
necessarily be taken as being representative of
longer-term performances.
North Warwickshire went through the same
conversion as Chesterfield and North East
Derbyshire, changing from a paper-only to a
multi-material kerbside programme. The
kerbside yield rose dramatically in all 3 districts,
though the step change was proportionally
greater in North Warwickshire than in
Derbyshire. The historic performance
differentials between poorer and better
performing districts were maintained across the
conversion in North Warwickshire as they were
in the two Derbyshire districts. Neighbourhoods
that had performed relatively well before the
conversion were still amongst the best
performers afterwards, and vice versa.
In Derbyshire, there was some evidence that
the kerbside recycling base had expanded to
include more of the traditionally poorer
recycling groups, whilst in North Warwickshire
there was no evidence that the demographic
recycling base had changed significantly.
Overall, the general picture emerging is that
both county-wide and also within districts the
majority of „discrete area‟ kerbside
performances are all reasonably similar, and
reasonably consistent with the local
demographic variations. Demographic factors
only appear to be important in explaining the
finer-scale local variations in recycling
performances. Demographic factors appear to
be much weaker global predictors. By and
large, recycling performances are quite uniform
and of an „intermediate standard‟ across areas.
Whilst 80% or more of the collection round
performances within a district might be
considered to behave broadly similarly, some
10 - 20% of all rounds may need to be
considered as being „different‟. A similar pattern
was seen at coarser resolution between
districts as well. It is further noted that the
pattern of 80% predictable, 20% different has
also been seen at much finer spatial resolutions
– right down to enumeration district levels of
around 250 households (Tucker, 2001). The
key to understanding lies in understanding the
outliers.

6. Relationships Amongst
Behaviours
6.1 Introduction
Most surveys have shown that most people
claim to recycle something, though different
levels of recycling activity are seen for different
materials. However, few surveys have gone on
to investigate the relationships between the
recycling of different materials, or the
relationships between recycling behaviour and
the other pro-environmental waste
management behaviours. Understanding those
connections [if any] will assist the national
waste awareness campaigns to develop and
target the most appropriate messages that
might encourage the whole spectrum of waste
minimisation activities. This section looks at the
evidence for any such connections.
6.1.1 Review
While most previous waste management
attitude/behaviour research has concentrated
on the relationship between recycling
behaviours and their antecedent attitudes (see
section 8.1 or Tucker, 2001 for a review), some
researchers have attempted to find correlations
between different types of pro-ecological
behaviours. This research has shown mixed
results. Analysis by Siegfried et al. (1982) failed
to reveal a consistent pattern of predictors for
environmental behaviours. Lober (1996) could
find no correlations between recycling
behaviour and source reduction behaviour.
Cook and Berrenberg (1981) reviewed
research on conservation behaviour and
concluded that different behaviours are likely to
be related to different antecedent variables.
Granzin and Olsen (1991) confirmed this, using
recycling, donating to charity and walking to
save gasoline as examples. In contrast Folz
and Hazlett (1991) showed that offering a
community composting program could promote
the collection of recyclables in kerbside
collection schemes. Vencatasawmy et al.
(2000) found that households were more likely
to purchase environmental products if they
recycled glass but this relationship was not
found for those recycling paper. A reason for
these differing degrees of association is
suggested through the work of Tracy and
Oskamp (1984) who explored the interaction of
behaviours in some detail. Fifteen different
Understanding Recycling Behaviour 2
Waste Management
environmental behaviours were examined with
similar behaviours being placed into separate
groups recycling, energy conservation and so
on). Few correlations were found between
behaviours from different groups but strong
correlations were found amongst the
behaviours within each group, particularly
within the recycling group, and notably between
those recycling paper, glass and cans. It would
therefore appear that only very close
environmental behaviours might be correlated.
The results from Tracy and Oskamp's study are
backed up by two larger and more recent
studies. In Canada, Berger (1997) showed that
strong correlations were found between
individual recycling behaviours and also within
other groups of environmental behaviours.
Correlations between different groups of
behaviours were also seen in that study.
Recycling was found to have a strong
correlation with other environmentally-
conscious consumer behaviours (principally
composting, using one‟s own shopping bag and
buying recycled paper). Very much weaker
correlations were found between recycling and
water and energy conservation. Very few
correlations were found between any of the
other environmental groups. It was suggested
that this shows that environmentally
responsible behaviours are structured primarily
in terms of class-type activities.
The correlations between different waste
management behaviours were also studied by
Barr et al. (2001). Analyses using principal
components factor analysis showed that
recycling, reusing and reducing waste might be
classified as three fundamentally different
behaviours, with strong correlations amongst
individual behaviours within the three groups,
though not between groups.
Previous research at the University of Paisley
showed that newspaper and glass were the
main drivers of household recycling activities.
Almost nobody recycled magazines if they did
not recycle newspapers, and almost nobody
recycled cans if they did not recycle newspaper
or glass as well. Significantly fewer households
recycled magazines and cans as recycled
69

newspaper or glass. This was witnessed again
in Chesterfield and N.E. Derbyshire (section
4.6). Part of the difference, however, can be
ascribed to [some 20 - 40%] fewer households
consuming magazines and cans and
consequently having none of them to recycle.
6.1.2 New Analyses
The new analyses reported here are all based
on results from questionnaire surveys that had
been undertaken in the course of other
research by the University of Paisley. Three of
the surveys were carried out in areas with
6.2 Basic Correlations
6.2.1 Levels of Activity
In all areas, a large proportion of households
stated that they carried out the environmental
behaviours at least occasionally (Figures 6.1a-
e). Highest participations were found for
donating products to charity shops, with 90-
95% claiming that they did so at least
occasionally. A large proportion of households
(80-85%) also stated that they bought recycled
products, although only 15-25% said that they
did so frequently. In each area the lowest
participations were in can recycling and, where
asked, plastic recycling. This was partially
explained by some respondents having none of
those categories to recycle. However, facilities
to recycle cans and plastic were sparser than
those for paper and glass in all the surveyed
areas, making those the most difficult
environmental behaviours for residents to carry
out.
The numbers carrying out recycling varied
amongst all areas. The three areas with paper-
only kerbside schemes (Fylde, Lancaster and
Girvan) had high proportions of households
recycling paper though almost as many
households in Inverclyde, served only by paper
banks, recycled paper as did in Fylde.
Conversely for both glass and can recycling,
levels were highest in Inverclyde followed by
Blackwood, Lancaster, Fylde and Girvan. It
should be noted that these reported
percentages all refer to self-reported
behaviours. Actual recycling levels may be
somewhat lower (see section 9).
70
Understanding Recycling Behaviour 2
kerbside paper-only collections, three where
multi-material collections were in operation, and
three where there were only bring collections.
The three paper-only kerbside scheme areas
investigated were located in the boroughs of
Fylde and Lancaster in North West England
and Girvan in South West Scotland. The multi-
material collections were in Chesterfield, North
East Derbyshire and Renfrewshire. The other
areas investigated were the village of
Blackwood in South Lanarkshire and areas of
Inverclyde district and Clackmannanshire in
Scotland.
Recycling levels in Renfrewshire (not shown)
followed similar trends, with a high percentage
of paper recyclers, lower numbers of plastic
and can recyclers and relatively few glass
recyclers. Glass was not collected in the
kerbside scheme there. The picture in
Clackmannanshire was one of relatively low
levels of recycling of all materials. A detailed
breakdown of the materials recycled there is
given later in table 8.3, section 8.4.3.
Figures 6.1a-e Levels of Waste Minimisation
Activity in a number of districts.
Lancaster
Recycle paper
Recycle glass No
Recycle cans Yes
Recycle plastic
0% 20% 40% 60% 80% 100%
Girvan
Recycle paper
Recycle glass No
Recycle cans Yes
Recycle plastic
0% 50% 100%
 Understanding Recycling Behaviour 2

Blackwood

Recycle paper
Recycle glass
Recycle cans
Never
Buy recycled
Occasionally
Donate to charity
Frequently
Reuse waste
Decline packaging
Compost

0% 20% 40% 60% 80% 100%

Inverclyde

Recycle paper
Recycle glass
Recycle cans
Never
Buy recycled
Occasionally
Donate to charity
Frequently
Reuse waste
Decline packaging
Compost

0% 20% 40% 60% 80% 100%

Fylde

Recycle paper
Recycle glass
Recycle cans No recyclables
Recycle plastic Never
Donate to charity Sometimes
Reuse waste Usually

Decline packaging
Compost

0% 20% 40% 60% 80% 100%

6.2.2 Correlations between correlations overall were normally between

Behaviours
In all areas correlations were high between
individual recycling behaviours (paper, glass,
cans and plastic), and were particularly high in
the areas where no kerbside scheme was
running (tables 6.1 d-e). In the areas where
paper was collected by the kerbside scheme,
correlations were noticeably lower between
paper and cans and paper and plastic, although
they were still highly significant. The highest
paper and glass recycling. Although
correlations were not so high between other
behaviours around three quarters were
significant at 95% confidence and well over half
were significant at 99% confidence.
The results firstly demonstrate that paper
recyclers are very likely to be recyclers of glass
as well, even if paper is collected at kerbside
and glass recycling means traveling to a
recycling site. However, those recyclers

71
 Understanding Recycling Behaviour 2

serviced by a kerbside paper collection may Table 6.1a Correlation Coefficients between
recycle a smaller range of materials overall Waste Management Behaviours (Lancaster)
than those having to travel in order to recycle. Paper Glass Cans
This would support the notion of a generally
stronger recycling ethos occurring amongst the Glass 0.243**
participants of the „lower convenience‟ outlets. Cans 0.096* 0.385**
Thus, whilst a greater percentage of Plastic 0.072 0.286** 0.478**
households may recycle where a kerbside ** Significant at 99% confidence, * 95%
scheme is in place, their overall recycling confidence
activities might be somewhat more constrained
to only what the kerbside scheme collects.
Table 6.1b Correlation Coefficients between
Similar trends were seen amongst the Waste Management Behaviours (Girvan)
measured collection round yields in
Paper Glass Cans
Chesterfield and N.E. Derbyshire taken
together (table 6.1c). It is noted however that
Glass 0.197**
whilst the paper/ can and glass/ can
Cans 0.026 0.275**
correlations were significant in N.E.Derbyshire,
Plastic 0.105* 0.219** 0.473**
much lower correlations were found locally in
Chesterfield.

Generally lower correlations were found

between recycling activities and reduce Table 6.1c Correlations between Materials –
activities, though higher correlations were seen Collection Round Averages (Chesterfield
between the individual reduce/ reuse activities and N.E. Derbyshire combined)
themselves. This is perhaps seen most clearly Paper Glass Cans
in the analyses from Clackmannanshire (table Glass 0.786**
6.2) where a wider range of recycled materials Cans 0.460** 0.538**
and reduce/ reuse behaviours were Textiles 0.168 0.165 0.303
investigated. In the figure the relative depths of
shading indicate the relative strengths of the
associations.

Table 6.1d Correlation Coefficients between Waste Management Behaviours (Blackwood)

Paper Glass Cans Buy Rec. Charity Reuse Decline
donation packaging

Glass 0.534**
Cans 0.328** 0.397**
Buy Rec. 0.209* 0.301** 0.152
Charity 0.234** 0.230** 0.018 0.261**
Reuse 0.257** 0.363** 0.077 0.288** 0.346**
Decline 0.131 0.169* -0.055 0.295** 0.218* 0.329**
Compost 0.162 0.147 -0.112 -0.026 0.085 0.238** 0.140
** Significant at 99% confidence, * 95% confidence

Table 6.1e Correlation between Environmental Behaviours (Inverclyde)

Paper Glass Cans Buy Charity Reuse Decline
Rec. donation packaging

Glass 0.675**
Cans 0.378** 0.297**
Buy Rec. 0.194** 0.239** 0.332**
Charity 0.104 0.057 0.192** 0.227**
Reuse 0.253** 0.261** 0.185* 0.174* 0.297**
Decline 0.291** 0.259** 0.176* 0.177* 0.288** 0.395**
Compost 0.211** 0.265** 0.042 0.039 -0.034 0.198** 0.090

72
 Understanding Recycling Behaviour 2

Table 6.1f Correlation Coefficients between Waste Management Behaviours (Fylde)

Paper Glass Cans Plastic Charity Reuse Decline
donation packaging

Glass 0.408**
Cans 0.128* 0.335**
Plastic 0.154** 0.228** 0.336**
Charity 0.175** 0.228** 0.077 0.036
Reuse 0.176** 0.254** 0.147** 0.143** 0.271**
Decline 0.075 0.166** 0.169** 0.077 0.214** 0.392**
Compost 0.063 0.170** 0.144** 0.137** 0.036 0.196** 0.042
** Significant at 99% confidence, * 95% confidence

Table 6.2 Correlations amongst Recycling and Reduce/ reuse Behaviours Clackmannanshire)
G A F P J G A P B G H D R R O D
N B C T B C T M M J F F a W C C u f b p
Newspaper
Glass Bottles 5
Alu Cans 5 6
Food Tins 3 2 5
Plastic Bottles 3 2 4 7
Cardboard 3 4 4 5 4
Textiles 3 4 4 2 1 4
Magazines 9 5 5 4 3 3 3
Junk Mail 6 8 4 6 4 4 3 6
Glass Jars 4 8 5 3 2 4 3 4 3
Alu Foil 3 1 2 4 7 2 1 3 4 1
Plastic Film 4 2 5 3 3 1 4 3 4 2 3
Batteries 4 3 4 2 2 1 1 4 2 3 0 3
Garden Waste 4 3 5 5 6 3 2 4 6 3 6 5 2
Home Compost 2 2 3 2 1 2 2 2 3 2 1 1 0 3
Donate Charity 1 3 3 0 1 3 4 2 2 2 0 2 1 1 1
Reuse 2 3 2 0 1 3 3 2 2 3 2 3 2 3 2 3
Buy Refills 2 2 2 0 0 2 2 2 1 2 0 1 2 0 2 3 2
Own shop bag 1 2 3 3 3 4 2 1 3 3 1 2 1 2 3 4 3 3
Decline packing 2 2 2 2 1 3 2 2 2 3 1 2 2 1 2 5 5 4 5
Buy recycled 1 1 3 2 2 1 1 1 2 1 1 1 1 1 2 3 3 4 4 5

Note: The numbers 1-9 correspond to correlation coefficients from 0.1 to 0.9 respectively.

The strongest correlations, perhaps not
unexpectedly, are those between newspaper
magazine, and junk mail recycling and
between glass bottle and glass jar recycling,
though some other individual correlations are
also fairly high. Noticeably, those taking their
garden waste for recycling [centralised
composting] were strong recyclers of other
materials as well, the main exception being
textiles.
All the reduce/ reuse activities were relatively
strongly inter-correlated with the exception of
home composting. Reduce/ reuse activities
generally showed low correlations with all
recycling activities. The exception here was a
link between textile recycling and donating
waste items to charity. Again this might be
expected as many respondents take the two
activities to be synonymous.
However, whilst the correlations between
reduce/ reuse and recycling behaviours were
relatively weak individually, recyclers still
tended to practice more reduce/ reuse
activities than did non-recyclers (table 6.3).
These differences were statistically significant.

73

Table 6.3 Number of Reduce/ reuse activities practiced by Recyclers (Clackmannanshire)
No. of Reduce/ Reuse Activities Practised
0 1-2
Non-recycler 7 14
Recycler 1 10
The above analyses give good indications of
possible separations between different classes
of activities. Those separations will be explored
further in the next section using the statistical
technique of factor analysis.
Before that analysis, however, it is worthwhile
to look at one more correlation. That is to test
whether material recoveries are correlated
week-by-week as well as round-by-round in
kerbside collections. Tables 6.4 a-b show these
correlations for Chesterfield and N.E.
Derbyshire respectively. It is noticeable that, at
Table 6.4a Correlations between Materials –
Week by Week (N.E. Derbyshire)
Paper Glass Cans
Glass 0.289**
Cans 0.131 0.493**
Textiles 0.220* 0.145 0.277**
** Significant at 99% confidence, * 95%
6.3 Factor Analysis
6.3.1 Methodology
Factor analysis is a statistical technique that is
employed to look for similarities between
variables when undertaking multi-variable
comparisons. The main applications of factor
analysis are to reduce the number of
explanatory variables and to detect any
underlying structures in the relationships
between the observed variables. If there is a
high correlation between two items they can be
combined into a single factor. If we could then
define a variable that would approximate the
regression line in such a plot, that variable
would capture most of the "essence" of the two
items. In a sense we have reduced the two
variables to one factor. Extending this
argument to multiple variables, the technique
ends up with all the original variables being
associated to varying degrees with each of the
identified factors (termed factor loadings).
74
Understanding Recycling Behaviour 2
3-4 5-6
18 7
25 13
this level of resolution, the yields of most pairs
of materials are significantly and positively
correlated. That is, in individual rounds, in the
weeks when paper yields go up, the glass and
can yields tend to go up as well. However all
the correlation coefficients are relatively low
indicating considerable variations in the
strengths of these relationships between
rounds. In other words, whilst there is a general
trend linking the temporal ups and downs of all
the materials, the ratios of those ups and
downs can vary considerably.
Table 6.4b Correlations between Materials -
Week by Week (Chesterfield)
Paper Glass Cans
Glass 0.520**
Cans 0.278** 0.459**
Textiles 0.263** 0.356** 0.150
The decision of when to stop extracting factors
basically depends on when there is only very
little random variability left. The nature of this
decision is arbitrary; however, various
guidelines have been developed, the most
popular being to retain only factors with
eigenvalues greater than unity. The eigenvalue
essentially provides a measure of the marginal
increment in the „variability explained‟ that is
gained by adding a new factor.
When the number of factors to be retained are
found, we can then rotate the factor structure to
give the best fit using just these factors (termed
extracting „Principal Components‟). Various
rotational strategies have been proposed. The
goal of all of the strategies is to obtain a clear
pattern of loadings, that is, factors that are
somehow clearly marked by high loadings for
some variables and low loadings for others.

Typical rotational strategies are varimax (used
here), quartimax, and equamax.
The outcome is a pattern of the loadings of
each variable on the factors. Where different
variables all load strongly onto one factor, then
statistically those factors could share a
common explanation. However, it must be
recognised that the factors themselves may not
necessarily have any direct physical meaning
or basis. They should only be considered as
arbitrary constructs. However, often they are
amenable to some kind a posteriori labelling
that typifies the characteristics of the variables
most heavily loaded onto them.
6.3.2 Results
The results from the principal component factor
analysis were similar in the three areas tested
(Blackwood, Inverclyde and Fylde). The three
factors that explained the most variance were
extracted. These are listed in tables 6.5a-c.
The figures highlighted in bold denote the
principal factor associated with each behaviour
(i.e. the factor that it loads on to most heavily).
The results clearly distinguish two consistent
and distinct groups of activities: (i) recycling
and (ii) reduce and reuse. Composting would
not appear to be strongly associated with
reduce and reuse behaviours, however it is
equivocal whether it might be considered
alongside recycling or should be distinguished
as a separate group. The answer may depend
on location.
Table 6.5a Principal Component Factor Analysis of the Correlation Matrix (Fylde)
Factor 1 Factor 2
Paper 0.014
Glass 0.169
Cans 0.085
Plastic -0.032
Charity 0.555
Reuse 0.755
Decline 0.790
Compost 0.146
Variance explained 19.5%
Understanding Recycling Behaviour 2
It is also noted that paper and glass recycling
always loaded most heavily onto the same
factor, whereas can recycling was much more
inconsistent in its associations, though tended
to show its strongest links with „buy recycled‟.
This may suggest an “inside the supermarket
factor”, however that supposition must remain
unconfirmed.
Before leaving this section, it is worthwhile to
look briefly at the other way of grouping the
variables - cluster analysis. This analysis was
undertaken on the Clackmannanshire data set.
In cluster analysis, the parameters that are
most similar are progressively joined, either
pair-wise to form a new cluster or are added
into an existing cluster. The analyses can be
examined progressively at differing levels of
similarity between the variables. This is most
easily visualised in a dendrogram (figure 6.2).
The results clearly show: (i) the strong
newspaper/ magazine relationship and, with
less similarity, junk mail relationship, and (ii) the
glass jar/ glass bottle relationship. The results
also reveal similarities between food can,
aluminium foil and plastic bottle recycling and
between drink can recycling and carrier bag
recycling. Perhaps the latter reflects the twin
recycling points for cans and carrier bags that
are found in-store in some supermarkets.
Curiously, the glass cluster only links with the
paper cluster at a relatively low level of
similarity, even though glass bottle and
newspaper recycling are strongly inter-
correlated. Battery recycling is a definite outlier
that does not link strongly with any other
recycling behaviour.
Factor 3
-0.028 -0.848
-0.323 -0.717
-0.701 -0.200
-0.744 -0.154
0.173 -0.405
-0.188 -0.133
-0.129 0.052
-0.510 0.047
18.6% 18.5%
75
 Understanding Recycling Behaviour 2

Table 6.5b Principal Component Factor Analysis of the Correlation Matrix (Blackwood)
Factor 1 Factor 2 Factor 3

Paper 0.736 0.145 0.252

Glass 0.753 0.264 0.153
Cans 0.803 -0.106 -0.281
Buy Rec. 0.270 0.655 -0.271
Charity 0.095 0.642 0.051
Reuse 0.225 0.667 0.261
Decline -0.128 0.712 0.074
Compost 0.087 0.069 0.918

Variance explained 23.9% 23.8% 14.5%

6.5c. Principal Component Factor Analysis of the Correlation Matrix (Inverclyde)

Factor 1 Factor 2 Factor 3

Paper 0.691 -0.106 -0.482

Glass 0.744 -0.090 -0.422
Cans 0.177 -0.088 -0.756
Buy Rec. -0.035 -0.192 -0.702
Charity -0.266 -0.714 -0.278
Reuse 0.296 -0.746 -0.015
Decline 0.214 -0.708 -0.119
Compost 0.685 -0.140 0.265

Variance explained 21.7% 20.6% 20.5%

Figure 6.2 Similarities between Recycling Behaviours (Clackmannanshire)

Similarity

70.56

80.37

90.19

100.00
News
s Mags il
s Jmail ti n
FeTin t
PlaBot il
AlFoil Card
d GlaBot
t GlaJar
r xt
Textile AlCan
n PlaFilm
g tt
Battery
new g a maj bo
p l fo
a ar
c bo g jag te ca u ba ba
m al c

Variables

76

6.4 Correlations and Length of Experience
A possibility that needs to be considered is
whether gaining experience in practising one
activity can influence the participations in other
activities, perhaps through some kind of „knock-
on‟ effects. This section examines the evidence
for such effects. These investigations are
centred on the correlations between recycling
and home composting behaviours, and
between kerbside recycling and bring recycling
respectively.
Table 6.6 compares the recycling behaviours of
recently started composters (new recruits) of
less than 2 years standing with non composters
and with more experienced (prior) composters,
including those who have since ceased
composting.
The results show that the new composters
have similar recycling behaviours to the non-
composters, but have substantially different
(weaker) behaviours to the more experienced
composters. Those who composted in the past
but have since dropped out were found almost
as likely to be recyclers as those who continued
to compost.
The inverse effect would be whether
composting behaviours differed significantly
between new and more experienced recyclers.
This was investigated during surveys of a newly
introduced multi-material kerbside collection (9
months previously) in Renfrewshire, Scotland
(table 6.7). The new recyclers were those who
had taken up recycling for the first time (with
the kerbside scheme), whilst the prior recyclers
comprised those who were already recycling
(via bring systems) prior to the scheme
introduction. The results show that composting
levels were significantly higher amongst the
Table 6.6 % Recycling Behaviours amongst Categories of Composters
Non-composters New Recruits
Never recycle 19%
Sometimes recycle 37%
Frequently recycle 44%
Results combined for Blackwood, Inverclyde and Fylde
New vs. Prior, significantly different at 95% confidence
New vs. Non, not significantly different
Understanding Recycling Behaviour 2
prior recyclers than amongst the newer recruits
to recycling, though the newer recruits were in
turn more likely to practice home composting
than were those who were not recycling at all.
Also, after nine months recycling experience,
18% of the new recyclers also claimed to be
recycling their glass [via bottle banks] in
addition to the materials targeted in the
kerbside scheme, though the percentage of
glass recyclers in the group was still
significantly below that seen within the prior
recycling group (table 6.8). The complementary
statistic is whether the previous (bring)
recyclers took up recycling more materials
when the kerbside scheme became available.
The results are shown in table 6.9. Of those
who had recycled previously, a large proportion
had recycled newspaper but less had taken
their magazines and cans to the recycling
banks. Plastic bottles were very rarely recycled
previous to the scheme with local drop-off
facilities for plastic bottles being poor.
Overall, it would appear that there may be
much stronger correlations between waste
management behaviours for those who have
practiced those activities for some time
compared with those newly engaging in the
activities. It would appear that there is no
immediate knock-on effect between different
behaviours, though active practice of other pro-
environmental behaviours becomes evident
amongst those with nine months to two years
experience. If this is a „knock-on‟ effect, the
evidence would suggest that it may take a
rather longer timeframe in order for it to
develop fully.
Prior Lapsed
Composters Composters
18% 5% 11%
34% 25% 25%
48% 70% 64%
77
 Understanding Recycling Behaviour 2

Table 6.7 % Composting Garden Waste amongst Categories of Recyclers

Composting Non-recyclers New recruits Prior recyclers

Composted
No 92% 81% 67%
Yes 8% 19% 33%

Table 6.8 % Bring Scheme Glass Recycling Table 6.9 Previous Behaviours of Prior
by Kerbside Recyclers Recyclers
Recycle New recruits Prior % Recycling
glass recyclers
Newspaper 79%
No 81% 49% Magazines 63%
Yes 19% 51% Plastic 27%
Cans 64%

6.5 Behaviours and Demographics

6.5.1 Significant Relationships
The analyses from Chesterfield and North East
Derbyshire (section 4.5) suggested that can
and textile recyclers may be drawn more from
the poorer, younger households than compared
with paper and glass recyclers. Whether that
hypothesis is borne out more widely in practice
is now tested. Those tests are based on
surveys carried out in Fylde, Blackwood and
Inverclyde, Girvan, Renfrewshire, Kilmarnock
and Clackmannanshire (tables 6.10 - 6.12).
The results show that newspaper and glass
recycling are sometimes correlated with
demographics, whilst can recycling rarely did.
The significant factors that were found
generally conformed to traditional wisdom that
it was the older residents in the more affluent
housing who were the better newspaper and
glass recyclers. Young adults without children
and families with young children were often
amongst the poorer recyclers. Fewer residents
of the older terraced housing and flats tended
to recycle. Single person households were also
relatively poor recyclers as were large
households of five or more persons. Social
class never proved to be a significant factor.

Table 6. 10 Significance Test of Behaviour versus House Type

Girvan Fylde Black- Kilmar- Renfrew- Clack- Inver-
wood# nock shire mannan clyde

Newspams X *** - - *
Newspaper X **
Magazines X -
Glass *** ** X - X * *
Plastic - X X -
Cans X X X X - X
Donate X X X -
Reuse X X X X
Decline X X X X
Compost *** X *** X -
Buy rec X X **
*** Significant at 99% confidence; ** 95%; * 90%; X = Not significant
- relationship tested but insufficient data for statistical analysis
# Only semi-detached and small detached housing in the sample

78
 Understanding Recycling Behaviour 2

Table 6. 11 Significance Test of Behaviour versus Family Life Stage

Girvan Fylde Black- Kilmar- Renfrew- Clack- Inver-
wood nock shire mannan clyde

Newspams *** *** X X X

Newspaper X -
Magazines X -
Glass X ** X X *** - X
Plastic - X - * -
Cans X X X X - X
Donate ** - - -
Reuse X - - -
Decline X X - X
Compost X X *** X X
Buy rec - - -
*** Significant at 99% confidence; ** 95%; * 90%; X = Not significant

Table 6. 12 Significance Test of Behaviour versus Household Size

Girvan Fylde Black- Kilmar- Renfrew- Clack- Inver-
wood nock shire mannan clyde

Newspams * ** *** X -
Newspaper *** -
Magazines *** -
Glass X X X X ** - -
Plastic - ** *** -
Cans X * X X - -
Donate X - - -
Reuse X - ** -
Decline * - ** -
Compost X X *** X X
Buy rec - X -
*** Significant at 99% confidence; ** 95%; * 90%; X = Not significant

Home composting generally showed a strong 6.5.2 Relationships between

correlation with housing type, not unexpectedly
being weak amongst flat dwellers and those in
older terraced housing. Young adults and
single person households were also relatively
poorly represented amongst home composters.
Reduce/reuse behaviours only appeared
significantly associated with demographics in
one out of the four areas in which they were
tested. In that area, significantly weaker
behaviours towards reusing waste materials
and in declining excess packaging were found
in the larger households of four or more
persons. The strongest proponents of the
tabulated reduce/reuse behaviours were two
person households. That group were most
likely to use their own bag for shopping, though
were no more likely than any other group to
seek out refillable containers.
Materials
Whilst the previous analyses showed some
differences in the demographic associations of
the different behaviours, other behaviours
appeared quite similar amongst all
demographic groups. For some behaviours
(e.g. donating to charity chops), it was because
nearly everyone was doing it, whilst for others it
was because relatively few from any group
were doing it (e.g. can recycling). Participations
in reduce/ reuse activities were less polarised
than the participations in recycling activities.
Many more respondents claimed to reduce or
reuse at least occasionally, though not
necessarily at every opportunity. Those
behaviours are illustrated in figures 6.3 a-e.
They refer to the housing type differences in
the Fylde sample. The results for other
demographic factors and for the other localities
all show broadly similar features.

79
 Understanding Recycling Behaviour 2

Figure 6.3 a-e Levels of Activity in Different

Household Waste Management Behaviours Decline packaging
(by Housing Type in Fylde)
Terraced/flat
Recycle paper
Semi

Terraced/flat Detached

Semi 0% 50% 100%

Detached
Never Sometimes Usually
0% 50% 100%

Donate to charity shops

Never Sometimes Usually

Terraced/flat
Recycle glass
Semi

Terraced/flat Detached

Semi
0% 50% 100%
Detached
Never Sometimes Usually
0% 50% 100%

Never Sometimes Usually The final picture is provided by comparing the

behaviours side by side. Figures 6.4a-c portray
these comparisons for the Renfrewshire
Recycle cans sample.

Terraced/flat The results show that, by and large, the

differences amongst the demographic fractions
Semi are broadly maintained across materials.
Perhaps there is some evidence of a slight
Detached
convergence of behaviours around can
0% 50% 100%
recycling (which would fit in with the evidence
from Chesterfield and N.E. Derbyshire).
Never Sometimes Usually

Figure 6.4 a-c Recycling Levels for Different Materials (Renfrewshire)

Acorn

100
Note: lines
90 are drawn in
80 the figure
70 simply to aid
% Recyclers

A visualisation.
60
B They do not
50 imply any
D
40 continuous
EF
30 relationship.
20
10
0
News Mags Plastic Cans Glass Compost

80
 Understanding Recycling Behaviour 2

Family life stage

90
80
70
YA
% Recyclers

60
FYC
50
FOC
40
MM
30
RET
20
10
0
New s Mags Plastic Cans Glass Compost

Household size

100
90
80
70 1
% Recyclers

60 2
50 3
40 4
30 5+
20
10
0
New s Mags Plastic Cans Glass Compost

6.5.3 Recycling, Reduce / Reuse
and Demographics
The factor analyses and correlation analyses
presented earlier in this section indicate that
whilst recycling behaviours and reduce/ reuse
behaviours might be weakly correlated, they
can show significant differences as well. That is
they may have significantly different roots and,
as such, could potentially be associated with
different segments of the community. A test is
now made to ascertain whether recyclers and
reduce/ reusers might be distinguishable as
separate groups demographically
The test looks at the four extremes of
behaviour: high reduce/ low recycle, high
reduce/ high recycle, low reduce/ low recycle,
and low reduce/ high recycle. The results
shown in tables 6.13a-c highlight where specific
demographic factions are substantially over or
under represented amongst the groups.
The findings would appear to be quite
inconclusive for three of the sectors. Different
factors take on different importances for each
community. The only sector showing consistent
behaviours is the low reduce/ low recycle
sector where larger households, younger
residents and lower social classes are
generally over-represented with the older
residents and two to three person households
being under-represented.

81
 Understanding Recycling Behaviour 2

Table 6.13a Demographic Profiles of Table 6.13c Demographic Profiles of

Extremes of Behaviour – Housing Type Extremes of Behaviour – Household Size
Reduce high Reduce high Reduce high Reduce high
Recycle high Recycle low Recycle high Recycle low
F BI C F BI C F BI C F BI C

Det  - -  1 -   -  
Semi -  -  2 - -  - - 
Ter/fl     3 -   -  -
B*   4     - -
C* - - 5+ -   -  
F*  -
Reduce low Reduce low
Reduce low Reduce low Recycle high Recycle low
Recycle high Recycle low F BI C F BI C
F BI C F BI C
1     
Det  -  -  2  -  - 
Semi    - - 3 -     
Ter/fl    - - 4     - 
B -  5+ -     
C  
F  
F = Fylde, BI = Blackwood + Inverclyde, C =
Clackmannanshire
,  under-represented, - representative
, over-represented
* Acorn categories

Table 6.13b Demographic Profiles of

Extremes of Behaviour – Family Life Stage
Reduce high Reduce high
Recycle high Recycle low
F BI C F BI C

YA -     
FYC  -   - -
FOC -    - 
MM -  -   
RET  -   - -

Reduce low Reduce low

Recycle high Recycle low
F BI C F BI C

YA -     -
FYC  - -   -
FOC    - - 
MM - -  -  
RET  -   - 

82

6.6 Discussion and Conclusions
The results generally show that those who
practice one type of household waste
minimisation behaviour are more likely to
practice other waste minimisation behaviours
as well. However such correlations can also
depend as much on local factors, notably the
extent of local recycling provision (and garden
size with respect to home composting).
However whilst a recycler may be more likely to
practice reduce/ reuse activities than a non-
recycler, the research shows that the two
classes of behaviour may have significantly
different roots and, as such, may be driven by
fundamentally different motivations. Research
into how specific attitudes might motivate waste
minimization behaviours is presented in section
8 of this report.
Much attention has been given to researching
the role of demographics in differentiating
between different behaviours. The results of
that research have been inconclusive. Whilst
older, better-off residents are often over-
represented amongst recyclers in a community,
with younger residents and single person
households under-represented, many of the
differences are found to be not significant
statistically. Even fewer dependences have
been found that might link reduce/ reuse
behaviours to demographics. Neither is there
any clear demographic feature that can
distinguish a strong source reducer and reuser
from a strong recycler.
The correlations between different pro-
environmental behaviours tend to weaken as
the activities become more disparate, so
stronger correlations are shown between [say]
recycling paper and recycling glass than
between recycling paper and home composting
or recycling glass and source reduction. The
correlations between different behaviours can
also differ amongst the different segments of
the population engaged in those behaviours,
with those with a longer experience of one or
more of the behaviours being more likely to
practice the other behaviours as well.
Understanding Recycling Behaviour 2
Indications are that such correlations may only
develop fully over a relatively long time frame.
Immediate „knock-on‟ effects, whilst positive,
may be small. Separate pro-active interventions
may be needed to develop each individual
behaviour within a shorter time frame.
The research points strongly to the need for
specific promotions that are targeted closely to
specific individual activities. That is, a general
awareness campaign to encourage reduce,
reuse and recycle may need fundamentally
different and separate messages for the
reduce/ reuse component and the recycling
component. The current research was not
designed to establish what those messages
might be, and indeed there is still a dearth of
scientific published analyses on the efficacy of
promotional campaigns in general. The national
waste awareness campaigns now provide us
with exceptional opportunities to undertake
such analyses and to delineate the
fundamental attitudinal and other factors that
are essential to increased and sustained public
participation in pro-environmental household
waste management activities.
Overall, whilst there was a large degree of
inter-correlation between many waste
management behaviours, which suggests that
there may be a common antecedent factor, i.e.
some sort of underlying attitude, the
correlations were not similar between all pairs
of behaviours. Correlations were higher
between different recycling behaviours than
between recycling and composting or recycling
and reduce / reuse. However, the results from
principal components factor analysis suggested
that recycling and non-recycling waste
management behaviours may have
fundamentally different roots. Whilst these two
sets of results would appear conflicting, they
could be reconciled through assuming that
there is a two-tier system in effect, a basic pro-
environmental waste management component
and an activity specific component.
83

7. Time Series Analyses
7.1 Introduction
Whilst waste management behaviours can
normally be regarded as being intrinsically
stable, events do happen that can switch on
and switch off individual behaviours. Adverse
experiences with a recycling scheme can lead
to the cessation of use of that scheme. Positive
prompts, on the other hand, might encourage
new users. The aggregated effects from all
individuals will show up in the performance
monitoring data. As collection tonnages are the
only data normally monitored in recycling
programmes, we need to look for the effects in
that weight data.
Whilst it is well known that collection tonnages
go up and down from collection to collection,
the magnitude of those variations has never
been fully explained (Tucker, 1997–98). Part
may be due to some recyclers not participating
every collection. There are also bank holidays
weeks, which normally give rise to a low
followed by a high in collection yield. Waste
arisings can also fluctuate with time (Jones et
al., 1998b)
The questions posed here are “What can
collection round time series tell us about
recycling behaviours?” and “Can that
information be used to aid and enhance
recycling scheme management?”
This section now examines the changes in
district-wide performances and individual
collection round yields over time. The main
results reported in the section are based on
time series supplied by the Boroughs of Wyre,
Fylde and Burnley in Lancashire. The Fylde
7.2 Between District Comparisons
The most remarkable feature of these
comparisons is that the paper yield from all the
districts investigated appeared to be
significantly correlated (figure 7.1). When one
district recorded a high, others did likewise, and
similarly for the lows. The coherences seemed
to be maintained across the country and were
still evident even between widely separated
85
Understanding Recycling Behaviour 2
and Wyre analyses were able to consider
discrete time horizons at one month [or 4 week]
intervals from the period April 1999 to March
2002 and beyond. The Burnley data set ran for
20 months. Further data was made available by
Chesterfield, North East Derbyshire and North
Warwickshire districts for their paper only
schemes. At the time of writing, the multi-
material data sets from those districts sets are
not long enough for statistical interpretation.
The data sets that were used in the analyses
were „as received‟ from the local authority
records. Overall, all the data sets were quite
„noisy‟, with considerable month to month
variations. Some of that noise could be
attributable to data errors (see section 11.2.3).
The interpretation of any significant temporal
trends or discrete events from such raw data
can be subject to much uncertainty as many
different factors may be in play. To help
facilitate a better overall interpretation, it
becomes necessary to try to deconvolute the
individual effects of some of the major factors.
This was achieved through the statistical
technique of decomposition. In the analyses
reported here, the time series were
decomposed into three factors: (i) an overall
trend, (ii) seasonal effects, and (iii) the
residuals. The analyses sought to establish
how coherent each of those factors was across
different rounds within an authority, and as far
as we were able, across different authorities as
well. The final part of the analysis sought to
investigate reasons for identified non-
coherences.
districts, though the degree of correlation did
appear to drop away with increasing distance.
The correlation coefficient between Wyre and
South Ribble was 0.54, between Chesterfield
and N.E. Derbyshire 0.82 and 0.55 and 0.46
between North Warwickshire and Chesterfield
and N.E. Derbyshire respectively.
 Understanding Recycling Behaviour 2

Figure 7.1 Correlations between Districts Collection Round Yields

250

200
Tonnage

150 Chesterfield
N.E.Derbyshire
100 N.Warwickshire

50

0 10

13

16

19

22

25
1

4-week period

300

250

200
T o n n ag e

150

100

50
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33

M o n th

W yre S .Ribble

The observation that the seasonal variations in
kerbside paper recovery can be quite similar
across many district authorities could have
important consequences for scheme
management. If, as it appears, there some kind
of national factor affecting the month by month
variations in yield, then that national factor
could turn out to be quite predictable. Clearly
such a factor will be influenced significantly by
collection logistics (especially the impact of
bank holidays), but will also reflect coherences
of life style (summer holiday periods) and any
seasonal variations in [news]paper weights as
well.
Any week by week commonalities within and
between rounds can aid in forecasting future
performances. If certain aspects of round
performance match, or if those aspects are
repeated with the same seasonal variations
year-on-year, then we have the substantive
basis for the development of a predictive tool
that can foretell future performances. That also
provides us with a significant performance
monitoring tool whereby actuals can be
compared with forecasts – and alarms could
then by triggered when those two statistics
diverge.
To test the feasibility of being able to make
sensible forward predictions, it is first

86
 Understanding Recycling Behaviour 2

necessary to extract the shorter-term seasonal
components from the longer-term trends. That
was achieved through decomposition analysis.
Figure 7.2 illustrates the use of the technique
on the Wyre data set.
advertising (in the nationals at least), and
paginations have consequently fallen a little.
So, newspaper yield might be expected to be
lower in 2001/02 than in 2000/01 for a stable
recycling population.

The curves show the residuals from removing
the constant linear trend, removing the monthly
seasonal variations, and from removing both.
It is noticed, however, that the assumption that
there may be a linear year-on-year average
growth in tonnage collected may not be strictly
valid for this data set. Inspection of the raw
data shows that the growth experienced during
1999/00 and 2000/01 may now be levelling off
(from month 20 onwards in the diagram). This
would imply that any recently experienced
marginal increases in participation and capture
with time might not be sustainable into the
future. Alternatively, it could simply mean that
there was less paper in the system in 2002
than in 2001. In the late 1990s, newspaper and
magazine paper consumption showed a year-
on-year rise of some 2 to 3%. However
2001/02 saw a downturn in newspaper
The overall decomposition fit for Wyre is shown
in figure 7.3. This fit is built up from the
computed long-term trend and computed
monthly seasonal indices averaged across
years. Results show that the decomposition fit
is reasonably close, indicating that much the
same seasonal variation may have happened
over each of the three years. Figure 7.4a plots
the monthly seasonal indices. Comparable
indices from South Ribble show much similarity
for most of the year, though do record some
differences over the late summer period.
To enable the forward predictions, the trend
and seasonal indices were computed from the
first two years data only. Those statistics were
then extrapolated to provide the monthly
forecasts for year 3. The results are shown in
figure 7.5a. Corresponding predictions for Fylde
and South Ribble are given in figures 7.5 b-c.

Figure 7.2 Decomposition of the Wyre Collection Yield Time Series

Component Analysis for Wyre

Original Data Detrended Data

250 50

200
0

150
-50

0 10 20 30 40 0 10 20 30 40

Seasonally Adjusted Data Seasonally Adj. and Detrended Data

230
220 30
210 20
200 10
190 0
180
170 -10
160 -20
150 -30
140 -40
0 10 20 30 40 0 10 20 30 40

87
 Understanding Recycling Behaviour 2

Figure 7.3 Decomposition Fit for Wyre

Figure 7.4 Seasonal Indices for Wyre and South Ribble

Wyre

Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar
40
20
0
-20
-40
-60

South Ribble

40
20
0
-20
-40
-60

88
 Understanding Recycling Behaviour 2

Figure 7.5a A Year‟s Forward Prediction for Wyre based on the Previous 2 Years‟ Data

300
250
200
150
100
1 2 3 4 5 6 7 8 9 10 11 12

Actual Forecast

Figure 7.5b Fit Between Forecast and Actual Tonnages (Fylde)

150

100

50
1 2 3 4 5 6 7 8 9 10 11 12 13

Actual Forecast

Figure 7.5c Fit Between Forecast and Actual Tonnages (South Ribble)

250

200

150

100
1 2 3 4 5 6 7 8 9 10

Actual Forecast

The correlations between the forecasts and the
actuals were all high and highly statistically
significant. Correlation coefficients were 0.806,
for Wyre, 0.751 for Fylde and 0.725 for South
Ribble.
It must be noted that the forecasts were made
using a simple linear trend model with seasonal
decomposition. The assumption that the year-
on-year „trend‟ is actually linear may be quite
questionable. More sophisticated forecasting
models do not rely on that assumption. Models
such as ARIMA (Autoregressive integrated
moving average) could potentially improve the
forward projections.

89
 Understanding Recycling Behaviour 2

Basically, the results add support to the notion way, “You will get next year what you got last
that “The past is the key to the future”. Current year” … unless, of course, something happens.
performance may indeed provide a good The next stage is to examine what the
predictor of future performance. Or put another happenings might be.

7.3 Intra-district Variations

The relative scales of the year-on-year trend
are shown for Wyre in figure 7.6a. The yields 7

Number of rounds
from nine out of the ten collection rounds 6
increased with time whilst one reduced. The 5
growth distribution by collection round for Fylde 4
is shown in figure 7.6b. Here, 3 out of 20 3
rounds showed reducing yields with time with 2
the other 17 increasing. Many of the changes 1
0
were quite significant at around 10% per
annum, with the highest year-on-year changes <-20 -20 to -10 to 0-10 10-20 >20
-10 0
reaching up to 20% per annum.
Year-on-year % change

Figure 7.6a Summary of Trends in Year-on-

year Paper Collected in Wyre (10 Rounds)
In Burnley (figure 7.6c), there was more
5 variation amongst the individual rounds with
more rounds showing [positive and negative]
Number of rounds

4 changes in excess of 20%. Whilst it is not

possible to identify reasons for the contrasting
3 behaviours, it is noted that both Fylde and
Wyre operate mature schemes that have been
2
running for over 5 years, whilst the scheme in
1 Burnley was a much more recent introduction.
It could then be hypothesised that performance
0 adjustments at round-level are more likely to
<0 0-5 5-10 > 10 happen in the early stages of a schemes
history, with greater stability setting in with
Year-on-year % change maturity.

Overall, whilst relatively stable and uniform

behaviours appear to be seen at district level,
Figure 7.6b Summary of Trends in Year-on- there would appear to be many more
year Paper Collected in Fylde (20 Rounds) systematic variations amongst the individual
collection rounds inside those districts. The
question is “Why?” Are there any features that
7
might distinguish the most rapidly improving
Number of rounds

6
rounds from rounds with deteriorating
5
performances? Those rounds may have
4
experienced different historic event chains or
3 alternatively they could have just reacted
2 differently to a common set of events.
1
0 It is probably impossible now to identify any
<0 0-5 5 - 10 10 - 15 > 15 events in retrospect. What can be done,
Year-on-year % change however, is to identify whether there are any
differences in the round profiles, i.e. in their
demographic make up. Comparing the
demographic profiles of the strongest improving
Figure 7.6c Summary of Trends in Year-on- rounds and the weakest improving rounds will
year Paper Collected in Burnley (20 Rounds) furnish this information.

90
 Understanding Recycling Behaviour 2

In Wyre, the largest distinguishing feature of Figure 7.7a Demographic Profiles of

the poorest performers was the higher than Rounds with the Lowest Annual Growth in
average proportion of semi-detached houses Paper Yield (Burnley)
and owner occupations. The best performers
were in many ways quite similar
demographically. Slightly elevated proportions rented -
of detached housing, smaller household sizes other
and more older residents were common rented -
amongst these rounds. However, no definite privately
conclusions could be drawn.
ow ner
occupied
Much more pronounced effects were seen in
Burnley. The demographic profiles of the four 2 cars
rounds showing the highest average
improvements over the last 18 months are
shown in figure 7.7b. Figure 7.7a shows the 1 car
corresponding profiles for the four rounds with
the largest decreases in yield over that period. No car
The demographic profiles are plotted in the
form of differences from the district mean. The RET
results show that the most deteriorating rounds
tend to have higher proportions of detached
MM/CLH
and, possibly, semi-detached housing, high car
ownership, and higher than average levels of
owner occupation. In contrast the highest F/OC
improving rounds have lower than average
proportions of detached housing, higher than F/YC
average proportions of purpose-built flats, more
single person households, low car ownership YA/NC
and high levels of public sector rented
accommodation. These trends greatly amplify
what was seen in Wyre, with the yield from the 5+ pers
traditional recycling base declining with time
whilst that from the less traditional recyclers 4 pers
improving. Overall this shows that there may
have been a progressive convergence of paper 3 pers
yields with time over the last 18 months, with
demographic differences becoming eroded with
2 pers
time.

This effect is also seen through the much 1 pers

tighter distribution of yields across all the
collection rounds during the more recent period Flats b
(figure 7.8b). This effect was not apparent in
the Wyre data (figure 7.8a). Flats a

Terr

Semi

Det

-0.5 0 0.5

YA/NC = Young adults, no children, F/YC = Families

with young children, F/OC = families with older
children, MM/CLH = more mature/ children left
home, RET= retired. Flats a = purpose built flats,
Flats b = converted flats

91
 Understanding Recycling Behaviour 2

Figure 7.7b Demographic Profiles of Figure 7.8a Yield Distribution across collections
Rounds with the Highest Annual Growth in rounds for 3 successive years (Wyre)
Paper Yield (Burnley)

rented - 2.5
other

rented -
privately

Kg/hh/w
ow ner
1.5
occupied

2 cars

1 car
0.5

0 1 2
No car
Year

RET

MM/CLH Figure 7.8b Yield Distribution across collections

rounds for 2 successive years (Burnley)
F/OC
1.4
F/YC 1.3

1.2
YA/NC 1.1
Kg/hh/w

1.0

5+ pers 0.9

0.8

4 pers 0.7
0.6

3 pers 0.5

1 2
2 pers Year

1 pers

Flats b Figure 7.9 1999-2001 Paper Collection

Times Series Chesterfield (Whole District)
Flats a
Chesterfield
Terr
0.95

Semi 0.9
0.85
0.8
Det
Kg/hh/w

0.75

-0.5 0 0.5 0.7

0.65
0.6
0.55
0.5
Aug/Sep01
May/Jun01
Sep/Oct00

Dec-00
Jan-00

Mar-01
Oct-99

Apr-00

Jun/Jul00

92

In Chesterfield, most of the old paper-only
collection rounds experienced declines in yield
over their last years of their operation (figure
7.9). During that period many of the rounds had
been converted from fortnightly to 4-weekly
collections.
Stepwise regression was used to test whether
year 2000/01 recyclers were different
demographically to the recyclers of 1999/00
(see section 4.5 for explanation of the
methodology). The results show relatively little
change in the significant contributory groups
over the two years, though there were
(perhaps) marginally less contributions from the
„poorer group‟, Acorn F, and also (not shown)
„no car‟ and rented properties in the latter year.
However, overall, from the data, the decline in
contributions in Chesterfield may have been
quite pervasive amongst all demographic
categories.
7.4 Summary and Conclusions
The results show that similar temporal trends in
paper yields may be occurring amongst
individual districts, with quite similar seasonal
indices in yield, but with some variations in
year-on-year trends. Future yields should be
reasonably predictable from the past
performances given the status quo in recycling
provision and scheme promotion.
The temporal variabilities will be linked to
variations in waste arisings, individual recycling
frequencies, and any changes in participation
levels. For effective performance monitoring
and good scheme management, it is essential
to spot any irreversible changes in participation
as early as possible. The difficulties are in
extracting that signal from an intrinsically „noisy‟
record. The analyses presented here show that
monthly seasonal fluctuations might be quite
predictable from year to year. As such, monthly
records can be de-seasonalised allowing their
early evaluation.
93
Understanding Recycling Behaviour 2
1999/00 Regression (Acorn classes)
Kg/hh/w = 0.0125  A + 0.0141  B +
0.0079  D + 0.0043  F
2000/01 Regression (Acorn classes)
Kg/hh/w = 0.0140  A + 0.0132  B +
0.0101  D
In these regressions, a collection frequency
term (4-weekly as opposed to 2-weekly) was
introduced into the pool of possible predictor
variables. Collection frequency did not appear
as a significant term in the regressions. That is
there was no significant difference in yield
declines between the monthly and fortnightly
collection regimes. The changes in collection
frequency did not exacerbate the situation.
The yearly changes in yield tend to be relatively
small (<5%) at district level, though much larger
swings can be seen within the individual
collections rounds of the districts. It is
conjectured that more variation could happen in
relatively immature schemes before they settle
down into more stable behavioural patterns.
Whilst the analyses of section 4 and 5 of this
report showed that traditional demographic
indicators may offer some explanation of the
performance differences amongst individual
rounds, the time series analyses provide some
evidence that those differentials may be
changing. In Burnley, there is evidence that the
differentials are being eroded; however in other
districts evidence was far less clear.
Finally, it is stressed that any conclusions
reached here must be considered quite
tentative due to the small number of data sets
upon which they were based, and because of
possible limitations and uncertainties in those
data.

8. Attitudes, Demographics and Behaviour
8.1 Literature Review
8.1.1 Attitudes
Much of the evidence supporting waste
management policy decisions has been derived
from householder surveys. Many questions
have been asked, but how valuable are the
answers? Most surveys have tried to explore
the reasons why some people do not recycle.
The answers have come out quite consistently
amongst surveys: “inconvenience”, “time”,
“poor information”, “not having enough waste”,
or “no recycling container” appear quite
ubiquitous. The surveys often then go on to
ask: “What would make you recycle more?” or
“What would make you start to recycle?” Three
main classes of answer are normally
volunteered: “Give me a kerbside collection” [if
there isn‟t one], “Improve the kerbside
collection” [if there is one], or “Nothing”. Yet, if
the respondents‟ wishes are subsequently
honoured by the programme provider, many
households do not respond and still do not
recycle. Are there then some more fundamental
attitudes underpinning their recycling
behaviours? If so, what are those attitudes?
And how might they be changed for the better?
Waste management research has been
seeking such answers for two decades. Much
of that research has been reviewed in volume 1
of Understanding Recycling Behaviour (Tucker,
2001). A short summary and update of that
review is presented here.
Attitude researchers would argue that lasting
changes in behaviour can not be accomplished
without first changing the attitudes towards that
behaviour, or even changing the more
fundamental beliefs and values that underlie
those attitudes.
General pro-environmental attitudes are
normally found to exert little or no influence on
recycling behaviours (e.g. Vining and Ebreo,
1992; Oskamp et al., 1991; Shrum et al., 1995).
It is considered that attitudes must be more
specific towards recycling in order to influence
behaviour (Shrum et al., 1995; Oskamp et al.,
1995). Specific pro-environmental reasons for
recycling, volunteered by recyclers, include
resource conservation, litter reduction, energy
conservation, reduction of landfill space (Vining
and Ebreo, 1989); satisfaction from saving
Understanding Recycling Behaviour 2
natural resources, reducing litter, decreasing
landfills, saving energy (Hopper and Nielsen,
1991); saving landfill space, reducing pollution,
saving natural resources, reducing litter and
saving energy (Kilner, 1992); or reducing
pollution, avoiding waste, and improving the
future (Howenstine, 1993).
Negative attitudes about inconvenience and
perceived effectiveness appear to demonstrate
much clearer links with behaviour than do
positive attitudes (see Shultz et al., 1995;
Taylor and Todd, 1995). The reason why direct
links are much weaker or absent with positive
attitudes have been discussed by a number of
authors. Guagnano et al. (1995), for example,
suggest that the attitude-behaviour link may
only operate in certain social contexts. Other
authors consider that the link is moderated by
the presence or absence of other variables.
Several authors consider that the mediating
variable „intention to behave‟ provides the
crucial link between attitude and behaviour.
Attitudes directly affect the intention to behave.
External factors may then prevent a positive
intention from being converted into behaviour.
This consideration follows directly from the
„Theory of Planned Behaviour‟ proposed by
Azjen and Fishbein (1977, 1985).
8.1.2 Attitude and Behavioural
Models
The theory of planned behaviour considers that
attitudes, subjective norms and perceived
behavioural control (that is, belief that one has
full control over one‟s actions) are the three
direct antecedents of an intention to behave.
Perceived behavioural control is also
hypothesised to have an additional direct
influence on behaviour (figure 8.1). Perceived
behavioural control is itself made up of two
components: self-efficacy and facilitating
conditions.
Several research studies into recycling
behaviours have taken the theory of planned
behaviour as their starting point (e.g.
Goldenhar and Connell, 1992-93; Jones, 1989;
Boldero, 1995; Taylor and Todd, 1995; Pieters,
1989). In applying the theory, however, Boldero
(1995) still found that negative attitudes rather
95

than positive attitudes appeared to be the main
antecedents of behaviour. Other workers,
however, have confirmed that there are strong
links between attitude and intention, e.g. Jones
(1989) and Taylor and Todd (1995).
Figure 8.1. Simplified Schematic of the
Theory of Planned Behaviour
Self-
Facilitating efficacy
Attitudes
conditions
Perceived
Subjective
behavioural
norm
control
Intention to
behave
Behaviour
Another school of thought considers that
recycling is more fundamentally an altruistic
behaviour. Effort or cost is borne by the
individual whilst the benefits are gained by
society at large. The altruism model is strongly
advocated in the studies of Nielsen and
Ellington (1983), Granzin and Olsen (1991),
Vining, Linn and Burdge (1992), and others.
Altruism, however, can never be fully
decoupled from other motivational factors.
Altruism is usually founded on strong
environmental beliefs, or could alternatively
result from moral or religious principles (Hallin,
1995). Hopper and Nielsen (1991) also argue
that altruistic behaviour is by definition
normative behaviour with norms being
developed through social interaction.
The Schwartz (1977) altruism model (figure
8.2) when applied to recycling behaviour
highlights the concept of a personal norm that
„recycling is the right thing to do‟. The personal
norm is a highly internalised moral attitude that
governs individual behaviours. Social norms,
on the other hand, relate to the values and
beliefs held to be significant by others, and the
ensuing social actions. For behaviour to
96
Understanding Recycling Behaviour 2
happen, the social norm „to recycle‟ needs to
be adopted as a personal norm. This adoption
will only take place when the personal norm is
both relevant and applicable to the situation
(Schwartz, 1977). For this to happen the
individual must be aware of the consequences
of action and ascribe a personal responsibility
for those consequences. This implies that basic
knowledge and the rationale for recycling [or
the consequences of not recycling] must be in
place and that the individual must not only
perceive that his/her contribution is effective but
must also believe that his/her non-participation
will have negative consequences for others.
Figure 8.2. Schematic of Altruism Model
Social norm
Personal norm
Ascription of
responsibility
Awareness of
consequences
Behaviour
The literature contains several references to
whether the above antecedents to altruistic
behaviour exist within the recycler. Vining and
Ebreo (1992), for example, showed that a pro-
recycling personal norm was more strongly
endorsed by recyclers than non-recyclers and
that recyclers saw greater perceived
effectiveness from recycling. However, they
found that awareness of consequences was
equally high amongst recyclers and non-
recyclers alike, and that awareness of
consequences and social norms could have
direct effects on behaviour. Oskamp et al.
(1991) found no significant relationships
between personal efficacy or the effectiveness
of recycling with recycling behaviours, whilst
Gamba and Oskamp (1984) and Vining and
Ebreo (1989) found that recyclers did hold
stronger beliefs in the effectiveness of
recycling. Ellen (1994) and Granzin and Olsen
(1991) showed that recyclers held strong
beliefs in the effectiveness of their action and
that they had internalised the personal
responsibility for that action.

More recently, Yule et al. (2001) undertook a
survey of 252 residents of the central belt of
Scotland. The survey explored past behaviours
(recycling paper, glass, aluminium and plastic
containers) and the intentions to recycle those
materials in the future. Thirty-two attitudinal
questions were asked, and nine reasons for
recycling and 15 reasons for not recycling were
offered. A principal components analysis of the
reasons for recycling produced two factors that
the authors identified as showing „social
responsibility‟ and „habit and belief‟
respectively. Four factors were identified from
the reasons for not recycling: „excuses‟, „lack of
practicality‟, and „a feeling that recycling was
not important for the individual‟; the fourth
factor had no obvious interpretation. A factor
analysis on attitudes also identified four factors:
(i) positive feelings towards recycling and social
endorsement, (ii) opportunities and ease of
recycling plus a sense of identity that the
authors interpreted as „perceived behavioural
control‟, (iii) social pressure, (iv) financial
considerations. Out of the four factors, only
perceived behavioural control proved to be a
significant predictor of recycling behaviour.
8.1.3 Experience and Habit
In most households, practices for waste
disposal tend to be automatic responses, with
no thought being given to them. Participation in
any new practice requires this habit to be
changed and new habits to be formed and
maintained. Consumers may forget to act upon
their new intentions simply because of the force
and persistence of their old habits (Pieters,
1991). Pieters considers that a decrease in
participation with time can be due to the
„uniqueness‟ stimulus wearing off and old
habits reappearing.
Basically, household waste management
behaviours tend to be intrinsically stable.
Pieters (1989) showed that prior performance
data were very good predictors of current
performances, and Dahab et al. (1995) and
Bagozzi et al. (1992) found that past behaviour
was very important in strengthening current
intentions. Dahab et al. considered that
previous experience will have lowered the
perception of effort and have already corrected
any misconceptions. This is echoed by Werner
et al. (1995) who considered that participation
through experience can serve to increase pro-
recycling attitudes, and by Pieters (1991) who
showed that mis-perceptions about
„inconvenience‟ may be quickly corrected once
real experience is gained.
Understanding Recycling Behaviour 2
Ronis, Yates and Kirscht (1989) developed a
model of repeated behaviours. Although this
was based on health-related behaviours, it was
considered to apply equally to other repeated
behaviours such as home composting (Aberg
et al., 1996). The research of Ronis et al.
concluded that habits are characterised by
initiation and persistence, and that different
determinants may apply for each of those
components. Variables including skills,
memory, and obstacles were considered to
have stronger effects on persistence whilst
expectancies, values, and attitudes were
considered to have stronger effect on initiation.
It has also been argued that in time prolonged
experience in a scheme may serve to stabilise
behaviour that was originally externally induced
(e.g. Pallak et al., 1980; Katzev and Pardini,
1987-88), and that new factors may be
internalised as reasons for adopting that
behaviour (Vining et al., 1992). Vining et al.
(1992) also suggest that, in time, participants
will tend to simplify their reasons why they
participate and condense their rationale into a
small number of terms (i.e. they will simplify
their cognitive load). Questioning people why
they recycle after the event may not reveal the
real reason that actually got them started.
The length of experience needed to effect
lasting attitude change is poorly defined. Some
argue that „prolonged‟ experience is necessary
whilst Pieters (1991) demonstrated that certain
„inconvenience‟ attitudes might become fixed
within a relatively short time-scale.
8.1.4 Knowledge
The possession of knowledge, particularly
procedural knowledge is also important to
recycling. Studies by Vining and Ebreo (1990),
Oskamp et al. (1991), Gamba and Oskamp
(1994), and others all found that recyclers had
greater knowledge [or felt better informed]
about recycling than did non-recyclers. Mis-
perceptions of one‟s knowledge can also be
important. A person who intends to participate
but does not know how to, or has an incorrect
knowledge of the rules, will not participate
properly. If the perceived task knowledge
exceeds actual task knowledge, low quality
performance will follow irrespective of personal
motivation (Pieters, 1991). On the other hand,
when perceived knowledge is less than actual
knowledge, people may not feel confident in
making the right decisions and may under-
perform.
97

8.2 Introduction to the Experimental Study
8.2.1 Pointers from Home
Composting Behaviour
The story continues, not with recycling, but with
home composting. In 1999, researchers at the
University of Paisley launched a promotional
free distribution of home compost bins for
University staff. Staff were interviewed
immediately prior to receiving their bin and the
cohort were then followed through two years
composting experience. Follow up surveys
were taken in 2000 and 2001. The attitudes
tested in the surveys were the negative
perceptions held about composting. Also tested
alongside were social norms and activity-
specific knowledge.
The results expressing strength of influence of
each of the factors were recorded on a scale of
1 to 5. Profiles were developed separately for
four classes of composter:
 New recruits
 Prior composters still composting
 Prior composters now dropped out
 „Never‟ composters
Figure 8.3 compares the „attitude‟ profile of the
new recruits, who were about to embark on
composting for the first time, with the other
groups.
The results show firstly that the new recruits
have more gardening interest than those who
declined to take up composting. They also had
more gardening knowledge but had equally
weak knowledge about composting, compared
to those who were already composting or had
composted previously. The new recruits had
similar perceptions about the time, effort and
knowledge required as those declining to take
up composting. Those perceptions were
significantly different to the perceptions held by
those who already had composting experience.
The perceptions of the new recruits about the
amount of waste required before composting
became worthwhile, of vermin and fly problems,
and about the unsightliness of a compost bin
were much more in line with the experienced
composters than with those who had not
composted before and who had not taken up a
promotional bin.
This perhaps provides evidence that two
distinct classes of attitude or perceptions may
98
Understanding Recycling Behaviour 2
be instrumental in determining the composting
activity. The first class, which includes
perceptions about time, effort and knowledge,
may not be set until an initial experience is
gained, though may be set very soon after that
initial experience is gained. The second set of
perceptions, including perceptions about waste
requirements, flies and vermin and aesthetics
of the compost bin may already be established
prior to any real experience and, as such, may
be significant in forming a pre-disposition to
participate given the right opportunity.
This second set of perceptions was also
relatively strong amongst the composting drop-
outs, and the perception that a lot of effort was
required was also relatively strong amongst this
group. It was seen in the follow up surveys that
a negative shift in one or more of those
attitudes was often a precursor of drop-out.
The attitude profiles of the prior composters
remained reasonably stable over the three
years. The perceptions that composting took a
lot of time, effort and knowledge never relaxed
for the University drop-outs, whilst perceptions
that a lot of waste was required, that
composting attracted flies or vermin and that
compost bins were unsightly strengthened over
time amongst the drop-outs.
Social influences, whether from the household,
from friends or from neighbours were similar
across all groups, and did not appear to be
significant determinants of composting
behaviour.
Overall, the analyses support the earlier
findings ((section 8.1.3), that initiation and
persistence may be driven by two different sets
of attitudes. The most important attitudes are
those that form a pre-disposition towards the
activity. With those attitudes in place, behaviour
should follow if there is a suitable trigger.
Without those prior attitudes in place, behaviour
is unlikely to ensue.
A fuller description of this research, and of
related home composting research, can be
found in the companion monograph
„Understanding Home Composting Behaviour
(Tucker and Speirs, 2001).
 Understanding Recycling Behaviour 2

Figure 8.3 „Attitude/knowledge Profiles of Different Composting Groups

4
Score

Unsightly
Vermin

Household
Effort
G Interest

Time

Space

Lot waste

Friends

Neighbours
Know
G Know

Comp know

Nevers Dropouts Currents Recruits

GInterest = Gardening interest, GKnow = Gardening knowledge, CompKnow = Composting knowledge, Time =
Composting takes a lot of time, Effort = Composting takes a lot of effort, Know = Composting requires a lot of
knowledge, Space = Composting needs a lot of space, Lotwaste = Composting needs a lot of waste to be
worthwhile, Vermin = Composting attracts vermin or flies, Unsightly = Compost bins are unsightly, Household =
Household influence, Friends = Friends influence, Neighbours = Neighbours influence

8.2.2 Focus of the Current Study have tested the constructs from different
models together. The current research
Whilst the above results refer to home programme was designed to co-evaluate the
composting behaviours, it is considered likely relative importances of particular features of
that similar considerations could apply to altruism (awareness of consequences and
recycling behaviours as well (although the acceptance of responsibility) against particular
specific attitudes and specific negative features of planned behaviour (self-efficacy and
perceptions involved are likely to be facilitating conditions respectively). The
fundamentally different for the two behaviours – emphasis of the research was on
see section 6). understanding the role of those attitudes on
behaviours and on behavioural change.
Previous research has already gone a long way
to delineating which specific attitudes might be An initial survey was carried out in 2001 in the
important in determining recycling behaviour. town of Elderslie in Renfrewshire, Scotland.
Consideration of the psychological models This survey of 450 residents was carried out
suggests that the important constructs should nine months after a [multi-material] kerbside
include: collection had been introduced into the
community. Before the introduction, the
 Local facilitation residents were serviced by a relatively low
 Self efficacy density of bring recycling facilities. The aim of
 Acceptance of responsibility that study was to compare the acceptance of
responsibility between the new recyclers
 Awareness of consequences
recruited from the kerbside roll-out with those
 Negative perceptions (time, effort,
already recycling prior to its introduction.
convenience, need a lot of waste, etc., see
Tucker, 2001)
The second survey, undertaken in 2002/03 was
designed to track a cohort of householders
In the past researchers have tended to opt a-
across the introduction of a new multi-material
priori for a given model (e.g. the theory of
kerbside recycling scheme. That survey was
planned behaviour or the altruism model) and
undertaken in the district of Clackmannanshire,
have tested only that model. Few researchers

99
 Understanding Recycling Behaviour 2

Scotland. As with Elderslie, residents only had was introduced. All identified classes of
access to bring facilities before the scheme attitudes were tested in the survey.

8.3 Attitude/ Behaviour Research in Renfrewshire

8.3.1 Introduction
The scheme that was studied operated in the
town of Elderslie, Renfrewshire. It comprised a
weekly kerbside collection of paper, plastic
bottles, aluminium and steel cans. Most of the
households served were owner-occupied,
detached and semi-detached housing, in a
mixture of new residential estates and older
housing. At the time of the study, the scheme
serviced 900 households and had been running
for approximately 9 months.
A major aim of the study was to investigate the
attitudes and behaviours of the new recruits to
recycling (who had started recycling when the
scheme commenced) and to compare those
attitudes and behaviours with the attitudes and
behaviours of the prior recyclers and the non-
recyclers respectively.
Other major aims were to gather information on
relationships between the materials recycled
(see section 6.2), and on the differences
between self-reported and measured
behaviours (section 9).
The overall study comprised participation
monitoring of all serviced households for a
period of 10 consecutive collections, along with
weight and compositional analyses of
recyclables set out. A questionnaire survey was
administered to 450 of the households. The
return rates from the questionnaire survey were
very good with 70% of the targeted households
answering the questionnaire.
8.3.2 Demographics and
Participation
The first aim of the Renfrewshire study was to
explore any possible demographic differences
between the relatively new converts to
recycling and those who had been recycling for
a greater time. Results are shown in tables 8.1
to 8.4. For explanation of Social Classes see
Appendix A, table A2. In these tables, the
column labelled „bring‟ separates a small group
of active recyclers (numbering 9) who had not
joined the kerbside scheme but were still
recycling elsewhere.

Table 8.1 % Distribution by Social Class

All groups Prior New recruits Bring Non-
recyclers recyclers recyclers

1 17% 18% 17% 0% 14%

2 44% 52% 43% 0% 34%
3n 18% 14% 17% 50% 28%
3m 13% 6% 16% 50% 17%
4 8% 10% 7% 0% 7%

Table 8.2 % Distribution by Family Life Stage

All groups Prior New recruits Bring Non-
recyclers recyclers recyclers

Young adults 7% 3% 17% 22% 11%

Families/ young children 22% 22% 25% 11% 16%
Families/ older children 26% 24% 31% 0% 16%
More mature 17% 16% 16% 0% 25%
Retired 28% 35% 21% 67% 32%

100
 Understanding Recycling Behaviour 2

Table 8.3 % Distribution by Number in House

All groups Prior New recruits Bring Non-
recyclers recyclers recyclers

1 13% 11% 6% 56% 32%

2 33% 39% 31% 33% 34%
3 22% 17% 26% 0% 20%
4 26% 30% 28% 11% 14%
5+ 6% 3% 9% 0% 0%

Table 8.4 % Distribution by Acorn Category

All groups Prior New recruits Bring Non-
recyclers recyclers recyclers

A 11% 15% 10% 22% 7%

B 35% 40% 35% 11% 25%
D 45% 37% 48% 33% 60%
E/F 9% 8% 7% 33% 8%

The results highlight that families with young now using the kerbside scheme had longer
children were amongst the weakest residencies than the new recruits, who in turn
participants. Single person household were had longer residencies than the non-recyclers.
also relatively poor users of the kerbside
scheme. However, several single person
households were recycling using bring Figure 8.4 % Distribution by Length of
schemes instead. There was little difference Residency
between recyclers and non-recyclers in terms
of their social class. Non-recyclers were drawn
120
Cumulative percentage

most heavily from Acorn category D (settling).

100
A slightly different picture emerged when
80
comparing new and old kerbside recyclers.
More new recyclers were drawn from the non- 60
car owners (not shown) and possibly from 40
families with older children. Fewer new
recyclers were drawn from the retired residents. 20
Proportionally more prior recyclers were drawn 0
from Acorn categories A and B whilst
proportionally more recruits were drawn from 1 5 9 13 17 21 25 29 33 37
Acorn D. However, because many of the Years of Residence
traditional demographic base (i.e. retired
residents, Acorn A) were already recycling, Prior Recruits
those sectors would be severely depleted of
„would be‟ recyclers leaving relatively few to Bring Non
recruit from. Consequently proportionately
more recruits would be expected, on statistical
grounds, to come from the younger age groups
and less affluent housing stock. It is probably 8.3.3 Attitudes
not because the individuals from these groups
were any more susceptible to behavioural The only attitude measured in Renfrewshire
change. was the acceptance of a personal responsibility
for recycling. The attitude scale that was used
Other features that were found to impact on included the 8 statements listed in box 8.1,
take up were: (i) habit, or resistance to change, section 8.4.5, together with an additional
and (ii) the stability of population. Figure 8.4 statement: “You should always recycle”. Each
shows how recyclers with long residencies in statement was scored on a scale 1: strongly
the area were less likely to switch from drop-off
to kerbside recycling. Prior recyclers who were

101
 Understanding Recycling Behaviour 2

disagree to 5:strongly agree. This provided a Figure 8.5a-b Acceptance of Responsibility

total score in the range of 9 to 45.
120

Cumulative percentage
The results show that the attitude scores were 100
slightly higher for those who had recycled
previously compared to those who only started 80
recycling with the kerbside scheme (figure
8.5a). The lowest scores of all were found 60

amongst the non-recyclers. Curiously, the bring

40
recyclers who were not using the new scheme
had much weaker attitudes than those who did
20
move with the change. However, further
analysis showed that four out of the group of 0
nine bring recyclers were, in fact, drop-outs
20 23 26 29 32 35 38 41 44
from the scheme. Overall scheme drop-outs
had intermediate attitudes between the current Attitude score
recyclers and the „never‟ recyclers (figure 8.5b).
The same trend has been noticed amongst Prior Recruits
home composting behaviours, with lapsed
composters showing intermediate attitudes Bring Non
between current composters and „never‟
composters (Tucker and Speirs, 2001).
100
Further analysis showed that there was little
Cumulative percentage

difference in attitude scores amongst different 80

demographic groups. There were no
differences between family life stages, marginal 60
decreases with increasing household size and
marginal decreases towards lower social 40
classes. The only statistically significant
difference was a stronger acceptance of 20
responsibility amongst car owners than
amongst non-car owners. 0
20 23 26 29 32 35 38 41 44
Attitude score

User Never Dropout

8.4 Attitude/ Behaviour Research in Clackmannanshire: Part 1 Before

Change

8.4.1 Introduction had accompanied the kerbside introduction.

The Clackmannanshire study was designed to
test features of a number of attitude/ behaviour
models. It was also designed as a before and
after survey to be implemented prior to the
introduction of a new kerbside recycling
programme, then to be administered again a
few months after the kerbside scheme had
been introduced. Differences between the
before and after responses would quantify the
extent of behavioural and attitude changes that
The comparative results of the before/after
survey are given in section 8.5 of the report.
The current section concentrates on developing
an in-depth analysis of the before survey.
The study specifically looked at an area where
a new kerbside collection scheme was about to
be introduced. This was an expansion of the
normal kerbside collections in the district of
Clackmannanshire, in central Scotland. For the
research, 200 households were targeted from

102

three distinct demographic groups. These were
drawn from (i) an area in the south of the town
of Alloa where 57 recently built semi detached
houses were being added to the programme.
Those properties were all from Acorn groups
B11 („Affluent Working Couples with
2
Mortgages) , (ii) 96 households in the nearby
town of Tullibody, comprising local authority
and ex-local authority housing stock, with an
Acorn group of F40 („Council Areas, Older
People, Health Problems‟). In addition 48 flats
near the centre of Alloa (Acorn group C17 -
Flats & Mortgages, Singles & Young Working
Couples) were also targeted for the initial
survey.
The „before‟ questionnaires were undertaken in
August/ September 2002, one month before the
3
kerbside scheme was due to be introduced,
and before households had been informed that
the scheme was to be introduced in their area.
In the event, there were operational delays that
led to the roll out being put back to January
2003. At the time of the before survey, the
household sample simply had access to bring
recycling points.
The questionnaire was designed to examine
other waste management behaviours as well as
recycling behaviours (see section 6), and to
examine key attitudes and waste management
knowledge alongside behaviours. The
questions were designed to give quantitative
performance indicators which could (i) test the
differences between recyclers and non-
recyclers in the before survey, and (ii) test the
changes that occurred across the intervention.
The questionnaire was delivered to each
household with the surveyor first making a
personal contact to introduce the questionnaire.
The survey form was then left at the household
to be picked up one or two days later. At the
outturn, 99 completed questionnaires were
recovered from the 200 targeted households.
The return rates were substantially lower in the
Alloa flats than in the town‟s houses and the
houses in Tullibody. The return rate was at the
lower end of the normal response rate (50-
75%) normally achieved by the University with
that delivery method.
2
See Appendix A for full details of the Acorn
classification
3
The kerbside scheme being introduced used a blue
box to collect glass, cans and textiles, and a
separate blue bag to collect paper. The scheme was
of the same design as that deployed in Chesterfield
and North East Derbyshire (section 4)
Understanding Recycling Behaviour 2
8.4.2 Demographics and Behaviour
Recycling
For these analyses, a „recycling‟ household has
been defined as a household that claims to
recycle at least some of one or more of the
following materials: newspaper, glass bottles,
glass jars, drinks cans, food tins, plastic bottles,
magazines, cardboard, textiles, aluminium foil,
carrier bags, junk mail, or batteries. That list
appeared to cover all materials recycled.
Nobody volunteered recycling any material that
was not on that list.
The results showed a difference in the numbers
of residents claiming to recycle in the housing
scheme in Tullibody (44%) compared with
households in Alloa (58%). However this
difference was not statistically significant. It is
also noted here that, at the time of the initial
survey, Alloa offered a greater range of
recycling facilities than were provided at
Tullibody This may have contributed to the
difference. Social class and family life stage
were not significantly related to whether
households recycled or not. The significant
demographic predictors were car ownership
and household size, with greater
representations of car owners and smaller
households amongst the recycling households.
However it is noted again that the larger
households were concentrated more in
Tullibody than in Alloa.
Table 8.1 Demographic differences between
Recyclers and Non-recyclers.
Statistic Significance
House type 
Social class 
Family life stage 
Household size ***
Car ownership **
*** significant at 99% confidence, ** 95%,
* 90%,  not significant
Reduce / Reuse
In responding to a check list of reduce and
reuse behaviours, most households stated they
generally carried out three or four of the offered
activities, at least some of the time, with 90% of
household claiming to practice at least one of
the activities. The differences in scores were
difficult to confirm statistically amongst
demographic groups, though flat dwellers,
smaller households, and particularly older
103
 Understanding Recycling Behaviour 2

residents appeared to carry out the highest The questionnaire additionally posed an open
number of reduce/reuse activities (table 8.2a- question to list ways other than recycling and
4
c) . composting by which they might reduce their
household waste. Only 11 constructive answers
were produced by the respondents, with 73
Table 8.2a-c Number of Reduce/ reuse respondents returning a “Don‟t know” response.
Activities Practiced Seven inappropriate responses were given,
including “recycle it”, “burn it in the garden”,
a. House Type “fold it more carefully”, and “put it in next door‟s
Average number of bin”.
reduce/ reuse
activities practiced
8.4.3 Material Capture Rates
Semi Detached 2.97
Table 8.3 shows the proportions of available
LA /ex-LA 3.14
material recycled by individual households. It is
Flat 3.53 seen that for food tins, cardboard and textiles in
particular, less than half the recyclers were
recycling all that was available to them.
b. Family Life Stage
Average number of Tables 8.4a-b list the most common reasons
reduce/ reuse expressed for not recycling all available
activities practiced material. Table 8.5 lists the volunteered
incentives that would make the respondents
Young adults 2.80 recycle more.
Families/ young 2.83 There is no need to make detailed comment on
children these expressed reasons. They are quite
Families/ older 2.07 similar to the reasons expressed in many other
children surveys [by the University of Paisley and by
More mature 3.77 many other researchers as well].
Retired 3.64
The overall material capture rates across the
sample were separately computed for 5 marker
c. Household Size materials. These were estimated from the
Average number of formulae:
reduce/ reuse
Capture rate (%) =  (Items consumed 
5
activities practiced 6
Percentage recycled )
1 3.18 The material capture rate was obtained by
2 4.04 summing responses over all residents. A
3 3.56 second indicator was obtained by only
4 2.44 summing over the recyclers. This second
indicator tells us how much of their available
5+ 1.82
material a recycler recycles. In the results
presented in table 8.6, the second indicator has
been broken down by geographical district.
The most commonly practiced reduce/ reuse
activities were donating to charity shops (86%), The results show reasonable correspondences
buying recycled (84%), and buying between material capture rates and the
concentrated products or refills (82%). The percentages of people recycling, showing that
other three offered activities were much less the combined effects of consumption
practiced: only 55% avoided over-packaged
5
goods, 52% used their own shopping bag and The number of items consumed was determined
50% reused waste materials. through direct questions in the Clackmannanshire
survey. The results are available and are listed in
4
The reduce/ruse activities tested were: Donate section 10 of this report.
6
materials to charity, seek out refillable containers, Percentage recycled was obtained on the discrete
decline excess packaging, use own shopping bag, scale: 0, 25, 50, 75, 100
buy recycled, and reuse waste.

104
 Understanding Recycling Behaviour 2

differences and selectivities in recycling might where there were no immediate local recycling
be quite small. facilities for the materials.

The depressed capture rates for newspaper Overall, the losses of available materials from
and steel tins might be due to the relatively recycling were substantial, even amongst
poor capture of those materials in Tullibody, practising recyclers.

Table 8.3 Proportions of Available Materials Recycled

None to None A quarter About About three- All or nearly
recycle or less half quarters all

Newspaper 10 57 5 3 2 18
Glass bottles 2 56 3 6 2 21
Drinks cans 7 66 2 5 1 9
Food Tins 1 78 4 3 0 4
Plastic bottles 1 82 2 1 0 4
Cardboard 1 79 1 5 2 2
Textiles 3 64 3 3 7 10
Magazines 5 64 3 3 2 13
Junk Mail 0 77 3 2 1 6
Glass jars 0 61 5 5 3 16
Aluminium Foil 4 81 0 1 1 2
Carrier bags 0 67 3 6 2 12
Batteries 2 83 1 2 1 1
Green waste 1 76 4 2 1 5
Others 0 90 0 0 0 0

Table 8.4a Reasons for Not Recycling All (% Responses)

I do Recycle all I can 13%
I don't recycle all - no reason given 22%
I don't recycle all - reason given 65%

Table 8.4b Reasons for Not Recycling All (% Responses)

Facilities (lack of kerbside, materials not taken, poor facilities etc.) 30%
Knowledge (lack of knowledge on recycling) 19%
Laziness (can't be bothered to recycle) 19%
Access (can't get to recycling bins) 12%
Storage (nowhere to store recyclables, not enough space) 7%
Time (recycling is too time consuming) 4%
Other (don't think about it, forget, lack material, sorting) 9%

Table 8.5 What would Make you Recycle More (% Responses)

Better facilities (easier access to recycling facilities) 43%
Kerbside (the introduction of a kerbside scheme) 28%
Separate bins (an additional bin to put recyclables in) 14%
More information 11%
More time 2%
Financial incentive (a reduction in council tax) 2%

105
 Understanding Recycling Behaviour 2

Table 8.6 Material Capture Rates (%)

News- Glass Alu cans Fe tins Plastic
paper bottles

% Population recycling 28% 36% 18% 11% 7%

Capture rate (%) 20% 36% 18% 8% 8%
Capture rate – recyclers 42% 73% 49% 19% 17%
Capture rate–recyclers–Alloa 62% 80% 48% 31% 35%
Capture rate–recyclers-Tullibody 24% 68% 49% 8% 2%

8.4.4 Knowledge and Information Table 8.7a. Knowledge of waste

hierarchy
General Waste Knowledge All

The questionnaire set out to test general waste Don't know 96%
knowledge as a potential performance Answer wrong 4%
indicator. The questions that were set to Partly correct 0%
provide this indicator were: Totally correct 0%

 What is meant by the waste hierarchy?

 How much waste an average
household produces each year? Table 8.7b Cost of Disposal per Household
 How much of that waste is recyclable? per Week
 How much it costs the Council each All %
week to collect and dispose of a
household‟s waste? Don't know 88%
£0 – 0.75 0%
Results are summarised in tables 8.7a-d. In the £0.75-£1.99 0%
tables, the correct responses are shaded. The £2 - 4.99 1%
results demonstrate that there is an extremely £5+ 11%
poor knowledge across all four statistics,
amongst recyclers and non-recyclers alike.

Table 8.7c How Much Waste per Household per Year?

Non-recycler % Recycler % All %

Don't know 85% 74% 79%

<500 kg/year 2% 2% 2%
500-1500 kg/year 4% 8% 6%
>1500 kg/year 9% 16% 13%

Table 8.7d How Much is Recyclable?

Non-recycler % Recycler % All %

Don't know 55% 35% 45%

<25% 6% 6% 6%
25-49% 9% 8% 8%
50-74% 19% 26% 23%
75%+ 11% 24% 18%

106
 Understanding Recycling Behaviour 2

Local Awareness recyclers knew where the recycling points were

To test the awareness of local recycling
provision, residents were asked to name a
location, or locations, where they could recycle
each of the following materials: Newspaper,
glass, aluminium cans, cardboard, garden
waste, and batteries. The responses were then
checked against the actual provisions (table
8.8).
Overall, a maximum of 34% respondents knew
where to recycle their paper, whilst 61% knew
where glass could be recycled. 43% knew
where to recycle aluminium cans, 6%
cardboard, 14% green waste, and 19%
batteries. It is noticeable from these results that
increasing local recycling knowledge appeared
to go hand in hand with increasing local
recycling provision.
but did not use them. Amongst the recyclers [of
at least something], those who were not
recycling newspaper, glass or drinks cans split
more evenly between having no knowledge of
the appropriate recycling points, and having the
knowledge, but choosing not to recycle those
materials. For green waste, cardboard and
batteries, the lack of knowledge of appropriate
recycling outlets always constituted the major
reason for not recycling them.
The most problematic category of people,
however, were those who confessed no
knowledge (or had incorrect knowledge) of
where to recycle specific materials but still
claimed to recycle those materials. Those
individuals are designated by the shaded row in
figure 8.9. It is considered that these
inconsistent responses might be due to:

A comparison was then made of each (i) respondents falsely claiming to be

individual‟s knowledge of recycling points recyclers, or
against their stated recycling behaviours. The (ii) recyclers claiming to recycle more
results of that comparison are listed in table materials than they actually do.
8.9. It is clear from the results that the majority
of non-recyclers had no knowledge of where Their stories will be taken up further in section
they could recycle. Only a minority of non- 9.2.

Table 8.8 Where to Recycle Specific Materials

News- Glass Alumin- Card- Green Batteries
paper bottles ium board waste
cans

Don't know / nowhere 52 34 50 69 57 71

Stated correctly 20 38 26 4 5 13
Yes – but location not 11 17 13 1 8 4
stated*
Stated incorrectly 7 1 1 16 20 1
* Or answer was vague

Table 8. 9 Recycling specific materials vs. Knowledge of Where to Recycle them

Know- Recycle News- Glass Drinks Cardboard Green Batteries
ledge material paper bottles cans waste
Where to
Non-recyclers
No No 39 28 36 45 37 42
Yes No 8 19 11 2 10 5
Recyclers
No No 11 2 14 35 23 33
Yes No 10 6 18 1 14 10
No Yes 8 5 2 9 5 1
Yes Yes 20 36 15 4 7 5
Inconsistent behaviours are shown as a shaded row.

107
 Understanding Recycling Behaviour 2

8.4.5 Attitudes For these initial analyses, a „recycler‟ was

simply defined as a household claiming to
The attitudes tested in the Clackmannanshire recycle any quantity of at least one of 13
survey were grouped into four categories: different listed materials. The responses were
taken at their face value. However, it was noted
(i) Acceptance of responsibility above that some respondents may have made
(ii) Negative perceptions false or exaggerated claims about their
(iii) Self-efficacy and awareness of recycling or about the extent that they recycle.
consequences It was also noted above that the non-recyclers
(iv) Local facilitating conditions fell into two distinct categories: those who were
unaware of where to recycle, and those who
The full set of questions that were asked is were aware but did not recycle. Based on these
listed in boxes 8.1 to 8.4. considerations, a new behavioural
categorisation was constructed:
Respondents were asked to indicate their
strength of agreement with each offered Non recyclers
statement on a scale of 1 (strongly disagree) to Category 0: Unaware & do not recycle
5 (strongly agree). The total score for each Category 1: Aware and do not recycle
category was computed separately for each
respondent. For negatively phrased questions, Recyclers
the scores were reversed. The results are Category 2: Aware and recycle
summarised in figure 8.1a-d in the form of box
and whiskers plots. Inconsistents
Category 3: Unaware but recycle some
The results demonstrate that recyclers do tend materials
to hold more positive attitudes across all four Exaggerators
attitude categories, though only one attitude: Category 4: Aware and recycle most
„acceptance of responsibility‟ was found to be materials, but unaware and
significantly different between recyclers and recycle others
non-recyclers.
The revised results for the new classifications
are shown in figures 8.2a -d.

Boxes 8.1 - 8.4 Attitude Questions

Acceptance of Responsibility
A1 It is your duty to recycle
A2 It is not worthwhile recycling unless everyone does it
A3 You should try to encourage others to recycle
A4 You feel guilty when you do not recycle
A5 The responsibility for recycling should lie with the local council
A6 It doesn‟t matter if you don‟t recycle as long as others do
A7 You should not have to sort your rubbish into recyclable and non recyclable waste
A8 You would feel embarrassed if a neighbour saw you throwing out waste rather than
recycling it

Negative Perceptions
N1 Recycling takes a lot of time
N2 Recycling takes a lot of effort
N3 Recycling needs a lot of storage space
N4 Recycling is not worthwhile if you do not have much waste
N5 Recycling can be unpleasant
N6 Recycling needs a lot of knowledge

Local facilitating conditions

F1 The local recycling facilities are not convenient for me to use
F2 The local recycling facilities are poorly serviced

108
 Understanding Recycling Behaviour 2

Self-efficacy & awareness of consequences

E1 You have enough information to enable you to recycle correctly
E2 You are happy that you know the environmental reasons why you should recycle
E3 You are happy that you know why the Government wants you to recycle
E4 You understand what happens to your waste if you recycle it
E5 You are happy that you know where to recycle your material
E6 You are happy that when people give waste for recycling that the waste is actually recycled

Figure 8.1a-d. Attitude Distributions between Recyclers and Non-

recyclers
Acceptance of Responsibility
40

30
acres

20

0 1
recycler

Self Efficacy / Awareness of Consequences

30

20
effaoc

10

0 1
recycler

Note: The box delineates the inter-quartile range of the observations, with the horizontal line across the box
showing the median value. The whiskers extend out to the lowest and highest observations that are still inside
the region defined by the following limits: Lower Limit: Q1 - 1.5 (Q3 - Q1); Upper Limit: Q3 + 1.5 (Q3 - Q1).

109
 Understanding Recycling Behaviour 2

[Absence of] Negative Perceptions

30
negper

20

10

0 1
recycler

Facilitating Conditions

10

7
fcon

0 1
recycler

Figure 8.2 of]

[Absence a-dNegative
AttitudePerceptions
Distributions between Different Behaviour/ Awareness Categories
Acceptance of Responsibility

40

30
acres

20

0 1 2 3 4
category

110
 Understanding Recycling Behaviour 2

Self Efficacy/ Awareness of Consequences

30

20
effaoc

10

0 1 2 3 4
category

[Absence of] Negative Perceptions

30
negper

20

10

0 1 2 3 4
category

Facilitating Conditions

10

7
fcon

0 1 2 3 4
category

111

Figure 8.3 Environmental Concern and Categories of Recycler
5
environmental concern
0 1 2 3
category
The two categories of non-recycler show quite
comparable levels of acceptance of
responsibility, though those with the awareness
of where to recycle (category 1) tend to hold
greater negative perceptions against recycling
and greater negativity against the local
facilitation. They also demonstrate greater self-
efficacy than the „non-awares‟.
It is considered that the category 1 profile may
be indicative of some of its members being
drop-outs from recycling, though this remains to
be tested.
The „non-aware‟ but still recycling category
(category 3) show more negative attitudes than
„normal‟ recyclers (category 2) across all four
attitudes, though only the difference in negative
perceptions was significant. The category 4
recyclers (exaggerators) showed a significantly
higher self-efficacy than normal recyclers,
though returned comparable responses over
the other three attitudes. This shows a much
higher perceived knowledge (probably
exceeding actual knowledge) amongst the
category 4 recyclers.
It is also noticeable that the expressed
environmental concern was stronger amongst
the recyclers than amongst the non-recyclers
(figure 8.3). This contradicts the results of some
previous research investigations (section
8.1.1).
112
Understanding Recycling Behaviour 2
4
8.4.6 Attitudes and Demographics
Finally, the hypothesis that there may be a
relationship between attitudes and
demographic factors was explored. The results
that may be statistically significant are shown in
tables 8.10a-c. The residents of the housing
scheme in Tullibody perceived a significantly
poorer facilitation of recycling than did their
counterparts in Alloa. This may well be due to
the relatively poorer local recycling provision
there rather than being any direct feature
associated with housing type. Families with
children expressed the lowest environmental
concern, whilst stronger negative perceptions
and weaker self-efficacy/ awareness of
consequences were common to all three of the
younger groups. The levels of environmental
concern and acceptance of responsibility for
recycling were lowest amongst the larger
households. Self-efficacy was weak in the
single person households and in the largest
households.
Table 8.10a Mean Attitude Scores versus
House Type
Facilitating
Conditions
Semi-detached 6.8
Local authority (& ex-LA) 4.5
Flat 5.7
 Understanding Recycling Behaviour 2

Table 8.10b Mean Attitude Scores versus Family Life Stage

[Absence of] Self-efficacy Environmental
Negative AOC concern
perceptions

Young adults no children 18.6 14.8 3.4

Families with young children 19.3 18.9 2.6
Families with older children 19.0 18.1 2.4
More mature 23.7 20.8 3.3
Retired 21.3 22.1 3.6

Table 8.10c Mean Attitude Scores versus Number in Household

Acceptance Self-efficacy Environmental
responsibility AOC concern

1 29.1 17.8 3.5

2 29.0 22.0 3.6
3 27.8 22.7 2.9
4 25.5 17.3 2.6
5+ 22.0 19.1 2.2

8.4.7 Further Considerations of that specifically indicate that somebody else

Attitudes should be doing something as well.

Whilst acceptance of responsibility, negative
perceptions, self-efficacy/ awareness of
consequences, facilitating condition and
environmental concern have been considered
as distinct categories of attitudes, it is of course
possible that they may be mutually correlated in
some way. The similarities amongst the
responses to the variables were investigated,
firstly using cluster analysis then using factor
analysis (which picks out common factors
across the variables – see section 6.3 for
further details).
Figure 8.4 shows the results of the cluster
analysis. The results show how the two
attitudes (e2 and e3) that express the
knowledge of why recycling should be done
cluster with the acceptance of personal
responsibility for doing it (a1, a3, a4) and with
concern for the environment. Self-belief and
trust of others all cluster together (e1, e4, e5,
e6). All negative perceptions remain in a single
cluster. Facilitating conditions cluster loosely
with the acceptance of responsibility variables
The factor analysis also returned the same
general picture.
Overall, the four prime clusters or factors that
were identified might be described as:
F1. Belief that recycling is necessary and
that one should do something
personally
F2 Negative perceptions about doing the
activity
F3 Belief that one can do the activity and
trust that it will be followed through by
others
F4 Belief that others need to do something
as well
The statistics regrouped for each of the above
four factors are given in table 8.11. Factors 1,
3, and 4 are all significant at 95% confidence,
with „negative perceptions‟ showing a weaker
discrimination, but nevertheless still significant
at 90% confidence.

113
 Understanding Recycling Behaviour 2

Table 8.11 Statistical Summary for Derived Attitude Factors

Statistic F1 F2 F3 F4
„self „Negative „Locus of „Collective
commitment‟ perceptions‟ Control‟ responsibility‟

Mean (Recycler) 19.1 22.1 13.5 20.6

(Non-recycler) 16.7 19.7 11.3 17.2

Stdev (Recycler) 4.2 5.9 4.3 4.8

(Non recycler) 4.5 5.4 4.1 3.6

Figure 8.4 Cluster Analysis of Attitude Variables

Similarity

56.69

71.13

85.56

100.00
env
m a1
a1 a4
a4 a3
a3 e2
e2 e3
e 3 e1
e1 e5
e5 e4
e4 e6
e6 a8
a8 aa2
2 aa5
5 aa6
6 aa7
7 f1 f1 f2f2 nn1
1 n3n3 nn2
2 n5
n5 nn4
4 nn6
6
on
nvi r
e

Variables

Whilst the above analyses show that the Table 8.12 Results of Discriminant Analysis
distributions of attitudes are different between Put into True group
recyclers and the rest of the population, it does group Non- Recycler
not necessarily provide a model that can recycler
discriminate between recyclers and non-
recyclers on the basis of their attitudes. The Non-recycler 33 12
predictive power of the four factors was Recycler 9 23
assessed using the statistical technique of
discriminant analysis, the results of which are N correct 33 23
summarised in table 8.12. (79%) (66%)

The derived model predicted 73% of

behaviours correctly from the four revised
attitude factors. The proportion of correct
predictions was slightly higher for predicting
8.4.8 Performance Indicators
not-recycling compared with predicting The set of performance indicators from the
recycling (79% correct versus 66%). „before‟ survey are summarised in table 8.13.

114
 Understanding Recycling Behaviour 2

Table 8.13 Performance Indicators

INDICATOR DESCRIPTION P.I. VALUE

Actual Knowledge ALL Recycler Non-rec
P1 Knowledge of basic waste facts % of questions answered within acceptable tolerance 7.3 8.5 5.9
limits
P2 Knowledge of [local] recycling % of 6 key commodities correctly placed* 33.2 48.9 19.5
provision
P3 Knowledge of ways to reduce / Number of correct ideas volunteered without prompting 1.8 2.4 1.1
reuse waste (marked out of 6 & expressed as %)
Claimed Behaviour
P4 % recycling Percentage of sample recycling at least one of 6 marker 47 (43) 96 (88) 0
materials**
P5 % garden owners not putting % not using dustbin or general waste skip 35.6 39.5 31.4
garden waste in refuse
P6 Reduce/ reuse activities undertaken % of the 6 marker activities carried out 50.8 58.3 43.6
Claimed Diversion
P7 % of designated wastes recycled {(N100  100)+(N75  75)+ (N50 Newspaper 20.2 41.8 0
 50)+ Glass bottles 36.2 73.2 0
(N25  25)}/ {N100 +N75 + N50 Alu cans 18.0 48.5 0
+N25 + N0} Fe tins 8.4 19.5 0
Plastic bottles 8.0 17.2 0
Attitudes
P8 Acceptance of responsibility Average score per question –1, as a % of max possible 59.4 64.7 54.1
P9 [Absence of] Negative perceptions Average score per question –1, as a % of max possible 61.3 65.0 57.5
P10 Self-efficacy/AOC Average score per question –1, as a % of max possible 57.4 61.3 53.3
P11 Perception of local facilitating Average score per question –1, as a % of max possible 54.5 58.7 50.0
conditions
* newspaper, glass, aluminium cans, cardboard, batteries, green waste
** newspaper, glass bottles, drinks cans, food tins, cardboard, plastic bottles, bracketed figures give % recycling at least 50% of at least one marker

115
 Understanding Recycling Behaviour 2

8.4.9 Discussion and Conclusions  Negative perceptions about doing the

activity
The before survey in Clackmannanshire has  Belief that one can do the activity and trust
derived a set of recycling performance that it will be followed through by others
indicators that quantify the key behaviours and
attitudes of the residents before they were
 Belief that others need to do something as
well
provided with a kerbside recycling service. Only
two out of 97 households replying expressed
Effectively these portray “Commitment”,
an intention not to use a kerbside recycling
“Constraints”, “Control“, and “Collectiveness”
scheme if it was introduced, though not every
respectively.
respondent would seek to use the scheme for
every material.
However, despite clear differences in attitude
distributions between recyclers and non-
Just over half the sample claimed to already
recyclers, the differences were less clear cut at
recycle at least a small proportion of
an individual level. Only 73% of behaviours
something. The differences between those self-
could be predicted correctly based on the
claimed recyclers and the non-recyclers were
selected attitudes.
reflected in the performance indicators.
Additionally, it should be noted that even
The general waste knowledge was extremely
amongst the recyclers, the capture rates for
low amongst recyclers and non-recyclers alike.
individual materials were well below their
Knowledge in this area did not appear to be
maximum potential, either through selectively
any kind of prerequisite for recycling.
recycling just a few materials, or by not
recycling all available material. Whilst nearly
Not surprisingly, the recyclers amongst the
75% of glass bottles were recovered from
population could correctly name more sites
recyclers, only around 50% of newspapers and
where they could recycle specific materials,
drinks cans were recovered, and less than 20%
though 15 out of 49 self-claimed recyclers
of food tins and plastic bottles. These
actually gave incorrect answers. Over half of
differentials may be linked to local recycling
the 46 self-claimed non-recyclers had no
provision, though there may be other
knowledge of where to recycle anything.
influencing factors as well.
Whether this is the cause or symptom of their
not recycling is not yet answered. The non-
Whilst there have been several competing
recyclers who did know where they could
theories of recycling behaviour based on their
recycle tended towards having greater negative
own specific constructs, few studies have
perceptions about recycling as well, and also
attempted to run comparative tests on the
less favourable assessments of local provision.
various constructs taken from those models.
It is conjectured that some members of that
The Clackmannanshire pilot is amongst the first
group may be drop-outs from recycling.
to do so. All the attitudes tested appear to have
played some part in determining recycling
All four tested attitudes {Acceptance of
behaviours. That does not deny the possibility
responsibility, Negative perceptions, Self-
that other attitudes (as yet undefined) may be
efficacy and awareness of consequences, and
important as well.
Perceptions of Facilitation} were stronger
amongst recyclers than amongst non-recyclers.
Finally, it must be realised that the results
It was then shown that an alternative
drawn from the study have been based on
classification of attitudes could be considered.
stated attitudes and self-claimed behaviours.
The four prime attitudes were found to map
There is no guarantee that those claims
onto four separate constructs:
actually reflect the truth. Some discrepancies
have been noted in the claims made in this
 Belief that recycling is necessary and that study. These will be discussed in more detail in
one should do something personally section 9 of the report.

116

8.5 Attitude/ Behaviour Research in Clackmannanshire: Part 2
After Change
8.5.1 Introduction
The second questionnaire deployed in
Clackmannanshire targeted a cohort of 84
residents drawn from the respondents of the
first waste attitude/ behaviour survey (section
8.4). The flatted properties were excluded from
the second survey, as they had not been
offered the kerbside scheme.
The first survey was administered ahead of
kerbside recycling being introduced. The
second survey was undertaken four months
after the scheme was introduced, and 8 months
after the first survey.
Unfortunately the response to the second
survey was very disappointing, with only 35 of
the 84 targeted households (42%) completing
and returning the questionnaire. All those 35
households claimed to be using the new
kerbside scheme.
The second questionnaire was also designed to
examine other waste management behaviours
as well as recycling behaviours and attitudes.
All the questions were structured to provide
quantitative performance indicators that would
quantify any changes that occurred when the
new scheme was launched.
The developed indicator set used in the study is
described in full in table 8.13 in section 8.4.8 of
this report. All indicators are expressed as
percentages of the maximum possible „scores‟
for each performance category.
The whole study had been designed to test
features of a number of attitude/ behaviour
models, specifically how attitudes, attitude
changes and knowledge linked into behavioural
changes. The original intentions for the after
surveys were to provide comparative analyses
across three distinct categories of behaviour:
 Kerbside recyclers who had recycled
prior to the introduction of the kerbside
scheme
 New recruits to recycling
 Continuing non-recyclers
In the event, only the first two groups could be
studied. There were no respondents from the
third group. In total, nineteen new recruits and
sixteen prior recyclers were identified from the
questionnaire returns.
Understanding Recycling Behaviour 2
8.5.2 Whole Sample Analyses
Table 8.14 lists the results for the cohort
sample before and after the kerbside
introduction. The full results from the first
survey (99 households) are reproduced from
table 8.13 for comparison.
The results clearly show pronounced increases
in material capture across all the materials
collected (paper, glass, aluminium and ferrous
cans). Significant increases were seen for the
prior recyclers as well as the new recruits.
Respondents also claimed to be recycling more
plastic bottles, which were not targeted by the
kerbside scheme. The majority of this increase
was due to five respondents claiming to recycle
plastic bottles through the kerbside scheme. It
is not known whether those respondents were
simply exaggerating the extent of what they
recycled, or were genuinely introducing
contrary material into the kerbside collection.
The answer can only be determined through
compositional analysis of their recyclate.
Overall, the data provided no substantive
evidence that there had been any knock-on
effect on plastic recycling at bring sites.
Eighty-eight percent of the sample claimed to
recycle newspapers, 77 - 83% recycled bottles
and cans, 63 - 72% recycled magazines, junk
mail and glass, jars, whilst only 26% of
respondents recycled textiles through the
scheme. However, a significant proportion of
the overall sample (20%) were still recycling
textiles elsewhere. The other main materials
being recycled elsewhere were carrier bags (by
20% of the sample) and green waste (by 26%).
Less than 10% of the sample claimed to
recycle any other materials.
The main contrary materials introduced into the
kerbside collection were plastic bottles (claimed
by 14% of respondents) and aluminium foil
(claimed by 8%). However these results may
be skewed by the claims from two respondents
(6% of the sample) who appeared particularly
suspect in what they claimed to recycle. As well
as plastic bottles and aluminium foil, they also
claimed to recycle cardboard, carrier bags,
batteries and all of their garden waste through
the kerbside scheme.
117
 Understanding Recycling Behaviour 2

The knowledge of basic waste facts and of
ways to reduce and recycle [tested through an
unprompted open question] were very low
throughout. These types of knowledge appear
almost irrelevant to recycling behaviours. The
claims to participate in specific designated
reduce/ reuse behaviours indicated much
higher levels of activity than would be expected
according to the expressed knowledge. Whilst
there may be some exaggeration amongst the
replies, it is also likely that respondents might
not have made the connection between those
behaviours and the principles of reduce and
reuse. Overall, the claimed participation in
reduce/ reuse activities showed a marginal
increase from the before to the after survey.
Management practices for garden waste did not
change between the two surveys.
Not unexpectedly, the perceptions of local
recycling facilitation improved with the new
kerbside scheme. The combined indicator for
self-efficacy and awareness of consequences
also increased between the two surveys,
mainly through large increases in the self-
efficacy components of the indicator.
Respondents now felt more comfortable with
their own procedural knowledge, though they
did not necessarily claim any better knowledge
of the wider ramifications of recycling. There
were only marginal changes in the respondents
[negative] perceptions about recycling and in
their acceptances of personal responsibility for
recycling. Overall, no attitude change proved to
be statistically significant.
Three out of the seven respondents who
claimed to recycle contrary material through the
kerbside scheme also returned quite extreme
attitudes towards recycling. One respondent
scored maximum marks for all four attitudes
tested. That respondent‟s only digression was
to put foil into the kerbside collection. He/she
did not engage in high levels of reduce/ reuse
activities nor participate in environmentally
conscious garden waste management. The two
respondents who claimed to be using the
kerbside scheme for nearly everything scored
very high marks for their self-efficacy and
awareness of consequences. Both also shared
strong negative perceptions about recycling,
particularly in terms of the effort, storage space
and knowledge required. Only one of the above
three respondents has recycled prior to the
kerbside scheme being introduced.
Segmentation of the sample by demographic
groups is not reported for the second survey.
The results were not statistically representative
due to the small sample sizes involved.

Table 8.14 Recycling Performance Indicators Across the Introduction of Kerbside Recycling –
All Indicators Expressed as Percentages
Performance Indicator Cohort Sample Full first survey
Before After Recyclers Non-
recyclers

P1 Waste Facts 8.6 9.3 8.5 5.9

P2 Local provision - - 48.9 19.5
P3 Reduce/reuse knowledge 2.9 3.8 2.4 1.1

P4 Recycling participation* 45 (31) 100 (93) 96 (88) 0

P5 No garden waste to bin 26 24 39.5 31.4
P6 Reduce/reuse activity 52.9 58.6 58.3 43.6

P7 Newspaper capture 30.9 92.7 41.8 0

Glass bottle Capture 27.4 78.8 73.2 0
Aluminium can capture 11.9 73.3 48.5 0
Ferrous can capture 3.0 70.6 19.5 0
Plastic bottle capture 1.5 15.8 17.2 0

P8 Acc. of responsibility 58.8 61.2 64.7 54.1

P9 [Lack of] Negative perceptions 66.9 65.3 65.0 57.5
P10 Self eff./ aware of cons 63.9 72.5 61.3 53.3
P11 Facilitating conditions 42.6 50.4 58.7 50.0
* Bracketed figures give % participating with at least 50% of at least one material

118
 Understanding Recycling Behaviour 2

8.5.3 Prior Recyclers and New
Recruits
Table 8.15 lists the before and after
performance indicators separately for the prior
recyclers and for the new recruits. It is very
noticeable from these results that the current
recycling performances and attitudes of the
prior recyclers still exceed those of the new
recruits. Capture levels amongst the prior
recyclers are all higher than those recorded for
the new recruits. The pre-existing attitude
differentials were maintained across the
scheme introduction with weaker attitudes still
being recorded for the new recruits after the
event. In contrast, the garden waste
management behaviours of prior recyclers and
new recruits remained comparable throughout,
and the new recruits appeared to catch up in
their practice of reduce/ reuse activities.
It is also noticeable that, prior to the
introduction, the new recruits held stronger
prior attitudes of self-efficacy and awareness of
consequences and held weaker negative
perceptions about recycling compared to the
full complement of non-recyclers in the first
survey (see table 8.14). The new recruits also
expressed poorer perceptions of local
facilitation than did the average non-recycler.
These observations add some supporting
evidence that lack of negative perceptions and
strong levels of relevant knowledge (and
awareness) could help determine take-up,
when necessary facilitations are put into place.
However, nothing more substantive can be
said, as we have no knowledge about the prior
attitudes of the non-takers. They cannot be
distinguished separately from the other non-
respondents to the questionnaire.
Tables 8.16a-d illustrate the scales of attitude
changes recorded for the prior recyclers and for
the new recruits, now broken down into
individual attitudes. Whilst the total scores for
acceptance of responsibility and for negative
perceptions might have changed very little with
the new kerbside scheme, the individual
components that make up those constructs
could have undergone more substantial
changes.
Prior recyclers strengthened their attitudes that
it is ones duty to recycle and that one should do
it even if others do not. Perhaps those changes
could have arisen out of the newly experienced
visible messages put out by the kerbside
scheme. Recyclers would realise that many of
their neighbours were not recycling. New
recruits relaxed their expectations that the
council should be solely responsible for
recycling and appeared to become more
conscious of their own behaviours. New
recruits felt knowledge to be important to
recycling whereas prior recyclers saw their
household storage demands increasing with
the new kerbside scheme.

Table 8.15 Recycling Performance Indicators across the Introduction of Kerbside

Recycling: Prior Recyclers versus New Recruits – All Indicators expressed as Percentages
Performance Indicator Prior Recyclers New Recruits
Before After Before After

P5 No garden waste to bin 26 23 26 25

P6 Reduce/reuse activity 59 57 47 59

P7 Newspaper capture 45 100 0 85

Glass bottle Capture 66 82 0 75
Aluminium can capture 29 84 0 62
Ferrous can capture 7 79 0 54
Plastic bottle capture 4 21 0 10

P8 Acc. of responsibility 64 64 54 59
P9 [Lack of] Negative 71 71 63 61
perceptions
P10 Self eff./ aware of cons 69 75 60 70
P11 Facilitating conditions 47 55 38 47

Not surprisingly, both groups perceived a receiving procedural information on how, what
general improvement in local recycling and where to recycle.
facilitation and felt more self-confident after

119
 Understanding Recycling Behaviour 2

Table 8.16 a-d Attitude Changes across the Introduction of a Kerbside Recycling Scheme
Acceptance of Responsibility Prior New
recyclers Recruits

It is your duty to recycle ++

It is not worthwhile recycling unless everyone does it 00
You should try to encourage others to recycle
You feel guilty when you do not recycle 0 +
The responsibility for recycling should lie with the local council + ++
It doesn‟t matter if you don‟t recycle as long as others do +
You should not have to sort your rubbish 0
You would be embarrassed if a neighbour saw you throwing out ++
recyclable waste
++ Strengthened by >10% absolute, + = > 5%, 0 = weakened by > 5%, 00= weakened by > 10%

[Lack of] Negative Perceptions Prior New

recyclers Recruits

Recycling takes a lot of time 0

Recycling takes a lot of effort
Recycling needs a lot of storage space 00
Recycling is not worthwhile if you do not have much waste +
Recycling can be unpleasant +
Recycling needs a lot of knowledge 00

Self-Efficacy and Awareness of Consequences Prior New

recyclers Recruits

You have enough information to enable you to recycle correctly + ++

You are happy that you know the environmental reasons why you should +
recycle
You are happy that you know why the Government wants you to recycle ++
You understand what happens to your waste if you recycle it 0
You are happy that you know where to recycle your material +++ +++
You are happy that when people give waste for recycling that the waste is
actually recycled
+++ = > 20% change

Local Facilitating Conditions Prior New

recyclers Recruits

The local recycling facilities are not convenient for me to use + +

The local recycling facilities are poorly serviced ++ +

8.5.4 General Comments
Conclusions
Overall, 86% of respondents rated the new
scheme to be either excellent or good.
Seventy-four percent considered that the
information provided was also good or
excellent. When asked what would make them
recycle more, 31% of the sample considered
and that plastic and cardboard could be included in
the kerbside scheme. All except one
respondent expressed their intention to
continue to use the scheme. That lone
respondent was under the misapprehension
that households would be charged for its use in
the future.

120

Perhaps the main outcome from the before-
after survey was the finding that some attitudes
had changed substantially between the two
events. All the observed changes appear to
have plausible explanations, though such
explanations cannot be proven at this stage.
The major attitude change that was observed
was a large step change in self-efficacy,
presumably brought about through the
information provision at the launch. Attitude
changes appear to have occurred just as much
8.6 Normative Influences
8.6.1 Literature Review
The potential influence of others in shaping
recycling behaviours was reviewed in depth in
volume 1 of Understanding Recycling
Behaviour. The main points of that review are
summarised here. The reader is referred to
volume 1 for a more detailed analysis.
In general, studies showed that recyclers
perceived greater social pressures to recycle
than did the non-recyclers. Oskamp et al.
(1991) report that both friends and neighbours
can exert significant social pressures. Vining
and Ebro (1989) did not consider family
pressures to be important, however Granzin
and Olsen (1991) found household influences
to be the strongest influence. Jackson et al.
(1993) saw friends exerting greater influences
than family or others in their study. Spaccarelli
et al. (1989), however, consider that
neighbours may be the most important source
of peer group pressure.
Vining, Linn and Burdge (1992) found that
social pressure was not a significant motivation
to recycle in their study of a mature recycling
scheme with a very high participation. Taylor
and Todd (1995) argued that with a mature
programme, people have already had time to
develop strong [internal] attitudes and are less
susceptible to external influence. It was
considered that normative influences would
operate most during the early stages of a
recycling programme.
Normative influences can operate directly
through social dialogue with friends and
neighbours, or indirectly through observation
and copying. The second effect can be
particularly pronounced in kerbside recycling
schemes where the visibility of the behaviour is
Understanding Recycling Behaviour 2
amongst already practicing recyclers as
amongst the new recruits. The attitudes of the
prior recyclers remained stronger than those of
the new recruits, both before and after the
change.
However, when assessing these results, it
should be borne in mind that they were based
on a relatively small sample group. Further
research based on larger sample populations
will be needed to confirm the findings.
high (Oskamp et al., 1991; Vining and Ebreo,
1992). However, if may take a certain number
of homes to start putting out recyclables before
the rest see recycling is popular and start to
recycle themselves (Salimando, 1987). At that
stage, people might feel embarrassed not to
recycle.
Normative influence interventions have been
utilised to try to increase recycling rates. The
approach involves local prompting by
indigenous committed, and locally-respected
individuals. It is often termed the „block leader‟
approach
Block leaders are thought to be effective
because they serve as initiators of social norms
and provide a high degree of personal contact
(Shultz et al. 1995). It has also been suggested
that block leaders might act as role models or
provide an incentive of social recognition for
pro-recycling behaviour (Burn (1991)).
8.6.2 Previous Research at the
University of Paisley
Questionnaire survey results produced mixed
results on the importance of social influences
and pressures. When asked, the majority of
respondents said that they did notice whether
or not their neighbours had set out a container
on recycling day. However, the stimulus that
friends or neighbours had encouraged them to
recycle almost always ranked bottom of the
expressed reasons why they recycled.
Normative influence is difficult to measure other
than through direct questioning. Otherwise, it is
only possible to draw indirect inferences that
such influences have occurred. Positive
121

inferences follow from the following
observations (Tucker, 2001):
 Neighbourhoods of relatively „unfavourable‟
demographics located next to, or co-mingled
with, high [kerbside] participation areas
have shown participation rates much higher
than expected from their demographics.
 Areas with very high set out have been
found to contain a much higher than normal
number of households setting out very small
quantities of material. (It was considered
that the desire to be seen to set out may
increase recycling frequency more than
yield or participation).
 Closely-contained communities such as
blocks of flats and cul-de-sacs were often
very polarised in their behaviours – almost
all setting out or almost none setting out.
8.6.3 New Results
A new study of normative influences has been
carried out in Renfrewshire, Scotland. Other
aspects of the Renfrewshire survey have been
presented elsewhere in this report. In the new
survey, the questions asked were simply
whether or not households talked to their
neighbours about recycling.
In total, a quarter of the households stated that
they did talk to their neighbours about
recycling. However dialogue was relatively
infrequent. Very few talked more regularly than
once a month (table 8.17). Very few non-
recyclers talked about recycling. The highest
amounts of dialogue were between recyclers of
longest experience. Those differences were
statistically significant.
The percentages of households talking to
neighbours were then compiled by postcode
area. The results show a degree of polarisation
Table 8.17 Talking to Neighbours on an Individual Basis
All groups Prior*
recyclers recruits*
Never 220 (75%) 53 (62%)
Less than monthly 64 (22%) 30 (35%)
More than monthly 9 ( 3%) 3 ( 3%)
* Kerbside collections were introduced nine months before the questionnaire was administered. Prior recyclers
were scheme users who had recycled previously. New recruits had never recycled before.
122
Understanding Recycling Behaviour 2
into low dialogue areas (0-20% of residents
talking), and high dialogue areas (35-60%
talking) (figure 8.5). Sixty-one percent of the
non-recyclers resided in the low dialogue areas
compared to 46% of the recyclers.
The lowest levels of dialogue were found in
families with young children (15%) or with older
children (10%) and the highest levels in the
more mature but not yet retired households
(38%) and amongst Acorn group 9 – private
flats, elderly people (38%). It is noted that the
latter group are likely to be a reasonably
closely-knit community. Larger households
were significantly less likely to enter into social
dialogue about recycling than were single or
two-person households.
It was also found that households who had
experienced problems with the scheme, and
who were talking to their neighbours, were
more likely to be continuing to use the scheme
than those who were not talking (table 8.18).
Sixty-five percent of problem holders who were
talking voiced problems concerning the
collections and the service, whereas 67% of
problem holders not talking had grievances of
personal difficulties with the recycling container.
It may be that it is easier to talk about the
shortcomings of the council than to talk about
one‟s own difficulties.
In conclusion, the new results illustrate how the
levels of social dialogue about recycling can
vary demographically and spatially across the
community. Families with children present
appear to engage in the least dialogue. Not
surprisingly, almost no non-recyclers talk about
recycling. Perhaps a chat with a recycling
neighbour could make all the difference.
Talking about problems does seem to alleviate
drop out though people seem more reticent to
discuss personal problems with the recycling
box than they are to discuss problems with the
collection service provided.
New Bring Non-
recyclers recyclers
119 (78%) 7 (78%) 41 (91%)
28 (18%) 2 (22%) 4 ( 9%)
6 ( 4%) 0 ( 0%) 0 ( 0%)

Table 8.18 Talking to Neighbours – Problem
Holders
Talking Still recycling Dropped out
No 29 (59%) 11 (73%)
Yes 20 (41%) 4 (27%)
8.7 Summary and Conclusions
8.7.1 General
The research reported in this section has tried
to look beyond the customary excuses and
other reasons for recycling or for not recycling.
Doubt is raised in the research literature as to
whether the reasons voiced in surveys are the
true or complete justifications of peoples‟
behaviours. Psychological research suggests
that more fundamental attitudes, values and
beliefs are instrumental in determining
behaviours. Attitudes showing an acceptance
of responsibility and an awareness of
consequences of recycling are thought to be
important if recycling is to be considered an
altruistic behaviour. An alternative model of
planned behaviour highlights self-efficacy and
facilitating conditions to be important. Research
into repeated behaviours has shown that
different attitudes might affect initiation and
persistence of behaviours.
Having a strong personal norm towards
recycling effectively governs one‟s recycling
behaviour. A strong social norm towards
recycling can help develop that personal norm,
as can the internalisation of key supporting
attitudes. The central research question is
“Which attitudes?”
The study focused on testing the four attitudes
identified in the psychological models, together
with a fifth construct „[Absence of] negative
perceptions [about recycling]‟.
Understanding Recycling Behaviour 2
Figure 8.5 Distribution of Levels of Social
Dialogue by Postcode
Number of Instances
4
3
2
1
0
1 6 11 16 21 26 31 36 41 46 51 56 61
% of households talking
8.7.2 Recyclers versus Non-
recyclers
Whilst most attitudes tested were stronger
amongst recyclers than amongst non-recyclers,
not all differences were significant, with
considerable overlaps in attitudes between the
two groups. A full separation of recyclers from
non-recyclers could not be made on the basis
of their attitudes or of any combination of their
attitudes. Around three quarters of the
behaviours could be predicted from the
attitudes held.
Many attitudes appeared to be reasonably
long-lived and persistent. There were
indications that some of those attitudes might
be pre-determinants of whether behaviours
would be taken-up, including strongly-held
negative perceptions about home composting,
and possibly the acceptance of a personal
responsibility for recycling, though the latter
was far less clear.
Other attitudes appear to become shaped
quickly upon gaining experience. Notions of
personal cost (time, effort, storage demands
etc.) tend to be set mainly through initial
experiences. Other attitudes can also evolve as
circumstances change. Moving from a relatively
low visibility behaviour (bring recycling) to a
much higher visibility behaviour (kerbside
recycling) will convey new messages (i) to
others about your own recycling behaviour, (ii)
to you about others‟ recycling [or lack of
recycling] behaviours. Such messages might
affect one‟s perceptions about individual and
collective responsibility.
123

The attitude of self-efficacy (the self-belief that
one is able to recycle effectively and correctly)
is also important though it is not clear whether
this is a pre-determinant for the initiation of new
behaviour. Having no prior composting
knowledge was found to be no barrier to taking
up composting.
An over-inflated self-efficacy, however, can
lead to situations where perceived knowledge
is greater than actual knowledge („know-it-alls‟)
which could compromise recycling performance
and cause exaggerated claims.
A weakening of certain attitudes can be a pre-
cursor to dropping-out from undertaking the
behaviour. Strengthening of negative
perceptions was identified as a cause for drop-
out in home composting. Recycling drop-outs
showed weaker levels of acceptance of
responsibility than continuing users. However,
in such circumstances, it is difficult to ascertain
124
Understanding Recycling Behaviour 2
whether those attitudes weakened because of
an adverse experience or whether they were
low anyway, making the householder more
susceptible to dropping out when an adverse
experience was faced. The latter hypothesis
would tie in with the observations that new
recruits might have weaker attitudes than are
held by experienced recyclers and that the
main period of drop-out is during the early
months of a recycling programme.
Whilst the research has cast some new light on
attitude/ behaviour dynamics, there is still much
to understand. That understanding can be
complicated by the exaggerated and erroneous
claims made by some people. The extent and
implications of those claims are examined in
more detail in section 9 of this report.
Further discussion on the dynamics of recycling
attitudes and behaviours can be found in
sections 11.4 and 11.5.

9. How Accurate are Waste Behaviour Surveys?
9.1 Introduction
Determining attitudes can only be
accomplished through direct questioning.
Normally that involves questionnaire surveys.
Indeed, many waste attitude/ behaviour
surveys have been undertaken by many
organisations for many purposes. But how
accurate are the results? The information
returned by surveys depends heavily on
interpretation and honesty. Why should
recycling surveys fare any better than [say]
opinion polls on voting intentions?
9.1.1 Review
In most recycling attitude/ behaviour surveys,
self-reported behaviours are taken at their face
value. However, it must be recognised that the
results may be subject to error and bias. Firstly,
respondents often tend to exaggerate the
extent of their behaviours when those
behaviours are perceived to be morally good
(e.g. Geller, 1981; Warriner et al., 1984).
Secondly, recyclers are, in general, more likely
to complete a questionnaire on recycling
compared with non-recyclers. Both effects can
lead to an over-estimation of the levels of
recycling activity. In addition, accuracy of recall
can introduce further errors (Warriner et al.,
1984). The respondent must be able to answer
the question correctly and the effects of time
and memory and/ or a simple lack of
knowledge can lead to incorrect responses. If
questions are phrased in a way that is
confusing to the respondents, the likelihood of
inaccurate responses increases. Differences
between self-reported and actual recycling
behaviours recorded by Werner and Makela
(1998), for example, may have been due to the
householders‟ misinterpretation of their survey
questions.
Despite a general recognition that these effects
can happen, there has been relatively little
investigation into the extent to which they do
happen for waste management behaviours.
The few studies carried out have produced
mixed results. In two studies by Corral-Verdugo
(1997) and Corral-Verdugo and Figueredo
(1999) the reuse of materials was examined by
both self-reporting (quantitative and qualitative)
and direct measurement. Correlations were
found to be significant, albeit not particularly
high, between measurements and quantitative
self-reporting. Correlations were higher in the
Understanding Recycling Behaviour 2
1999 study where the respondents were told in
advance that their behaviour would be
observed. McGuire (1984) found no
correlations between self-reported recycling
rates and refuse analysis data for paper and
aluminium can recycling. It was concluded that
this was because self-reported responses may
be more likely to reflect attitudes rather than
behaviours. Rathje (1984) also considered that
self-reports primarily reflect beliefs and
perceptions more than actual behaviour, and
went on to demonstrate disparities between
„what we should do‟, „what we want to do‟, and
„what we actually do‟ (Rathje, 1989). Warriner
et al. (1984) amplify this further, considering
that if respondents feel, even slightly, that the
survey instrument is evaluating them as
individuals, the possibility of exaggerating more
favourable qualities increases, so intentions
and wishes rather than real behaviours get
reported.
In other studies, Gamba and Oskamp (1994)
found that 99% of respondents claimed to
participate at least once in every five kerbside
recycling collections, whilst 52% claimed to
participate every week. The city administration
estimated that the real figures were around
91% and 20% respectively. Perrin and Barton
(2000) compared the reported and monitored
participation rates for two kerbside collection
areas in Leeds and found discrepancies
between claimed and actual behaviours of
20.2% and 12.0% respectively. Those making
the false claims were found to be more likely to
state they used bring schemes, prefer the
collection to be more frequent and be more
wealthy. Williams and Kelly (2003) noted that
self-reported participations in the Borough of
Wyre blue-bag recycling scheme exceeded
66% whilst the council‟s own figures recorded
participations of just 37%.
Barker et al. (1994) performed a controlled
experiment to gauge the extent of over
estimation of self-reports. In this experiment
attitudes and self-reported recycling behaviours
of 493 college students were surveyed. Around
95% of students professed holding pro-
recycling attitudes and 88% claimed to use the
college recycling facilities and to actively seek
out those facilities. The students were then
mailed a bogus circular. Of the 308 circulars
later recovered, only 14% were found in the
recycling bin, the remainder were recovered
125

from trash cans or found on the floor or left
elsewhere in the vicinity.
Misrecall of past behaviours was also noted in
an investigation of the range of materials being
home-composted (Tucker and Speirs, 2001).
Volunteers logging all additions into their
compost bin recorded a wider range of
materials being composted than they
confessed to in a parallel questionnaire survey.
Misrecall, however, may be more than simple
forgetfulness. Vining, Linn and Burdge (1992)
considered that in time, participants will tend to
simplify their reasons why they participate and
condense their rationale into a small number of
terms (i.e. show a tendency to simplify their
cognitive load). In their studies, they found that
reasons given for recycling had been simplified
as altruism and convenience. Similar effects
were noticed in survey undertaken by Tucker
(1999).
Discrepancies between measured and claimed
numbers of materials recycled were shown in
the study of Perrin and Barton (2000). In that
study, householders were asked to specify
which of 14 materials they recycled through the
kerbside multi-material collection and which
they consigned to the residual waste bin. On
analysing of the contents of the recycling and
residual waste bins, the actual to claimed
recovery ratios were found to be 0.7 for
newspapers and magazines, and 0.5 or less for
plastic bottles, card and cans. Discrepancies
between the numbers of each item observed to
be recycled and the amounts claimed to be
recycled have been noted in surveys of both
kerbside and bring recyclers (Tucker et al.,
1998). Kerbside recyclers in South Ayrshire
claimed on average to recycle 90% of their
newspapers and 53% of their magazines,
though only 85% and 42% of their stated
consumptions were seen to be recycled. At a
drop off point in the same district, the survey
found an 18% discrepancy between the
numbers of glass bottles recycled and the
numbers claimed to be recycled.
The survey of the kerbside collection in South
Ayrshire additionally showed a strong bias in
the percentage returns from scheme users and
non-users. Postal questionnaire returns were
received from 464 scheme users and 111 non-
users of the scheme, representing 73% of
scheme users but just 12% of non-users
determined through a parallel participation
monitoring study. If taken at its face value, the
questionnaire would imply a participation rate of
80%. The monitored participation rate was 41%
(Tucker, 2001).
126
Understanding Recycling Behaviour 2
9.1.2 New Evaluations
The results of the waste attitude surveys
described in section 8.4 of this report were
reported at their face value although it was
considered that some of the claims could be
suspect. Those claims will now be scrutinised
in more detail to assess the extent of any
exaggeration. Opportunity has also been taken
to re-visit other surveys undertaken by the
University of Paisley, to bring in further
evidence of the likely levels of exaggeration.
The other surveys included studies undertaken
in two village communities in Fylde Borough,
Lancashire in 2000, and the survey in
Renfrewshire, Scotland in 2001 (which is
referred to throughout this report)). The Fylde
villages were serviced by a kerbside paper
collection whereas the Renfrewshire sample
had multi-material kerbside provision.
The 2000 and 2001 surveys were run in
tandem with extensive set out monitoring
programmes, which allowed direct comparisons
of actual and reported behaviours to be made.
This was not possible in the Clackmannanshire
[before] survey as, at that time, there was no
kerbside provision in the survey area, and it
would be impossible to monitor the
respondents behaviours at bring sites. To trap
possible false claims indirect measures had to
be taken. The trap was as follows: A paired
question, well separated in the survey, asked
respondents to name sites where they could
recycle named materials, then later asked what
materials the respondents actually recycled.
Inconsistencies between recycling a given
material and not being able to name a valid
recycling point for that material was taken as an
indicator of suspect behaviour.
Whilst the overall contents of the three surveys
differed according to their designed purpose,
some key questions were commonly asked
across all surveys. These related to the socio-
demographics of the household and to the held
attitudes on “Acceptance of responsibility for
recycling”.
In addition to asking whether households used
their kerbside recycling service, the
Renfrewshire sample were asked how many
newspaper, magazines, plastic bottles and
drinks cans they brought into the house each
week and the proportion of those they recycled
through the scheme. Compositional analyses of
the contents of the kerbside boxes were
undertaken in Renfrewshire, which allowed a
direct comparison to be made between items
claimed to be recycled and those found to be

recycled. In the study 126 boxes were analysed
from households who completed the
9.2 Participation Behaviours
9.2.1 Results
In Renfrewshire, 15% of self-reported recyclers
were not observed to recycle over 10
consecutive kerbside collections (table 9.1). In
Fylde, the percentages differed between the
two communities, with 44% and 16%
respectively of those claiming to use the
kerbside scheme not being monitored as doing
so over a 16 week period. In
Clackmannanshire, fourteen percent of
confessed recyclers could not name the
location of any recycling outlets.
In Clackmannanshire, the knowledge of valid
recycling outlets varied between materials, with
the highest awareness being shown for glass
banks (which were most plentiful locally) and
the lowest awareness for cardboard recycling
(for which there were no local recycling outlets)
(table 9.2). The highlighted results, which
depict the suspect claims, show that cardboard
and newspaper were the two materials with the
highest numbers of invalid claims. A more
detailed breakdown of the inconsistent
responses is given in table 9.3, where it is
shown that the 15 individuals making
inconsistent statements split into two distinct
groups. The first group did not know where to
recycle any material. This group mostly claimed
to recycle 3 or less materials, usually recycling
50% or less of those materials. The second
group specified invalid locations for one or
more recycling points (usually cardboard). This
group generally claimed to recycle 3 or more
materials and could correctly name the
recycling points for most of them. When
recycling, they tended to recycle 50% or more
of their available material. The first class of
Table 9.1 Numbers (Percentages) of self-reported Recyclers who may have made False Claims
Fylde 1 Fylde 2
Monitored as recycling 29
Not monitored 23
Aware of outlets - -
Unaware of outlets - -
Percentage of suspects 44%
Understanding Recycling Behaviour 2
questionnaire.
behaviours are hypothesised to be non-
recyclers who are falsely claiming to do so,
whilst the second class are hypothesised to
comprise bona-fide recyclers who may be
exaggerating the extent of their recycling
activities. This second class of behaviour will
be discussed later in section 9.3. The
remainder of this section concentrates on
claimants who are suspected to be non-
recyclers.
Table 9.3 showed that the suspected non-
recyclers („inconsistents‟) generally claimed
only to recycle fairly modest proportions of their
recyclable materials. This trend was also seen
clearly in Renfrewshire survey. Tables 9.4a-d
compare the claimed proportions of four
materials recycled by those not monitored
recycling with the proportions claimed by the
overall recycling population. The returns from
those claiming “none to recycle” have been
omitted from the tables. The results show that
lower percentages of inconsistents claimed to
recycle all their available materials, with higher
percentages of them claiming to recycle only
part of that material or none of it. The
inconsistents also claimed to recycle less
materials, on average 2.6 out of the 4 marker
materials compared with 3.1 materials recycled
by the overall recycling population (table 9.4e).
Overall, the picture of the inconsistent
behaviour is one where relatively modest levels
of recycling participation and recycling activity
are claimed. If the inconsistents are indeed
non-recyclers making false claims, then those
claims are relatively modest, effectively just
enough to register the claimant as a recycler.
Renfrew Clackmannan
21 199 -
4 36 -
- 42
- 7
15% 15% 14%
127
 Understanding Recycling Behaviour 2

Table 9.2 Self-claimed Recyclers Recycling Specific Materials versus Knowledge of Where to
Recycle Them (Clackmannanshire)
Know- Recycle News- Glass Drinks Cardboard Green Batteries
ledge material paper bottles cans waste
where to

No No 11 2 14 35 23 33
Yes No 10 6 18 1 14 10
No Yes 8 5 2 9 5 1
Yes Yes 20 36 15 4 7 5

Table 9.3 Proportion of Wastes Recycled at Inconsistently-specified Locations (15 cases,

Clackmannanshire)
Newspaper Glass Drinks Cardboard Green Batteries
bottles cans waste
1 0 0 0 0 0
0.25 0.25 0 0 0 0
0.25 0.25 0.25 0 0 0
0.75 0.5 0 0.75 0 0
0 0.25 0 0 1 0
0 0.25 0.25 0.25 1 0
0.25 0 0.25 0 1 0
0.75 1 0 0.5 0 0
0.5 0.75 0.5 0.5 0 0
0 1 0 1 0 0
1 1 1 1 1 0.25
0.5 0.5 0.5 0.5 1 0.5
1 1 0 0 0 1
1 1 0 1 0 0
1 1 1 0.5 1 0
Bold = Don‟t know/ nowhere to recycle
Bold italic = Incorrect location given

Table 9.4a Proportions of Newspaper Table 9.4c Proportions of Plastic Bottles

Recycled (Renfrewshire) Recycled (Renfrewshire)
% all % % all %
recyclers inconsistents recyclers inconsistents

None 6 11 None 16 20
A quarter 1 0 A quarter 5 9
About half 6 17 About half 6 20
Three quarters 6 17 Three quarters 3 3
All or nearly all 81 56 All or nearly all 71 49

Table 9.4b Proportions of Magazines Table 9.4d Proportions of Drinks Cans

Recycled (Renfrewshire) Recycled (Renfrewshire)
% all % % all %
recyclers inconsistents recyclers inconsistents

None 12 21 None 16 20
A quarter 4 8 A quarter 5 9
About half 6 17 About half 6 20
Three quarters 6 8 Three quarters 3 3
All or nearly all 73 46 All or nearly all 71 49

128
 Understanding Recycling Behaviour 2

Table 9.4e Numbers of Materials Recycled 9.2.2 Determinants Of Participation

(Renfrewshire) Behaviour
% all %
recyclers inconsistents There was no significant difference in social
class amongst recyclers, non-recyclers and
1 6 20 inconsistents in the Renfrewshire study.
2 16 9 However, sixty-nine percent of the
3 39 20 inconsistents were households with children
4 39 3 present, principally with young children in the
household. This loading of inconsistents into
Mean number 3.0 2.6 the younger family life stages was significant at
of materials 90% confidence.

The acceptance of responsibility of the

Also, inconsistents claimed to recycle less
frequently than did the overall recycling
population (table 9.4f), with significantly fewer
claiming every week participation in the
scheme. All the differences were statistically
significant.
Table 9.4f Frequency of Recycling
(Renfrewshire)
% all % sample sizes, significant differences could not
recyclers inconsistents
inconsistents was intermediate between those
of the scheme users and those of the non-
users of the schemes (figure 8.2, table 9.5,
figure 9.1), and these trends were consistent
across all three communities. The differences
between the inconsistents and the confirmed
users and between the inconsistents and the
confirmed non-users were both statistically
significant in the Renfrewshire survey. In the
other two communities, with their much smaller
be confirmed.

Less frequently 4 14 The other attitudes tested in the

Fortnightly 6 25 Clackmannanshire study, self-efficacy and
Most weeks 17 36 awareness of consequences, and negative
Every week 63 25 perceptions, gave less clear trends.

It is additionally noted here that the Fylde

communities showed anomalously high
patronage of their local bring facilities rather
than using the provided kerbside service (see,
Tucker, 2001 for details), so the attitudes of
kerbside non-users there may be elevated.

Table 9.5 Acceptance of Responsibility Attitude Scores

Community Category Mean Median Standard
deviation
Fylde 1
n = 34 Non-user 30.3 30 7.1
20 Inconsistent 32.0 32 4.6
28 User 33.9 33 4.6

Renfrewshire
n = 38 Non-user 29.3 29 6.6
32 Inconsistent 32.3 33 4.9
178 User 34.5 35 5.2

Clackmannan
n = 40 Non-user 25.3 24 4.4
7 Inconsistent 26.8 27 5.6
33 User 28.8 31 5.5

129
 Understanding Recycling Behaviour 2

Figure 9.1 Distributions of “Acceptance of Responsibility Attitudes” (Renfrewshire)

Cumulative % of sample

120
100
80
60
40
20
0
9 12 15 18 21 24 27 30 33 36 39 42 45
Attitude score

Non-users Inconsistents Users

9.3 Exaggerated Behaviours

9.3 Exaggerated Behaviours newspapers (and also plastic bottles) did so. A
9.3.1 Results larger proportion of those who stated they
9.3 Exaggerated Behaviours would recycle cans (and also magazines) did
It was shown earlier, that some households, not do so [at least on the collection monitored].
who may well be recycling quite strongly, may The levels of magazines brought in to the
also be exaggerating the extent of their households were low in most cases and it could
recycling activities. Whilst the be that there might not be any available for
Clackmannanshire study could only trap those recycling on some collections. Most self-
exaggerating the number of materials they reported can-recycling households however
recycled, it is recognised that there could also claimed to consume several cans each week
be an exaggeration in the claimed amounts of and claimed to recycle many of those though
the materials they actually recycled. the kerbside scheme. As such, a weekly
presence in the recycling bin would be
In Renfrewshire, there was the opportunity to expected for most can recycling households in
compare the expected quantities recycled the survey.
based on questionnaire returns with the
materials actually found in the recycling boxes.
The first part of the investigation looked at Table 9.6a Numbers of Residents
inconsistencies between the claimed and actual Recycling Newspapers
numbers of different materials recycled. The Survey Monitored
second part of the investigation looked at In box Not in box
inconsistencies between the expected and
measured numbers of items of each material Recycle it 110 6
found in the recycling box. However, in Don't recycle it 5 5
assessing the results of these analyses, it is
stressed that the compositional analysis was a
„one-off‟ investigation for one particular week. It Table 9.6b Numbers of Residents
did not take account of any temporal Recycling Drinks Cans
fluctuations in the amount of materials being Survey Monitored
recycled. In box Not in box

Tables 9.6 a-b highlight the inconsistencies Recycle it 37 35

between numbers of materials claimed to be Don't recycle it 7 48
recycled and those found in the recycling box.
Most who stated they were recycling

130

At the other extreme, some households who
stated they did not have any of a certain
material to recycle were actually found to have
recycled those materials. The reasons for such
behaviours remain unclear. It is considered that
plain forgetfulness is unlikely. More likely, the
discrepancy may be due to residents‟ personal
interpretations on what constitutes recycling, or
what „none to recycle‟ strictly means. For
example, a household that does not buy
newspapers may overlook the free newspaper
when making a claim that they have none to
recycle.
A significant difference was also seen in the
number of different materials households stated
they recycled through the kerbside scheme and
how many materials were found in their
recycling boxes (table 9.7). From an average of
3.03 materials expected to be present in the
recycling boxes, only an average of 2.36 were
found, i.e. only 78% of what was expected. Just
31% of the households were observed to
recycle exactly the same number of materials
that they claimed to recycle. Fifty-eight percent
were found to recycle less than they claimed,
whilst 11% were actually observed to recycle
more than they claimed (figure 9.2). Overall,
however, it seems that there may be a general
tendency by many to exaggerate the levels of
their recycling activities, though it should again
be noted that some materials might not be
available for recycling in every collection, which
could account for some of the discrepancies.
Table 9.7 Number out of Four Marker
Materials Recycled - Stated v Monitored
Number of % Recycling
Materials Questionnaire Monitored
0 2 0 estimation of recycling activities.
1 5 18
2 21 36
3 30 37
4 41 8 counts which were derived from a combination
Mean 3.03 2.36
At the individual household level, there was
considerable variation in the number of items of
each material retrieved from individual recycling
boxes compared with the expected numbers of
those items. In undertaking these comparisons,
the expected item counts were estimated using
the following formula:
Understanding Recycling Behaviour 2
N = Consumed per week  Proportion of
available material recycled  Weeks between
set-outs
Figure 9.2 Exaggeration in Number of
Materials Recycled
40
35
% of respondents
30
25
20
15
10
5
0
-1 0 +1 +2 +3
Number of Materials
Exaggerated
Scatter plots of the relationships are shown in
figure 9.3 a-d, which demonstrate very poor
correspondences between the recovered and
the expected figures, particularly for the
magazine fraction. The fitted trend lines provide
a model that predicts 92% of expected
newspapers are recovered, 65% of the plastic
bottles, 50% of the drinks cans, and 37% of the
magazines. The total levels of the under-
recoveries aggregated over the whole sample
showed that 107% of expected numbers of
newspapers were recovered, 70% of
magazines, 81% of plastic bottles, and 49% of
drinks cans. Overall, the results confirmed a
systematic tendency towards an over-
In making these interpretations, it must be
recognised that these figures for under-
recovery are estimates: (i) based on expected
of three self-reported statistics, and (ii) based
on one „snapshot‟ measurement of actual
behaviour. There may be errors of recall or bias
within the former, and temporal anomalies
within the latter. Nevertheless a reasonable
correspondence was achieved for newspapers,
when averaged over the whole sample, though
significantly lower than expected recoveries
were seen for the other three materials. It
should be borne in mind here that it might be
easier to be more precise in enumerating
newspaper consumptions than it is to recall the
131
 Understanding Recycling Behaviour 2

numbers of plastic and metal containers that

one consumes. Plastic bottles y = 0.6519x
2
R = 0.208
It should also be noted, that on an individual 30

Number recovered
level, there were significant levels of under-
claiming as well as over-claiming. Part of the
20
discrepancies was found to be related to the
rather coarse assumptions made about
recycling frequencies where recycling „most 10
weeks‟ and „every week‟ were taken literally. A
comparison with the set-out monitoring results 0
showed that several of those claimants had not
actually recycled the previous week. Temporal 0 10 20 30
anomalies such as holidays and illness can Number expected
easily affect the regularity of individual
participations. Other anomalous events, e.g.
parties, might produce spikes in [say] drinks
container arisings, and so on. However it is Drinks cans y = 0.505x
considered that errors of recall may be the R2 = 0.305
major contributory factor to the observed

Number recovered
30
scatter in the household level results. The
25
questions posed may have asked for more
detail than could be remembered easily. 20
15
10
Figure 9.3 a-d Relationships between 5
Numbers of Items Recovered from and
0
Expected in Kerbside Boxes (Renfrewshire)
0 10 20 30
Newspapers y = 0.925x
2
Number expected
R = 0.2249
30
Number recovered

20
9.3.2 Determinants Of Exaggerated
10 Behaviours

0 For the analyses presented in this section, the

0 10 20 30 household behaviours of the Renfrewshire
sample were classified into three broad
Number expected
behavioural categories:

1. Where the observed item count was

Magazines y = 0.3741x
greater than 3 times higher than the
2 expected count
R = -0.0169
2. Where the observed and expected item
Number recovered

15
counts were comparable (within a factor of
3).
10 3. Where the expected count was greater
than 3 times higher than the observed
5 count

0 Category 3 thus records those who may have

0 5 10 15 made gross exaggerations in the extent of their
recycling activities. It filters out any marginal
Number expected
discrepancies that might be determined more
strongly by inaccuracies in recall.

132
 Understanding Recycling Behaviour 2

There was no significant difference in the social
class of those grossly over-claiming the
recycling of at least one of the four marker
materials. There were some indications that
there were higher proportions of over-claimants
amongst the „more mature but not yet retired‟
age group, and indications of relatively fewer
over-claimants amongst the retired population.
There was no indication of differences for any
of the other life cycle stages.
Table 9.8 shows the differences in the attitude
„Acceptance of responsibility‟ amongst the
three classified groups. A consistent trend is
seen between the strengths of attitudes and the
level of claim for all the materials investigated.
Over-claimants showed the strongest attitudes
whilst under-claimants showed the weakest,
however those differences were only
statistically significant for the newspaper and
magazines fractions. Over-claiming can and
plastic bottle recycling would appear to be
determined by factors other than acceptance of
responsibility.
A similar strengthening of attitudes is also seen
amongst the Clackmannanshire suspected
over-claimants (table 9.9). „Self-efficacy and
awareness of consequences‟ or „perceived
knowledge‟ was the most significant
discriminator between over-claimants and all
other behavioural categories. The holding such
an attitude would appear logically consistent
with that type of behaviour.

Table 9.8 Mean Acceptance of Responsibility Score (Range: 9 to 45) - Renfrewshire

Newspapers Magazines Plastic bottles Drinks cans

Category 1 32.9 (n=14) 33.0 (n=12) 34.7 (n=16) 33.0 (n=8)

Category 2 34.1 (n=92) 33.7 (n=66) 33.9 (n=73) 34.3 (n=78)
Category 3 36.1 (n=12) 35.8 (n=40) 35.1 (n=29) 35.0 (n=32)

Table 9.9 Attitude Differences between Suspected Over-claimants and Other Recyclers
(Clackmannanshire)
Acceptance of Self-efficacy & Negative
responsibility awareness of perceptions
(range 8-40) consequences (range 6-30)
(range 6-30)

Over-claimants (n=7) 30.9 25.5 21.6

Other recyclers (n=48) 28.7 20.7 21.7

9.4 Bias in Questionnaire Returns

This section now examines the hypothesis that
recyclers may be more likely than non-recyclers
to answer questionnaires on recycling, thereby
providing a further bias towards the over-
estimation of participations. The number of
returns from monitored recyclers and those not
monitored as recycling have been cross-
checked with the monitored recycling rates.
The Renfrewshire survey shows a 68%
participation rate based solely on the returns.
The monitored participation there was 63%. In
the two Fylde communities the figures were
41% based on returns compared to 42%
monitored, and 26% based on returns
compared to 23% monitored respectively. The
latter two differences were not statistically
significant. Overall, across the three surveys,
the biases in response rates proved to be
relatively modest (5% or less) which is much
less than that recorded previously in South
Ayrshire (80% to 41%; Tucker, 2001). It should
be noted that the South Ayrshire survey was
postal whereas the surveys giving the low
discrepancies were through doorstep delivery
and collection.

133

9.5 Summary, Discussion and Conclusions
This research has highlighted four factors that
can lead to misrepresentative information being
reported in waste attitude/ behaviour surveys,
and sought to quantify the scales of the errors
that might occur. The research centres on three
surveys addressing kerbside paper-only
recycling, kerbside multi-material recycling, and
bring recycling respectively. The factors
identified were:
(i) A bias amongst those answering the
survey with an under-representation of
non-recyclers.
In the surveys of the investigation, that bias
was found to be less than 5%, thought it was
noted that some surveys in the past returned a
much stronger bias. However, their respective
survey techniques differed. A direct contact
approach was employed in the current surveys,
whereas a postal survey provided the more
highly biased returns. The delivery mechanism
may be very important.
(ii) An exaggeration in the numbers of
households claiming to recycle.
Typically, the research found that around 15%
of self-reported recyclers might be suspected of
making false or exaggerated claims. An
independent study by Perrin and Barton (2000)
returned much the same conclusion. A study by
Williams and Kelly (2003) put the figure nearer
to 40%. Evidence remains equivocal as to
whether the suspects might be drawn
preferentially from given demographic
groupings. Whilst the possible motivations for
making false claims was not investigated fully,
it is considered likely that part of that motivation
may come from:
(a) the belief that recycling is morally „good‟
behaviour, and
(b) the belief that they should be undertaking
that behaviour.
This hypothesis gains support from the
reported attitudes of the suspects towards their
personal responsibility for recycling. Attitudes of
the suspect group were higher than the
attitudes of the self-reported non-recyclers,
though lower than the confirmed recycling
group.
It was also noted that the suspect groups only
claimed to participate in reasonably modest
levels of recycling activity. They did not claim to
134
Understanding Recycling Behaviour 2
recycle a lot of materials, nor did they claim to
recycle many of those materials, nor to recycle
them often. It was as if they were claiming just
enough to be registered as recycling. However,
it is still conceivable that they could be telling
the truth but had simply not recycled during the
period over which they were monitored (i.e. 16
and 10 weeks respectively in two of the surveys
considered here). However, in the third survey,
a suspicion of false claiming comes back to the
fore, through the claimants who could not
identify where they could recycle.
(iii) An exaggeration in the numbers of
different materials claimed to be
recycled.
There was a systematic tendency to overstate
the numbers of different materials recycled. For
the samples considered, up to 22% in over-
estimation was inferred. Previous studies have
also noted that recyclers claim to recycle a
wider range of materials than they actually do
when intercepted (Tucker, 2001). However,
those previous results all related to bring
recycling where it is conceivable that recyclers
may selectively take different materials with
them on each trip. It is harder to reconcile this
reasoning towards the mixed kerbside
collections of the current study, although it is
still plausible that some waste materials will not
be generated regularly at every collection.
(iv) An exaggeration in the numbers of items
of each material recycled.
There was a systematic tendency towards
over-estimating the numbers of items recycled
for separate material. Over-estimations of 50%
were inferred for drinks cans, 20% for plastic
bottles, though the numbers of newspapers
being recycled were estimated reasonably
correctly. Similar levels of discrepancy have
also been found in earlier research (Tucker et
al., 1998), Perrin and Barton (2000).
At the individual household level, there was a
considerable scatter in results, with many
households providing under-estimates as well
as over-estimates of their levels of activity.
Such variations can be expected when the
more detailed aspects of behaviour are
investigated. At that level, accuracies of recall
can significantly affect any questionnaire-
derived data, whilst week-by-week fluctuations
in waste arisings can have significant effects on
the observed data. This latter point was also

noted by Cote (1984) in a survey of food
consumption behaviours, whilst the former
corroborates the findings of Warriner et al.
(1984).
Stronger attitudes towards recycling were held
by those suspected of exaggerating their
recycling behaviour than were held by recyclers
less prone to exaggeration. There was some
evidence that a high perceived knowledge, or
expressed, knowledge may correlate strongly
with exaggeration. It could be hypothesised that
such respondents feel a need to demonstrate
that they are highly knowledgeable and highly
active, rather than admitting that they are not
quite that perfect. Again, perhaps they are
expressing what they perceive they should be
doing. This returns to the point of McGuire
(1984), Rathje (1984) and Warriner et al.
(1984), that self-reported responses are more
likely to reflect attitudes rather than behaviours.
On this assumption, it would appear that the
necessary attitudes to support increased
recycling performance may already be in place
for many households. Measures (or
interventions) to increase performance may
simply need to find ways to mobilise those
attitudes into behaviours without needing to
stimulate significant attitude changes in order to
do so. The relationship between attitudes in the
dynamics of behavioural change has been
addressed in section 8.5 and is discussed
further in sections 11.4 and 11.5 of this report.
Waste attitude and behaviour surveys are
normally undertaken either to provide baseline
statistics “to support policy”, or to perform a role
in the longitudinal monitoring of performance. If
the statistics and opinions derived from the
surveys are misrepresentative and exaggerate
real activity levels, that could set and sustain
false levels of expectation. If strategic policy,
operational or funding decisions are made on
those false expectations, the decisions incur a
higher risk of failure of meeting the real targets.
It also becomes difficult to produce and target
effective management actions, campaigns and
promotions if the market research is unreliable,
and it becomes difficult to judge the efficacy of
the campaigns if the feedback is misleading. If
the survey results are published for wider public
consumption, the contained over-optimism
could fuel a false complacency, defeating the
objective of giving that feedback.
Whilst many practitioners recognise that
household surveys tend to paint an over-
Understanding Recycling Behaviour 2
optimistic picture of recycling activity, there
have been few attempts to quantify the degree
of that over-optimism. The research reported
here is one of the pioneering studies. It
provides working estimates of the likely scale of
overestimation that waste management
professionals can now use with their own
judgement to help understand the true levels of
household recycling activity. It also provides a
stark lesson to lay readers of household
recycling surveys: “The statistics presented in
the surveys may not necessarily be close to the
truth”. The discrepancies will of course vary
from survey to survey depending on the who is
being surveyed, the design of the survey, the
questions asked, how it is implemented, and so
on. The fact that surveys can be inaccurate
must always be stressed.
In order for surveys to become more effective
and reliable sources of information, those
surveys need to be quite carefully designed.
Any well-designed survey should include in-
built consistency checks that can help judge the
reliability of the contained information. Also the
analyses of that information should be subject
to rigorous statistical significance testing to
ensure the validity of the findings. Sadly many
household waste surveys contain neither.
Normally only academic surveys are designed
to that rigour.
Finally, based on the light of the findings
presented in this research, it would be
instructive to reanalyse some of the more
recent national household waste surveys in
order to ascertain their likely scales of error and
the potential implications. Whilst a full analysis
is beyond the scope of this report, one survey
is illustrated here by way of an example. In a
recent national survey in Scotland, 50% of the
sample claimed to recycle, with 65% of that
recycling public recycling newspapers (Waste
Aware Scotland, 2002a). On average, those
recyclers claimed to recycle around 95% of
their newspapers (Waste Aware Scotland,
2000b), which implies an overall newspaper
recovery of just over 30%. This compares with
an estimate that just 14% of post consumer
newspaper was recovered by Scottish local
authorities in 1999 (SEPA, 2002). That SEPA
figure of 14% is itself somewhat inflated as
magazine recoveries were counted as
newspaper recoveries in the analysis. The
updated estimate for Scotland for 2001/02 is
still less than 20%.
135

10. Waste Arisings
10.1 Review
10.1.1 Context
Much of the discussion in the earlier chapters
focussed on recycling yields - the tonnages
collected in individual recycling programmes
and in individual neighbourhoods within those
programmes. The reason why so much
emphasis was given to yield is simply because
yield is the only performance indicator that is
routinely measured in normal practice.
Participation rates and material capture rates
are too costly or too difficult to measure in
routine operations.
Recycling yield is not the ideal parameter for
monitoring household recycling behaviours
although it serves well for monitoring financial
returns. Yield is compounded from three
factors: (i) waste arisings, (ii) participation
rates, (iii) capture rates from participants. To
„unpick‟ the two recycling behaviours –
participation and capture - from the yield
requires a knowledge of waste arisings. As
such, it is important to develop an
understanding of the factors affecting waste
generation, and how they might vary across the
community.
The compounding effects of waste generation
and recycling participation may be seen in the
case study results presented in sections 4 and
5 of this report. The yields from retired
residents were relatively moderate, yet retired
residents are traditionally seen as a high
participation group in recycling. Could this be
reconciled simply through their being low
consumers.
Waste generation is normally considered to be
highly lifestyle dependent. Waste arisings are
built up from the residuals from all the
necessities and luxuries consumed by the
household. It is normally considered that those
lifestyle tastes and household purchasing
powers can be linked to socio-economic and
socio-demographic factors. On that premise,
researchers over the last two decades have
sought to delineate and quantify those links.
137
Understanding Recycling Behaviour 2
10.1.2 Methodologies
Basically there are three ways to estimate
household waste arisings:
 Weigh all waste generated by each
household before any segregation takes
place;
 Weigh each of the segregated waste
streams;
 Analysis the disposable items purchased
by the householder.
Two decades ago, when recycling activities
were relatively low, the analysis of dustbin
waste was taken to be a relatively good
measure of household waste arisings.
Nowadays, however, the arisings generally
need to be reconstructed from paired
measurements of the residual waste and the
recyclate.
Waste arisings and compositions are very
variable amongst households. To ensure the
necessary statistical accuracy, figures must be
aggregated across many households. Two
main methodologies have been tried:
 Individual household analyses from
many households;
 Analyses of the bulked waste from
multiple households.
The first approach is now generally the
preferred method.
Despite the importance of having accurate
statistics on waste arisings, relatively few
authorities have up-to-date waste analyses.
The analyses that do exist are invariably based
on relatively small samples of a few hundred
households or less.
Research over the last three decades has
sought to explain waste generation rates on the
basis of demographic factors and other
household characteristics. This section
concentrates on the results obtained from those
analyses, and how far they have led towards a
predictable model.

10.1.3 Early Research
It was recognised as early as 1926 that the
type of waste produced by households varied
according to the type of property they inhabited
(Dawes, 1926). Dawes attempted a 10.1.4 Subsequent Research
classification based on rateable value.
Higginson (1965) considered rateable values,
housing type, and smoke control zoning as
predictor. In the 1970s and 1980s, variables
other than property type were considered as
indicators of waste generation. This extended
approach is typified by the work of Rufford
(1984). The fundamental premise was that the
patterns of household waste generation could
only be understood at the individual household
level.
Rufford (1984) analysed the waste from 1277
households into 15 compositional categories.
He found that nineteen out of the twenty
household characteristics that he investigated
were significantly correlated with at least one of
the 15 material categories. However, Rufford
noted that the explanatory household
characteristics were themselves highly
correlated. He then selected just four variables
that he considered to be the most powerful
predictors and also the least inter-correlated:
 Household size
 Family life stage
 Whether the head of household was
economically active
 Tenure
Rufford found that these variables could
account for around a quarter of the variation of
total waste arisings of individual households.
In a revised model, MEL (1993) chose
household size, tenure and mode of collection
as the independent variables. Two other
variables, car ownership and housing type were
also considered but dropped because they
were too highly inter-correlated with the other
variables.
Aspinwall and Co. (1991) adopted the
alternative area-based approach rather than
referring to individual households. That
approach was taken forward under the UK
National Household Waste Analysis
Programme (WSL and Aspinwall, 1994).
The WSL and Aspinwall model (WSL and
Aspinwall, 1994) was based on the Acorn 1981
classification (different to the Acorn 1991
classification used elsewhere in this report).
Aspinwall claimed a distinct association
Understanding Recycling Behaviour 2
between waste arisings and Acorn group, with
the better-off groups generating the most
waste.
Further data compilations, under the National
Household Waste analysis programme, were
carried out amongst Open University students
Jones et al., 1998a 1998b). However, the most
significant study into the determinants
household waste generation was the study
carried out by MEL (1996).
The MEL (1996) investigation comprised two
elements: (i) analysis of waste from individual
households, and (ii) a questionnaire survey.
Sample households were selected from
communities in Birmingham, Coventry and
Wolverhampton, to cover the full range of
socio-demographic variables that might be
expected over the whole community. Overall
780 households were sampled, with small
supplementary studies looking at households
serviced by kerbside collection or high density
bring recycling.
The links between purchasing and waste
generation produced the following findings
amongst the conclusions:
Those cooking one or more pre-packed meals
per week tended to generate more card
packaging than those cooking less than one a
week. Card packaging also increased with
household size. In contrast, no discernable
relationship was found between the
consumption of tinned food and ferrous metal
waste arisings, nor between consumption of
fresh vegetables and putrescibles in the waste.
Use of fresh vegetables, however, was
negatively coupled with card packaging.
Households buying two or more daily
newspapers averaged 1.9 Kg/hh/w newsprint
waste, whilst those buying just one produced
1.25 Kg newsprint waste per week, with those
not purchasing any averaging 0.8 Kg/hh/w
newsprint in their dustbins.
Overall, the study concluded that much of the
variations amongst households appeared
random with no obvious explanatory factors.
Seasonal variations were found to be at a scale
far smaller than the random variation.
Some positive correlations between waste
arisings and socio-demographic variables were
found in the study. Paper and card arisings
138

were positively correlated with mode of refuse
collection, household size, and presence of
infant children, and negatively correlated with
age. Paper and card arisings were not
correlated with housing type, socio-economic
class, ethnic status or car ownership. Glass
and non-ferrous metal arisings correlated
positively with household size and negatively
with the proportion of retired residents.
Skilled manual workers were the largest weight
generators. Retired households were relatively
low waste producers (for most fractions).
Owner-occupier households with a mortgage
and council tenants were also relatively high
waste producers. Maisonnette and flat dwellers
were amongst the lowest waste producers
whilst residents in bungalows generated more
waste that might be expected from their
household size.
Whilst different associations were found for
different materials, household size and mode of
10.2 New Results
10.2.1 Items Consumed
In the current research, household waste
arisings were estimated from the self-reports of
the numbers of materials brought into the
household each week. Results are shown in
tables 10.1 and 10.3 for the Renfrewshire
sample and the Clackmannanshire sample
respectively. In these tables, per-household
consumptions are tabulated for the major
demographic variables: housing type, social
class, family life stage and household size.
Results are also shown graphically in figure
10.1. Tables 10.2 and 10.4 show the
significances of any relationships between
household consumptions and demographics.
In Renfrewshire, the most significant predictors
were household size and family life stage.
Consumptions of aluminium cans, plastic
bottles and magazines all increased with
increasing household size. The trend for
newspaper consumptions was less clear. The
strongest step change in consumption was
seen between single person households and
multi-person households.
Newspaper consumptions were lowest
amongst young adults and families with young
children. Can consumptions and, to a lesser
139
Understanding Recycling Behaviour 2
refuse collection (wheeled bin versus plastic
sack) appeared most generally amongst the
most important predictors of waste arisings and
composition. The relationships with other socio-
economic descriptors were either very material
specific or were very tenuous.
Associations between the numbers of items
disposed each week and the demographic
factors revealed a general increase in the
arisings of glass bottles, plastic bottles, cans
and food packages with household size. In
contrast magazine arisings showed no
association with household size and newspaper
arisings actually decreased with increasing
household size.
Finally, it is noted that the review of the MEL
study has been quite selective. The reader is
referred back to the original report for the full
details.
extent, bottle consumptions, peaked amongst
the middle lifecycle stages, with young adults
and retired residents being the lowest
consumers.
In Clackmannanshire, consumptions of all
materials were highest in the housing schemes
and lowest in the flats. Glass, cans, and plastic
bottle arisings were, as in Renfrewshire,
elevated amongst the middle lifecycle stages
and depressed amongst the young adults and
retired residents. All the container
consumptions increased with household size,
whilst newspaper consumptions again revealed
that single against multi-person households
may be the key discriminator.
The results from earlier University of Paisley
surveys in Girvan and Glasgow revealed
significant household size differences for glass
bottle consumption and newspaper and
magazine consumptions. Newspaper
consumptions in Glasgow were highest
amongst detached households and lowest in
the flats. Higher glass consumption was again
associated with the middle lifecycle stages and
with the larger household sizes. Newspapers
again showed a single versus multi-person
distinction.
 Understanding Recycling Behaviour 2

Table 10.1 Mean Number of Items brought into the Household Each Week (Renfrewshire)
Newspaper Magazines Alu. cans Plastic
bottles
Acorn A 9.1 1.5 5.2 5.3
Acorn B 8.5 1.8 6.8 6.4
Acorn D 9.1 1.6 5.6 4.8
Acorn E 10.3 2.2 4.4 5.4
1 7.5 2.4 6.4 7.1
2 7.1 1.5 6.0 5.1
3n 9.9 2.0 6.9 6.3
3m
10.0 2.0 6.4 6.5
4/ 5
11.8 1.4 7.4 4.8
YA 6.5 2.0 4.9 5.7
FYC 7.5 1.8 6.7 7.5
FOC 9.9 1.9 8.2 5.9
MM
11.0 2.2 7.7 5.8
RET
9.4 1.2 2.7 3.5
1 6.7 0.8 1.9 2.8
2 9.6 1.6 4.1 4.7
3 8.5 2.0 6.9 5.5
4 9.6 2.0 8.7 7.3
5+
10.1 2.4 5.8 7.1
YA = Young adults, no children; FYC= Families with young children; FOC = Families with older children;
MM = More mature, RET = Retired

Table 10.2 Significance Test of Differences in Consumption amongst Groups (Renfrewshire)

News- Magazines Alu. cans Plastic
paper bottles
House type    **
Social class    
Family life stage **  ** **
No. in house ** ** ** **
** significant at 95%,  not significant

Table 10.3 Mean Number of Items brought into the Household Each Week (Clackmannanshire)
Newspaper Glass Alu. cans Steel tins Plastic
bottles
Semi 8.0 7.9 15.1 12.5 5.9
LA/ ex-LA 9.7 8.9 14.7 14.2 9.2
Flat 6.9 3.3 2.7 4.4 4.6
2 7.2 8.9 10.2 11.6 6.5
3n 8.4 8.1 18.6 11.7 10.6
3m 10.0 7.5 10.6 15.7 6.5
4/ 5 9.2 8.2 15.5 14.3 6.6
YA 10.4 4.4 7.0 9.9 8.0
FYC 7.4 8.7 18.5 16.6 9.8
FOC 7.9 10.2 15.9 15.3 10.8
MM 10.6 7.5 11.8 11.2 6.0
RET 8.3 5.9 7.1 5.7 3.5
1 6.6 3.7 1.4 4.3 3.4
2 10.7 6.7 12.2 8.7 5.1
3 8.5 9.6 12.9 14.7 7.1
4 7.7 8.0 17.9 15.7 9.3
5+ 10.2 13.1 21.9 21.1 15.7

140
 Understanding Recycling Behaviour 2

Table 10.4 Significance Test of Differences in Consumption amongst Groups

Clackmannanshire)
News- Glass Alu. cans Steel tins Plastic
paper bottles
House type  ** ** ** **
Social class     
Family life stage   ** ** **
No. in house ** ** ** ** **
** significant at 95%,  not significant

Figure 10.1a Newspaper Consumptions

Newspaper/Renfrewshire Newspaper/Clackmannanshire

12 12

Items per week

Items per week

10 10
8 8
6 6
4 4
2 2
0 0
A B D E Semi LA/ ex-LA Flat

Acorn Housing type

Newspaper/Renfrewshire Newspaper/Clackmannanshire

12 12
Items per week

Items per week

10 10
8 8
6 6
4 4
2 2
0 0
ya fyc foc mm ret ya fyc foc mm ret
Family lfe stage Family life stage

Newspaper/Renfrewshire Newspaper/Clackmannanshire

12 12
Items per week

Items per week

10 10
8 8
6 6
4 4
2 2
0 0
1 2 3 4 5+ 1 2 3 4 5+
Household size Household size

141
 Understanding Recycling Behaviour 2

Figure 10.1b Aluminium Can Consumptions

Al cans/Renfrewshire Al cans/Clackmannanshire

8 20

Items per week

Items per week

6 15

4 10

2 5

0 0
A B D E Semi LA/ ex-LA Flat

Acorn Housing type

Al Cans/Renfrewshire Al cans/Clackmannanshire

10 20
Items per week

8 Items per week 15

6
10
4
2 5
0 0
ya fyc foc mm ret ya fyc foc mm ret
Family life stage Family life stage

Al Cans/Renfrewshire Al cans/Clackmannanshire

8 25
Items per week

Items per week

6 20
15
4
10
2 5
0 0
1 2 3 4 5+ 1 2 3 4 5+
Household size Household size

10.2.2 Consumptions and Recycling

Behaviour
In earlier research into paper-only kerbside
recycling, it was noted that, in general,
recyclers tended to be significantly higher
consumers of newspaper than were non-
recyclers (Tucker, 2001). This effect has also
been noticed in other communities and other
countries (e.g. Boldero, 1995).
In both the Renfrewshire and
Clackmannanshire samples, there were no
strongly significant differences in the average
consumptions of recyclers and non-recyclers
for the majority of the materials studied7. The
only marginally significant difference (at 90%
confidence) was in slightly higher newspaper
7
Non-consumers were omitted from these analyses

142
 Understanding Recycling Behaviour 2

consumptions amongst newspaper recyclers content to add in their newspapers as well.

than amongst non-recyclers of newspaper in Others, however, may still judge their
Renfrewshire (figure 10.2). newspapers to be too few, particularly if they
are collected in a separate container to the
cans and bottles.
Figure 10.2 Cumulative Distribution of
Newspaper Consumption (Renfrewshire)
Table 10.5a Average Material
Consumptions of Recyclers and Non-
Cumulative % of sample

120
recyclers (Renfrewshire)
100 Items Consumed per
80 Week
60
Recyclers Non-
recyclers
40
20 Newspaper 9.5 7.9
Magazines 2.5 2.5
0
Plastic bottles 6.2 4.8
0 4 8 12 16 20 24 Alu. cans 9.0 8.1
Newspapers per week

Recyclers Non-recyclers Table 10.5b Average Material

The major difference in consumptions between
recyclers and non-recyclers was the significant
percentages of non-recyclers who did not
consume certain materials, notably magazines
and cans.
Table 10.5a summarises the average
consumptions in Renfrewshire and table 10.5b
records similar data for the Clackmannanshire
sample. Generally, the average consumptions
were higher amongst recyclers, although it is
noted that newspaper consumptions are
actually higher amongst the non-recyclers of
newspaper in the Clackmannanshire sample.
However, as indicated in section 8.4.2, this may
be due to a geographical divide between the
owner-occupied properties (who had
reasonable access to local paper banks) and
local authority housing for which paper facilities
were further afield.
The incidence of lower material consumptions
amongst non-recyclers has been considered to
be tied in with an attitude that “you need a
certain amount of materials before recycling
becomes worthwhile” (Tucker, 2001). On that
premise, low consumers of newspaper might
not consider it worthwhile to join a paper-only
recycling scheme. However, in a multi-material
collection, recyclers might base their judgement
on what constitutes a minimum amount more
on the totality of their contribution. That is, even
if they have few newspapers, they may have
enough bottles and cans to render the scheme
worthwhile to them. Some could then feel
Consumptions of Recyclers and Non-
recyclers (Renfrewshire)
Items Consumed per
Week
Recyclers Non-
recyclers
Newspaper 8.2 9.7
Glass bottles 8.0 7.9
Alu. cans 16.8 14.0
Overall, consumption behaviours may be
showing a significantly different pattern of
association with demographic factors than are
recycling participation behaviours (see section
6.5 for an account of these). Retired residents
are often amongst the highest participants,
though they are also amongst the lowest
consumers. Large households are often poor
participants but are generally the highest
consumers. This leads on to the premise that
variations in recycling yields amongst different
demographic groups may follow quite a
different pattern to the variation in their
participation behaviours. Yields from high
consuming/ low recycling groups could turn out
to be quite similar to those from low consuming/
high recycling groups.
Such behavioural differences can be seen
through comparing the recycling yields with
material consumptions. Figure 10.3 shows a
selection of those relationships for the
Renfrewshire and Clackmannanshire samples.
A trend line has been plotted on the figures to
aid visualisation. It is based on a linear

143
 Understanding Recycling Behaviour 2

regression constrained to pass through the
origin, and represents the average material
capture of the included group. Points above the
trend line indicate higher capture rates (and
implicitly higher participations). Points below
the line indicate the lower participating groups.
Whilst these plots show that some larger
households might indeed be under-performing,
the evidence remains equivocal. No definite
pattern emerges. Single person households, for
example, show different newspaper recycling
behaviours in the two samples.
The pictures for the other demographic
indicators and for the other materials (not
shown) also failed to reveal any consistent
patterns. Young adults produced poor can
yields in both communities, poor newspaper
yields in Renfrewshire but contributed the
highest newspaper yields in Clackmannanshire.
Families with older children produced very poor
yields in Clackmannanshire but returned
average performances in Renfrewshire.

Figure 10.3a Yield versus Consumption by

Household Size

Renfrewshire/ newspaper Clackmannanshire/ newspaper

10 Items recycled 4
Items recycled

8 1 person hh
3
6
2
4 1 person hh 4+ person hh
2 1
0 0
0 5 10 15 0 5 10 15
Items consumed Items consumed

Renfrewshire/ cans Clackmannanshire / cans

2 & 3 person hh
6 4
Items recycled

Items recycled

3
4
4 person hh 2
2
1 4+ person hh

0 0
0 5 10 0 10 20 30
Items consumed Items consumed

10.3 Conclusions and Implications

The research of Rufford (1984) and MEL consumer goods, for example with food
(1996) both stress the importance, and perhaps packaging and newspapers.
the dominance, of the household size variable
as a predictor of household waste arisings. The The current study provides further supporting
association is clearly seen for most waste data that household size is a key determinant
materials associated with the purchase of of the amount of food packaging materials likely

144

to enter the household waste stream. Arisings
increase roughly in proportion with the numbers
in the household. Magazine arisings show a
similar trend. Newspaper arisings, however,
were found to show a weaker dependence on
household size. The main step change in
consumption appeared to be between single
and two person households, with only modest
further increases in consumption in larger
households.
The research has shown social class to be a
relatively poor determinant of household
consumption, but has shown that family life
stage may be significantly linked with
consumption, with the middle life stages being
the heaviest consumers.
However, whilst relationships with different
census and purchasing variables can be
identified, it is difficult to account for more than
a minority part of the variation amongst
households using such variables.
Part of the problem in deriving a good
explanatory model lies in the strong inter-
dependencies between most of the predictor
variables. For example families with children
present correlate positively with household
size, as does detached housing. Car ownership
correlates positively with social class and so
on. These inter-correlations tend to obscure
any robust physical interpretation through
8
statistical analysis . In deriving their predictive
model, MEL chose to alleviate the problem by
using only a small [nearly independent] subset
of predictors.
8
Note these are the identical problems to those
encountered earlier in interpreting regression
analyses of recycling behaviours (see section 4).
145
Understanding Recycling Behaviour 2
Another problem lies in the constraints of
sample size. Few studies are resourced to
explore more than a few hundred households.
Even the Government‟s National Household
Waste Analysis Programme of the early 1990s
has been strongly criticised as being
inadequate to derive robust and meaningful
statistics (Parfitt and Flowerdew, 1997).
Until better data becomes available, it remains
expedient to take the household size statistic
as the main predictor of household waste
arisings. However, as MEL (1996) point out, it
does not necessarily apply equally to all
categories of household waste. Garden waste
and miscellaneous non-combustibles for
example do not normally show any relationship
to household size.
Consumption behaviours generally show little
correspondence with recycling behaviours.
Their demographic bases of the two behaviours
would appear to be different. There is some
evidence, however, that larger households who
tend to be the highest consumers also tend to
be amongst the poorest recyclers. Increasing
recycling rates amongst those households
would therefore appear to present the greatest
opportunities for increasing overall recycling
yields.
Overall differences in consumption between
recyclers and non-recyclers have been found
for newspapers though not for any other
material, newspaper consumption being higher
amongst the paper recyclers. The differences
were marginal in the current research which
centred on bring recycling and multi-material
kerbside recycling respectively. Previous
research has shown that larger differences may
occur in paper-only kerbside collections.

11. Discussion
11.1 Government Targets and Human Behaviour:
Bridging the Gap
11.1.1 What are the Endpoints?
The obligation to meet Government targets to
recycle or compost 25% of household waste by
2005 (2006 in Scotland) from 2000/01 levels of
11.2% (England) and 7.0% (Wales) and
2001/02 levels of 7.4% in Scotland, imposes
severe and immediate challenges on most local
authorities. The major responses to the
challenge have been: (i) to increase the
supporting infrastructure, which normally
means increasing kerbside collections or
increasing the number of materials collected,
and (ii) to raise public awareness through
educational and promotional campaigns.
Current advice additionally recommends the
use of financial inducements or penalties, such
as variable waste charging, as additional levers
for participation (Strategy Unit, 2002).
However Government targets represent quite
arbitrary levels of household behaviour. If we
really want to understand recycling behaviour,
we really need to be asking a different set of
questions:
 “What levels of behaviour could our
imperfect society aspire to?” and “What
help is needed on the way?”
 “Given our imperfect society, what
levels of behaviour are expected with the
help that is being given?”
Effectively this represents planning and
monitoring from the people perspective.
This section of the report examines the
planning issues.
In work commissioned by the Government
Strategy Unit, the University of Paisley
Integrated Household Waste Management
Model was used to forecast the recovery
endpoints that could be achieved given a
scenario of voluntary household participation in
an „imperfect‟ England - starting from today‟s
recycling behaviours. The analyses
concentrated on distinguishing the relative
contribution that enhanced infrastructure might
make, and how much would still need to be
addressed through education and promotion.
Understanding Recycling Behaviour 2
The results showed that increasing
convenience and engendering attitude change
were both essential to realising recycling rates
of 40% and above.
The maximum possible material capture rates
based on the most advantageous set of
circumstances were predicted to be of order:
80 - 82% (Newspapers, pamphlets and
magazines)
80 - 82% (Glass bottles)
75 - 78% (Glass jars, aluminium cans, steel
cans)
71 - 73% (Cardboard)
71 - 73% (Plastic beverage bottles)
These captures would be achieved by multi-
material kerbside collections of all the above
materials coupled with high levels of effective
promotional and education campaigns. In such
scenarios 85-87% of households would be
engaging in some kind of recycling activity, with
75-80% of them using kerbside schemes.
11.1.2 Designing the Programme
Whilst the above analyses provided guidance
on the staged enhancements that would be
needed to achieve the highest recycling levels,
the finer details of how to progress locally must
be worked out locally by individual local
authorities. Most local authorities have now
made their major strategic decisions on what to
collect and how to collect it, by bag, bin or box,
co-mingled or source separated, fortnightly or
weekly. But how much will those decisions
matter? Does one collection system have
advantages over another? These questions
were addressed by running statistical analyses
on the published performance statistics for
alternative kerbside collection regimes. The
analyses were undertaken specifically for the
paper and card fractions of the recyclate.
The results indicated that, on average, kerbside
yields of paper and card will be higher in:
(a) schemes that collect other materials
alongside that paper and card, and
147

(b) schemes that use collection boxes (or
bins) rather than bags or sacks.
The average difference in yield could be as
much as 20% between multi-material box
collections and paper-only bag collections.
Collection frequency did not significantly affect
the yield.
Although programme design details do appear
be important to scheme performance, research
the US has shown that the top performing
schemes all have comparable performances
irrespective of their programme design.
However, more „convenient‟ programmes still
give the edge in poorer recycling communities.
Overall, it is apparent that simply making
recycling more convenient will not, on its own,
achieve the Government‟s recycling targets.
Whilst the infrastructure enhancements that are
needed are clear (e.g. nation-wide household
kerbside collection of recyclables and organic
wastes, augmented by increased home
composting and waste reduction and reuse
measures (Strategy Unit, 2002)), the details of
the complementary „educational‟ elements
remain much less clear. Normally, campaign
strategies hinge on stimulating behavioural
change, through conveying „awareness‟ and
giving „procedural instruction‟, often backed up
by persuasive messaging intended “to change
attitudes”. But what attitudes? Some pointers
on this are now emerging (see sections 8, 11.4)
though much further research is still required.
The launch of a new waste management
programme may be the crucial time to shape
attitudes. At that time, the saliency of waste
management issues to individuals‟ life styles
might become relatively high, increasing their
susceptibility to change. Documented evidence
shows that significant positive step changes in
recycling performances have almost always
been linked to changes in infrastructure
provision. In comparison, even the most
intensive promotional campaigns have only
realised quite modest improvements in
recycling rates (see section 1.4).
11.1.3 Making the Changes
New research investigations were launched in
Chesterfield and North East Derbyshire
(section 4), North Warwickshire (section 5.2)
and Clackmannanshire (sections 8.4. and 8.5)
to investigate the detailed effects of
infrastructure change on recycling behaviours
and attitudes.
148
Understanding Recycling Behaviour 2
The above four districts have all introduced
multi-material kerbside collections during the
last two years. All chose to collect paper, glass,
mixed cans and textiles using a box and bag
system, the paper and textiles being collected
by bag and the other materials in the box.
Before the new recycling programmes were
introduced, the two Derbyshire districts and
North Warwickshire had been serviced by
paper-only kerbside collections. The yield from
the paper-only collection in North East
Derbyshire was 1.31 kg per serviced household
per week, which was above the national
average for paper collections. The yields in
Chesterfield and North Warwickshire were
below the national average at 0.68 and 0.63
kg/household/week respectively. The
conversions to multi-material collections
resulted in significant improvements in dry
recyclate yield in all three districts, recording
yields of 2.38, 1.79 – 2.9, and 2.31 kg/serviced
household/week in North East Derbyshire,
Chesterfield and North Warwickshire
respectively.
The increases in yield were not solely
attributable to new materials being collected.
The paper yields also increased. Those
increases were by factors of 1.1 to 1.4 in N.E.
Derbyshire, and by factors of 1.5 or more in
Chesterfield. In North Warwickshire the paper
yields doubled.
Whilst the increased kerbside recoveries were
to some extent at the expense of decreased
yields from the local bring sites, the combined
gains were substantial. Bring site yields
dropped by around 25% (paper) and 25 - 50%
(glass) at recycling banks situated close to the
converted kerbside rounds. However, there
were less than 10% changes in the can and
textile bank yields.
Overall, in the Derbyshire conversions, there
were lower can recoveries than expected
across all recycling points (both kerbside and
bring) and a seeming reluctance to transfer
allegiance for textile recycling away from the
bring sites onto the kerbside collection.
Poor can recoveries appear to be quite a
common feature of many kerbside recycling
schemes (e.g. Renfrewshire – section 9;
Clackmannanshire – section 8.5; Ayrshire –
Tucker, 2001, 2003; Babergh – section 1.4.5,
and elsewhere). Previous surveys have
indicated that discard in non-household waste
streams and forgetfulness were amongst the
main reasons given for not recycling aluminium

cans, whilst “too much effort” and “too
unpleasant to wash out” were major reasons for
not recycling ferrous cans (Tucker, 2001). In
Renfrewshire, a misunderstanding that drinks
cans were not allowed in the box dominated the
reasons for not recycling drinks cans. “Not
enough to make it worthwhile” and “not enough
room in the box” were also amongst the given
reasons. It is also noted that these were also
amongst the strongest reasons for not recycling
plastic bottles.
Whilst district-wide recoveries of all materials
increased when the option to recycle at
kerbside is presented, the responses for paper
and glass appear stronger than those realised
for other materials. Paper and glass yields are
normally strongly and significantly correlated.
Those correlations are apparent across
different areas and also week-by-week within
those areas. That is, when paper yields are
high, glass yield are normally high as well.
Correlations between glass and paper yields
and can and textile yields were much weaker.
It was shown in figure 2.1 how the introduction
of district wide multi-material collections should
provide a significant step forward towards
achieving the ultimate recycling levels. The
next obvious step is to provide green waste
kerbside collections. Whilst that has occurred in
both Chesterfield and North Warwickshire, the
results have not been analysed in this
research. However, the green waste collection
in Chesterfield is extremely interesting for other
reasons.
Parallel green waste collections were
introduced in Chesterfield in nine of the initial
fifteen multi-material collection rounds. The
green waste collections in Chesterfield utilise a
twin bin system for green and residual wastes,
the bins being collected on alternate weeks.
That means the frequency of collection of
residual waste is halved.
Reducing the frequency of residual waste
collection or reducing the capacity of the
residual waste bin are well-known strategies for
Understanding Recycling Behaviour 2
attempting to increase the segregation of
recyclables form the residual waste. The
intention is that householders will need to seek
alternatives, preferably recycling, for disposing
of their excesses. Providing the solution (i.e.
the multi-material box) should steer
householders into adopting that solution.
The introduction of the alternate week green
and residual waste collections at the same time
as the multi-material collections produced a
four-fold increase in the total dry recyclables
recovered in Chesterfield (to 2.9 kg/serviced
household/week). In other areas of
Chesterfield, the alternate week collections
were introduced prior to the multi-material
collections. Although the dry recyclate yields
were still elevated compared to the areas with
weekly residual collections, those differences
were not significant when demographic
differences were taken into account. Timing is
clearly crucial. Imposing the collection
frequency change prior to providing the solution
rendered that solution almost irrelevant when it
subsequently came. The necessary life-style
adjustments had already been made.
It is also noted that significant increases in dry
recyclate yield have also been achieved by
other authorities adopting fortnightly residual
waste collections. Introducing the changes on
top of an established multi-material collection
also appears to be a successful strategy
(Woodard et al, 2001).
Now that the new programmes are established
in Chesterfield and North East Derbyshire the
best opportunities for synergistic promotions
and education have already occurred.
Nevertheless, there is still potential to improve
scheme performances further. Scenario
modeling using the University of Paisley
Integrated Household Waste Management
Model predicted that well orchestrated
awareness and [procedural] information
campaigns could increase yields by 5 - 8%
overall, and metal yields by up to 20%.
149

11.2 Performance Monitoring and Evaluation
11.2.1 Introduction
Most local authorities have now taken the main
strategic decisions for increasing their recycling
rates. Many new kerbside programmes have
been started, or are about to start up. The issue
now facing the authorities is “Are the results
fulfilling expectations?” The need for analysis is
now moving away from a programme design
role and becoming increasingly concerned with
monitoring and evaluation.
“Given our imperfect society, what levels of
behaviour are expected with the help that has
been given?”
The questions on whether performances meet
expectations are not easy to answer. What
actually are the expectations for a given
programme design, given the imperfect
behaviours of the residents, and other local
peculiarities. What should the local behavioural
norm be? All we know are the aspirational
goals. These are the audited targets that must
be met. Programme designers will have
provided the assurances. Auditing will judge
their success.
Performance monitoring by district authorities is
generally audit driven. However those audit
statistics actually tell us little about what is
really going on inside the districts. They provide
insufficient information for performance
diagnostics, for targeting effective remediation
strategies or campaigns, and for realising
continual and sustained improvements.
Effective recycling management is more than
providing statutory statistics; it is also about
understanding and developing the best
behaviours amongst the residents.
Given that the chosen recycling scheme may
already be up and running, the manager may
now have relatively little scope to manipulate
programme design. The role is more about
encouraging the residents to use the
programme to best advantage. That does not
mean simply relying on national or regional
campaigns to do the job. Experience has
shown that those campaigns have so far failed
to produce significant step changes in recycling
behaviours. Rolling out generic campaign
material in a more local application may also
not achieve the solution. The solution is likely to
lie in the sympathetic design of interventions
that are geared specifically to local
circumstances and problems. Understanding
those local circumstances and issues is the first
crucial step. That means effective performance
150
Understanding Recycling Behaviour 2
monitoring. Any factors that inhibit locals to
recycle almost invariably have rational causes
behind them. Diagnosing those causes is the
second crucial step.
Effective local performance monitoring must be
designed to provide the key information upon
which key decisions can be based – in order to
provide the most effective solutions most cost-
effectively with minimum risk. However, it must
also be borne in mind that the monitoring
activity in itself can incur a significant financial
cost.
The key indicators such as household
participation rates are too difficult and too
resource intensive to monitor routinely. Per-
collection [kerbside] set-out rates are more
practical to measure though they can
significantly under-estimate true participations.
However, research by the University of Paisley
(Tucker, 2001) and others has produced some
guidance on the likely scale-up factors between
set-out rates and participation rates, and such
„multipliers‟ might be usefully applied in regular
performance monitoring. However, even
counting set-outs can be labour intensive, so
an expedient could be adopted to select and
concentrate on a small number of „marker‟
streets for those analyses. A bar coding of
containers (which could arise in some
scenarios of variable waste charging) could
provide the ultimate solution.
Currently, participations only tend to be
monitored through separate research
investigations, and are generally limited to short
time spans and to relatively small samples of
the local population. Waste analyses are only
undertaken in exceptional circumstances, often
as one-off exercises, again over restricted
populations.
Routine monitoring is normally limited to
recording the recovered tonnages, broken
down where appropriate by saleable
commodity. Much of that recorded information
can be quite coarse. Most authorities keep
records on the monthly tonnages of paper and
card, glass by colour, dense plastic, ferrous
and aluminium cans etc. that are collected
district wide. However, [quite surprisingly] a
significant number of authorities do not keep
records at any better resolution than that. The
impact that this has on monitoring is discussed
in section 11.2.3.
Another mechanism of monitoring is through
eliciting the opinion of householders. The

mechanism of citizens‟ juries is well
established, though is geared more towards
performance audit than performance
diagnostics. Wider, and more closely targeted
household surveys might be undertaken but,
like other forms of in-depth monitoring, surveys
are normally only undertaken in exceptional
circumstances. Whilst those surveys can
furnish a wealth of information, they can also
furnish misinformation. The advantages and
limitations of surveys as monitoring tools are
discussed further in section 11.2.4.
11.2.2. Methodological Limitations
Monitoring essentially needs to furnish the
information to answer the questions “Are
performances meeting expectations?” and “If
not, why not?” As stated earlier, we do not
know precisely what should constitute normal
behaviour within a given district. Our perception
of expected behaviour is invariably based on a
conceptual model. That model is probably
based on the following reasoning, or something
similar:
 Borough „X‟ achieves 18% recycling with
programme design „Y‟, so I should get that
as well;
 80% will participate with 80% of their
material, so I should get 64% of what I ask
for;
 The consultants tell me that programme
design Y will give me a 20% recycling rate.
In reality, the complex interplay of local factors
and unknowns renders the outcomes far less
predictable. Even the most sophisticated
forecasting models, such as the University of
Paisley Integrated Household Waste
Management Model, do not always get it right.
It is very much like weather forecasting!
Normal behaviour for given local circumstances
can only really be defined after that behaviour
has been measured. Effectively „normal‟ is what
you have got. However, household recycling
performances are not homogeneous and can
vary markedly in space and in time. So given
that different recycling performances will occur
in different neighbourhoods, and that those
individual neighbourhood performances may
change with time, the monitoring task firstly
needs to resolve the better performing
neighbourhoods from the poorer performing
neighbourhoods, then to track those
differentials with time. Given that in normal
circumstances the only data available will be
the recyclate yields, any classification in terms
Understanding Recycling Behaviour 2
of „better‟ or „poorer‟ must then be based on
those yields.
Yield = Arisings  Participation  Capture (from
participants)
It should be borne in mind that the high yielding
areas are not necessarily the strongest
recycling areas. They could have high arisings
but mediocre participations. Nevertheless, in a
commercial sense, they are the best areas to
go for.
Monitoring must be carried out against some
objective variable, and that variable must be
some readily and affordably measurable
characteristic. For comparative monitoring of
different areas, the obvious variable is the
demographic profile of those areas. Most waste
behaviour research over the past two decades
has been based on the premises that
demographic factors can form useful proxies for
accounting for differences in household
consumption and recycling behaviours. On the
premise that demographic factors can account
for at least some of the variabilities, the logical
step is to filter off the demographic effects, then
to look for the factors that might explain any
residual effects. Here again we become
dependent on models.
Fitting a demographic model to area
performance data can potentially realise
information on:
1. The demographic factors that are most
important to explaining performance
differences;
2. Identifying the areas that perform
significantly better or worse than the
behavioural norm.
This clearly goes some way to answering the
prime questions on who is the better recycler
and where the main causes for concern may
lie. However, the analyses can never be that
clear-cut. The first need is to pick out the
effects of individual demographic factors, e.g.
housing type, from data sets that contain the
aggregated behaviours of many different
housing types. Ideally we would like to derive a
model of the form:
Area yield =
Proportion of detached houses  CDetached
+ Proportion of semi-detached  CSemi-
detached
+…
where Ci is the average yield from house type i
151

In practice, when fitting such a model, the
coefficients Ci will not take up such an obvious
physical explanation. They are simply scaling
coefficients that give the best empirical fit to the
data, given that other terms are also included in
the model. As such, all that can be said is that
the variables with the larger, more positive
coefficients have greater, more positive
influences on overall yield than the variables
with smaller, or more negative coefficients. The
analyses can be thought of more as providing a
ranking of the influence of the variables rather
than a quantification of those influences. It must
also be borne in mind that whilst a high ranking
may be achieved, that ranking may not
necessarily be statistically significant.
Another limitation is that many of the possible
demographic predictor variables are highly
correlated. This is not a problem if variables are
mutually exclusive such as a model based on
different types of housing. It is important,
however, for mixed classes of variables such
as housing type and household size. For
example flat dwellers tend to comprise smaller
households than those living in detached
houses. Such inter-relationships can obscure
any physical interpretation.
A further complexity is that a model based on
housing type variables will not be the same as
one based on household sizes. Areas that
appear to perform better than normal on a
housing type model may perform worse than
normal on a household size model. A truly
better or worse performing round may only be
identified if it consistently performs better or
worse across a number of different
demographic models.
Clearly, these limitations restrict the amount
and the quality of the information that can be
extracted from the analyses. Nevertheless,
based on the monitored data that is currently
available, and probably for the data available
for the foreseeable future, little more may be
achievable. The current investigation set out to
apply these types of analyses to real data sets
and to establish whether they would render
meaningful performance diagnoses that could
contribute usefully to the evaluation of their
respective recycling programmes.
The alternative would necessitate a redesign of
performance monitoring practices to supply
monitoring data more closely aligned to the
needs of performance evaluation. This would
almost certainly incur a significant extra cost.
152
Understanding Recycling Behaviour 2
As well as the regression techniques described
above, other statistical techniques such as
cluster analysis can be used effectively to
identify areas of the community that might be
behaving significantly differently from the
whole.
Finally, it must be highlighted that as well as
the inherent methodological limitations to
filtering out “what you would like to know” from
“what you have measured”, there can also be
many uncertainties in what you have actually
measured. Those uncertainties all add to the
„noise‟ in the measurement, making the
extraction of “what you want to know” even
more difficult. Some of the problems with
measurement are considered in the next
section.
11.2.3 Some Current Data
Limitations
Performance Monitoring Data
The fundamental units of assessment (or
performance indicators) that need to be derived
from monitoring [given that only the total yields
will be measured] are the weights of material
collected per household served, normally
expressed in terms of Kg/household/week.
Computing these statistics relies on three
figures: (i) the weight recovered per area
covered, (ii) the number of houses serviced in
that area, (iii) the time period spanned by the
recorded information. Some of those figures
are not always certain or unambiguously
recorded. Before discussing those
uncertainties, however, it is important to
consider what should be defined as the
monitoring „unit‟.
In the course of the current research it was
surprising that a significant number of
authorities could only supply records on a
whole district basis aggregated over a monthly
period. Often those results were presented by
calendar month. Monitoring is only useful if you
can compare like with like. There may be two
collections from one area one month and three
the next. Records need to be distinguishable by
the actual collection cycle – weekly, two weekly
or four weekly as appropriate.
Each collection cycle normally consists of a
series of daily rounds each servicing a different
area. The research has shown how different
the performances of individual rounds can be,
and perhaps more importantly how the
performances of some individual rounds can

change dramatically with time for the better of
for the worse), even though the performance of
the whole district appears quite stable. Put
another way, significant „natural‟ changes may
be occurring at neighbourhood levels, though
these local „ups‟ and „downs‟ may balance out
over a whole district. Good practice will be in
sustaining the „ups‟ whilst preventing the
„downs‟. Localised actions will then provide the
best management solutions. However, before
that can be done it becomes essential to
understand the local pictures in detail. In this
respect it is important to routinely monitor
performances at those local levels [and to
assure that the recorded data is as attributable
as possible to local boundaries]. Good practice
has to take a bottom-up approach. To monitor
and evaluate performances effectively means
monitoring the performances of all parts of the
community individually. The smallest part of the
community that is practicable to monitor
routinely is the daily kerbside collection round.
“How can you manage it if you haven‟t
measured it!”
Of course, many districts are already
monitoring at this resolution; however, even
then uncertainties and ambiguities can enter
the records:
 Sometimes weighbridge tickets do not
coincide with round beginnings or round
ends, or records are assigned to the day
following rather than the actual day of
collection.
 Some councils subsume bring site
recoveries into the kerbside collection
figures.
 Sometimes data are missing.
 There can be uncertainties in the numbers
of houses actually serviced within a given
round.
Demographic Data
There are a number of alternative systems
available for classifying socio-demographic
factors. These basically comprise (i) the
proprietary market research indicators such as
Acorn and Mosaic, and (ii) raw census data,
though it must be remembered that both Acorn
and Mosaic are mainly derived from that same
base census data. Effectively Acorn and
Mosaic cluster a wide range of census statistics
into a relatively small number of categories
which can be regarded as providing
segmentations according to different „life-styles‟
Understanding Recycling Behaviour 2
The Acorn classifications are listed in Appendix
A of this report.
The problem with all the descriptors is that, at
the time of the research, they were over ten
years out of date, all being based on the 1991
census data. At the time of the research the
2001 census data was not yet available. There
has also been some debate on whether Acorn
and Mosaic are actually good descriptors for
accounting for household waste management
activities. The argument against is that they are
based on multiple factors, many of which
appear to be quite unrelated to waste
management. However, there is as yet no
conclusive evidence as to whether Acorn [or
Mosaic] indicators are better than raw census
data and vice versa.
Census data are only available down to the
enumeration district level approximately 250
households) whilst Acorn data goes down to
the postcode level (10-50 households).
Uncertainties in matching kerbside coverage
with these base units of the census
demographic descriptors have to be tackled
and noted as a possible source of error.
11.2.4 Household Surveys as
Monitoring Instruments
Whilst it is possible to make many direct
measurements of household recycling
performance at least in theory, and at a price),
some data can only be obtained through direct
questioning of householders. These data fall
into two main categories:
 Where comparative data is available from
other monitoring methods;
 Where it is not.
Examples of the first category are waste
arisings. A householder can be asked to
recount their level of household purchasing
see section 10), or co-measurements can be
made of the weights and compositions of all
household waste streams. The two sets of data
have very different bases but should convey
similar thrusts of information. Asking
householders whether they recycle can check
against participation monitoring data. However,
as shown in section 9, the two sets of derived
data can be very different. The underlying
tendency is for self-reported behaviours to be
exaggerated. Asking householders what they
recycle can also produce very different results
to compositional analyses of their recyclate.
Again, the main tendency is towards over-
153

exaggeration, though problems with accuracy
of recall can also be significant. Another
problem with survey responses, in this respect,
is in the interpretation of the questions. A
householder who buys one or two cans a
month may not consider it worthwhile to
mention it. On the other hand, a householder
who recycles one or two cans a year may claim
to be an active can recycler. Overall, some
interviewees can take up a morally defensive
standpoint, overplaying desirable traits such as
recycling and underplaying less desirable ones
like alcoholic beverage consumptions.
Whilst checks between monitored and self-
reported behaviours can easily be made, most
monitoring exercises just do one or the other.
Checks between the two normally only arise in
specialist research investigations.
Nevertheless, household surveys are still
valuable instruments is canvassing opinion and
in acquiring customer feedback. To be effective
and reliable sources of information, those
surveys need to be quite carefully designed.
Any well designed survey should have in-built
consistency checks that can help judge the
reliability and of the contained information. Also
the analyses of that information should be
subject to statistical significance testing to
ensure the validity of the findings. Sadly many
household waste surveys contain neither.
Normally only academic surveys are designed
to that rigour.
Effective monitoring and evaluations should
encompass more than performance
assessment. The drivers behind those
performances need to be investigated as well.
These can only be assessed through dialogue
with householders. Surveys are the principal
instruments for establishing those dialogues.
Establishing what the drivers are is not easy.
Straightforward questions like “Why don‟t you
recycle?” may not always uncover the true
reasons for non-participation. People often tend
to rationalise their reasons for behaving in a
certain way into a small number of terms that
do not necessarily reflect the original reasons
for the initiation of those behaviours (see e.g.
Vining et al., 1992). Asking “Why don‟t you
recycle?” tends to generate a small number of
stock responses: like “No bag”, “No
information”, “Takes too much time”, “Forget”,
“Too lazy”, “Poor service” and so on (see
Tucker, 2001). Asking “What would make you
recycle?” generates a list of compatible
reasons: “Give me a recycling bag”, “Give me
more information”, “Provide better service”, etc.
However, even when those requests are
154
Understanding Recycling Behaviour 2
fulfilled, most people still do not recycle.
Surveys need to probe deeper and uncover the
true attitudes that fundamentally drive people‟s
recycling behaviours. The research presented
in section 8 of this report was designed to try to
delineate those attitudes.
A further role for household waste attitude/
behaviour surveys is to derive recycling and
waste minimisation performance indicators that
cannot be obtained by any other technique.
Section 8.4.8 proposes a set of performance
indicators and a survey protocol that can be
used to monitor waste knowledge, waste
attitudes and household waste management
behaviours. The protocol was specifically
designed for monitoring across episodes of
change, i.e. before and after the
implementation of a new recycling programme,
or before and after the application of a
promotional campaign. However, it is
considered that the indicators can serve equally
in any longitudinal study to monitor recycling
activities over time.
11.2.5 Longitudinal Monitoring
It is very important that recycling performances
are analysed on a regular and frequent basis.
This can help identify the onset of gross
behavioural changes as early on as possible.
Often performances are only assessed at year-
end at which stage it may be too late to remedy
some of the problems that have developed
throughout the year. Spotting the onset of true
problems, however, can be difficult with the
performance monitoring statistics that are
normally available. The problem, once more, is
extracting the „signal‟ from the „noise‟.
It is well accepted that recycling yields can vary
enormously from collection to collection. These
short scale [seasonal] fluctuations arise through
a number of factors. The most obvious of these
are non-collection weeks (on bank holidays).
Those non-collections usually lead to a „high‟
on the next collection, followed by a „low‟ on the
subsequent collection. Basically this arises
because recyclers who normally recycle every
other collection tend to become forced into
phase.
Other causes relate to the seasonal variations
in consumption (see e.g. Jones et al., 1998b,
MEL, 1996). These may relate to voluntary
purchasing decisions like purchasing fewer
magazines in summer or may be involuntary
outcomes, for example, from seasonal changes
in newspaper pagination. The research has

shown that these seasonal factors may be
reasonably consistent from year to year. As
such they might then be quite predictable. The
statistical technique of seasonal decomposition
allows those seasonal variations (i.e. much of
the „noise‟) to be extracted from the total signal.
This helps the identification of any underlying
longer-term trend. The research has shown that
the seasonal tends in paper arisings might, in
fact, be quite similar across different
authorities. However, it is not yet established
whether the same phenomenon apply equally
to other recycled materials as well. That
assessment still needs further research.
The underlying trends, post deseasonalisation,
convey important information on the evolution
of recycling behaviours. On their own those
trends can be difficult to interpret. They may be
due as much to changes in consumption as to
changes in recycling behaviour. Overall year on
year changes in consumptions are reckoned to
be increasing by 2 – 3% per annum; however
different rates may apply to different materials.
For example newspaper consumptions
(paginations) decreased between 2001 and
2002, recovering slightly in 2003. These affects
were mainly the consequence of changes in the
advertising revenues of the major national
newspaper titles.
The district-wide trends were found to be far
less pronounced than the changes in their
constituent daily collection rounds, i.e.
individual collection rounds showed much more
variation than the whole district averages.
Yields from some rounds fell over the
monitoring period whilst others rose. This
clearly emphasises the need to conduct
performance monitoring at the individual
collection round level. The reasons for the
differences in individual round performances
will be discussed further in section 11.3.
Understanding Recycling Behaviour 2
11.2.6 Conclusions
Overall, the research results lead to the
conclusion that performance monitoring must
be undertaken at the highest resolution that is
practical. Effectively this implies that routine
monitoring should be conducted [ideally] at the
daily collection round level. Temporal
fluctuations at that level can be relatively large
compared with district-wide statistics. Problems
or improvements within individual collection
rounds can easily be masked in the district-
wide analyses. Identifying those localised
problems or improvements is crucial. The
whole is no more than the sum of the parts.
Managing the whole also means understanding
the parts.
Performance evaluation is a comparative
exercise. Monitored performances must be
compared against something for assessments
to be made. The prime criteria are, of course,
the statutory targets and other audited criteria
set by Central Government. At the more local
level comparisons will be made against
forecasts or against historical, model, or
baseline performances. Often this only takes
place at infrequent intervals. If monitoring and
evaluation is to evolve from a primarily audit-
driven task to one of promoting continual
improvement, more regular performance
evaluations become essential. Those will only
prove effective if the necessary comparisons
can be made reliably at an increased
resolution. That will in turn demand better
forecasting models. The research has shown
how a simple forward extrapolation of recycling
yield time series may serve (at least to some
extent) in this respect. More sophisticated
models could be built, for example based
around the principals of the University of
Paisley Integrated Household Waste
Management Model. Those models could be
easily configured effectively using an expert
system approach, to provide automated
assessments, trigger alarms and so on. The
development of such decision support tools for
recycling programme monitoring and evaluation
is seen as an important next step for further
research.
155

11.3 Do Demographics Really Matter?
11.3.1 The Underlying Premise
Household waste management behaviours are
quite diverse. Everyone is different – though
some are more similar than others. The notion
that those with similar waste management
behaviours might share some other
fundamental characteristics as well has
provided the cornerstone for waste behaviour
research for the last two decades. It is
fundamentally appealing and reassuring to
believe that different waste management
behaviours might be separately identifiable for
separately distinguishable sets of people.
Socio-demographic criteria are used widely in
marketing to distinguish consumer purchasing
power, habits and lifestyle preferences.
Implicitly this means that similar criteria should
apply equally to consumer waste generation as
well. The link with recycling performance is
more tenuous, though supported by similar
arguments about lifestyle differences and
priorities, educational levels, frugality of
upbringing, and personal transportation. Many
studies have sought to correlate recycling
behaviours and demographics, however the
results have been quite equivocal (see Tucker,
2001 for a review). However, overall, the notion
that the older or more affluent members of the
population are the better recyclers generally
seems to hold.
The current research has given much emphasis
to exploring the demographic bases of
recycling behaviours, and to assessing how far
demographic considerations might aid recycling
programme management.
Of course, demographic factors will inevitably
lose their explanatory power as more and more
people start to recycle. This is already the
situation with donating waste goods to charity,
where the practice is now pervasive across all
segments of the community.
11.3.2 Who is Now the Recycler
The new data gathered in the course of the
questionnaire-based research demonstrated
that newspaper and glass recycling sometimes
correlated with demographic factors, though
aluminium and steel can recycling rarely did.
The significant factors generally conformed to
traditional wisdom that it was the older
residents and the more affluent households
156
Understanding Recycling Behaviour 2
who were the better newspaper and glass
recyclers. Young adults without children and
families with young children were often
amongst the poorer recyclers. Fewer residents
of the older terraced housing and flats tended
to recycle compared to residents of detached
and semi-detached housing Single person
households were also relatively poor recyclers
as were large households of five or more
persons. Social class never proved to be a
significant factor.
The case studies in Chesterfield, North East
Derbyshire, Lancashire and North
Warwickshire produced more mixed results
with respect to demographics. In the
Lancashire and Derbyshire districts,
demographic models fitted to the recyclate
yields usually identified the expected factors
amongst those contributing most positively to
material yield – that is, detached houses, more
mature residents, Acorn A etc. Similarly the
factors with the most negative influence on
yield were also much as expected, e.g. non-car
owners, private sector rentals, Acorn E and F.
In North Warwickshire, however, Acorn E
housing (essentially the better-off local
authority housing schemes) and families with
young children had the most positive influence
on yield, with the most affluent housing (Acorn
A) exerting a surprisingly weak influence.
Different demographic factors were identified
as significant in the recycling of different
materials. A strong contrast between paper and
glass recycling and textile and can recycling
was seen in Derbyshire, with the older and
more affluent households exerting more
influence on the former, and the younger
poorer households exerting more influence on
the latter. Partly this could be due to
[demographically] linked differences in
consumption of these materials, i.e. having
more to recycle, or it may be due to the
traditional recycling base selective choosing not
to recycle such materials. Demographic
associations for can recycling tended to be
weak across most communities.
Reduce/ reuse behaviours appeared even less
strongly associated with demographics than did
recycling behaviours. The weakest reduce/
reuse behaviours were generally found
amongst the largest households whilst the
strongest of those behaviours were most
commonly associated with two person
households.

Whilst there is strong evidence that the
recycling ethos and the reduce/ reuse ethos
might be separately and preferentially
associated with different sectors of the
community, there is no evidence that those
differences are of demographic origin. There
was no consistent demographic factor that
identified a recycler from a waste reducer or
reuser and vice versa. The only commonality
was in the group who practiced none of the
activities, that group being preferentially drawn
from the younger life stages and the larger
households.
Overall, it can be concluded that whilst there
appears to be a general pattern between
observed recycling behaviours and
demographic profiles, the performances of
individual neighbourhoods show much scatter
around that general trend. A general
demographic predictor cannot explain all of the
variance between neighbourhoods. Different
demographic factors appear to take on different
importances in different districts.
Perhaps, more importantly, the overall picture
appears to be one of reasonable conformity
across the majority of areas (both districts and
neighbourhoods), with quite similar behaviours
usually found for around 80% of the sample
irrespective of their demographics. The other
20% showing the outlying behaviours
sometimes, though by no means always, were
associated with extreme demographic profiles.
Often, it was the most highly affluent areas that
stood out from the rest.
This paints the general picture of a district-wide
(or even county-wide) behavioural norm that is
largely independent of demographics, at least
for areas of reasonably well-mixed
demographics, with local demographic
variations only contributing quite small-scale
variations about the norm. It might take quite
extreme demographics before an area shows
up as having a significantly different recycling
behaviour to the rest of the community.
11.4 Attitudes and Behavioural Change
11.4.1 Fundamental Attitudes
Over recent years there has been increasing
research into household waste management
attitudes and behaviours. However, most
studies have concentrate more on exploring
„Who recycles?‟ and „How much they recycle‟
Understanding Recycling Behaviour 2
However, having a significantly different
recycling behaviour to the rest of the
community can also be due to many other
factors other than demographics.
11.3.3 Who will Become the Recycler
By and large, it is of less strategic importance
to know who is recycling now to knowing who
could be recruited to recycling in the future. The
heaviest recruitment events happen when new
recycling programmes are implemented. But
how far do such events encourage the poorer-
recycling demographic groups to recycle more?
In Chesterfield and North East Derbyshire,
there was some [weak] evidence that the
traditional demographic bias of recycling was
eroded slightly upon the conversion of their
paper-only kerbside collections to multi-
material.
Stronger evidence was seen in the
Renfrewshire survey, where a new kerbside
collection preferentially recruited non-car
owners and families with older children to
recycling. Proportionally fewer new recyclers
were drawn from the retired residents and from
Acorn A households.
However, it must be borne in mind that many of
the traditional recycling base (i.e. retired
residents, Acorn A) were already recycling prior
to the new schemes being introduced. As such,
those sectors would be severely depleted of
„would be‟ recyclers/ thereby leaving relatively
few to recruit from. Consequently
proportionately more recruits would be
expected, on statistical grounds, to come from
the younger age groups and less affluent
housing stock. It is probably not because the
individuals belonging to any groups were any
more susceptible to behavioural change.
rather than the more pertinent question „Who
does not recycle currently, but could be
stimulated to do so?‟ Psychological models of
behaviour, e.g. The theory of planned
behaviour (Azjen, 1985), altruism (Schwartz,
1977), and repeated behaviours (Ronis et al.,
1989) can all help explain parts of the observed
157

variations in householders‟ recycling
behaviours. The various constructs contained
in each of those models may be equally
important. Attitudes relating to self-efficacy,
awareness of consequences, and acceptance
of responsibility continue to be found to be
determinants of recycling participation, together
with negative perceptions about recycling and
perceptions about the local recycling
programmes (see section 8.4). The current
research found that levels of environmental
concern were also stronger amongst recycling
participants. This contrasts with some of the
earlier research studies that found levels of
general environmental concern were not
significant to recycling activity.
Despite all the above attitudes being stronger
amongst recyclers than amongst non-recyclers,
no one attitude or any combination of attitudes
could achieve a clear separation of the
recyclers from the non-recyclers. Only 73% of
behaviours could be predicted correctly from
the attitudes held. It may be that other, as yet
unidentified, attitudes are important in
supporting recycling behaviour. Alternatively,
simply holding the right attitudes may not be
sufficient, on its own, to support recycling
behaviour. Some other factor or factors could
be necessary as well. This point will be
discussed further in the next section. However,
it must not be overlooked that another
possibility that could explain the imperfect
mapping of attitudes onto behaviours may be
the imperfect responses of the residents. It was
suspected that several respondents might be
making exaggerated claims about the levels of
their recycling activity, and falsely claiming to
recycle (section 9). This could confound the
delineation of attitude behaviour links. Of
course, the expressed attitudes could have
been falsely reported as well. However, on
taking those attitudes at face value two logically
consistent features were revealed:
 The acceptance of personal responsibility
attitudes of those suspected of falsely
claiming to recycle were intermediate to
those held by the non-recyclers and those
held by the confirmed recyclers.
 Self-efficacy and awareness of
consequences attitudes were elevated in
those suspected of claiming to recycle
more materials than they actually did.
The first group therefore had some beliefs that
they ought to be recycling and appear to have
made their claims of recycling based on that
attitude rather on than what they actually did.
158
Understanding Recycling Behaviour 2
The second group voiced the highest levels of
self-confidence and knowledge. It was as if
these respondents felt a need to demonstrate
that they were highly knowledgeable and highly
active, rather than admitting that they were not
quite that perfect. Again, they appear to be
expressing what they think they should be
doing rather than what they actually are doing.
Overall, the attitudes held were generally self-
consistent. Those living in areas of poorer
facilitation recognised that fact. Those
perceiving poor facilitation or holding the most
negative perceptions were non-recyclers with
the knowledge of the local recycling
infrastructure. Perhaps they were previous
recyclers who had dropped out. Those without
knowledge of where to recycle were much
more non-committal towards those attitudes.
Weaker attitudes were normally found amongst
the younger family life stages.
Whilst the research focused on the four specific
attitudes of acceptance of responsibility, self-
efficacy and awareness of consequences,
negative perceptions and local facilitation, it
was found that the public‟s responses fitted
more logically into four slightly different
constructs:
 Belief that recycling is necessary and
that one should do something
personally;
 Negative perceptions about doing the
activity;
 Belief that one is able do the activity
and the trust that one‟s actions will be
followed through by others;
 Belief that others need to do something
as well;
11.4.2 Initiation and Persistence
Attitudes
Different attitudes might influence recycling
behaviours at different times. There is growing
evidence that different sets of attitudes control
the initiation of new behaviours compared to
those that influence the persistence of those
behaviours (section 8.1.2, 8.2). For example,
holding negative predispositions about the
activity can prevent the take-up of that activity
even when all other conditions might be
favourable. These negative mind-sets could
have their roots in historical experiences or
could be prejudices developed outside of
experience. Persistence, however, appears to
be strongly coupled to the perceptions gained
through experience. If those perceptions are

favourably set then stable positive behaviours
should ensue. However, if negative perceptions
develop, they can lead to immediate or
eventual drop-out, depending on the severity of
the causal experience. Such negative
perceptions comprise two distinct categories:
 Perceptions of personal cost;
 The results of bad experiences.
There is growing evidence that many long-term
attitudes might be set in the very early stages
of gaining experience, i.e. during the first few
months of a new recycling programme. The
initial experiences with those programmes will
quantify the personal costs that are involved
(e.g. time and effort that need to be spent, and
the conflicts with established lifestyle priorities).
The conflicts that are recognised might also be
related to time and effort (too busy) or could
relate to more physical difficulties, e.g. storage
demands, or the recycling box being too heavy
or too difficult to handle.
Untenable personal costs and conflicts might
also develop later. Here, a triggering event
could render what was tenable previously to
being untenable thereafter. Moving to a smaller
house or rearranging the garden could conflict
with the motivation to home compost. The
physical demands of home composting or
carrying the recycling box to the gate might
become too demanding after an onset of illness
or infirmity. A new baby in the house could
upset previous routines and priorities.
Bad experiences usually relate to poor services
delivered by the recycling scheme provider or
to unfortunate incidents such as theft of the
recycling box, the lid blowing away, vandalism
and so on.
Whilst the cause-effect links of drop-out are
now reasonably well understood, the specific
attitudes that govern take-up of recycling are
less well explained. Further research is still
needed to isolate the key attitudes that control
the initiation. Acceptance of responsibility
would appear to be one of the more important
of those attitudes. New recruits to recycling
were found to have a weaker acceptance of
responsibility than those who had recycled for
several years (sections 8.3, 8.5). The same
attitude was also weaker in those claiming to
be recyclers rather than being to the true
recyclers (section 8.4), and also amongst the
recycling drop-outs (section 8.3).
Complementary to this, new recruits, false
claimants and drop-outs all showed a stronger
acceptance of responsibility than that held by
Understanding Recycling Behaviour 2
the confirmed non-recyclers and the „never‟
recyclers.
In Clackmannanshire there were distinct
attitude differences between the long-term
recyclers in the community, and those who
were about to become recyclers. Attitudes were
all stronger amongst the long-term recyclers.
After the enabling event of a new kerbside
collection, attitudes of both groups increased.
The differentials between their attitudes were
maintained. The biggest change was a sharp
rise in self-efficacy amongst both groups. Other
attitude components changed as well, but to a
lesser degree. The more pronounced of those
other changes may have been related to the
new visibility of the behaviour (see the section
on norms below). All the observed changes
appeared to have plausible explanations,
though those explanations could not be
verified.
Norms
In section 11.4.1, it was mentioned that other
factors apart from attitudes may be involved in
initiation of new behaviours. Normative
influence is one clearly identified factor.
Normative influence or social pressure to
recycle can act directly through dialogue with
friends and neighbours, or indirectly through
observation of neighbours‟ behaviours. The
latter effect may be particularly strong in
kerbside recycling schemes where one‟s
behaviour (i.e. the set out of recycling
container) can be clearly visible. The
introduction of kerbside recycling to
Clackmannanshire produced attitude changes
that were consistent with putting one‟s
behaviour under increased scrutiny (for the new
recruits) and with scrutinising one‟s neighbours‟
behaviours (for the prior recyclers).
Research has produced little hard evidence to
confirm that significant „natural‟ recruitment
may be taking place through normative
influences, however interventions through
setting up local champions do seem to work in
some circumstances. The links between more
natural normative influences and recycling
behaviours are assumed largely by inference.
Normative influences would appear to be
stronger in the more closely-knit parts of the
community, and if they can stimulate changes
in recycling behaviour, those changes would be
more likely early on in the life of a recycling
programme. There is some evidence that
159

talking through problems with the recycling
scheme may help alleviate drop-out.
Engaging in social dialogue about recycling
could lead to change of attitude in one or both
parties, propagate awareness, and also correct
[or spread] any misconceptions.
Almost nobody surveyed in the course of the
research admitted to becoming a recycler
because of receiving encouragement from
friends or neighbours, or because they saw
their neighbour recycle. However,
circumstantial evidence remains strong that it
does happen. Maybe it is a case of
respondents making a post-rationalisation of
why they now recycle (see Vining, Linn and
Burdge, 1992).
Triggers and Convenience
The companion monograph Understanding
Home Composting Behaviour showed how the
recent take-up of home composting has sprung
mainly from special triggering events such as
local compost bin promotions. The concept of
the „latent composter‟ was introduced to
describe those who were recruited to
composting through such promotions. It was
considered that positive attitudes towards home
composting were already in place amongst the
„latent composters‟, but that some kind of a
trigger was needed to switch on those attitudes
and turn them into actions.
A similar trigger may operate when people are
given a new kerbside recycling collection. That
trigger can deliver a cocktail of different stimuli:
 a step change in convenience
 persuasive messaging
 opportunism
Guagnano et al. (1995) proposed a simple ABC
model of behaviour, that essentially stated that
if:
Attitudes + Convenience > Threshold then
Behaviour will occur
As such, increasing attitudes or increasing
convenience offer two alternative routes to
switching on behaviours. However, as shown in
section 2, the „either-or‟ situation may not hold
through to the highest levels of behaviour. The
improvements that are achievable through
convenience alone or through attitudes alone
will be limited. Increasing both will be essential
160
Understanding Recycling Behaviour 2
if the highest levels of behaviour are to be
reached.
On the Guagnano et al. (1995) model, drop-out
can be considered to occur when the negative
attitude shifts from adverse experiences move
„attitude + convenience‟ total back across the
participation threshold. This will be discussed
further in section 11.5.
11.4.3 The Role of Knowledge
The importance of knowledge to participation is
not clear-cut. Procedural knowledge about how
to carry out the behaviour would appear to be
the most relevant class of knowledge
supporting participation. Having that knowledge
strengthens the attitude of self-efficacy, which
has been shown to be a significant determinant
of behaviour. However, that knowledge may not
be as relevant to initiation than it is to
persistence. Many new recruits to home
composting took up the activity despite
professing poor knowledge of how to compost
at home. Awareness of consequences provides
a general attitude conveying knowledge of the
reasons why the activity is important and what
would happen if the activity did not occur. This
was also important to recycling, though most
recyclers felt that they had sufficient information
to justify their decisions to recycle. Specific
technical details, and statistics about recycling,
however, were poorly known amongst recyclers
and non-recyclers alike. In-depth knowledge at
that level would appear almost superfluous or
irrelevant to the decision to recycle.
Procedural knowledge is also important to
material capture. Some recyclers do not recycle
some materials because they think those
materials are not wanted in the recycling
programme. Conversely some recyclers include
contrary material because they believe that
those materials are wanted.
11.4.4 „Natural‟ Change
Behaviour changes can be classified either as
„engineered‟ changes that result from local
management interventions - both programme
enhancements and promotional campaigns, or
as „natural‟ changes which are not specifically
planned. „Natural‟ changes comprise the
adverse experiences that result in drop-out,
together with the „background‟ recruitments that
are not directly attributable to local
management actions. These could include the
national awareness campaigns, local and
 Understanding Recycling Behaviour 2

national media coverage, or the results of local
social influences.
The results presented in section 7 indicate that
far stronger and varied changes occur at the
more local (collection round) level than are
apparent at the district level. This would
indicate that the drivers of behavioural change
might in fact be highly localised. This reinforces
the impression that broader regional or national
campaigning might be contributing relatively
little to behavioural change. Both positive and
negative local changes have been witnessed at
the local level. Incidents of localised problems
or developing local norms may both be
involved. The specific cause-effects were not
investigated in the current study.
There was some evidence that experiencing
specific behaviours may lead to closely allied
behaviours being taken up as well (e.g. bring
recycling of glass after experiencing kerbside
recycling of other dry recyclables). However,
the timescales may be protracted and there is
no guarantee that it will happen. Further
evidence needs to be assimilated before any
more definite conclusions can be drawn.

11.5 A Conceptual Model of Recycling Behaviour

11.5.1 Basic Concepts behaviours (section 6) has provided much

further evidence to support that notion. A fuller
Perhaps the most successful model for schematic of the likely hierarchical attitude
accounting for recycling behaviours has been structure is shown in figure 11.2
the Theory of Planned Behaviour (see section
8.1.2). The research findings at the University Figure 11.1 Hypothesised Participation
of Paisley have usually proved to be quite Model
consistent with that theory. With minor
modifications, the theory was adopted as the Specific barriers Scheme-specific
kernel of the conceptual model for the
Awareness appeal
University of Paisley Integrated Household
Waste Management Model. Figure 11.1 Perceived
reproduces the original conceptual model [for effectiveness Specific recycling
the Paisley model] as was presented in volume Have container attitudes
1 of Understanding Recycling Behaviour. Lifestyle factors
Pro-environmental
Whilst specific barriers to recycling have been General waste management
identified, the „attitude‟ input to the model has recycling attitudes
never been defined. It has always been barriers
assumed that it was just some kind of Social norms
anonymous recycling attitude. Different
researchers have investigated the various
attitude components that could potentially
contribute to that general attitude; however
there has been little coherence across that Intention to recycle
research and no general consensus has
emerged. The current research programme
also investigated the possible components of Personal
the pro-recycling attitude (section 8.4) though, difficulties
like other investigations, the results were not
conclusive. The research produced pointers System
rather than answers. difficulties

In developing the original conceptual model it

was well recognised that the supporting Recycling
attitudes might follow a tiered structure, from
general to specific. The new research into the
correlations between waste minimisation

161
 Understanding Recycling Behaviour 2

Figure 11.2 Postulated Hierarchical Structure of Household Waste Minimisation Attitudes

Kerbside Bring CA
Appeal Appeal Appeal

Recycling Attitudes & Reduce/ reuse Home Compost

Perceptions Attitudes Attitudes

Energy
Waste Management Attitudes Conservation
Attitudes

Environmental Concern

On moving up the tiers in the hierarchical
structure, the attitudes become more specific,
more focused and more pertinent to the activity.
All attitudes can be positive or negative, and
are not necessarily confined to being
„environmental‟ or „moralistic‟ in nature. For
example, gardening benefits (in home
composting), and saving money (by reuse) both
focus on personal gains.
Problems affecting one tier do not normally
compromise the attitudes held in the lower
tiers. For example, problems with a kerbside
collection could lead to drop out from that
scheme, but will not necessarily damage the
more generally held attitudes about recycling.
The individual experiencing the problems may
simply switch to using recycling banks –
provided of course that their scheme appeal is
sufficient (see Tucker, 2001 for a case study).
convenience] and with self-efficacy [through
awareness and procedural knowledge].
Whilst the specific attitude components within
the aggregated general recycling attitude
cannot be delineated at the current time, a
slightly enhanced classification is now possible,
based on the considerations of the current
research. The updated behavioural model is
shown in figure 11.3. The feedback loops
should be specially noted. Those loops show
how behavioural experience can revise
attitudes and/or reinforce the personal norm.
The latter effect is essentially habit formation.
On this model, householders will participate if
the balance between pro-recycling attitudes,
barriers and norms is favourable, i.e. when:
Attitudes + Norms – Barriers > 0

Scheme appeal also links with facilitation This formulation is essentially an extension of
[through infrastructure availability and the ABC model of Guagnano et al. (1995).

Figure 11.3 Updated Behavioural Model

Personal Norms Intention Behaviour

Social

Attitudes & Barriers

Personal
 Bad experience  Scheme difficulties
 Awareness
 Personal cost appeal
 General  Perceived
 Other post- System
recycling effectiveness
rationalised difficulties
 General  Have container
 Acceptance of
waste  Lifestyle factors
responsibility
 Procedural
 Pre-dispositions min.
information
 Other

162
 Understanding Recycling Behaviour 2

11.5.2 Visualising Behavioural basic attitude/convenience balance will become

Change
Here we assume that the aggregated effects of
all causal factors (attitudes, barriers and norms)
provide some kind of an overall „motivation‟ to
recycle. Whilst we still need to disentangle how
individual attitudes and other factors add up to
that motivation, the precise knowledge of that
make-up is not crucially important at this stage.
It does become important, however, when we
need to consider how to increase that
motivation (section 11.6). At this stage we
simply assume that such a motivation can be
defined.
sufficient on its own for participation to occur
normally, and additional normative
enhancements would no longer be necessary
for participation.
The above point may be quite important. The
factor governing participation has shifted. The
recycler may now be recycling for quite
different reasons than when they initiated the
behaviour. The recycler may also have
rationalised new attitudes to vindicate their
behaviour: e.g. “I recycle because it is good for
the environment” even if environmental reasons
played no part in why they started.

On this simplified model, an individual will

recycle if that individual‟s motivation exceeds a
given participation threshold. It follows that
individuals will only change their behaviour
when their motivation crosses that participation
threshold. A central tenet of the research is that
the motivational changes will only occur when
triggered by an event or events.
Figure 11.4 Effects of Attitude Change
a) Major catastrophe
b) Major catastrophe (stronger initial
attitude)
c) Cumulative effect of minor set-backs
Strength of Motivation

The way this works can be visualised best

through an example. Consider what could
happen if a recycler has an adverse experience
with the recycling programme. The recycler is
then likely to take a decision on future
involvement. The outcome for future motivation
will depend on the perceived severity of the
causal experience. Time --- 
Time

The magnitude of the motivational change may

be very large, if the event is construed as a
major catastrophe, or it could be quite small. A
large change in motivation could shift the
attitude/convenience balance from a positive to
a negative intention to recycle (figure 11.4a). A
small change, however, would not necessarily
do so, but a succession of small changes could
eventually do so (figure 11.4c). This could
equate to repeat incidents of missed Behavioural
collections. Conversely, if the initial attitude is change
strong enough then the individual might come
through quite a major hiatus without changing
behaviour (figure 11.4b).

Prolonged experience in a recycling scheme

might forge new fundamental attitudes towards
that scheme even though the original
motivation to recycle was externally induced
(e.g. triggered by normative influence or
reward). On the conceptual model, it can be
considered that each recycling episode that is
triggered by local social pressures causes a
positive small step change in motivation
(through positive augmentation of the personal
norm). After a succession of such events, the
The drop off in participation that is often seen
during the first few months of a new recycling
programme is often put down to the novelty
wearing off, or recycling fatigue. However, in
reality it is likely to be due to more complex
attitude dynamics.

163

For example, a step change in one attitude
[say, for sake of argument, a new found
acceptance of responsibility] may recruit a
person to recycle. Another attitude may then
cause a rapid drop-out [e.g. realising untenable
personal costs]. The attitude profile of the
individual might be quite different now to what it
was a few weeks ago. That person was a non-
recycler then and is still a non-recycler now.
This type of effect was seen in the
Clackmannanshire survey (section 8.4) where
two main classes of non-recycler were
identified as having significantly different
attitudes.
It was also apparent in Clackmannanshire that
a raft of individual attitude changes was
associated with an episode of behavioural
change (section 8.5). Many attitudes might shift
with time.
An individual‟s current pool of attitudes will
have been shaped through that individual‟s
cumulated personal experiences. Each
experience will have been associated with an
event. Those events might have been planned
(e.g. changes in the offered recycling
programme, promotional interventions and
awareness campaigns) or might have
happened outside the control of the local
recycling manager (e.g. through national
advertising, collection problems, theft,
vandalism, social dialogue, moving house and
so on). Each event can give rise to changes in
one or more attitudes. It has been found that
those changes are likely to follow a
characteristic profile (figure 11.5).
Figure 11.5 Characteristic Attitude Profile in
Response to an „Event‟
Strength of Attitude
A
R
T
Event Time 
The event is assumed to cause a step-change,
A, in attitude, which may be positive or
negative. The effects may „wear off‟ over time,
164
Understanding Recycling Behaviour 2
T,9 though could leave a residual effect, R,
in the attitude. The decay time is likely to be „a
few months at most‟ for most promotional
interventions, though with some such as giving
„rewards for recycling‟ the decay is usually
instantaneous as soon as the rewards are
withdrawn (and the residual is zero as well).
Normally, little decay happens after adverse
experiences (R = A; A being negative).
The strength of the attitude change, i.e. the
individual‟s reaction will vary from individual to
individual. It will depend firstly on the
susceptibility of the individual to take heed of
the event. It will then depend critically on the
perceived relevance of the event to that
individual‟s lifestyle, that is the salience of the
event.
With bad experiences, the strength of the effect
will also be determined by its salience but may
be mitigated through resilience or finding a way
of overcoming the problem (e.g. asking the
council for a new recycling container when
yours is stolen, seeking advice when the
compost goes slimy, and so on.).
A change to the local recycling programme will
generally produce a change in motivation given
by:
M = C  salience  susceptibility
where C is the step change in convenience
Model analyses showed that the changes from
paper-only collections to multi-material
collections in Chesterfield and North East
Derbyshire produced identical step changes in
motivation, M, across the two communities
section 4.6.1). This is an important result,
showing that an identical intervention can result
in the same „attitudinal‟ outcome in different
communities. As such the changes in recycling
performance resulting from specific programme
changes should be quite predictable. However,
although the motivational change was the same
in the two areas, the actual recycling
performances did, of course, differ. A larger
step change in performance was found in
Chesterfield than in North East Derbyshire.
This eroded but did not eliminate the historical
performance differential between the two
districts. The reasons for these outcomes are
easy to visualise using the conceptual model
figures 11.6 a-b).
9
In reality the decay could be due more to other
[conflicting] attitudes building up.
 Understanding Recycling Behaviour 2

Figure 11.6 a-b Changes in Motivation right, the higher the motivations. P0 designates
Resulting from a Change from Paper-only to the proportion of recyclers before the
Multi-material Kerbside Collection conversion whilst P shows the number of new
recruits resulting from the conversion. The
percentage of recyclers in the new scheme is
N.E. Derbyshire given by P0 + P.
Threshold
100 In Chesterfield, much larger step changes in
performance were achieved when the residual
Cumulative % of population

waste collection frequency was halved when

the multi-material collection was introduced
P0 section 4.3). It was considered that the
imposed change in the residual waste
collection raised the saliency of the recycling
issue significantly amongst the affected
P households. The outcome was that the
motivational change towards recycling was
significantly increased figure 11.6c).

0
Figure 11.6c Changes in Motivation
Motivation Resulting from a Change from Paper-only to
Multi-material Kerbside Collection Coupled
Before After with a Reduced Frequency Residual Waste
Service

Chesterfield +Alt Week Residual

Chesterfield
100 Threshold
100 Threshold
Cumulative % of population

P0
P0
Cumulative % of population

P

0
0 Motivation
Motivation
Before After
Before After

As everyone is different, individual motivations

to recycle will also differ. As such the
motivations amongst any community take on a
distributed range of values rather than being
single valued. This can be visualised in figures
11.6 a-b. These plots show the cumulated form
of the probability distributions. In the curves,
the y-axis shows the proportion of the
population with recycling motivations above a
given value. The further the curve is to the

165
 Understanding Recycling Behaviour 2

11.6 Promotional Campaigns

11.6.1 Visualisation of Promotional

Campaigns D No Change in Behaviour
(Preaching to the converted)
Following the arguments of the previous
section, a promotional intervention can be
visualised conceptually in terms of the
attitudinal changes it produces. These are
assumed to follow the characteristic profile
shown in figure 11.5. An intervention may
target and increase one or more attitudes. On
the conceptual model, it is assumed that the
new combination of new attitudes will then E Habit Formation/ Post-rationalisation
aggregate in some way into an increased
motivation to recycle. If the new motivation is
below a given threshold, the individual will not
recycle. If it is above, they will.

Figures 11.7a-e illustrate the five main possible

outcomes of promotions.

Figure 11.7 Possible Outcomes of

Promotional Interventions
Figure 11.7F Repeat Interventions to
A Recruitment & Sustained Participation Prolong Effects

Threshold
Event

B Recruitment & Early Drop-out

The ultimate success of the promotion will be
determined by its strength and decay
characteristics. A key objective of the
promotional strategy should be to keep the
effects going long enough for habits to form
and for the supporting attitudes to become fully
internalised (figure 11.7e) (see Pallak et al.,
1980; Katzev and Pardini, 1987-88, Werner et
al., 1995). Repeat stimuli may be needed to
C No Change in Behaviour keep behaviour going long enough for these
effects to be achieved (figure 11.7f). However,
the research literature offers no clear guidance
on exactly how long it takes for habits to form. It
should also be noted that the follow up
promotion may need to address different
attitudes to the first as new attitudes may have
come into play at that stage. Providing early
feedback could also reinforce attitudes at this
stage (Seligman et al., 1981, Katzev and
Mishima, 1992). Making the feedback

166

consequences salient is critical to its success
(De Young et al., 1995).
11.6.2 Salience and Susceptibility
More broadly, saliency would appear to be a
critical factor in making any promotional
intervention successful. The governing
relationship in the conceptual model is:
Attitude change = Strength of message 
Salience of message  Susceptibility to
assimilating the message
Salience means of personal relevance to the
recipient. For the communication to be
effective, recipients must be able to relate the
message content to their own experiences and
lifestyles. Salience also means receiving the
message at the right time and at the right place
(see below).
Susceptibility to assimilating the message firstly
means actually receiving the message. That
depends on the delivery medium. Those
methods include:
- mailing leaflets,
- hand delivered fliers,
- personal contacts,
- newspaper announcements and
advertisements,
- television or radio commercials,
- reminder signs,
- exhibits at fairs, etc.
Everett, Jacobs and Pierce (1991) ranked the
effectiveness of commonly used promotional
media. The most effective delivery mechanisms
were all high on contact area and could reach a
large proportion of potential participants. The
top ranked promotion (personal contact) was
also high on contact level. It must also be borne
in mind that different segments of the
population tend to get their information from
different media. It is considered, for example,
that more educated people are most likely to be
influenced by newspaper coverage, whilst TV
advertisements may be the best medium to
reach those not specifically looking for recycling
information (Vining and Ebreo, 1990).
However, even if the message is received, it
may not be assimilated. Leaflets might not be
read. Newspaper advertisements and articles
will, at best, only be observed by readers of
that newspaper. Even then, readers may not
assimilate or remember the information. There
is also evidence that, to be effective, the point
of delivery of the communication must be
Understanding Recycling Behaviour 2
congruent with the point of action. Promotions
for recycling household recyclables should be
delivered at home rather than, for example, in
the work place (Needleman and Geller, 1992).
Spatial separation of the message to the action
can diminish the desired response. Even if the
information is read, it may not be accepted.
Vining and Ebreo (1990), for example, consider
that non-recyclers might selectively ignore or
discount information that is irrelevant to their
behaviour, because holding such information
creates a dissonance with the non-recycler‟s
self-concept of being a responsible member of
society.
Market segmentation principles are advocated
to ensure that the delivery mechanism is best
tuned to suit the target (Katzev and Mishima,
1992). Care must be taken to tailor the
message to suit individuals holding different
sets of values; however care must also be
taken not to offend viewers by messages
directed at other market segments (Granzin
and Olsen, 1991). As such, addressing „micro-
markets‟ may be difficult to achieve in practice.
11.6.3 Message Content
Very little will be said in this report about what
the persuasive messages should or should not
say and how that information should be
presented. There are many opinions on this
subject, though there has been very little hard
practical information to back up any of the
hypotheses, theories and conjectures.
The research in this report has given emphasis
to investigating which specific attitudes might
be most important to determining recycling
behaviours. If the deficient attitudes can be
identified, then interventions can be designed
to address their shortcomings. Whilst this
research has not reached its conclusion,
certain classes of attitudes were identified as
being significant. As different attitudes might
affect the initiation and persistence of recycling
behaviours, it follows that the type of
communication and its content may need to
evolve as the recycling scheme matures.
Different messages might be needed to
promote different behaviours. The research
presented in section 6 of this report adds
further evidence that recycling, reduce/ reuse,
and home composting behaviours might have
fundamentally different roots, and might need
fundamentally different messages.
Recycling attitude/ behaviour dynamics are
complex and still need much understanding.
That understanding will only be gained through
167

further research. However, with the
understanding, it should become possible to
design more effective promotional campaigns
that can be deployed with greater certainty of
their outcomes.
When considering promotions and campaigns,
it important to understand that the entire
package is important, not just its components.
Managed combinations of techniques are
generally found to be more successful than
individual techniques applied on their own. It is
conjectured here that a single stage of
intervention may be quite likely to be destined
to fail. Can a single message convey salience,
information, and action points all at once
168
Understanding Recycling Behaviour 2
without the dilution of any one of those
aspects? Conveying one single point per
message can be much more hard-hitting. A
combination of techniques can be used to raise
saliency [first] then hit home with the action
points. A multiple approach will also emphasise
the seriousness of the message.
Finally, it must still be borne in mind that if there
are fundamental shortcomings existing
elsewhere in the recycling programme, those
deficiencies are unlikely to be overcome by
promotions (Shrum et al., 1995). Getting the
programme working right is always the first
concern.

12. Concluding Remarks
This section does not seek to provide any
further summary, analysis or conclusions on
the research. Those aspects are fully covered
in the executive summary and in section 11 of
the report. The section simply provides
additional commentary on the research
12.1 Research Direction
The start of third Millennium has seen stringent
recycling targets being set by Government.
Local authorities have now set out on their
challenge to meet those targets. Whilst the
Government targets graphically portray what
the country needs to achieve, the targets are
quite arbitrary in terms of the behavioural levels
they imply. We live in an imperfect society
which has a relatively low recycling ethos
compared to our neighbours in Europe. What,
then, can our country aspire to, given the
investment, help, cajoling and penalties that the
authorities are likely to impose, with the
investments that they are able to make. The
research presented in this monograph does not
focus on targets or investments. Investments
too are quite arbitrary in terms of behavioural
payback. Nobody has yet sought to quantify the
key indicator „Kg recycled [or diverted from
landfill] per pound invested‟. The research
presented in this monograph starts from a
rather different viewpoint – the people
themselves. It looks at our imperfect society
and asks the question “What can the people
achieve?” given, of course, that the local
authorities, campaigners and other bodies will
provide some help along the way.
The research started off to establish the limits
of what „our people‟ could practically achieve
assuming they were all provided with
convenient recycling programmes and that the
campaigners delivered their promises. That
research delineated the relative contributions of
the two main formative contributions to
behaviour - convenience and education. The
research showed that both were essential and
the application of both together provided the
most effective means to stimulate a large step
change in behaviour. The effects of a step
change in convenience were found to be
reasonably predictable. The impacts of
education were found to be less certain.
Understanding Recycling Behaviour 2
direction that underpins the University of
Paisley‟s research, and examines some of the
opportunities and decisions that could be taken
in the future that might further our
understanding of household recycling
behaviour.
In defining the behavioural limits we assume
that intensive and well orchestrated campaigns
can [in theory] remedy deficiencies in
awareness and procedural knowledge, raise
general pro-recycling attitudes, and correct
misperceptions such as it needs a lot of waste
to make recycling worthwhile. The more difficult
attitudes to change will be any entrenched
predispositions against recycling and the
perceptions of untenable personal cost. The
latter may well have been set through the
actual experience of recycling and will not be
overturned easily through persuasive
messaging from campaigners.
The key issue is not about who is [and who is
not] recycling now, it is about “who is not
recycling now but could do so in the future”. It is
about understanding the mechanisms and the
dynamics of behavioural change. A large
section of the research in this monograph has
concentrated on investigating those
mechanisms and those dynamics. The roles of
attitudes and attitudinal changes to behavioural
changes have been studied. However, whilst
many new insights have been achieved, a full
understanding is still to be realised. This
remains a key issue that still demands further
research.
Two years ago, the watchword in waste
management was strategic planning. Local and
regional waste plans were being written.
Councils were making decisions on the routes
they would take to climb towards their targets. It
was a time to consider the relative merits of
bags, bins and boxes, co-mingled or source
separated collections, and many other details
of new recycling programmes. In a perfect
society, those decisions would not matter… but
we live in an imperfect society!
169

The decisions now taken have led to the start
up of many new kerbside schemes. But are the
results from those schemes meeting
expectations? The emphasis has now shifted
from programme design to monitoring and
evaluation. For most people, monitoring and
evaluation means targets and audits.
Unfortunately those statistics tell us very little
about the people and how they are behaving.
Effective recycling management is more than
providing statutory statistics; it is also about
understanding and developing the best
behaviours amongst the residents.
A large section of the research in this
monograph has focused on the questions of
monitoring, evaluation and diagnostics of
household recycling behaviours. How can you
get the information that you really need to know
out of what you can afford to measure?
12.2 Barriers and Opportunities
12.2.1 Getting Sufficient and
Representative Data
Before assessing what the next steps should
be, it is worthwhile to flag up the major barriers
that have inhibited progress to date.
The central issue lies with sampling and
statistics. Fundamental research investigations
have been severely constrained in their scope
due to resource limitations. Sample sizes have
been much smaller than the ideal. Researchers
usually concentrate on getting good samples of
small populations. But how representative are
they of the country?
As well as the representational issues, small
sample sizes also restrict the depth of data
analysis that can be undertaken. This is
particularly acute analysing sub-classifications
[say by demographics]. Each time a sub-
classification is performed, the data becomes
spread more thinly, over more and more
categories, making the statistics meaningless.
A special problem was encountered in the
cohort studies. A decrease in response
between the measurement horizons is always
expected. In the current research, that
decrease was far more acute than anticipated.
Cohort studies must start from a relatively large
sample base.
170
Understanding Recycling Behaviour 2
Overall, three central issues have come out of
the „people-centred‟ way of thinking:
The Decision Support Issue: How to improve
performance evaluation and diagnostics;
The Research Issue: Understanding the
dynamics of behavioural change;
The Operational Issue: Getting the educational
message right.
The research at the University of Paisley under
the Newspaper Industry Environmental
Technology Initiative has actively addressed
the first two issues and is now starting to
address the third.
So what should the next steps be?
Obtaining large, consistent and reliable data
sets is essential to improving our understanding
of recycling behaviours. Currently the data
being collected in the UK tends to be
fragmented, based on many different
methodological approaches and of very mixed
scientific value. There is very little proactive
dissemination of data. What is needed is more
consistency in measurement, more detail in
monitoring and recording, and more open and
widespread dissemination of results [both good
and bad].
12.2.2 Opportunities
The next few years will provide us with an
unprecedented (and perhaps unrepeatable)
opportunity to gain new data on the drivers of
behavioural change. New recycling schemes
are still being rolled out. Awareness,
promotional and educational campaigns are
being intensified. Because of the country‟s
inexperience in mounting successful
campaigns at this point in time all campaigns
have to be viewed as experiments. We cannot
predict their outcomes. However by measuring
their outcomes we can learn. But we do need to
make those measurements scientifically.
„Getting the message right‟ still needs much
further research. That research needs a twin
track approach:

 Comparative evaluations of the behavioural
effects of different campaign messages and
their means of delivery;
 Continuing fundamental studies of
attitudes, attitude change and behavioural
change.
The other key issue is performance monitoring.
Monitoring protocols need to be designed to
measure behaviours as well as audit statistics.
To be most effective, that monitoring needs to
be frequent and at a high spatial resolution.
12.3 The Final Words
The research of the Newspaper Industry
Environmental Technology Initiative has led to
significant advances in understanding
household recycling behaviours, and has
established and developed the framework for
accounting for those behaviours. However, it
has not yet completed our understanding. Key
issues and priorities for future research have
been identified, and they are discussed openly
in this document.
The monograph has been structured as a
technical document. It has aimed to develop
the scientific basis for understanding household
waste management behaviours.
It has aimed to deliver the mission of the
Newspaper Industry Environmental Technology
Initiative and its principal sponsors, Bridgewater
Paper Co. Ltd, Cheshire Recycling, Stora Enso,
and Sun Chemical:
Understanding Recycling Behaviour 2
Individual weighings of barcoded containers or
containers fitted with scannable microchips
would provide much of the data needed for
behavioural monitoring. It would also add
significantly to the scale and quantity of
research data. Of course, coarser expedients
could still be adopted.
Analysis, evaluation and diagnostics of the
monitoring data could be automated through
computer-based decision support tools, or
expert systems.
 To gain new or improved data sets and to
place these data sets in the public domain,
and to disseminate the findings widely;
 To research, develop and exploit new
methods, processes and procedures to
promote continual environmental
improvement and enhance environmental
decision making;
 To carry out objective investigations, of
high scientific integrity.
The monograph started out by posing the
question: “Everyone is different, but how
significant are those differences?”
The answer is probably that: “Most of us are
quite similar, but some of us are quite different”.
Our goal now is to understand the different
ones.
171

References
Aberg, H., Dahlman, S., Shanahan, H. and
Saljo, R. (1996) Towards sound environmental
behaviour: Exploring household participation in
waste management. Journal of Consumer
Policy, 19, 45-67
Aspinwall & Co. Ltd. (1991). The Analysis and
Prediction of Household Waste Arisings. Report
No. CWM/037/91, Department of the
Environment: London.
Azjen, I. (1985), From Intentions to Actions: A
Theory of Planned Behaviour. In Action
Control: From Cognition to Behaviour, (Kuhl
and Beckmann, eds.), Springer Verlag: New
York.
Azjen, I. and Fishbein, M. (1977). Attitude-
Behavior Relations: A Theoretical Analysis and
Review of Empirical Research. Psychological
Bulletin, 84, 888-919.
Bagozzi, R.P., Baumgartner, H., Youjai, Y.
(1992). State versus Action Orientation and the
Theory of Reasoned Action: An Application to
Coupon Usage. Journal of Consumer
Research, 18, 505-508.
Barker, K., Fong, L., Grossman, S., Quinn,
C.and Reid, R. (1994). Comparison of Self-
reported Recycling Attitudes and Behaviours
with Actual Behavior. Psychological Reports,
75, 571-577.
Barr, S., Gilg, A.W. and Ford, N.J. (2001)
Differences between household waste
reduction, reuse and recycling behaviour: a
study of reported behaviours, intentions and
explanatory variables. Environment and Waste
Management, 4(2), 69-82
Barton, J., Perrin, D. and Barton, J. (2001). The
Millennium Recycling Scheme. University of
Leeds: Leeds
Berger, I. E. (1997). The demographics of
recycling and the structure of environmental
behaviour. Environment and Behaviour, 29,
515 - 526
Boldero, J. (1995). The Prediction of Household
Recycling of Newspapers: The Role of
Attitudes, Intentions and Situational Factors.
Journal of Applied Social Psychology, 25(5),
440-462.
173
Understanding Recycling Behaviour 2
Burn, S. (1991). Social Psychology and the
Stimulation of Recycling Behaviours: The Block
Leader Approach. Journal of Applied Social
Psychology, 21, 611-629.
CIPFA, (2001). Waste Collection and Disposal
Statistics, 1999-00 Actuals. CIPFA: London.
Clamp, F. (2000). Kerbside Collection of Paper
for Recycling. Hertsmere Borough Council.
Cook, S.W. and Berrenberg, J.L. (1981).
Approaches to Encouraging Conservation
Behavior: A Review and Conceptual
Framework. Journal of Social Issues, 37, 73-
107.
Corral-Verdugo, V. (1997). Dual realities of
conservation behaviour: Self reports vs
observations of re-use and recycling behaviour.
Journal of Environmental Psychology, 17, 135-
145.
Corral-Verdugo, V. and Figueredo, J.A. (1999).
Convergent and divergent validity of three
measures of conservation behaviour.
Environment and Behaviour, 31 (6), 805-820.
Cote, J.A. (1984). Use of household refuse
analysis to measure usual and period specific
food consumption. American Behavioural
Scientist, 28, 129-138.
Dahab, D.J., Gentry, J.W. and Su, W. (1995).
New Ways to Reach Non-recyclers: An
Extension of the Model of Reasoned Action to
Recycling Behaviours. Advances in Consumer
Research, 22, 251-256.
Dawes, J.C. (1926). Report into an
Investigation into the Public Cleansing Service
in the County of London. HMSO: London.
De Young, R., Boerschig, S., Carney, S.,
Dillenbeck, A., Elster, M., Horst, S., Kleiner, B.
and Thomson, B. (1995). Recycling in Multi-
Family Dwellings: Increasing Participation and
Decreasing Contamination. Population and
Environment, 1995, 16(3), 253-267.
Ellen, P.S. (1994). Do we Know What we Need
to Know? Objective and Subjective Knowledge
Effects on Pro-Ecological Behaviors. Journal of
Business Research, 30, 43-52.

Everett, J.W., Jacobs, T.L. and Pierce, J.J.
(1991). Recycling Promotional Strategies:
Statistical and Fuzzy Set Comparisons.
Journal of Urban Planning and Development,
117(4), 154-167.
Feiock, R.C. and Kalan, L.G. (2001). Assessing
the performance of solid waste recycling
programs over time. American Review of Public
Administration, 31(1), 22-32.
Folz, D.H. (1991). Recycling program design,
management and participation: A national
survey of municipal experience. Public
Administration Review, 51(3), 222-231.
Folz, D.H. and Hazlett, J.M. (1991). Public
Participation and Recycling Performance:
Explaining Program Success. Public
Administration Review, 51(6), 526-532.
Folz, D.H. (1999). Municipal recycling
performance: A public sector success story.
Public Administration Review, 59(4), 336-345.
Gamba, R.J. and Oskamp, S. (1994). Factors
influencing community residents‟ participation
in commingled curbside recycling programs.
Environment and Behaviour, 26 (5), 587-612.
Geller, E.S. (1981). Evaluating Energy
Consumption Programs. Is Verbal Report
Enough. Journal of Consumer Research, 8,
331-335.
Goldenhar, L.M. and Connell, C.M. (1992-93).
Understanding and Predicting Recycling
Behavior: An Application of the Theory of
Reasoned Action. Journal of Environmental
Systems, 22(1), 9-103.
Granzin, K.L. and Olsen, J.E. (1991).
Characterising Participants in Activities
Protecting the Environment: A Focus on
Donating, Recycling and Conservation
Behaviours. Journal of Public Policy and
Marketing, 10(2), 1-27.
Guagnano, G.A., Stern, P.C. and Dietz, T.
(1995). Influences on Attitude-Behavior
Relationships. A Natural Experiment with
Curbside Recycling. Environment and
Behavior, 27(5), 699-718.
Hallin, P.O. (1995). Environmental Concern
and Environmental Behavior in Foley, A Small
Town in Minnesota. Environment and Behavior,
27(4), 558-578.
Understanding Recycling Behaviour 2
Higginson (1965). The Analysis of Domestic
Waste. The Institute of Wastes Management:
Northampton.
Hogg, D. and Mansell D. http://www.foe.co.uk/
resource/reports/maximising_recycling_rates
_report.pdf
Hopper, J.R., and Nielsen, J.M. (1991).
Recycling as Altruistic Behavior: Normative and
Behavioral Strategies to Expand Participation in
a Community Recycling Program. Environment
and Behavior, 23(2), 195-220.
Howenstine, E. (1993). Market Segmentation
for Recycling. Environment and Behavior,
25(1), 86-102.
Jackson, A.L., Olsen, J.E., Granzin, K.L. and
Burns, A.C. (1993). An Investigation of the
Determinants of Recycling Consumer Behavior.
Advances in Consumer Behavior, 20, 481-487.
Jones, R.E. (1989-90). Understanding Paper
Recycling in an Institutionally Supportive
Setting: An Application of the Theory of
Reasoned Action. Journal of Environmental
Systems, 19(4), 307-321.
Jones, A., Nesaratnam, S. and Porteous, A.
(1998a). Regional Variations in Household
Waste arisings in the UK Environmental and
Waste Management, 1(2), 97-106.
Jones, A., Nesaratnam, S. and Porteous, A.
(1998b). Enumerating the Sources of Variation
in Household Waste Arisings in the UK.
Environmental and Waste Management,
1998b, 1(4), 235-241.
Katzev R.D. and Pardini A.U. (1987-88). The
Comparative Effectiveness of Reward and
Commitment Approaches in Motivating
Community Recycling. Journal of
Environmental Systems, 17(2), 93-113.
Katzev, R.D. and Mishima, H. (1992). The Use
of Posted Feedback to Promote Recycling.
Psychological Reports, 71, 259-264.
Kilner, S.M. (1992). Participating in a Recycling
Program: Does it Change Attitudes towards
Solid Waste. Proc. 15th Annual Madison Waste
Conference, 97-108.
Lober, D.J. (1996). Municipal Solid Waste
Policy and Public Participation in Household
Source Reduction. Waste Management and
Research, 14(2), 125-143.
174

McGuire, R.H. (1984). Recycling - Great
expectations and garbage outcomes. American
Behavioural Scientist, 28, 93-114.
MEL (1993). Trends in Household Waste
Arisings and the Analysis of Recycled
Materials. Interim Report to the Department of
the Environment. MEL Research: Birmingham.
MEL (1996). Analysing Household Waste: A
New Method for the Analysis and Estimation of
Household Waste Arisings. MEL Research:
Birmingham.
MEL (1999). Project Integra: Waste analysis
and questionnaire survey results, Unpublished
report for Project Integra.
Needleman, L.D. and Geller, E.S. (1992).
Competing Interventions to Motivate Work-site
Collection of Home-generated Recyclables.
American Journal of Community Psychology,
20, 775-785.
Nielsen, J.M. and Ellington, B. (1983). Social
Processes and Resource Conservation: A
Case Study in Low Technology Recycling.
Environmental Psychology: Directions and
Perspectives. (Feiner and Geller, eds.),
Praeger: New York.
Noehammer, H.C. and Byer, P.H. (1997). Effect
of design variables on participation in
residential curbside recycling programs. Waste
Management & Research, 15, 407-427.
Oskamp, S., Harrington, M., Edwards, T.,
Sherwood, P.L., Okuda, S.M. and Swanson,
D.L. (1991). Factors Influencing Household
Recycling Behavior. Environment and
Behavior, 23, 494-519.
Oskamp, S. (1995). Resource Conservation
and Recycling: Behavior and Policy. Journal of
Social Issues, 51(4), 157-177.
Pallak, M., Cook, D. and Sullivan, J. (1980).
Commitment and Energy Conservation. In
Applied Social Psychology, Annual 1 (L.
Bickman, ed.), 235-253.
Parfitt, J. and Flowerdew, R.(1997).
Methodological Problems in the Generation of
Household Waste Statistics. Applied
Geography, 17(3), 231-244.
Parfitt, J.P., Lovett, A.A., Sunnenberg, G.
(2001). A classification of local authority waste
collection and recycling strategies in England
175
Understanding Recycling Behaviour 2
and Wales. Resources, Conservation &
Recycling, 32, 239-257.
Perrin, D. and Barton, J.R. (2000). If only
households recycled what and when they said
they would. ISWA World Congress 2000, 255-
263.
Perrin, D. and Barton, J. (2001). Issues
associated with transforming household
attitudes and opinions into materials recovery:
a review of two kerbside recycling schemes.
Resources, Conservation & Recycling, 33, 61-
74.
Pieters, R.G.M. (1989), Attitudes and Behavior
in a Source Separation Program. A Garbology
Approach. Eurobon: Delft, The Netherlands.
Pieters, R.G.M. (1991). Changing Garbage
Disposal Patterns of Consumers: Motivation,
Ability and Performance. Journal of Public
Policy and Marketing, 10(2), 59-76.
Rathje, W.L. (1984). Where‟s the Beef?
American Behavioural Scientist, 28, 71-79.
Rathje, W.L. (1989). The Three Faces of
Garbage – Measurements, Perceptions and
Behaviors. Journal of Management and
Technology, 17, 61-65.
Read, A.D. (1998). The recycling roadshow.
Communicating local government services to
residents. Environmental & Waste
Management, 1(2), 113-124.
Read, A.D. (1999). “A weekly doorstep
recycling collection, I had no idea we could!”
Overcoming local barriers to participation.
Resources, Conservation and Recycling, 26,
217-249.
Ronis, D.L., Yates, J.F., and Kirscht, J.P.
(1989). Attitudes, Decisions, and Habits as
Determinants of Repeated Behaviors. In
Attitude, Structure and Function (Pratkanis,
A.R., Breckler, S.J., and Greenwald, A.J. –
eds.). Erlbaum: Hillside, NJ, 213-239.
Rufford N.M. (1984) The Analysis and
Prediction of the Quantity and Composition of
Household Refuse. Unpublished PhD Thesis,
The University of Aston: UK.
Salimando, J. (1987). Camden County Sets the
Recycling Pace. Waste Age 18(7), 48-53.
Schultz, P.W., Oskamp, S., Mainieri, T. (1995).
Who Recycles and When? A Review of

Personal and Situational Factors. Journal of
Environmental Psychology, 15, 105-121.
Schwartz, S.H. (1977). Normative Influences on
Altruism. Advances in Experimental Social
Psychology, (Berkowitz, L. ed.), Vol 10,
Academic Press: New York.
Seligman, C., Becker, L.J. and Darley, J.M.
(1981). Encouraging Residential Energy
Conservation though Feedback. Advances in
Experimental Psychology, Vol 3, Energy:
Psychological Perspectives. (Baum and Singer,
eds.), Erlbaum: Hillsdale, NJ.
SEPA. (2002). Waste Data Digest 2002. SEPA:
Stirling, 2002.
Shrum, L.J., Lowrey, T.M. and McCarty, J.A.
(1995). Applying Social and Traditional
Marketing Principles to the Reduction of
Household Waste. American Behavioural
Scientist, 38(4), 646-657.
Siegfried, W.D., Tedeschi, R.G. and Cann, A.
(1982). The Generalizability of attitudinal
correlates of pro-environmental behaviour.
Journal of Social Psychology, 118, 287 - 288
Smith, D.N., Harrison, L.M. and Simmons, A.J.
(1999). A survey of schemes in the United
Kingdom collecting plastic bottles for recycling.
Resources, Conservation & Recycling, 25, 17-
34.
Spaccarelli, S., Zolik, E. and Jason, L.A. (1989-
90). Effects of Verbal Prompting and Block
Characteristics on Participation in Curbside
Newspaper Recycling. Journal of
Environmental Systems, 19(1), 45-57.
Strategy Unit. (2002). Waste Not Want Not,
Cabinet Office: London.
Taylor, S. and Todd, P. (1995). An Integrated
Model of Waste Management Behavior: A Test
of Household Recycling and Composting
Intentions. Environment and Behavior, 27(5),
603-30.
Thomas, C. (2001). Public understanding and
its effect on recycling performance in
Hampshire and Milton Keynes. Resources,
Conservation and Recycling, 32, 259-274.
Tracy, AP and Oskamp, S. (1983-4)
Relationships among ecologically responsible
behaviours. Journal of Environmental Systems,
13, 115-126
Understanding Recycling Behaviour 2
Tucker, P, (1997-98). Accounting for Temporal
and Spatial Variabilities in Household Recycling
Schemes: A Simulation Approach. Journal of
Environmental Systems, 26(4), 1997-98, 365-
390.
Tucker, P., Lamont, J., Murney, G. and Smith,
D. (1998). Material Capture Rates in Household
Waste Recycling Schemes. Environmental and
Waste Management, 1(3), 169-181.
Tucker, P. (1999). A Survey of Attitudes and
Barriers to Kerbside Recycling. Environmental
and Waste Management, 2(1), 55-63.
Tucker P. (2001). Understanding Recycling
Behaviour, University of Paisley: Paisley,
Scotland. ISBN 1-903978-01-7.
Tucker, P., and Speirs, D. (2001).
Understanding Home Composting Behaviour.
University of Paisley: Paisley, Scotland. ISBN
1-903978-07-6.
Vencatasawmy, C.P., Ohman and M.,
Brannstrom, T. (2000) A survey of recycling
behaviour in households in Kiruna, Sweden.
Waste Management and Research, 18, 545-
556
Vining, J., and Ebreo, A. (1989). A. Public
Response to Recycling Education. Society and
Natural Resources, 2, 23-36.
Vining, J. and Ebreo, A. (1990). What Makes a
Recycler? A Comparison of Recyclers and
Non-recyclers. Environment and Behavior,
22(1), 55-73.
Vining, J. and Ebreo, A. (1992). Predicting
Recycling Behavior from Global and Specific
Environmental Attitudes and Changes in
Recycling Opportunities. Journal of Applied
Social Psychology, 22(20), 1580-1607.
Vining, J., Linn, L. and Burdge, R.J. (1992).
Why Recycle? A Comparison of Recycling
Motivations in Four Communities.
Environmental Management, 16, 785-797.
Warriner, K.G., McDougall, G.H.G. and
Claxton, J.D. (1984). Any data or none at all?
Living with inaccuracies in self-reports of
residential energy consumption. Environment
and Behaviour, 16 (4), 503-526.
Waste Aware Scotland (2002a). Public
Attitudes to Reduce Reuse Recycle in
Scotland. Waste Aware Scotland: Stirling.
176

Waste Aware Scotland (2002b). Area Summary
Reports, http://www.wascot.org.uk
Waste Research Ltd., AEA Technology. (2001).
Assessment of kerbside collection schemes for
dry recyclables. Resource Recovery Forum.
Werner, C.M., Turner, J., Shipman, K.,
Twitchell, S., Dickson, B.R., Brushke, G.V. and
Von Bismarck, W.B. (1995), Commitment,
Behavior and Attitude Change: An Analysis of
Voluntary Recycling. Journal of Environmental
Psychology, 15, 197-208.
Werner, C.M. and Makela, E. (1998).
Motivations and Behaviors that Support
Recycling. Journal of Experimental Psychology,
18, 373-386.
Williams, I.D. and Kelly, J. (2003). Green
Waste Collection and the Public‟s Recycling
Behaviour in the Borough of Wyre, England.
Resources, Conservation and Recycling, 38(2),
139-159.
177
Understanding Recycling Behaviour 2
Woodard, R., Harder, M.K., Bench, M. and
Philip, M. (2001). Evaluating the performance
of a fortnightly collection of household waste
separated into compostables, recyclates and
refuse in the south of England. Resources,
Conservation and Recycling, 31, 265-284.
WRAP (2002). Kerbside Collection of Glass,
WRAP: Banbury, UK.
WSL and Aspinwall (1994). National Household
Waste Analysis Project - Phase 2 Volume 1:
Report on Composition and Weight Data.
Report CWM 082/94. Department of the
Environment: London.
Yule, F., Knussen, C. and MacKenzie, J. (2001)
Environmental Attitudes and Household
Recycling Behaviour. Report to SNIFFER.
Foundation for Water Research: Marlow,
Bucks.
 Understanding Recycling Behaviour 2

Appendix A
Table A1 Acorn Market Research Indicators

Group Description Sub- Description

group

A Thriving 1 Wealthy Suburbs, Large Detached Houses

2 Villages with Wealthy Commuters
3 Mature Affluent Home Owning Areas
4 Affluent Suburbs, Older Families
5 Mature, Well-Off Suburbs
6 Agricultural Villages, Home Based Workers
7 Holiday Retreats, Older People, Home Based Workers
8 Home Owning Areas, Well-Off Older Residents
9 Private Flats, Elderly People
B Expanding 10 Affluent Working Families with Mortgages
11 Affluent Working Couples with Mortgages, New Homes
12 Transient Workforces, Living at their Place of Work
13 Home Owning Family Areas
14 Home Owning Family Areas, Older Children
15 Families with Mortgages, Younger Children
C Rising 16 Well-Off Town & City Areas
17 Flats & Mortgages, Singles & Young Working Couples
18 Furnished Flats & Bedsits, Younger Single People
19 Apartments, Young Professional Singles & Couples
20 Gentrified Multi-Ethnic Areas
21 Prosperous Enclaves, Highly Qualified Executives
22 Academic Centres, Students & Young Professionals
23 Affluent City Centre Areas, Tenements & Flats
24 Partially Gentrified Multi-Ethnic Areas
25 Converted Flats & Bedsits, Single People
D Settling 26 Mature Established Home Owning Areas
27 Rural Areas, Mixed Occupations
28 Established Home Owning Areas
29 Home Owning Areas, Council Tenants, Retired People
30 Established Home Owning Areas, Skilled Workers
31 Home Owners in Older Properties, Younger Workers
32 Home Owning Areas with Skilled Workers
E Aspiring 33 Council Areas, Some New Home Owners
34 Mature Home Owning Areas, Skilled Workers
35 Low Rise Estates, Older Workers, New Home Owners
36 Home Owning Multi-Ethnic Areas, Young Families
37 Multi-Occupied Town Centres, Mixed Occupations
38 Multi-Ethnic Areas, White Collar Workers
F Striving 39 Home Owners, Small Council Flats, Single Pensioners
40 Council Areas, Older People, Health Problems
41 Better-Off Council Areas, New Home Owners
42 Council Areas, Young Families, Some New Home Owners
43 Council Areas, Young Families, Many Lone Parents
44 Multi-Occupied Terraces, Multi-Ethnic Areas
45 Low Rise Council Housing, Less Well-Off Families
46 Council Areas, Residents with Health Problems
47 Estates with High Unemployment
48 Council Flats, Elderly People, Health Problems
F Striving 49 Council Flats, Very High Unemployment, Singles
50 Council Areas, High Unemployment, Lone Parents
51 Council Flats, Greatest Hardship, Many Lone Parents

179
 Understanding Recycling Behaviour 2

52 Multi-Ethnic, Large Families, Overcrowding

53 Multi-Ethnic, Severe Unemployment, Lone Parents
54 Multi-Ethnic, High Unemployment, Overcrowding

Table A2 – Social Classes

Social Class Example Occupations

1 Professional
2 Intermediate
3n Non-manual Skilled
3m Manual Skilled
4 Semi-skilled
5 Unskilled
Source: Registrar-General
Accountant, doctor, clergyman, university teacher
Pilot, farmer, manager, police officer, teacher, manager
Clerical worker, sales rep., shop assistant, secretary
Butcher, bus driver, electrician, miner
Bus conductor, bar person, postal worker, packer
Labourer, office cleaner, window cleaner

180