You are on page 1of 12

Finding and Counting Difficult to Reach

Population groups; and some


Observations on ‘Household’ Surveys
for
Understanding Households
Edinburgh, 13th March 2018i
Roy Carr-Hill
UCL Institute of Education and
Independent Consultant
Introduction
• Used administrative data extensively in the 1990s for
designing resource allocation formulae for care services
and health services in 4 nations
• Since 2000, have been preoccupied by the undercount of
the poorest in developing countries
• This will not be a perfectly constructed talk: (a) title is
changed because here are very few strictly non-
household populations (however defined) in the UK
• (b) difficult to reach population groups are, and have
been pervasive forever. But there are several non-
compatible categories.
• ©The term ‘counted’ is used loosely; most of those in
the institutional group considered below can be counted
but we have very little other information about them
    Home- Care  Mili- Pri- Refugees  Noma- Urban   TOTAL
less/  homes/  tary sons and Illegal  dic/  Slums
chil- hospi- Immi- Pastor-
dren tals grants alist 
(Gypsies)

Gobal  Min. n.e. 25.0 92.0 9.8 36.5 21.8 82.8 267.9

(millions) Max n.e. 35.0 92.0 9.8 73.0 43.5 175.6 438.9
.
Status of  B B A C B C
estimate

UK  83.0 678.0 108 90.0 238 150.0 Ne. 1,337


(thousands)

Status of  A A A A C B A
estimate
Poor or not in the UK?
• Military: not in UK or other developed countries
• Homeless / street children almost by definition
• Those in or just released from 446 prison (90,000) have
low or very low incomes, as have most (say 75 per
cent) of the 238,000 unofficial refugees (Doyle, 2008),
and an estimated 80 per cent of the 150,000 gypsies
and travellers (Cemlyn et al., 2009), making a total of
398,000.
• The income situation of the largest groups – those in
hospitals or care homes (678,000) – is less clear
• Care homes: depends on funding regulations
• Hospitals: morbidity confounded with poverty but, in
developed countries more with age, so levels of
Counting Institutional Populations
• Those in institutions (care homes/ hospitals, military
or prisons) can be head-counted but apart from
relevant institutional status, gender and perhaps
age, we have no other information about them.
• In principle, this can be relatively easily remedied by
making agreements with the appropriate
authorities; but actually carrying out the census may
be difficult, especially in hospitals where proxy
respondents may have to be used
• Another problem in many cases is that there is
potential double counting with their (temporarily)
broken household
Counting Homeless/ street children
• These are the only strictly non-household
groups (in terms of residence)
• On census night, ONS do carry out a count in
major cities
• There have also been several experiments
(including by ONS?) with the capture/
recapture method; but.
• I don’t know how to validate their results other
than perhaps by interviewing a sample of ‘The
Big Issue’ sellers; and e.g. Salvation Army
personnel
Refugees and Illegal Immigrants
• Probably the hardest group to track and understand
• Even official refugees in the UK would be more
reluctant than most to respond to surveys, and, they will
probably be poor.
• It almost goes without saying that illegal immigrants or
their ‘owners’ will be very reluctant to respond to
interviews. One could argue that they are not relevant
for social policy but, if they remain and have children,
they will contribute to the social policy relevant
population
• Reference Tyler, P. (2016) Radical Statistics Newsletter
115
Counting gypsy & traveller populations (1)
• Many Romany gypsies are in caravans or semi-formal
camps and so can be counted and interviewed; but
many are not and travelers are not and it will be
difficult to count let alone interview them. The
problem here is more the attribution of ethnic identity
• “policy-makers have failed to take into account the
intimidating and discouraging effects of current or
threatened policy on responses to questions.” For
example, questions about ‘Main reason for travelling’
allows only one response when many gypsies travel for
several reasons; also very few East European Roma
accept the term ‘gypsy’ (a direct translation of ‘Cigny’)
• Reference Acton et al (2016) Radical Statistics
Newsletter Issue 114
Counting those in Urban Slums
• We have few urban slums left in the UK but the
high rise blocks where lifts are not always
working will certainly include some unsurveyed
and uncensored people; our interviews with
interviewers demonstrate that they often
complete the questionnaire themselves;
• Even where the lifts are working the Grenfell
Towers experience shows that there is not
always a secure count of the numbers
Conclusion
• Globally, it is an international scandal that no
serious attempt has been made to count or
adjust for these missing populations; instead UN
statisticians etc have repeated, with no evidence,
that surveys provide good estimates
• Nationally, there was an ONS paper by Joloza in
2009 but the issues were substantially ignored by
Atkinson et al (2002) in discussing and proposing
measurements of social inclusion in Europe and
by Hills (2010) in analysing inequality in the UK
• The problem has to be taken much more
seriously.
Observations on Household Surveys
• Administrative data has a clear focus on activities
(e.g. alcohol consumed), dwellings (e.g. council
tax,), households (e.g. income tax, marriage and
divorce) or individuals (e.g. driving license, voting)
• But so called household surveys, whilst purporting
to cover all those issues do not. Their sampling
frame is nearly always dwellings; but the only
surveys which focus on dwellings are housing
surveys; and although comprehensive surveys list
all members of the household members and their
relationship to a sometimes fictive household
head, they rarely establish a relationship-based
household.
Observations on Household Surveys (2)
• In fact, they are nearly always focussing on collecting
information (activities, behaviours, practices; income;
or attitudes)
• Comprehensive surveys will weight their achieved
sample to the last Census age-sex profile.
• For activities and practices, we should take into
account household relationships as activities and
practices are often shared (e.g. both drinking and
smoking are strongly influenced by how the partner
behaves)
• For attitudes, shouldn’t we also weight by the
probability of someone of that age and sex being in a
dwelling of that size and composition?

You might also like