Arithmetical Difficulties: Developmental and Instructional Perspectives

EDUCATIONAL AND CHILD PSYCHOLOGY VOLUME 24, NUMBER 2

Guest Editors Penelope Munn Rea Reason

Contents
3 5 About the contributors Arithmetical difficulties: Developmental and instructional perspectives (extended editorial) Penny Munn & Rea Reason Strategy flexibility in children with low achievement in mathematics Lieven Verschaffel, Joke Torbeyns, Bert DeSmedt, Koen Luwel & Wim Van Doreen Early markers for arithmetic difficulties Pieter Stock, Annemie Desoete & Herbert Roeyers Assessing and teaching children who have difficulty learning arithmetic Ann Gervasoni & Peter Sullivan Assessing pupil knowledge of the sequential structure of numbers David Ellemor-Collins & Robert Wright What can intervention tell us about the development of arithmetic? Ann Dowker Investigating variability in classroom performance amongst children exhibiting difficulties with early arithmetic Jenny Houssart Language-based retrieval difficulties in arithmetic: A single case intervention study comparing two children with specific language impairment Tuire Koponen, Tuija Aro, Pekka Rsnen & Timo Ahonen Achieving new heights in Cumbria: Raising standards in early numeracy through mathematics recovery Ruth Willey, Amanda Holliday & Jim Martland Educational psychologists assessment of childrens arithmetic skills Susie Mackenzie Linking childrens home and school mathematics Martin Hughes, Pamela Greenhough, Wan Ching Yee, Jane Andrews, Jan Winter & Leida Salway Supporting children with gaps in their mathematical understanding: The impact of the national numeracy strategy on children who find arithmetic difficult Jean Gross

16 28 40 54 64 83

98

108

119 137

146

Educational & Child Psychology Vol 24 No 2

About the contributors

Timo Ahonen is a Professor of Developmental Psychology in the Department of Psychology, University of Jyvskyl, Finland. Jane Andrews is a Senior Lecturer in Education, University of the West of England, UK. Tuija Aro is a Senior Researcher in the Niilo Mki Institute, Jyvskyl, Finland. Bert De Smedt is a Postdoctoral Fellow of the Department of Educational Sciences of the Katholieke Universiteit Leuven, Belgium. Annemie Desoete is a Professor in the Department of Clinical Psychology, University of Ghent, Belgium and at the Artevelde Hogeschool, Belgium. Ann Dowker is a University Research Lecturer in the Department of Experimental Psychology, University of Oxford, UK. David Ellemor-Collins is a Researcher in the School of Education, Southern Cross University, Australia. Ann Gervasoni is a Senior Lecturer in the Faculty of Education, Australian Catholic University, Ballarat Campus, Australia. Pamela Greenhough is a Research Fellow in the Graduate School of Education, University of Bristol, UK. Jean Gross is the Director of Every Child a Reader, KPMG Foundation, UK. Amanda Holliday is a Numeracy Consultant with Cumbria County Council Childrens Services, UK. Jenny Houssart is a Senior Lecturer in the Institute of Education, University of London, UK. Martin Hughes is a Professor in the Graduate School of Education, University of Bristol, UK. Tuire Koponen is a Researcher in the Niilo Mki Institute, Jyvskyl, Finland. Koen Luwel is a Postdoctoral Fellow of the Fund for Scientific Research, Flanders (Belgium) associated to the Department of Educational Sciences of the Katholieke Universiteit Leuven, Belgium. Susie Mackenzie is an Educational Psychologist with Education Leeds, UK. Jim Martland is a member of the International Board of Mathematics Recovery and Director in the Mathematics Recovery Programme, UK. Penny Munn is a Reader in Education at Strathclyde University, UK. Pekka Rsnen is a Senior Researcher in the Niilo Mki Institute, Jyvskyl, Finland. Rea Reason is a Co-Director in the Doctoral programme for Practising Educational Psychologists, School of Education, University of Manchester, UK.

Educational & Child Psychology Vol 24 No 2

About the contributors Herbert Roeyers is a Professor of Developmental Disorders in the Department of Clinical Psychology, University of Ghent Belgium. Leida Salway was a Teacher-Researcher in the Graduate School of Education, University of Bristol, UK. Pieter Stock is a Researcher in the Department of Clinical Psychology, University of Ghent, Belgium. Peter Sullivan is a Professor of Science, Mathematics and Technology Education, Monash University, Clayton, Australia. Joke Torbeyns is a Postdoctoral Fellow of the Fund for Scientific Research, Flanders (Belgium) associated to the Department of Educational Sciences of the Katholieke Universiteit Leuven, Belgium. Wim van Dooren is a Postdoctoral Fellow of the Fund for Scientific Research, Flanders (Belgium) associated to the Department of Educational Sciences of the Katholieke Universiteit Leuven, Belgium. Lieven Verschaffel is a Professor of Educational Sciences in the Centre for Instructional Psychology and Technology, Katholieke Universiteit Leuven, Belgium. Wang Ching Yee is a Research Fellow in the Graduate School of Education, University of Bristol, UK. Ruth Willey is a Specialist Educational Psychologist with Cumbria County Council Childrens Services, UK. Jan Winter is a Senior Lecturer in the Graduate School of Education, University of Bristol, UK. Robert Wright is a Professor of Education in the School of Education, Southern Cross University, Australia.

Educational & Child Psychology Vol 24 No 2

Extended editorial

Arithmetical difficulties: Developmental and instructional perspectives

Penny Munn & Rea Reason
Abstract
This introductory paper sets the scene for the special issue by outlining the research that informs our understanding of the way children learn arithmetic, and the difficulties that they may encounter. Overviews of the papers contributed for this special issue reveal converging views on childrens arithmetical ability, the research uses of interventions, the influence of the social context, mathematics assessment and curriculum design, and the most effective forms of interventions for children who are struggling to learn. We draw several conclusions from our overview of the research and the contributions to this issue. Our main conclusions are that arithmetical difficulties are best understood within the framework of mathematical development, and that discussion of dyscalculia should take place as part of this broader context. Secondary conclusions are that educational psychologists need an understanding of current theories in mathematics education and that they can usefully contribute to a growing field in the individual and social psychologies of mathematics education.

DUCATIONAL PSYCHOLOGISTS work in the area between psychological theory and its applications in a range of contexts. They are responsible both for its direct application and for common understanding of psychology as it relates to childrens learning and development (Cameron, 2006; Farrell et al., 2006). Because of this it is both timely and relevant that we should present this collection of papers on arithmetical difficulties reflecting current research in the psychology of mathematics education. There are a range of reasons for the growing interest in arithmetical difficulties by educational psychologists, all related to changing social and educational attitudes towards numeracy problems. We can identify these changes as advances in psychological knowledge and research, the arrival in education of the term dyscalculia, changes in education due to the implementation of the National Numeracy Strategy and increased demands of employers for numeracy.

First and foremost, there is a critical mass of psychological and educational research that can inform knowledge of the development of numeracy. As outlined in this article and illustrated by the international contributions to this special issue, the combination of two research traditions cognitive-developmental work and the psychology of maths education provides the basis for examining childrens concepts and strategies from a constructivist standpoint. The migration of the term dyscalculia from neuropsychology to education provides a second reason for educational psychologists to be interested. Indeed, the initial impetus for the present issue was the publication of Primary National Strategy information that addressed the concept of dyscalculia (DfES, 2001). A comprehensive DfES commissioned report, entitled What works for children with mathematical difficulties (Dowker, 2004), argued that arithmetical ability was not unitary and there were vast individual differences. Dyscalculia major difficulties 5

Educational & Child Psychology Vol 24 No 2 The British Psychological Society 2007

Penny Munn & Rea Reason with basic number concepts was also mentioned in this report as applicable to a small minority of learners. Nevertheless, as in the present journal issue, the main thrust of the report was on understanding of the range of areas involved in childrens difficulties with number. The National Numeracy Strategy in England and Wales has had considerable impact on the way number is taught in primary schools, and on teachers perceptions and expectations of young childrens numeracy learning (Millet, Brown & Askew 2004). Although there is argument in some circles about its effect on mathematics education (Brown, 2005; see also Hughes et al. in this issue) there is no doubt that the NNS emphasis on computational aspects of arithmetic has had an effect on teachers and children, and has in some respects increased childrens abilities. It has also brought into the spotlight those children who have problems with numeracy and whose difficulties need to be understood and addressed (see the contribution of Gross in this issue). Education systems around the world have been changing in the direction of increasing their emphasis on numeracy, but they have not been doing this in a vacuum; they have been responding to a co-ordinated international demand for numeracy skills (see, for instance, the history page of the International Statistical Literacy Project 2007). Literacy problems have long been known to disadvantage adults in the workplace. Now we also know that adults with numeracy problems are seriously disadvantaged in the workplace (Bynner & Parsons, 1997). Maths and numeracy are basic life skills that make a major contribution to the five outcomes of education as set out in Every Child Matters (DfES, 2004): to be safe, healthy, enjoying and achieving, have economic security and make a positive contribution to the community. Similarly, numeracy education makes major contributions to the four major aims of the Scottish Curriculum for Excellence (to produce successful learners, confident individuals, responsible citizens and effective 6 contributors) (SEED Curriculum Review Group, 2004). For all these reasons, educational psychologists will find the papers in this special issue both important and informative. We shall also argue that the time is now ripe for their active professional engagement in the practical issues that the papers raise.

Background in educational and developmental psychology

Educational psychologists already have considerable understanding of research. This will enable them to get involved now with the practical and theoretical issues that have been raised over the past three decades in maths education and the psychology of numeracy. There have been two strands in the history of research in arithmetical difficulties, which have evolved separately from each other. One strand began with Skemps (1973) work on the psychology of maths education, which culminated in the development of an international forum for all involved in maths education research. This strand has been concerned with the link between theory and education practice across all sectors, and has been very productive (Gutierrez & Boero, 2006). The second strand follows on from Piagets work on number (Piaget, 1952). Piaget was not an eductionalist, but his extensive work on number development was given educational interpretations by many, and was influential until it was challenged by Gelman & Gallistels (1978) work on counting that showed that very young children do have some grasp of the principles of number. This apparent overthrow of Piagetian theory was followed by a comprehensive research programme that mapped the understanding of number in babies, young children and animals in ways that were not defined by Piagetian theory (for an overview of this research see Sophian, 1996; for an overview of its theoretical relation to Piagetian and Educational theory see Bryant, 1995). The cognitive strand of research is embedded in cognitive-developmental psychology and is not overly concerned with practices within maths education. Some Educational & Child Psychology Vol 24 No 2

Editorial researchers have worked on developing the applications to maths education implicit in the cognitive-developmental tradition (for example, Resnick, 1989). However, the applications of experimental cognitive research to education are almost a research field in their own right, since the high reliability and low validity involved in experimentation make it hard to apply such research directly to educational practice. Now, the volume of research in both of these traditions of maths research has reached a critical mass where psychological and educational theories can interact, and theory and practice can inform each other. It is at this point that educational psychologists, who have considerable research skills, can usefully begin to get involved in the point at which bridges between theory and practice can form. While most educational psychologists will know the cognitive-developmental work on number, they may not be so familiar with the work on the psychology of maths education. It may help here to summarise the main ways in which the field of maths education research (as it applies to young children) has developed since the early days of Skemp (1973). (i) There has been a decisive shift away from behaviourist accounts and a wholesale adoption in the psychology of maths education community of constructivist theories to explain development and learning in mathematics. (ii)Current research emphasises the simultaneous development of additive and multiplicative structures. (iii)The future direction of research in the psychology of maths education is theoretically driven. It emphasises the modelling and reasoning that underlies arithmetical performance, rather than focusing on the computation procedures themselves. (Mulligan & Vergnaud 2006) These characteristics mean that maths education research can offer a coherent theoretical framework that links to practice. The framework linking maths education studies is one that takes practical utility as its Educational & Child Psychology Vol 24 No 2 first priority; the uniting factor is that the implication of findings for educational practice is much clearer than is the case for findings derived from cognitive developmental research. The converse of this is that the latter type of findings usually have much clearer implications for theoretical developments. This has resulted in the two strands of research having markedly different characteristics. The cognitive developmental research produces information that adds to our knowledge of psychological models of number development, but it is not always clear what relevance these models have to curriculum development, teacher education, or classroom practice. The maths education research, by contrast, produces information that adds to our understanding of curriculum development, teacher education and classroom practice, but does not always contribute directly to a psychological model of childrens number development. We now know a great deal about arithmetical development in childhood, but this knowledge needs to be integrated into a framework that is concerned with extending both psychological models and educational understanding. One thing that is clear from both educational and psychological frameworks is that arithmetical development in children is non-linear and multidimensional. Young children draw together threads of quantitative experience and link these with growing verbal number strings as they slowly construct for themselves a rich understanding of number (for examples, see Fuson, 1988). The non-linearity of developments are evidenced by their slow pace and by the way that children construct their mathematical understanding across multiple domains. The complexity of mathematical understanding in childhood contrasts starkly with vernacular accounts of arithmetical development. By vernacular we mean the account that is implicit in the layout of many guides for (non maths specialist) teachers. (see for instance Biggs & Sutton, 1983; Busbridge & Womack, 1991). Such vernacular accounts can describe childrens difficulties (as in she 7

Penny Munn & Rea Reason cannot remember number bonds within ten) but they are not helpful in remedying such difficulties because they are devoid of both theory and explanation. Many inspired maths educators have struggled to upgrade the vernacular account by contributing a discourse of mathematics education to the primary sector (see for instance Clemson & Clemson, 1994; Hughes, Desforges & Mitchell, 2000). Others have directed their attempts at the social organisation of maths teaching across the school (see for instance Atkinson, 1996). The success of such discourse and organisation will be essential if teachers are to replace their intuitive descriptions of childrens arithmetical difficulties with analyses that reflect the complexity of mathematical development and the recent growth in psycho-educational theories. and learning of arithmetic, and how much they still have to contribute. To illustrate this, we have grouped the papers into three thematic groups. The first group of papers illustrates the deep psychological understanding of the complexity of childrens arithmetical ability that is current in the literature. The second group of papers shows how we can use intervention to increase knowledge of arithmetical difficulties. The third group of papers shows how psychology can contribute to our understanding of the social contexts of arithmetical difficulties. Papers illustrating the inherent complexity of childrens arithmetical ability Four of the papers describe the complexity of arithmetical learning in the primary school, and articulate a profound understanding of the ways in which childrens arithmetical ability can be understood. Verschaffel et al., Stock, Desoete and Roeyers; Gervasoni and Sullivan; and Ellemor-Collins and Wright all focus in very different ways on the complexity of childrens arithmetical ability, and on the implications of this for approaches to arithmetical difficulties. Vershaffel et al. discuss the importance of childrens adaptivity, or strategy flexibility, for the full development of their mathematical ability. They argue that such adaptivity might arise from childrens perceptions of classroom demands, and that we should therefore make sure that classrooms make complex and challenging demands from the earliest school years even of children who seem to be low in ability. They warn that such challenge cannot be added in to the curriculum once children have learned the basics. By then it will be too late, and children will already have lost the opportunity to develop adaptivity. Stock, Desoete and Roeyers work is aimed at discovering whether we can find precursors of arithmetical difficulties while children are in preschool. They compare the predictive power of variations in mathematical reasoning with variations in subitising ability. The findings show that it is Educational & Child Psychology Vol 24 No 2

Contributions to this issue

From an initial call for papers we have selected 11 papers that represent excellence in the application of psychological theory to the field of arithmetical difficulty. One very practical aspect of most of these papers is that the concrete examples in the papers provide suggestions for educational psychologists as to the kinds of assessments and interventions they can apply within their daily practice. For example, Ellemor-Collins and Wright give detailed illustrations of their four groups of assessment tasks that one of the editors was able to use immediately in her own assessments and suggestions for intervention. Gervasoni and Sullivan have illustrations within their framework of growth points that include creative use of concrete materials. Ann Dowker outlines the nine components within the Numeracy Recovery Scheme and her appendices include much detail that can be of immediate relevance to practice. Most of the other papers also contain information and appendices with illustrations of the materials being used. The papers that were contributed to this special edition form a wealth of description that shows both how much psychological approaches have achieved in the teaching 8

Editorial childrens number logic (their mathematical reasoning) that gives the better prediction from preschool number knowledge across the transition into school. This result supports the view that there is not an elementary building block from which arithmetical ability is fashioned. Rather, it is complex even at the beginning of education. Gervasoni & Sullivan conclude that there is no single formula for describing children who have difficulty learning arithmetic or for describing the instructional needs of this diverse group of children. One should not assume that, because a child has difficulty with one aspect of number learning, other areas will also be problematic. They argue explicitly that programmes need to include all number domains in tandem (addition, subtraction, multiplication and division) as childrens construction of number knowledge in a specific domain is not dependent on prerequisite knowledge in another domain, but dependent on being able to take advantage of a range of experiences in that domain. Ellemor-Collins & Wright examine the strategies of lower-attaining learners aged eight to ten years. They find that many pupils have not developed the strategy of jumping by tens and may therefore not use sequence-based understanding such as recognising that number sequences 8, 18, 28, 38 are always ten steps apart. The paper recommends that these pupils number learning should include a focus on number word sequences, skip counting and incrementing by tens and locating numbers in the range of 1100 on a number line. The practical relevance of their study supports the theoretical message that complexity in number learning should be respected even when intervening in the upper primary age groups. Papers illustrating how intervention can increase our understanding In this group of papers, we have three contributions that offer incisive analysis of childrens responses to mathematical situaEducational & Child Psychology Vol 24 No 2 tions, conclusions about how we understand developmental progressions, and reasoned programmes of intervention for the lowest achieving children. The thread that unites all these papers is their use of intervention to increase our understanding of arithmetical difficulties by adding to psychological theory. Dowkers account of numeracy recovery gives finely analysed detail of which aspects of her intervention were effective. Her findings support the view that arithmetic is made up of multiple components that are relatively independent of each other. She shows how arithmetical difficulties can be ameliorated and then uses the results of the intervention to analyse for predictors of improvement among the multiple components. She shows how her results add to our understanding of arithmetical difficulties. Houssart uses her position as researcher to reinterpret an aspect of behaviour that is troubling to teachers and often ignored by theoreticians. Her account of low ability childrens daily struggle with number highlights the dayto-day variability in what they appear able to do. Houssart comments that this variability is seen as a problem in teaching, since it challenges basic teaching principles. However, she shows how this variabilty can be seen as evidence of complexity in learning, and brings this neglected aspect of classroom learning into theoretical perspective. The contribution of Koponen et al. has a more clinical flavour in that they compare the performances of two children considered to have specific language impairment. However, what makes this contribution particularly relevant in this issue is that the focus is not on testing but on the response of the children to intervention over a period of some two months. The authors use this response to elaborate theories of the relation between language and arithmetic development. Willey, Holliday & Martlands account of Maths Recovery in Cumbria documents the effects of this programme on childrens conceptual development. They also to chart the effect of the intervention on teaching and 9

Penny Munn & Rea Reason support staff, concluding that because it inspires the teachers to work from principles rather than prescription, the programme has an impact on the teachers beliefs about maths teaching and learning as well as on their confidence in teaching number. This inclusion of the changes in teachers in a story about cognitive change in children is a valuable contribution to theory and extends our understanding of the role of maths teaching discourse in childrens development. All of these papers demonstrate the unique strengths that psychological theories have in helping teachers and other professionals to understand and intervene in arithmetical difficulties. They also demonstrate that the interventions themselves can in turn contribute to theory. Papers illustrating the social psychological contexts of arithmetic difficulties In this group of papers there is a common thread that illustrates the wider social psychological perspective on arithmetic difficulties. This perspective includes childrens beliefs and experiences concerning arithmetic, as well as the beliefs, guiding principles, and educational principles that education professionals bring to the class. These papers each tell an interesting story of their own, but they also indicate that there are ways in which educational psychology can be used to better understand and manage these out-of-class contexts. MacKenzie reviews the assessment tools and approaches that are available to educational psychologists, and notes the dearth of tools compared with those available for reading. However, her views on how educational psychologists might use the available tools in an interactive way, using their skills in observation, emphasises the role that the practitioners constructs play in their view of childrens abilities. Hughes et al.s account of out-of-school mathematics highlights the issue of authenticity in mathematical activity a characteristic that is extremely difficult to build into school maths activities, yet is often found in childrens 10 spontaneous activities at home. The authors suggest that assessment of maths ability should routinely include these out-of-school activities. They show that out-of-school maths has an emotional significance that is lacking in school maths, and makes an important contribution to the attitudes that children bring into class. Their account brings out-of-school settings into a story about how children establish the purposes of their learning. In particular, their discussion of parental dilemmas over styles of calculation produces a pragmatic solution that should become part of every schools numeracy curriculum. Grosss account of how lower attaining pupils have not benefitted from the National Numeracy Strategy includes an analysis of the use of diagnostic labels such as dyscalculia within the classroom ecology. She argues strongly against the effect that widespread use of such a term would have of taking responsiblity for low attaining children out of the teachers day-to-day classroom management. Her analyses inform our understanding of how policy contexts can shape the experiences of children with arithmetical difficulties. We have made three broad groupings of these papers for our own convenience in explaining how these authors can collectively take us forward in our understanding of psychologys role in maths education. However, there are common themes that cross the boundaries of these groups. One such theme is the importance of nonlinearity in curriculum design. That is, the curriculum should not be designed in a heavy-handed lock-step principle, in which supposedly simple aspects of maths are followed by supposedly complex aspects. One should also not assume that, because a child has difficulty with one aspect of number learning, other areas will also be problematic. We also find a thread running through many of the papers that urges us not to take precipitate decisions that limit the scope of early maths teaching, not to take poorly performing children out of the classroom teachers remit and make them somebody Educational & Child Psychology Vol 24 No 2

Editorial elses responsibility, and not to allow mechanical use of assessments to override our intuitions and direct experiences of childrens maths learning. All of these threads are drawn from current psychological understanding of arithmetic learning as a part of mathematics education. There is a further theme that echoes longstanding tensions over the differences between mathematics and computation. All of the papers address this theme in one way or another and all conclude that while computation may be important, it is not everything. Many of the papers implicitly point out that if educators behave as though computation is all that matters, we should not be surprised if children respond by only learning how to calculate. children are just beginning to construct the counting sequence and are still developing one-to-one correspondence in counting. Perceptual number stage: At this stage, children can only deal with adding together quantities that are visible. In the classroom, they look as though they understand early maths operations, because they can work with visible quantities. Figurative number stage: At this stage children seem to be able to deal with screened addition, but they are using a number sequence logic to achieve this. They cant yet operate with cardinal numbers. In the classroom they look as though they understand early maths operations because they can use ordinal strategies to work with hidden quantities. Initial number stage: At this stage children have acquired an adult-like understanding of number, and are able to comprehend number symbols and operations such as addition and subtraction. Children need to reach this cognitive stage before they can grasp the elements of the formal number curriculum cardinal number, written number symbols and number operations. Facile number stage: At this stage children are developing higher-order number concepts, and become able to focus on the relation between numbers, and on the relation between operations. Children need to reach this cognitive stage before they can grasp elements of the more advanced primary curriculum such as fractions, ratio and percentages. It takes children several years to grow from the emergent to the facile stage. Until children have reached the Initial Number stage they cannot grasp the mathematical nature of operations such as addition or subtraction and the formal written curriculum is lost on them. There is much variability in the length of time that children take to reach this stage, and there are a number of sources 11

Developmental progressions in number

Psychological models of number understanding are a combination of behaviourist and cognitivist description that somehow manage to exclude childrens own subjective experience of learning, and being taught, about number. Thus, there are many descriptions of changes in strategy as children age but no real account of the subjective experiences that must drive this sequence of development. Maths education research, by contrast, has used clinical interviewing techniques to produce a constructivist account of childrens typical journey through phases of number understanding (see, for instance, Steffe & Cobb, 1986). Many of the papers in this issue draw on Steffes research, which has outlined a clear progression in number concepts as they relate to mathematical activity. This progression has a step function in that movement from one stage to the next involves cognitive restructuring rather than a slow accretion of knowledge. (Steffe et al., 1983; Wright et al., 2000). Between the ages of roughly three and seven, most children pass through the following five broad stages in understanding arithmetical operations: Emergent number stage : At this stage, Educational & Child Psychology Vol 24 No 2

Penny Munn & Rea Reason of this variability (Ridler-Williams, 2003; Munn, 2006). While educational theorists have been describing the delicate and complex stages that children pass through to understand number, they have also created a rich understanding of the ways in which children construct the verbal string of number words. Piaget long ago pointed out that childrens number word string constructions do not necessarily co-incide with their number concepts, and present day research has given us a good understanding of why this is so. Fuson (1988) and Wright (1994) have done the most detailed research on this aspect of development, and both describe an active construction of the learned number word sequence. That is, children do not passively replicate the number word sequence and store it in memory; they actively construct it and reconstruct it as they learn it. This active construction has theoretical and practical implications. On the theoretical level, childrens number word sequences provide a window onto the developmental pathways they are taking towards the Initial Number Stage, at which number words become associated with cardinal meaning. On a practical level, it provides the understanding that childrens errors in the number word are never accidental, and can extend the practitioners understanding of the individual path that a particular child with difficulties is treading towards cardinal understanding or points beyond it. The papers by Willey et al. and by Wright et al. illustrate this approach. ing number facts and procedures. Within the progression outlined above, dyscalculia may be theorised as some kind of blockage at the perceptual and figurative stages of number construction that prevents children from moving on to operate within the later conceptual stages. However, contributions in this volume, such as those by Verschaffel et al.; Gervasoni and Sullivan; and Dowker, show us the dangers inherent in assuming that such a blockage is an insurmountable barrier to childrens progression to higher order mathematical thinking. If arithmetical understanding is made up of multiple components, and progress is not linear, then it is possible for children to develop conceptual understanding in some domains that compensate for slower progress in others. Difficulties with one aspect of number learning do not necessarily imply that other areas will also be problematic. There will also be more than one route to mathematical understanding. This view suggests that psychologists should assess exactly what children can or cannot do and plan appropriate interventions according to these assessments. Such an approach would look very different from one in which psychologists use a diagnostic label that has poorly articulated assumptions about the implications for intervention. In her extensive review of the concept of dyscalculia, Gifford (2005) concluded that an innate ability to apprehend precise numbers is unconfirmed and unlikely (p. 37). The true picture is probably much more complex. Visualisation, motor control, language and symbolic representations of numbers develop interdependently, suggesting that various areas of the brain work together to develop number understanding. One area of evidence points to problems with short term or working memory (Geary & Hoard, 2001) that might indeed explain why children struggling with basic literacy may also struggle with basic arithmetic. In the present issue, Koponen et al. argue that specific difficulty in retrieving verbal material and in forming verbal associations seems Educational & Child Psychology Vol 24 No 2

Implications of developmental progressions in number for the concept of dyscalculia

The term dyscalculia implies that there can be problems of a neurological origin affecting very basic capacity for understanding numbers (Butterworth, 2005). As stated in a document published by the Department for Education and Skills (2001), dyscalculic learners may have difficulty understanding simple number concepts, lack an intuitive grasp of numbers and have problems learn12

Editorial to be connected in the ability to learn arithmetical facts. However, the multiple sensory and symbolic pathways that children use to encode mathematical experiences mean that there is probably considerable plasticity in the neurological developments that underpin the growth of number understanding. This factor means that screening tests claiming to reflect the neurological basis of arithmetical difficulties should be interpreted in the context of all other aspects of the childs symbolic activity. MacKenzie (in this issue) argues that screening tests focussing on any particular cognitive area such as subitising do not have ecological validity in relation to the way a child uses number in everyday life and do not necessarily provide clues as to the nature of the childs actual computational difficulties, or gaps in understanding, that can lead to advice about appropriate interventions. Gross also critiques medicaldiagnostic models and argues convincingly for the importance of developing effective strategies that involve a teaching and learning response rather than debates about cognitive causation. whose number development is not well synchronised with classroom demands can develop a range of beliefs that locate arithmetical logic beyond the range of their own thought processes. Often, their beliefs about number lessons as a species of guessing game or rote-learned procedures clearly function to protect their concepts of themselves as learners. Without appropriate intervention, such children will continue to be left behind because they are constructing the arithmetical problems presented to them in radically different ways from their more successful classmates. Kelly (2004) outlined three different approaches to the curriculum. The curriculum as content approach privileges certain areas of knowledge and views education as the transmission of cultural knowledge. The curriculum as product approach is derived from early twentieth century scientific managerialism, and views education as the effective delivery of specified objectives. The curriculum as process approach is derived from child-centred education and views education as the development of the individual. The papers in this issue provide strong support for a curriculum that sets out to develop the individual rather than to deliver specified content or deliver mathematical products. Even if it were possible to specify objectively maths curricula that delivered specific outcomes, these would be experienced very differently by children at different stages of mathematical understanding.

A focus on learning opportunities

There may be purely experiential reasons for children becoming stuck in their number understanding and resembling those at earlier stages of development. It can be argued, for example, that because the traditional UK number curriculum moves rapidly into number operations in the first year, it is developmentally inappropriate for a significant proportion of children in the first year of school. Ridler-Wiliams (2003), for instance, found that a large proportion of Scottish schoolchildren had not progressed to the Initial Number Sequence by the end of their first year at school and were therefore not truly comprehending the number operations they had been taught. Follow-up study of these same children (Munn, 2006) highlights the effect that their experiences of the number curriculum can have on their beliefs about the nature of mathematics. Children Educational & Child Psychology Vol 24 No 2

Conclusion
The emphasis of this journal issue is on informing educational psychologists and their colleagues about current research regarding childrens number learning, the variability of that learning and the approaches available to develop the strategies and understandings of those children that struggle with the learning. In our view, educational psychologists are well placed to become involved in these areas both as practitioners and as theoreticians. We feel strongly that they should grasp this opportu13

Penny Munn & Rea Reason nity and avoid a precipitate narrowing of their focus to arguments about labels for arithmetical difficulties. Current research in the psychology of mathematical and arithmetical development is rightly concerned with computational processes, but there is considerable scope for the development of a social psychology of maths education. This would encompass the psychological study of the communicative and discursive contexts (at home and at school) in which mathematical thinking develops. Here too we think that educational psychologists could contribute worthwhile practitioner research on the social and individual psychology of arithmetical difficulties.

Address for correspondence

Penelope.munn@strath.ac.uk

References
Atkinson, S. (1996). Developing a Scheme of Work for Primary Mathematics. Avon: Hodder & Stoughton. Biggs, E. & Sutton, J. (1983). Teaching Mathematics 59, A classroom guide. London: McGraw Hill. Brown, M. (2005). The role of mathematics education research in influencing educational policy. (Closing Plenary) Proceedings of the Fourth Congress of the European Society for Research in Mathematics Education. Sant Feliu de Guxols, Spain, 2005. Butterworth, B. (2005). Developmental dyscalculia. In J.I.D. Campbell (Ed.), Handbook of mathematical cognition. New York: Psychology Press. Busbridge, J. & Womack, D. (1991). Effective Maths Teaching. Cheltenham: Stanley Thornes. Bryant, P. (1995). Children and arithmetic. Journal of Child Psychology and Psychiatry. 36(1), 332. Bynner, J. & Parsons, S. (1997). Does numeracy matter? London: The Basic Skills Agency. Cameron, R.J. (2006). Educational psychology: The distinctive contribution. Educational Psychology in Practice, 22(4), 289304. Clemson, D. & Clemson, W. (1994). Mathematics in the Early Years. London: Routledge. Department for Education and Skills (2001). Guidance to support pupils with dyslexia and dyscalculia (Ref: DfES 0512/2001). Nottingham: DfES Publications. Department for Education and Skills (2004). Every Child Matters: Change for Children. Nottingham: DfES Publications. Dowker, A. (2004). What works for children with mathematical difficulties? Research Report 554, Department for Education and Skills. Nottingham: DfES Publications. Farrell, P., Woods, K., Lewis, S., Rooney, S., Squires, G. & OConnors, M. (2006). A review of the functions and contribution of educational psychologists in England and Wales in the light of Every Child Matters: Change for Children. Research Report 792, Department for Education and Skills. Nottingham: DfES Publications. Fuson, K (1988). Childrens Counting and Concepts of Number. Chapter 1: Introduction and Overview of Different Uses of Number Words. NY: SpringerVerlag. Geary, D.C. & Hoard, M.K. (2001). Numerical and arithmetical deficits in learning-disabled children: Relation to Dyscalculia and dyslexia. Aphasiology, 15(7), 635647. Gelman & Gallistel (1978). The Childs Understanding of Number. Cambridge, MA: Harvard University Press. Gifford, S. (2005). Young childrens difficulties in learning mathematics: Review of research in relation to dyscalculia. Qualifications and Curriculum Authority: www.qca.org.uk/13809.html. Gutierrez, A. & Boero, P. (2006). Introduction. In A. Gutierrez, & P. Boero (Eds.), Handbook of Research on the Psychology of Mathematics Education: PME 19762006. (pp. 117146) Rotterdam: Sense Publishers. Hughes, M., Desforges, C. & Mitchell, C., with Carre, C. (2000). Numeracy and Beyond. Buckingham: OU Press. International Statistical Literacy Project (2007). History of the ISLP and the World Numeracy Project 19942006. http://www.stat.aukland.ac.nz/~iase/islp/hist. Kelly, A.V. (2004). Curriculum: Theory and Practice. London: PCP Sage. Millet, A., Brown, M. & Askew, M. (Eds.), (2004). Primary Mathematics and the Developing Professional: Multiple Perspectives on Attainment in Numeracy. Kluwer Academic Publishers. Mulligan, J. & Vergnaud, G. (2006). Research on Childrens Early Mathematical Development. In A. Gutierrez, & P. Boero (Eds.), Handbook of Research on the Psychology of Mathematics Education: PME 19762006. (pp 117146) Rotterdam: Sense Publishers. Munn, P. (2006). Mathematics in early childhood the early years maths curriculum in the UK and childrens numerical development. International Journal of Early Childhood, 38(1), 99112. Piaget, J. (1952). The childs conception of number. London: Routledge & Kegan Paul.

14

Educational & Child Psychology Vol 24 No 2

Editorial
Resnick, L.B. (1989). Developing Mathematics Knowledge. American Psychologist, 44(2), 162169. Ridler-Williams, C. (2003). Is Childrens Numeracy Development related to Communication in the Classroom? Unpublished PhD Thesis, University of Strathclyde. SEED Curriculum Review Group (2004). Purposes and Principles for the Curriculum 318. SEED Publications. Skemp (1973). The Psychology of Learning Mathematics. Penguin. Sophian, C. (1996). Childrens Numbers. Harper-Collins. Steffe, L. & Cobb, P. (with E. von Glaserfeld) (1988). Construction of Arithmetic Meanings and Strategies. New York: Springer-Verlag. Steffe, L., von Glaserfeld, E., Richards, J. & Cobb, P. (1983). Childrens Counting Types: Philosophy, Theory and Application. New York: Praeger Scientific. Wright, R.J. (1994). A Study of the Numerical development of 5 year olds and 6 year olds. Educational Studies in Mathematics, 26, 2544. Wright, R., Martland, J. & Stafford, A. (2000). Early Numeracy: Assessment for Teaching and Intervention. London: Paul Chapman Publishing.

Educational & Child Psychology Vol 24 No 2

15

Strategy flexibility in children with low achievement in mathematics

Lieven Verschaffel, Joke Torbeyns, Bert De Smedt, Koen Luwel & Wim Van Dooren
Abstract
Instructional psychologists and mathematics educators have for a long time emphasised the educational importance of recognising and stimulating flexibility in childrens self-constructed strategies as a major pillar of their innovative approaches to (elementary) mathematics education and have designed and implemented instructional materials and interventions aimed at the development of such flexibility. Especially in many curriculum reform documents from the last two decades as well as in many innovative curricula, textbooks, software, etc., there is a basic belief in the feasibility and educational value of striving for strategy flexibility, also for the younger and mathematically weaker children. However, systematic and scrutinised research that convincingly supports these basic claims is still rather scarce. In this contribution, we reflect on the flexible or adaptive choice and use of solution strategies in elementary school arithmetic. In the first part of this article we give some conceptual and methodological reflections on the adaptivity issue. More specifically, we critically review definitions and operationalisations of strategy adaptivity that only take into account task and subject characteristics and we argue for a concept and an approach that also involve the sociocultural context. Second, we address the question whether strategic flexibility is a parameter of strategic competence that differentiates mathematically strong and weak children. Finally, we discuss whether aiming for strategy flexibility is a feasible and valuable goal for all children, including the younger and mathematically weaker and weakest ones.

NSTRUCTIONAL PSYCHOLOGISTS and mathematics educators have for a long time emphasised the educational importance of recognising and stimulating flexibility in childrens self-constructed strategies as a major pillar of their innovative approaches to (elementary) mathematics education and have designed and implemented instructional materials and interventions aimed at the development of such flexibility (see e.g. Brownell, 1945; Freudenthal, 1991; Thompson, 1999; Wittmann & Mller, 19901992). Especially in many curriculum reform documents from the end of the previous century, such as the Curriculum and evaluation standards for school mathematics of the National Council of Teachers of Mathematics in the US (1989, 2000), the Numeracy Strategy in the UK (DfEE, 1999), the Proeve van een Nationaal Programma voor het Reken/wiskundeonderwijs

in The Netherlands (Treffers, De Moor & Feijs, 1990), the Handbuch Produktiver Rechenbungen in Germany (Wittmann & Mller, 1990), and the Ontwikkelingsdoelen en Eindtermen for elementary education in Flanders (1998), as well as in many innovative curricula, textbooks, software, and other instructional materials based on these reform documents, there is a basic belief in the feasibility and educational value of striving for strategy flexibility, also for the younger and mathematically weaker children (Baroody et al., 2003; Kilpatrick, Swafford & Findell, 2001; Verschaffel, Greer & De Corte, in press). However, systematic and scrutinised research that convincingly supports these basic claims is still rather scarce. In this contribution, we reflect on the flexible or adaptive1 choice and use of solution strategies in elementary school arithmetic. First, we present and discuss difEducational & Child Psychology Vol 24 No 2 The British Psychological Society 2007

16

Strategy flexibility in children with low achievement in mathematics ferent conceptions and operationalisations of strategy flexibility. Second, we address the question whether strategic flexibility is a parameter of strategic competence that differentiates mathematically strong and weak children. Finally, we discuss whether aiming for strategy flexibility is a feasible and valuable goal for all children, including the younger and mathematically weaker and weakest ones. tics. For example, Van der Heijden (1993, p. 80) defines it as follows: Flexibility in strategy use involves the flexible adaptation of ones solution procedures to task characteristics. He operationalised strategy flexibility by analysing whether children systematically use the 1010-procedure and the G10-procedure for, respectively, additions and subtractions in the number domain up to 1002. Exactly the same definition is used by Blte Van der Burg, and Klein (2001). In a similar way, Thompson (1999, p. 147) concurs in his plea for developing flexibility in elementary school arithmetic that mental calculation places great emphasis on the need to select an appropriate computational strategy for the actual numbers in the problem. So, these authors first distinguish different strategies for doing additions and subtractions, and, based on an analysis of the strengths and weaknesses of these different strategies vis--vis certain types of problems3, they then define certain problem type strategy type combinations as flexible and others as not. As we have illustrated elsewhere (see Verschaffel et al., submitted), this view on flexible or adaptive strategy use is also found in many (so-called) reformbased textbooks, such as the first years pupils book of a Flemish textbook series Nieuwe Reken Raak (Bourdeaudhui et al., 2002). In the textbook for Grade 1 children are taught three different strategies for doing additions with sums between 10 and 20: (a) the retrieval strategy (e.g. knowing by heart that 6 6 12), (b) the tie strategy (e.g. solving 6 7 by doing 6 6 1), and (c) the decomposition-to-ten strategy (e.g.

Conceptions and operationalisations of strategy flexibility/adaptivity

Strategy flexibility or adaptivity is sometimes defined and operationalised as using a variety of solution strategies, without any further qualification (e.g. Heirdsfield & Cooper, 2002). Although the availability of a variety of strategies and the ability to switch smoothly between these strategies can be considered an essential stepping-stone towards flexibility, the mere use of different solution strategies on a series of similar mathematical items or problems without any attention to the efficiency of these strategy choices can hardly be considered as evidence of adaptivity. After all, on the one hand, one can switch smoothly between different strategies in a completely arbitrary and random way while, on the other hand, the consistent use of one single strategy for a whole series of arithmetic tasks may sometimes be more adaptive than switching between a diversity of strategies available in ones repertoire (Verschaffel et al., submitted). More frequently, strategy flexibility or adaptivity has been defined and operationalised in relation to certain task characteris-

For the sake of simplicity, we will use both terms as synonyms, although, as we have explained elsewhere (Verschaffel, Luwel, Torbeyns & Van Dooren, submitted), some authors define both terms (slightly) differently. 2 The 1010 procedure involves splitting off the tens and the units in both integers and handling them separately (e.g. 47 15 .; 40 10 50; 7 5 12; 50 12 62). The application of the G10 procedure, on the other hand, requires the child to add or subtract the tens and the units of the second integer to/from the first unsplit integer (e.g. 47 15 .; 47 10 57; 57 5 62). 3 This analysis consists of a purely rational task analysis (wherein the procedural and conceptual complexities of the different strategies are compared), of an empirical analysis of the accuracy data from large groups of students for the different (types of) problems, or of a combination of both kinds of analysis.

Educational & Child Psychology Vol 24 No 2

17

Lieven Verschaffel et al. solving 6 7 by doing 6 4 3). Simultaneously, children learn to link each strategy to a particular type of sum over ten for which that strategy is considered most efficient: (a) the retrieval strategy for tie sums (e.g. 6 6 ), (b) the tie strategy for near-tie sums (e.g. 6 7 or 8 7 ) and (c) the decomposition-to-ten strategy for all other sums over ten (e.g. 6 8 or 3 9 ). Although this conception and operationalisation of flexibility/adaptivity can already be considered as more sophisticated than the first one, wherein it is simply identified with the (random) use of multiple strategies, it remains, in our view, highly problematic to define and operationalise strategy flexibility/ adaptivity in terms of task characteristics alone. Indeed, it is possible that, for a particular subject and/or under particular circumstances, the strategy choice process that the above-mentioned authors call flexible/ adaptive may become inflexible/inadaptive, and vice versa. Hereafter, we consider two other groups of factors that need to be incorporated into a more genuine concept of flexibility/adaptivity, besides task variables, namely subject and context variables. A first additional set of factors, which has been intensively and systematically investigated and modeled by cognitive psychologists such as Siegler and his associates (Shrager & Siegler, 1998; Siegler, 1996, 2000; see also Torbeyns, Arnaud, Lemaire & Verschaffel, 2004), are subject variables. Sieglers computer model of how childrens mastery of simple arithmetic sums develops, the Strategy Choice and Discovery Simulation (SCADS), indicates that whether a particular strategy (e.g. retrieval or an extensive counting strategy or a shortcut counting strategy) is chosen to solve a particular item by a particular child depends basically on how accurately and how fast that strategy is executed for that particular item and by that particular child, in comparison to other concurrent strategies available in the childs repertoire. In other words, SCADS always tends to select and apply the strategy that produces the most beneficial combination of speed and accuracy for a particular sum (on the basis of the accumulated data regarding speed and accuracy available in the systems data base)4. Irrefutably, the adaptivity concept underlying this computer model which claims to be an accurate simulation of how real children make strategy choices and develop new strategies in the domain of elementary arithmetic reflects a more complex and more subtle view on the strategy choice process, wherein affordances inherent in the task have to be considered in relation to, and balanced with, features of the individual who is solving the task, especially how well (s)he masters the distinct strategies. More recent theoretical developments, especially the rise of the sociocultural perspective, have revealed that the issue of flexibility/adaptivity is even more complicated than is suggested by cognitive (computer) models such as the one by Siegler and his colleagues (Shrager & Siegler, 1998; Siegler, 1996, 2000). More particularly, studies within the sociocultural view suggest that people may switch flexibly/adaptively between arithmetic strategies depending on not only task and subject variables, but also on variables related to the (sociocultural) context. Relying on the scarce theoretical and empirical sociocultural literature, Ellis (1997, p. 492) shows that, with age and experience, children develop (implicit) knowledge regarding what a given culture defines as appropriate, adaptive, and wise. This knowledge may also guide ones strategy choices at an explicit or implicit level. The idea of classroom situations

Actually, the selection mechanism is somewhat more sophisticated, as the system also awards novely points to newly discovered strategies, which increase the chance that a new strategy will be attempted (regardless of its efficiency in terms of speed and accuracy). This could be interpreted as a (first) attempt to include also a context variable into the (computer) model.

18

Educational & Child Psychology Vol 24 No 2

Strategy flexibility in children with low achievement in mathematics and psychological experiments as situations characterised by both social and task goals (Ellis, 1997, p. 508) or, to state it differently, as situations determined by, respectively, a didactical contract (Brousseau, 1997) or an experimental contract (Greer, 1997) can contribute, according to Ellis (1997), to our understanding of strategy choice, particularly to those choices that seem less than optimal at first glance (at least in terms of accuracy and speed, which dominate cognitive-psychological research). So, childrens strategy choices in elementary arithmetic are co-determined by characteristics of the sociocultural context in which they have to demonstrate their arithmetic skills: for example, what aspects of their strategic behaviour instead of, or in addition to, the salient aspects of speed and accuracy seem (most) valued in the classroom and/or testing context, such as simplicity, elegance, formality, generality, intelligibility, certitude, originality, etc. of the solution strategy (see also Ellis, 1997; Lave & Wenger, 1991; Rogoff, 1990). Based on the previous brief overview, we propose the following working definition of what it means to be adaptive in ones strategy choices: By an adaptive choice of a strategy we mean the conscious or unconscious selection and use of the most appropriate solution strategy on a given mathematical item or problem, for a given individual, in a given sociocultural context. By the phrase the most appropriate strategy we simply do not mean the strategy that leads most quickly to the correct answer (as in the strictly cognitive-psychological sense of the term), although we do not exclude that in a particular instructional or testing setting it may have that meaning (see also Ellis, 1997; Verschaffel et al., submitted). Are mathematically weak children less flexible than mathematically strong ones? Over the past few years, researchers have proposed and used several models to analyse and compare childrens strategic behaviour. An influential analytic model has been proposed by Siegler and associates (Shrager & Siegler, 1998; Lemaire & Siegler, 1995; Educational & Child Psychology Vol 24 No 2 Siegler, 1998, 2000). In their model of strategy change Lemaire and Siegler (1995) distinguish four dimensions of strategic competence: (1) the repertoire of strategies that people use to solve a set of items in a given task domain, (2) the relative frequency with which the different strategies are applied to solve this set of items, (3) the efficiency with which each strategy is executed (typically measured in terms of speed and/or accuracy of strategy use), and (4) the adaptivity with which the various strategies are chosen and applied on a given set of items, or stated differently, the efficiency of the strategy choices taking into account the relevant task and subject parameters. So, Sieglers theory depicts cognitive development as characterised by a continuously changing repertoire of coexisting strategies, which are applied with continuously changing frequencies and proficiencies and which are executed in an increasingly adaptive way. Whereas most research done within this framework has looked at strategy choice and strategy change processes in children with a normal arithmetic development, some researchers have started to use this model (or similar models) to analyse low achievers acquisition and use of arithmetic strategies in comparison to the strategic performance of their normally developing peers. Gray, Pitta and Tall (1997); Torbeyns, Verschaffel and Ghesquiere (2004a), and several others conducted detailed analyses of the strategies that mathematically weak children and children with mathematical difficulties (MD) apply on single-digit additions and subtractions like 5 8 and 12 5. Generally speaking, these analyses revealed that, compared to their normally achieving peers, low achievers and MD children use basically the same types of strategies. However, they rely more often on immature counting strategies and less often on more efficient mental strategies; in particular, they execute more advanced strategies, like the decomposition-to-ten strategy, less accurately and slower than typically developing children. Most of these comparative studies have, however, not addressed the 19

Lieven Verschaffel et al. fourth strategy parameter, namely flexibility, in their analysis of childrens strategy choices. Interestingly, some recent studies from our centre (Torbeyns et al., 2004a; Torbeyns, Verschaffel & Ghesquire, 2005) tried to address the flexibility parameter, by using the so-called choice/no-choice method (Siegler & Lemaire, 1997). This method requires testing each participant under two types of conditions. In the choice condition, participants can freely choose which strategy they use to solve each problem. In the no-choice condition, they must use one particular strategy to solve all problems. Ideally, the number of no-choice conditions equals the number of strategies available in the choice condition. The obligatory use of one particular strategy on all problems in the no-choice condition by each participant allows the researcher to obtain unbiased estimates of the speed and accuracy of the strategy (Siegler & Lemaire, 1997; Torbeyns et al., 2004b). The comparison of the accuracy and speed data of the different strategies gathered in the no-choice conditions, with the strategy choices made in the choice condition, allows the researcher to assess the adaptivity of individual strategy choices in the choice condition in a scientifically valid way: Does the subject (in the choice condition) solve each problem by means of the strategy that yields the best performance in terms of accuracy and speed on this problem, as evidenced by the information obtained in the no-choice conditions? Hereafter, we briefly and exemplarily review one study from our own centre wherein we have applied this method to compare the strategic performance, and especially the strategy flexibility/adaptivity, of the strategy choices of children of different mathematical ability in the domain of single-digit elementary arithmetic. Torbeyns et al. (2005) investigated the strategic performance of first graders of different mathematical achievement levels on arithmetic sums over ten by means of the choice/no-choice method. They characterized childrens strategy use with the four 20 above-mentioned parameters of Lemaire and Sieglers (1995) model of strategy change. Participants were 83 first graders with high, moderate or low mathematical abilities (hereafter abbreviated as HA, MA, and LA) and who had been taught two mental calculation strategies as part of their regular instruction namely the above-mentioned decomposition-to-10 strategy and the tie strategy (7 8 7 7 1 14 1 15). All children were administered a series of five near tie sums over ten (like 7 8 and 7 6 ) in different conditions. In the choice condition, children could choose between the decomposition-to-10 and the tie strategy on each near-tie sum. Afterwards they had to solve, in two no-choice conditions, the same sums with the decomposition-to-10 and with the tie strategy. Instructions and visualisations were used to force children to choose between the two strategies (in the choice condition) or to apply only the intended strategy (in the no-choice conditions). To assess whether children were able to solve the problems by means of retrieval (i.e., by knowing the answer to a sum by heart), a third nochoice condition was added to the design, in which the children had to solve the same near-tie items (together with some extra retrieval items) by means of retrieval. First, with respect to strategy repertoire, the results from the choice condition revealed that about half of the children applied both types of strategies to solve the near-tie items (57%), whereas the others either solved all sums by the decomposition-to-10 (25%) or the tie strategy (18%). There were many more children who applied both strategies in the HA (77%) and MA group (65%) than among the LA children (31%). Second, children used the tie strategy (50%) with the same frequency as the decomposition-to-10 strategy (50%) to solve the five near-tie sums in the choice condition, but the HA and MA children applied the tie strategy much more often than their LA peers. Third, the accuracy and speed data from the no-choice conditions showed that all children were quite efficient in performing both strategies Educational & Child Psychology Vol 24 No 2

Strategy flexibility in children with low achievement in mathematics (with accuracies .90), and that the HA and MA children applied these strategies more efficiently (in terms of speed) than the LA pupils. Finally, and for the topic of this paper most importantly, the comparison of the results of the choice and no-choice conditions did not reveal any group differences in strategy adaptivity. More specifically, in the choice condition mathematically HA and MA pupils did not solve more frequently the near tie sums with the most efficient strategy (according to a comparative analysis of the efficiency data from the two no-choice conditions) compared with the LA pupils. Torbeyns et al. (2005) provide several possible causes for this last unexpected finding. One of these explanations is of particular interest for the present article, as it reflects the growing awareness among researchers of the sociocultural (i.e., instructional) factors that co-determine childrens strategy choices in elementary arithmetic. According to the authors, the HA and MA pupils might have based their strategy selections more than the LA ones on (their interpretation of) the socio-mathematical norms and practices in their mathematics classroom (Yackel & Cobb, 1996) than on the strategy efficiency characteristics. The textbook that was used in these classrooms was Nieuwe Reken Raak. As explained above, this textbook strongly favours, or even imposes, one particular strategy for doing near tie sums, namely the tie strategy (e.g. 6 7 6 6 1 13), above the other strategy, namely decomposition-to-10 (e.g. 6 7 6 4 3 10 3 13). Interviews with the classroom teachers revealed that they had followed this textbook rigorously, at least for that part of the arithmetic curriculum. So, arguably, these norms and practices in these mathematics classes were strongly in favour of always solving neartie sums with the tie strategy. Stated differently, these HA and MA children may have demonstrated a level of adaptivity that was suboptimal for them (in the cognitive-psychological sense of the word), because they made strategy selections in line with (their Educational & Child Psychology Vol 24 No 2 interpretation of) the classroom norms and practices about what it means to behave flexibly with regards to this particular kind of arithmetic exercises. And these norms and practices reflected a notion of strategy flexibility that was purely defined in terms of task characteristics. So, if these pupils had been instructed in a mathematics classroom favoring a more sophisticated concept of flexibility (that also allowed subjective considerations besides task-related ones when making strategy choices), one might have found the expected ability-related difference in strategy flexibility. But this remains, of course, a hypothesis, which should be tested in future research. Engendering strategy flexibility: From when on and for whom? As stated before, many current reform-based documents and materials depart from the basic belief that striving for strategy flexibility is feasible and educationally valuable, also for the younger and mathematically weaker children (Baroody et al., 2003; Kilpatrick et al., 2001; Verschaffel et al., in press). However, systematic and scrutinised research that convincingly supports these basic claims is problematically scarce (Geary, 2003; Verschaffel et al., in press). Hereafter we briefly discuss the above-mentioned claims concerning the optimal age and the optimal public to strive for flexibility. First, there is the issue of the optimal moment for starting to strive for adaptivity/ flexibility. Several authors argue that it is better first, and above all, to teach for routine mastery of a given arithmetic skill, and afterwards to change ones aims and pedagogy in the direction of flexible or adaptive strategy use. This argument is supported by the widespread belief that without longterm memory of previously learned facts, procedures, models, and representational tools, there can be no flexible or adaptive thinking and by the observation that many children do not succeed in mastering fluency and/or automaticity in a given arithmetic strategy, even after numerous 21

Lieven Verschaffel et al. hours of regular classroom instruction and additional specific training (see e.g. Geary, 2003; Milo & Ruijssenaars, 2002; Warner, Davis, Alcock & Coppolo, 2002). This argument against a premature strive for flexibility is opposed by many advocates of the reform-based approach to mathematics education, who conjecture that the development of adaptive expertise is not something that simply happens after pupils have developed routine expertise in a given domain. On the contrary, these authors argue that education for flexibility should already be present from the very beginning of the teaching/learning process (see e.g. Baroody, 2003; Gravemeijer, 1994; Selter, 1998; Wittmann & Mller, 19901992). This idea is nicely expressed (albeit in more general terms) in the following quote from Bransford (2001, p. 3):
You dont develop it in a capstone course at the end of students senior year. Instead the path toward adaptive expertise is probably different from the path toward routine expertise. Adaptive expertise involves habits of mind, attitudes, and ways of thinking and organising ones knowledge that are different from routine expertise and that take time to develop. I dont mean to imply that you cant teach an old routine expert new tricks. But its probably harder to do this than to start people down an adaptive expertise path to begin with at least for most people.

Some authors go even further and warn against the rigidifying effects of years of diligent practice in routine expertise. According to Feltovich, Spiro & Coulson, (1997, p. 126), there are effects on cognition that come with such an extended practice that could lead to reduction in cognitive flexibility to conditions of relative rigidity in thinking and acting (while, we have noted, affecting other, more desirable goals, such as in efficiency and speed). Given the decrease in flexibility that might accompany increased routine experience in a certain domain, it seems very risky to design teaching/learning environments wherein one strives (only) for routine expertise first and postpones engendering flexibility/adaptivity until the moment routine expertise has been estab22

lished. Applied to the field of elementary mathematics education, an initial exclusive focus on procedural fluency of some explicitly taught procedures for particular problem types may not only be unhelpful, but also counterproductive for the development of adaptive expertise. Closely related to the issue when to start striving for strategy flexibility/adaptivity, is the question whether promoting flexible and adaptive strategy use is feasible and valuable across different levels of mathematical achievement, including the average and mathematically weaker ones. Threlfall (2002, p. 40) refers to the frequently heard claim that (because) only the more mathematically minded children will be capable of learning how to make good choices, (. . .) flexibility should be abandoned as an objective for the average and below average. The claim that working for flexibility is unfeasible for these latter groups of children is supported by cognitive psychological research that has documented lower working memory capacity in children with low achievement in mathematics. In particular, the central executive component of working memory, which is responsible for the control, regulation and monitoring of complex cognitive processes, appears to be related to individual differences in math ability (e.g. Bull & Scerif, 2001; McLean & Hitch, 1999; Swanson & Beebe-Frankenberger, 2004). This central executive component has been fractionated into separate, though overlapping, functions, one of which includes the ability to shift between tasks or strategies (Baddeley, 1996). Interestingly, some studies have reported that children with low achievement in mathematics have general difficulties in flexible shifting between solution strategies, as measured by standard cognitive shifting tasks (Bull, Johnston & Roy, 1999; Bull & Scerif, 2001; McLean & Hitch, 1999). Therefore, the goal of developing strategy flexibility might be unfeasible or even dangerous for these children. The claim that working for flexibility is unfeasible for these children is also Educational & Child Psychology Vol 24 No 2

Strategy flexibility in children with low achievement in mathematics supported by the results of several intervention studies (e.g. Baxter, Woodward & Olson, 2001; Geary, 2003; Milo & Ruijssenaars, 2002; Sowder, Philipp, Armstrong, & Schappelle, 1998; Woodward & Baxter, 1997; Woodward, Monroe, & Baxter, 2001) indicating that especially mathematically weaker children and/or children with MD profit more from instruction that pays major, if not exclusive, attention to developing efficiency (i.e., accuracy and speed) in one arithmetic strategy and that considers flexibility as a strategic quality of only second-rate importance. However, the results and conclusions of these intervention studies are contradicted by other studies that do support the claim that not only mathematically able children but also lower-achieving children benefit more from reform-based instruction (which strongly aims at the development of strategy variety and flexibility) than traditionally-oriented direct instruction (which aims at the development of one particular strategy) (Baroody, 1996; Bottge, 1999; Bottge, Heinrichs, Chan & Serlin, 2001; Bottge, Heinrichs, Mehta & Hung, 2002; Cichon & Ellis, 2003; Klein, Beishuizen & Treffers, 1998; Menne, 2001; Moser Opitz, 2001; Van den Heuvel-Panhuizen, 2001). Given the large differences in (a) the arithmetic tasks and solution strategies being addressed in these intervention studies, (b) the age and characteristics of the pupils being involved (and especially of the mathematically weaker ones), (c) the nature of the intervention (and especially what kind of strategy flexibility/adaptivity is aimed at and how it is realised through instruction), and (d) how the effects are measured (and especially how adequately flexibility/ adaptivity is assessed), the contradictory nature of the results and conclusions is not surprising and is an issue that implores further research. Anyhow, taking into account the theoretical issues discussed in the first part of this article, we believe that providing children with a (quasi-)algorithmic rule for linking problem types to solution strateEducational & Child Psychology Vol 24 No 2 gies and with systematic training in the fluent application of that rule (as in the above-mentioned Flemish method Nieuwe Reken Raak) is not the kind of approach that will yield flexible or adaptive expertise as we have conceived and defined it. The latter kind of instruction, which is based upon a notion of flexibility that merely looks at task variables without consideration of individual or contextual factors misjudges the quintessence of the notion of adaptivity. Indeed, adaptivity involves a personal and insightful choice based on weighing different kinds of affordances, not only task-related variables, but also subject- and context-related variables. The more one dismisses a notion of strategy flexibility that merely looks at linking welldefined strategies to task characteristics (as in Nieuwe Reken Raak), the more one will agree that there is no easy and direct shortcut to becoming adaptive, and that this is not something that can be trained or taught but rather something that has to be promoted or cultivated in a long-term perspective. Clearly, the more one adheres the latter view on adaptivity, the more difficult and challenging it might become to strive for it, especially with younger and mathematically weak children. Whether this will be feasible and educationally valuable remains an open issue requiring further empirical research. It is important that further intervention research about the feasibility and the optimal form of flexibility oriented instruction for younger and mathematically weaker children is clear with respect to its focus. If the major instructional aim is to obtain in the short term better gains in computational facility (operationalised as being able to solve familiar sums quickly and correctly), it might be more efficient to teach one single strategy for each arithmetic operation, which is used with all such problems, or to provide the children with a quasi-algorithmic rule to associate certain problem types with certain solution strategies and to train them in the (routine) application of that rule, with23

Lieven Verschaffel et al. out any concern for subject or context variables. If, however, the instructional goals are broader and more long-term, and comprise genuine strategy flexibility combined with good understanding of mathematical concepts and principles, pattern recognition skills, and appropriate beliefs, attitudes, and emotions towards mathematics and mathematics learning and teaching then other, less routine-based instructional approaches might be more appropriate, also for the younger and mathematically weaker children. So, the question which instructional approach is the best for children, and for children with mathematical difficulties in particular, is not purely an empirical one, but also, to a significant degree, dependent on the value system underlying ones view of the purposes of mathematics education. Furthermore, this question also has an ethical dimension: Is it equitable to expose children of average or low ability (as assessed by traditional criteria) to a mathematics education that is intellectually less stimulating (Greer, personal communication)? dence and too little on convincing evidence from empirical research. Especially the application of these ideas to mathematically weaker children and children with mathematical difficulties remains an open question, as supporting empirical evidence is problematically scarce (as is the counterevidence). If we want to make progress in our theoretical understanding and practical enhancement of strategy flexibility in elementary arithmetic of children with low achievement in mathematics (including children with MD), there is a great need for continued research efforts, especially to design experiments involving children from different ability ranges. Only when such studies have convincingly and repeatedly shown that the reform-based approach, which strives for varied and flexible strategy use effectively leads to the intended outcomes (without resulting in significant loss in computational accuracy and fluency), researchers will be in a good position to convince policy-makers, teachers, and parents of the feasibility and value of striving for adaptive expertise for children of all ages and all ability levels, rather than reserving it as a pinnacle for those who first have developed routine computational expertise.

Conclusion
In this article, we have argued that for children of all ages and all ability levels, the most valuable instructional approach will be not the one in which children receive drilland-practice in (quasi-)algorithmic techniques for selecting the most efficient procedures for particular types of problems. Instead, we favour an instructional approach wherein children are cultivated in developing their own preferences based on a personal reflection on task, subject, and context characteristics. However, we acknowledge that these ambitious pleas to strive for such a kind of adaptive expertise and the accompanying claims about the optimal road for reaching this ambitious goal, are still too much based on rhetoric and anecdotal evi-

Acknowledgements
This research was partially supported by Grant GOA 2006/01 Developing adaptive expertise in mathematics education from the Research Fund K.U. Leuven, Belgium

Address for correspondence

Lieven Verschaffel, Centre for Instructional Psychology and Technology, Katholieke Universiteit Leuven, Vesaliusstraat 2, 3000 Leuven, Belgium. Tel: 32 16/32.62.58, Fax: 32 16/32.62.74 E-mail: lieven.verschaffel@ ped.kuleuven.be

24

Educational & Child Psychology Vol 24 No 2

Strategy flexibility in children with low achievement in mathematics

References
Baddeley, A.J. (1996). Exploring the central executive. Quarterly Journal of Experimental Psychology, 49, 528. Baroody, A.J. (1996). An investigative approach to the mathematics instruction of children classified as learning disabled. In D.K. Reid, W.P. Hresko et al. (Eds.), Cognitive approaches to learning disabilities (third edition) (pp. 545615). Austin, TX: PRO-ED. Baroody, A.J. (2003). The development of adaptive expertise and flexibility: The integration of conceptual and procedural knowledge. In A.J. Baroody & A. Dowker (Eds.), The development of arithmetic concepts and skills (pp. 134). Mahwah, NJ: Lawrence Erlbaum Associates. Baroody, A.J., Wilkins, J.L.M. & Tiilikainen, S.H. (2003). The development of childrens understanding of additive commutativity: From protoquantitive concept to general concept? In A.J. Baroody & A. Dowker (Eds.), The development of arithmetic concepts and skills (pp. 127160). Mahwah, NJ: Lawrence Erlbaum Associates. Baxter, J.A., Woodward, J. & Olson, D. (2001). Effects of reform-based mathematics instruction on low achievers in five third-grade classrooms. Elementary School Journal, 101, 529547. Blte, A.W., Van der Burg, E. & Klein, A.S. (2001). Students flexibility in solving two-digit addition and subtraction problems: Instruction effects. Journal of Educational Psychology, 93, 627638. Bottge, B.A. (1999). Effects of contextualized math instruction on problem solving of average and below-average achieving students. The Journal of Special Education, 33, 8192. Bottge, B.A., Heinrichs, M., Chan, S. & Serlin, R.C. (2001). Anchoring adolescents understanding of math concepts in rich problem solving environments. Remedial and Special Education, 22, 299314. Bottge, B.A., Heinrichs, M., Mehta, Z. & Hung, Y. (2002). Weighing the benefits of anchored math instruction for students with disabilities in general education classes. The Journal of Special Education, 35, 186200. Bourdeaudhui, G., Caymax, J., Deboyser, G., Haepers, F., Jacobs, H., Matthys, L., Melis, J., Scheurweghs, M., Van den Eynde, R., Van der Avert, A., Vandeweyer, W., Verlinden, S., Westerlinck, W. & Wiecke, C. (2002). Nieuwe Reken Raak. Invulboek 1. Leuven, Belgium: Wolters. Bransford, J. (2001). Thoughts on adaptive expertise. (Unpublished manuscript); Available at http:// www.vanth.org/docs/AdaptiveExpertise.pdf. Brousseau, G. (1997). Theory of didactical situations in mathematics. In N. Balacheff, M. Cooper, R. Sutherland & V. Warfield (Eds. & Trans.). Dordrecht, The Netherlands: Kluwer. Brownell, W.A. (1945). When is arithmetic meaningful? Journal of Educational Research, 38, 481498. Bull, R., Johnston, R.S. & Roy, J.A. (1999). Exploring the roles of the visual-spatial sketch pad and central executive in childrens arithmetical skills: Views from cognition and developmental neuropsychology. Developmental Neuropsychology, 15, 421442. Bull, R. & Scerif, G. (2001). Executive functioning as a predictor of childrens mathematics ability: Inhibition, switching, and working memory. Developmental Neuropsychology, 19, 273293. Cichon, D. & Ellis, J.G. (2003). The effects of MATH Connections on student achievement, confidence, and perception. In S.L. Senk & D.R. Thompson (Eds.), Standards-based School Mathematics Curricula: What are They? What do Students Learn? (pp. 345374). Mahwah, N.J.: Lawrence Erlbaum Associates. Department for Education and Employment (1999). The National Numeracy Strategy Framework for Teaching Mathematics from Reception to Year 6. Londen: DfEE. Ellis, S. (1997). Strategy choice in sociocultural context. Developmental Review, 17, 490524. Feltovich, P.J., Spiro, R.J. & Coulson, R.L. (1997). Issues of expert flexibility in contexts characterized by complexity and change. In P.J. Feltovich, K.M. Ford & R.R. Hoffman (Eds.), Expertise in context: Human and machine (pp. 125146). Menlo Park, CA: AAAI Press. Freudenthal, H. (1991). Revisiting mathematics education. Dordrecht: Reidel. Geary, D.C. (2003). Arithmetical development: Commentary on Chapters 9 through 15 and future directions. In A.J. Baroody & A. Dowker (Eds.), The Development of Arithmetic Concepts and Skills: Constructing Adaptive Expertise (pp. 453464). Mahwah, N.J.: Lawrence Erlbaum Associates. Gray, E., Pitta, D. & Tall, D. (1997). The nature of the object as an integral component of numerical processes. In E. Pehkonen (Ed.), Proceedings of the 21st PME International Conference, 1, 115130. Gravemeijer, K. (1994). Developing realistic mathematics education. Utrecht, The Netherlands: Freudenthal Institute, University of Utrecht. Greer, B. (1997). Modelling reality in mathematics classrooms: The case of word problems. Learning and Instruction, 7, 293307. Heirdsfield, A.M., & Cooper, T.J. (2002). Flexibility and inflexibility in accurate mental addition and subtraction: Two case studies. Journal of Mathematical Behavior, 21, 5774. Kilpatrick, J., Swafford, J. & Findell, B. (2001). Adding it up. Helping children learn mathematics. Washington, D.C.: National Academy Press.

Educational & Child Psychology Vol 24 No 2

25

Lieven Verschaffel et al.

Klein, A.S., Beishuizen, M. & Treffers, A. (1998). The empty number line in Dutch second grades: Realistic versus gradual program design. Journal for Research in Mathematics Education, 29, 443464. Lave, J. & Wenger, E. (1991). Situated learning: Legitimate, peripheral participation. Cambridge: Cambridge University Press. Lemaire, P. & Siegler, R.S. (1995). Four aspects of strategic change: Contributions to childrens learning of multiplication. Journal for Experimental Psychology: General, 124, 8397. McLean, J.F. & Hitch, G.J. (1999). Working memory impairments in children with specific arithmetic learning difficulties. Journal of Experimental Child Psychology, 74, 240260. Menne, J.J.M. (2001). Met sprongen vooruit. Een productief oefenprogramma voor zwakke rekenaars in het getallengebied tot 100 een onderwijsexperiment. [A productive training program for mathematically weak children in the number domain up to 100 a design study]. Utrecht, The Netherlands: CD-beta Press. Milo, B. & Ruijssenaars, A.J.J.M. (2002). Strategiegebruik van leerlingen in het speciaal basisonderwijs: begeleiden of sturen? [Strategy instruction in special education: Guided of direct instruction?] Pedagogische Studin, 79, 117129. Moser Opitz, E. (2001). Mathematical knowledge and progress in the mathematical learning of children with special needs in their first year of school. In MATHE 2000. Selected papers (pp. 8588). Dortmund, Germany: University of Dortmund, Department of Mathematics. National Council of Teachers of Mathematics. (1989). Principles and standards for school mathematics. Reston, VA: National Council of Teachers of Mathematics. National Council of Teachers of Mathematics. (2000). Principles and standards for school mathematics. [http://standards.nctm.org/document/index. htm]. Ontwikkelingsdoelen en eindtermen. Informatiemap voor de onderwijspraktijk. Gewoon basisonderwijs (1998). [Standards. Documentation for practitioners. Elementary education.] Brussel: Ministerie van de Vlaamse Gemeenschap, Departement Onderwijs, Afdeling Informatie en Documentatie. Rogoff, B. (1990). Apprenticeship in thinking. New York: Oxford University Press. Swanson, H.L. & Beebe-Frankenberger, M. (2004). The relationship between working memory and mathematical problem solving in children at risk and not at risk for serious math difficulties. Journal of Educational Psychology, 96, 471491. Selter, C. (1998). Building on childrens mathematics. A teaching experiment in grade three. Educational Studies in Mathematics, 36, 127. Shrager, J. & Siegler, R.S. (1998). SCADS: A model of childrens strategy choices and strategy discoveries. Psychological Science, 9, 405410. Siegler, R.S. (1996). Emerging minds: the process of change in childrens thinking. Oxford: Oxford University Press. Siegler, R.S. (2000). The rebirth of childrens learning. Child Development, 71, 2635. Siegler, R.S., & Lemaire, P. (1997). Older and younger adults strategy choices in multiplication: Testing predictions of ASCM using the choice/no-choice method. Journal of Experimental Psychology: General, 126, 7192. Sowder, J., Philipp, R., Armstrong, B. & Schappelle, B. (1998). Middle-grade teachers mathematical knowledge and its relationship to instruction. Albany, NY: SUNY. Thompson, I. (1999). Getting your head around mental calculation. In I. Thompson (Ed.), Issues in Teaching Numeracy in Primary Schools (pp. 145156). Buckingham; U.K.: Open University Press. Threlfall, J. (2002). Flexible mental calculation. Educational Studies in Mathematics, 50, 2947. Torbeyns, J., Arnaud, L., Lemaire, P. & Verschaffel, L. (2004a). Cognitive change as strategic change. In A. Demetriou & A. Raftopoulos (Eds.), Cognitive Developmental Change: Theories, Models and Measurement (pp. 186216). Cambridge, UK: Cambridge University Press. Torbeyns, J., Verschaffel, L. & Ghesquire, P. (2004b). Strategy development in children with mathematical disabilities: Insights from the choice/nochoice method and the chronologicalage/ability-level-match design. Journal of Learning Disabilities, 37, 119131. Torbeyns, J., Verschaffel, L. & Ghesquire, P. (2005). Simple addition strategies in a first-grade class with multiple strategy instruction. Cognition and Instruction, 23, 121. Treffers, A., De Moor, E. & Feijs, E. (1990). Proeve van een national programma voor het reken/wiksundeonderwijs op de basisschool. Deel 1. Overzicht einddoelen. [Towards a national curriculum for mathematics education in the elementary school. part 1. Overview of the goals.] Tilburg, The Netherlands: Zwijsen. Van den Heuvel-Panhuizen, M. (Ed.) (2001). Children learn mathematics. A learning-teaching trajectory with intermediate attainment targets for calculation with whole numbers in primary school. Groningen, The Netherlands: Wolters Noordhoff. Van der Heijden, M.K. (1993). Consistentie van aanpakgedrag. [Consistency in solution behavior.] Lisse, The Netherlands: Swets & Zeitlinger. Verschaffel, L., Greer, B. & De Corte, E. (in press). Whole number concepts and operations. In F. Lester (Ed.), Handbook of Research on Mathematics

26

Educational & Child Psychology Vol 24 No 2

Strategy flexibility in children with low achievement in mathematics

Teaching and Learning (second version). New York: MacMillan. Verschaffel, L., Luwel, K., Torbeyns, J. & Van Dooren, W. (submitted). Conceptualizing, investigating, and enhancing adaptive expertise in elementary mathematics education. Manuscript submitted for publication. Warner, L.B., Davis, G.E., Alcock, L.J. & Coppolo, J. (2002). Flexible mathematical thinking and multiple representations in middle school mathematics. Mediterranean Journal for Research in Mathematics Education, 1(2), 3761. Wittmann, E.Ch., & Mller, G.N. (19901992). Handbuch produktiver rechenbungen. Vols. 1 & 2 [Handbook of productive arithmetic exercises. Volume 1 & 2]. Dsseldorf und Stuttgart, Germany: Klett Verlag. Woodward, J. & Baxter, J. (1997). The effects of an innovative approach to mathematics on academically low achieving students in inclusive settings. Exceptional Children, 63(3), 373388. Woodward, J., Monroe, K. & Baxter, J. (2001). Enhancing student achievement on performance assessments in mathematics. Learning Disabilities Quarterly, 24 (Winter), 3346. Yackel, E. & Cobb, P. (1996). Classroom sociomathematical norms and intellectual autonomy. Journal for Research in Mathematics Education, 27, 458477.

Educational & Child Psychology Vol 24 No 2

27

Early markers for arithmetic difficulties

Pieter Stock, Annemie Desoete & Herbert Roeyers
Abstract
The development of six prenumerical abilities and the relationship with numerical facility and arithmetic reasoning was explored in a longitudinal study (n 108) for children in preschool (age 5 to 6) and Grade 1. Procedural knowledge about counting was assessed using accuracy in counting objects. Conceptual knowledge about counting was assessed by asking children to make judgments about unusual rightto-left counts and reflections on why a procedure works. To assess the logical knowledge children completed seriation, conservation and classification tasks. The representation of number size was assessed by subitising or dot and number comparison tasks. The results showed that about half of the variance in mathematical reasoning (as measured by a standardised test at age 6 to 7) was associated with the prenumerical skills of children in preschool. Especially conceptual counting knowledge and the seriation performances in preschool, and the procedural counting knowledge in grade 1 were important for early arithmetic reasoning. Numerical facility in grade 1 (as measured by a standardised test) was associated for a smaller proportion with the prenumerical skills of children. Moreover, it was possible to correctly classify in almost ninety percent performance in grade 1 based on the six prenumerical skills in preschool. Again especially conceptual knowledge about counting and seriation were suitable early markers for arithmetic difficulties.

VER THE past few years, mounting empirical evidence suggests that the earlier we recognise vulnerable young children, the more likely we will be to support their subsequent development and prevent learning difficulties from occurring later on (e.g. Coleman, Buysse & Neitzel, 2006). However, especially children at-risk for reading disabilities have been at the forefront of research. Surprisingly, few studies have been conducted to explore initial learning and the development of childrens arithmetic knowledge in the transition from preschool to the primary grades (Aunola, Leskinen, Lerkkannen & Nurmi, 2004; LeFevre et al., 2005). In this article, we focus on preschool predictors, to add to our psychological understanding of initial development arithmetic skills and to help teachers respond to young children who may be at-risk for arithmetic learning disabilities as early as possible. Several cognitive antecedents have been suggested as factors that play a role in the development of initial arithmetic performance and eventually as early marker for arith-

metic difficulties. In 1941, Piaget postulated that logical abilities, namely seriation, classification, and conservation are conditional to the development of arithmetic (Piaget & Szeminska, 1941). However, up till now, the debate on the value of the Piagetian abilities for arithmetic remains unsolved (for a review, see Loureno & Machado, 1996). Besides the logical abilities, reseachers focused on the importance of procedural and conditional counting abilities in the development of arithmetic performance. Finally, since Landerl, Bevan & Butterworth (2004) suggested that the core problem of arithmetic disabilities is a subitising deficit, it may be interesting to explore if subitising or magnitude comparison can be used as early markers for arithmetic difficulties.

Cognitive antecedents of initial arithmetic performance

Logical knowledge: seriation, classification and conservation Piaget and Szeminska (1941) specified the logical abilities that children progressively Educational & Child Psychology Vol 24 No 2 The British Psychological Society 2007

28

Early markers for arithmetic difficulties acquire to master the concept of number. Seriation is the ability to sort a number of objects based on differences on one or more dimensions while ignoring the similarities. In contrast, classification is the ability to sort those objects based on their similarities on one of more dimension, making abstraction of the differences. Once a child masters seriation and classification, it develops the knowledge that the number of objects in a collection only changes when one or more objects are added or removed. This concept in logical thinking is called conservation (Piaget & Szeminska, 1941). Since the publication of the work of Piaget, there has been a substantial literature dealing with Grade 1 (age 6 to 7) predictors of subsequent development of arithmetic skills. For example, Arlin (1981) found that whether a child has reached the stage of concrete operations or not was an important component of a childs academic readiness. According to Kingma (1983), the combination of conservation and seriation was a predictor for number-language. Moreover, seriation tasks in preschool predicted the number-line comprehension in first-grade children (Kingma, 1984), whereas classification was found not suitable as predictor of computational skills in grade 1 (Kingma, 1983). Since the publication of the work of Piaget, researchers have criticized his theories (Donaldson, 1978; for an overview, see Loureno & Machado, 1996). Although the work of Piaget remains an essential reference for practitioners working with children with arithmetic problems (Grgoire, 2005), the Piagetian skills are no longer considered as conditional but rather as precursors of arithmetic development. Nevertheless, in initial arithmetic, classification was found relevant to know the cardinal of a set (e.g. how many balls can you see on this picture?), whereas seriation was needed when dealing with ordinal numbers (e.g. circle the third ball from the beginning). In addition, a childs acquisition of conservation reflected his ability to think in a reversible way. This Educational & Child Psychology Vol 24 No 2 way of thinking was especially beneficial for solving reversal addition and subtraction tasks (e.g. 2 5). Counting knowledge During the 1980s there was considerable interest in exploring procedural and conceptual knowledge in preschoolers counting (Le Corre, Van de Walle, Brannon, & Carey, 2006; LeFevre et al., 2006). Procedural knowledge is defined as childrens ability to perform an arithmetic task, for example, when a child can successfully determine that there are five objects in an array (LeFevre et al., 2006). Procedural knowledge can be assessed using accuracy in counting objects. Conceptual knowledge reflects a childs understanding of why a procedure works or whether a procedure is legitimate (LeFevre et al., 2006). For counting, conceptual knowledge includes understanding five principles (Gallistel & Gelman,1992): (a) stable-order principle according to which the order of number words must be invariant across counted sets; (b) one-one principle according to which every number word can only be attributed to one counted object; (c) cardinality principle according to which the final number word pronounced in a count represents the numerosity of the set; (d) abstraction principle according to which any kind of object can be counted; (e) order-irrelevance principle according to which objects can be counted in any order. Conceptual knowledge can be assessed by asking children to make judgments about types of counts they made (as in this study) or types of counts modeled by an animated frog (LeFevre et al., 2006). Some advocates of the continuity hypothesis (e.g. Gallistel & Gelman, 1992) claimed that children have conceptual knowledge before their procedural counting skills are well developed. Other researchers reported the opposite (e.g. Frye, Braisby, Lowe, Maroudas & Nicholls, 1989). The timing of the two types of knowledge may, however, largely depend on the particular task or the 29

Pieter Stock et al. development may be iterative (Rittle-Johnson, Siegler & Wagner, 2001). It is obvious that early arithmetic strategies for addition and subtraction involve counting in the count all or sum strategy in which the child first counts each collection and then counts the combination of two collections starting from one (i.e., 2 5 1,2 . . . 1, 2, 3, 4, 5 . . . 1, 2, 3, 4, 5, 6, 7). As practice increases, older children use more effective back-up strategies, such as the count-on strategy where they count up from the first addend the number of times indicated by the second addend (i.e., 2 5 (2), 3, 4, 5, 6, 7) or the min strategy where they count up from the larger addend the number of times indicated by the smaller addend (i.e., 2 5 5 2 (5), 6, 7). It is assumed that the retrieval strategy (2 5 7, I know this by heart) is made possible by the learning and progressive strengthening of memory associations between problems and answers as a result of the repeated use of algorithms (Barrouillet & Lepine, 2005; Torbeyns, Verschaffel & Ghesquire, 2004). Geary and colleagues (Geary, Hoard, Byrd-Craven & DeSoto, 2004) found that children with specific arithmetic difficulties had problems in counting. Moreover, it has been suggested that childrens basic conceptual understanding of how to count objects and their knowledge of the order of numbers play an important role in arithmetic performance because they promote the automatic use of arithmetic related information, allowing attentional resources to be devoted to more complex arithmetic problem solving (Aunola, 2004). Magnitude comparison Representation of number size is also involved in numerical competence ( Jordan, Kaplan, Olah & Locuniak, 2006). This numerical skill is involved in subitising (rapid apprehension of small numerosity) and in magnitude comparison (i.e., knowing which digit in a pair is larger). There are some arguments that problems encountered by pupils with arithmetic learning disabilities may be 30 due to a deficit in this skill (Butterworth, 2003; Gersten, Jordan & Flojo, 2005; Stock, Desoete & Roeyers, 2006).

Initial arithmetic and arithmetic difficulties

Initial arithmetic can be seen as a broad domain of various computational skills. Dowker (2005) differentiated between two domains: mathematical reasoning and numerical facility. Children might have problems with arithmetic due to an insufficient knowledge base. In addition, by executing arithmetic problems repetitively basic number facts (e.g. 5 2 _) become automatic. Children with arithmetic disabilities often lack numerical facility and do not know basic number facts by heart.

Method
Aim and research questions The present study was designed to examine if we can predict the level of childrens arithmetic from their performance in preschool (age 5 to 6). The second purpose of the study was to test if children at-risk for arithmetic learning disabilities in grade 1 (age 6 to 7) can be detected by their prenumerical skills in preschool. Participants In a nonselected population a total of 108 (54 girls, 54 boys) children from seven randomly selected preschools in the Dutch speaking part of Belgium participated. Informed consent was obtained from all the parents. All children were tested in May of preschool (M 5.9 years, SD 4.0 months) and one year later in grade 1. The original nonselected sample consisted of all the children (n 108). For some analyses out of the original sample a smaller sample (n 67) of low achieving (LA) and at least moderate achieving (MA) Caucasian, native Dutch-speaking children, without a history of ADHD, sensory impairment, brain damage, chronic poor health, serious emotional or behavioural problems, or a Educational & Child Psychology Vol 24 No 2

Early markers for arithmetic difficulties poor educational background were selected. The analysis was not run on all of the children, because the difference in group sizes was too big. The LA-children (10 boys and 11 girls) performed below the 10th percentile on at least one standardised arithmetic test and the low arithmetic performance level was confirmed by the form teacher of the child. The MA-children (22 boys and 24 girls) scored above the 50th percentile on both arithmetic tests, had an age-appropriate performance level (at least level B; 60%) according to the form teacher and no signs of any learning disability. All children and parents were fluent native Dutch-speakers. A socio-economic status was derived from the total number years of scholarship of the parents (starting from the beginning of elementary school), with a mean of 14.64 months (SD 2.90) for mothers and 14.62 months (SD 2.69) for fathers. Measurement The Kortrijk Arithmetic Test Revision (Kortrijkse Rekentest Revision, KRT-R) (Baudonck et al., 2006) is a Belgian test of arithmetic reasoning which requires that children solve mental arithmetic (e.g. 19 7 . . .) and number knowledge tasks (e.g. one less than eight is . . .). The psychometric value of the KRT-R has been demonstrated on a sample of 3,246 Dutch-speaking children from grade 1 till 6. In the study the standardised total percentile based on Flanders norms was used. The Arithmetic Number Facts test (Tempo Test Rekenen, TTR; de Vos, 1992) is a numerical facility test which requires that children in grade 1 solve as many number fact problems as possible within two minutes (e.g. 3 2 . . .). The psychometric value has been demonstrated for Flanders on a sample of 10,059 children (Ghesquire & Ruijssenaars, 1994). TEDI-MATH (Grgoire et al., 2004 Flemish adaptation) is a test designed for the assessment of arithmetic disabilities from preschool till grade 3. The psychometric value has been demonstrated on a sample of Educational & Child Psychology Vol 24 No 2 550 Dutch speaking Belgian children and has been proven to be a well validated and reliable instrument (Desoete, 2006). Procedural knowledge of counting (see Appendix A subtest 1) was assessed using accuracy in counting numbers, counting forward to an upper bound (i.e., up to 6), counting forward from a lower bound (i.e., from 3), counting forward with an upper and lower bound (i.e., from 5 up to 9), counting forward by number (e.g. what number you get when you count five numbers on from eight), counting backward given a starting number (i.e., 7), and counting by step (i.e., by 2) from it. One point was given for a correct answer. A sum score was constructed (maximum: 14 points). Cronbachs alpha was .73. Standardised percentiles were used. Conceptual knowledge of counting (see Appendix A subtest 2) was assessed with judgments about the validity of counting procedures. Children had to judge the counting of linear and random patterns of drawings and counters. To assess the abstraction principle, children had to count different kind of objects who were presented in a heap. Furthermore, a child who counted a set of objects was asked how many objects are there in total?, or how many objects are there if you start counting with the leftmost object in the array. When children had to count again to answer they did not gain any points, as this was considered to represent good procedural knowledge but a lack of understanding of the counting principles of Gelman & Gallistel (1978). One point was given for a correct answer with a correct motivation. A sum score was constructed (maximum: 13 points). Cronbachs alpha was .85. Standardised percentiles were used. Three logical operations were assessed (see Appendix A subtest 3). Children had to seriate numbers (e.g. Sort the cards from the one with fewer trees to the one with the most trees; maximum: 3 points). Cronbachs alpha for the subtest was .68. Children had to make groups of cards in order to assess the classification of numbers (e.g. Make groups with the cards that go together; maximum: 3 points). 31

Pieter Stock et al. Cronbachs alpha was .73. Counters were used to test the conservation of numbers (e.g. Do you have more counters than me? Do I have more counters than you? Or do we have the same number of counters? And why is this?; maximum: 4 points). One point was given for a correct answer with a correct logical motivation. Cronbachs alpha was .85. In the study standardised total percentiles based on Flanders norms were used. Magnitude comparison was assessed by comparison dot sets (6 items) and numbers (12 items). Preschool children and first graders were asked were they saw most dots. The first graders were also asked which number was closed to a certain target number (see Appendix subtest 4). One point was given for a correct answer. A sum score was constructed (maximum: 6 points in preschool and 18 points in Grade 1). Cronbachs alpha was .79. Standardised percentiles based on Flanders norms were used. Design and procedure All preschool children were assessed individually with the TEDI-MATH (Grgoire, Van Nieuwenhoven & Nol, 2004) in a separate and quiet room by a trained tester. In addition regular preschool teachers completed a teacher survey in the same period. One year later, in grade 1 of elementary school the children completed the KRT-R (Baudonck et al., 2006) and the TTR (de Vos, 1992) on the same day for about one hour in total, supervised by a trained tester. The children were also assessed individually in the same week by the same trained tester with TEDI-MATH. In addition regular elementary school teachers completed an elementary school teacher survey in the same period. The testers, all psychologists or therapists skilled in learning disabilities, received training in the assessment and interpretation of arithmetic difficulties. regression, two regression analyses were conducted in the nonselected sample to evaluate how well the prenumerical arithmetic abilities predicted arithmetic reasoning and numerical facility in grade 1. Six prenumerical abilities at age five to six were included simultaneously as predictor variables: procedural counting knowledge, conceptual counting knowledge, seriation, classification, conservation and magnitude comparison. The univariate F-tests were Bonferroni-adjusted to control for the number of comparisons. With six comparisons an adjusted alpha for each comparison is p .008. The linear combination of the prenumerical abilities was significantly related to arithmetic reasoning assessed in grade 1 (at age 6 to 7) with KRT, F (6, 107) 18.659, p .0005. R2 was .489. Conceptual counting knowledge and seriation were especially beneficial for beginning arithmetic reasoning (see Table 1). The second multiple regression analyses pointed out that the linear combination of prenumerical abilities at age five to six was also significantly related to numerical facility one year later (at age six to seven) assessed with TTR, F (6, 107) 9.655, p .0005. R2 was .327. Conceptual knowledge and seriation in preschool seem to be especially beneficial for both arithmetic reasoning and numerical facility in grade l (see Table 1). Concurrent assessment Two additional regression analyses were conducted to evaluate how well the concurrent prenumerical abilities in grade 1 (age 6 to 7) predicted arithmetic reasoning and numerical facility in the same grade. The same six prenumerical abilities were included simultaneously as predictor variables. The linear combination of prenumerical abilities was significantly related to mathematical reasoning in the same grade, F(6, 107) 6.067, p .0005. R2 was .221 (see Table 2). The linear combination of prenumerical abilities was also significantly related to numerical facility in the same grade, and

Results
Prospective assessment Since all variables were normally distributed and did meet the assumptions for multiple 32

Educational & Child Psychology Vol 24 No 2

Early markers for arithmetic difficulties Arithmetical reasoning (at age 6 to 7) Prenumerical skills at age 5 to 6 Constant Proc. Knowledge Conc. Knowledge Seriation Classification Conservation Subitising Unstandardised coefficients 1.95 .09 .14 .22 .07 .13 .02 .18 .31 .35 .13 .08 .02 t Numerical facility (at age 6 to 7) Unstandardised coefficients .60 .01 .11 .10 .05 .05 .05 .03 .39 .27 .16 .06 .12 t

.21 1.89 3.64 4.39 1.74 1.08 .29

.83 .06 .008* .008* .08 .29 .77

.09 .26 3.97 2.86 1.81 .64 1.51

.93 .80 .008* .008* .08 .53 .13

Table 1: Prediction of arithmetical reasoning and numerical facility in grade 1 (age 6 to 7) from prenumerical skills in preschool (age 5 to 6). Note: Proc. Knowledge procedural knowledge, Conc. Knowledge conceptual knowledge *p .008

Arithmetical reasoning (at age 6 to 7) Prenumerical skills at age 5 to 6 Constant Proc. knowledge Conc. knowledge Seriation Classification Conservation Subitising Unstandardised coefficients 9.33 .19 .02 .06 .09 .01 .16 .32 .03 .11 .19 .02 .17 t

Numerical facility (at age 6 to 7) Unstandardised coefficients .50 .07 .01 .04 .06 .02 .10 .19 .02 .11 .20 .07 .18 t

1.47 3.39 .33 1.10 2.13 .19 1.78

.15 .008* .74 .27 .04 .85 .08

.12 1.85 .25 1.02 2.09 .65 1.82

.90 .07 .81 .21 .04 .51 .07

Table 2: Prediction of arithmetic reasoning skills and numerical facility from prenumerical skills at age 6 to 7. Note: Proc. Knowledge procedural knowledge, Conc. Knowledge conceptual knowledge *p .008

F(6, 107) 3.056, p .009. R2 was .103 (see Table 2). To answer the second research question, and to investigate if children at-risk for arithmetic learning disabilities in grade 1 (age 6 to 7) can be detected by their prenumerical skills in preschool one year earlier a discriminant analysis on the sample (n 67) of low achieving (LA) and at least moderate achieving (MA) Caucasian, native DutchEducational & Child Psychology Vol 24 No 2

speaking children. Wilks lambda was significant, .51, 2 (6, N 67) 41.54, p .0005, indicating that overall the predictors differentiated among the low achieving and the at least moderate achieving group. In Table 3 the standardised weights of the predictors are presented. Based on these coefficients, conceptual knowledge of counting and seriation demonstrate the strongest relationships with initial arithmetic. 33

Pieter Stock et al. The mean prenumerical scores on the discriminant function were consistent with this interpretation. The at least moderate achieving group did better on conditional knowledge and seriation than the low achieving group. Based on the scores for these six predictors, 86.6 per cent was classified correctly into the low achieving or at least moderate achieving group, whereas 83.6 per cent of the cross-validated grouped cases were classified correctly. Based on the six prenumerical scores 95.7 per cent (or 44 out of the 46 children ) of the at least moderate achievers and 66.7 per cent of low achieving children (or 14 out of 21 children) were classified correctly. Finally, correlations were computed between the preschool results and the results in grade 1 (see Table 4). From Table 4 we can conclude a significant relationship between procedural and conceptual knowledge in preschool but not in Grade 1. Significant correlations were found between preschool and grade 1 scores for procedural counting skills, seriation and conservation. minants of childrens cognitive outcomes and school readiness skills. Nevertheless, currently most arithmetic learning disabilities are not detected until Grade 3 (Desoete, Roeyers & De Clercq, 2004). In this article, we focussed on the prediction of the level of childrens arithmetic in grade 1 from their performance in preschool, to add to our understanding of initial development arithmetic skills and to help teachers focus on early markers for arithmetic difficulties. About half of the variance in arithmetic reasoning skills in first grade can be predicted by assessing six prenumerical skills in preschool. Only about one fifth of the variance in arithmetic reasoning skills can be predicted by assessing the same skills in grade 1. Three markers showed significant contributions: conceptual counting knowledge and seriation in preschool and procedural counting knowledge in grade 1. The current results also suggest that more then one third of the variance in numerical facility in grade 1 can be predicted by assessing the prenumerical skills in preschool. Only about one tenth of the variance in fact retrieval skills can be predicted by assessing the same skills in grade 1, although no individual markers were significant. The analysis of low and at least average performing first graders was consistent with

Discussion and conclusions

Over the past few years, increasing attention has been paid by policymakers to the idea of early childhood programs as important deter-

Prenumerical skills at age 5 to 6

Procedural counting knowledge Conceptual counting knowledge Classification Seriation Conservation Magnitude comparison

Standardised canonical discriminant function coefficients .20 .62 .08 .64 .06 .02

Mean prenumerical score (SD) for LA (n 21) 39.19 (19.89) 24.67 (23.60) 56.86 (29.79) 62.55 (24.83) 75.49 (6.24) 55.91 (21.23)

Mean prenumerical score (SD) for MA (n 46) 69.28 (24.52) 63.78 (29.68) 75.59 (25.42) 89.65 (16.66) 77.97 (9.66) 62.91 (24.82)

Table 3: Standardised canonical discriminant function coefficients of the predictors at age 5 to 6 for arithmetic and the mean values for LA and MA at age 6 to 7. Note: LA low achievers at age 6 to 7, MA at least average achievers at age 6 to 7 34 Educational & Child Psychology Vol 24 No 2

Early markers for arithmetic difficulties Preschool skills (age 5 to 6) Preschool skills CK Cl Se Co Su PK .60* .35* .34* .07 .26* CK .13 .33* .05 .05 Cl .28* .09 .07 Se . 36* .15 Co .02 Su

Grade 1 skills (age 6 to 7) Grade 1 skills CK Cl Se Co Su PK .16 .17 .28* .14 .27* CK .16 .17 .12 .19 Cl .24 .08 .02 Se .19 .17 Co .07 Su

Preschool skills (age 5 to 6) Grade 1 skills PK CK Cl Se Co Su PK .37* .16 .17 .29* .14 .27* CK .32* .19 .17 .16 .12 .19 Cl .32* .31* .23 .24 .08 .02 Se .25 .00 .05 .26* .19 .17 Co .11 .09 .15 .12 . 28* .07 Su .07 .03 . 34* .14 .02 .00

Table 4: Intercorrelation matrix. Note : PK procedural knowledge of counting, CK conceptual knowledge of counting, Cl classification, Se seriation, Co conservation, Su subitizing *p .008 the previous analysis. Two thirds of the lowerelementary school children classified as at risk were classified correctly based on their prenumerical performances in preschool at the age five to six. Especially conceptual knowledge and seriation were suitable predictors of at-risk arithmetic performance. In line with research of Kingma (1983) the Piagetian model had some value added since children-at risk in grade 1 had lower scores seriation tasks, compared with at least average performing peers. In line with Geary and Hoard (2005) children at-risk also had less developed counting knowledge and especially lacked conceptual counting knowledge. Moreover, in line with the idea of Rittle-Johnson and colleagues Educational & Child Psychology Vol 24 No 2 (2001) arguing that procedural knowledge and conceptual knowledge develop iteratively, we found weak connections between those components. These results have as implication that in young children we should not only assess how accurately children can count (procedural knowledge) but also how they master the counting principles of Gallistel & Gelman (1992). It might be interesting to see if a controlled intervention focusing on conceptual counting knowledge and seriation skills in children at-risk can prevent learning difficulties from occurring later in these vulnerable children. In our dataset we found no significant con35

Pieter Stock et al. tribution of subitising skills in the prediction of early arithmetic scores. Nevertheless, in line with Landerl, Bevan & Butterworth (2004) children at-risk did worse on magnitude comparison tasks than at least average performing peers. It should be acknowledged that sample size is a limitation of the present study. Obviously sample size is not a problem for significant correlations or regressions. However, when analyses were not significant, a risk of type 2- or -error (concluding from the cohort that there were no differences although in reality there were differences in the population) can not be excluded. These results should be interpreted with care since the analyses are correlational in nature and numerosity skills might involve different cognitive skills and might be age-, and intelligence-dependent and still maturing. Also more research on other variables such as home and school environment, parental involvement but also the facets of numerical competence (such as the knowledge of the numerical system and the arithmetical operations) is needed. Such longitudinal studies with a multilevel design are currently being prepared. In addition, in our sample many children (about 20 per cent) could be classified, based on their test scores as pupils at-risk. All of those children had below critical cut-off scores (they scored pc 10 on at least one arithmetics test) in grade 1. It would be interesting to investigate if these deficits persist across two successive grades and if all these children will demonstrate a severe and resistant learning disability or if some of them will become mildly delayed arithmetic problem solvers having difficulties that are not related to learning disabilities. Research with larger groups of poor arithmetic performers and children with learning disabilities followed for longer periode of time during elementary school seems therefore indicated, to replicate the results of this study. In addition, not only the prenumerical but also the facets of numerical competence (such as the knowledge of the numerical system and the arithmetical operations) still need a full explanation. Such studies are currently being conducted. Summarising, our dataset support that we should assess procedural counting knowledge, the ability to seriate in preschool, and also the ability to distinguish essential form inessential counting characteristics (or the conceptual counting knowledge of young children). Especially the inability to do such things in preschool (at the age 5 to 6) may be a marker for later arithmetic disabilities.

Address for correspondence

Department of Clinical Psychology, Research Group Developmental Disorders, University of Ghent, Henri Dunantlaan 2, B-9000 Gent, Belgium E-mail: anne.desoete@Ugent.be

References
Arlin, P.K. (1981). Piagetian tasks as predictors of reading and math readiness in grades-K-1. Journal of Educational Psychology, 73, 712721. Aunola, K., Leskinen, E., Lerkkanen, M. & Nurmi, J. (2004). Developmental dynamics of math performance to Grade 2. Journal of Educational Psychology, 96, 699713. Barrouillet, P. & Lepine, R. (2005). Working memory and childrens use of retrieval to solve addition problems. Journal of Experimental Child Psychology, 91, 183204. Baudonck, M., Debusschere, A., Dewulf, B., Samyn, F., Vercaemst, V. & Desoete, A. (2006). De Kortrijkse Rekentest Revision KRT-R. [The Kortrijk Arithmetic Test Revision KRT-R]. Kortrijk: CAR Overleie. Butterworth, B. (2003). Dyscalculia Screener. London: NFER Nelson Publishing Company Ltd. Coleman, M.R., Buysse, V. & Neitzel, J. (2006). Recognition and response. An Early Intervening system for Young children at-risk for Learning Disabilities. Research synthesis and recommendations. UNIC FPG Child Development Institute. Desoete, A. (2006). Validiteitsonderzoek met de TEDIMATH9 [Validity research on the TEDI-MATH]. Diagnostiekwijzer], 9. Desoete, A., Roeyers, H. & De Clercq, A. (2004). Children with mathematics learning disablities in Belgium. Journal of learning disabilities, 37, 5061. De Vos, T. (1992). TTR. Tempotest rekenen [Arithmetic number fact test]. Lisse: Swets & Zeitlinger.

36

Educational & Child Psychology Vol 24 No 2

Early markers for arithmetic difficulties

Donaldson, M.C. (1978). Childrens Minds. London: Fontana. Dowker, A. (2005). Individual Differences in Arithmetic. Implications for Psychology, Neuroscience and Education. Hove, UK: Psychology Press. Frye, D., Braisby, N., Lowe, J. Maroudas, C. & Nicholls, J. (1989). Young childrens understanding of counting and cardinality. Child Development, 60, 11581171. Gallistel, C. & Gelman, R. (1992). Preverbal and verbal counting and computation. Cognition, 44, 4374. Geary, D.C. & Hoard, M.K. (2005). Learning disabilities in arithmetic and mathematics: Theoretical and empirical perspectives. In J.I.D. Campbell (Ed.), Handbook of Mathematical Cognition (pp. 253268). New York: Psychology Press. Geary, D.C, Hoard, M.K., Byrd-Craven, J. & DeSoto, M.C. (2004). Strategy choices in simple and complex addition: Contributions of working memory and counting knowledge for children with mathematical disability. Journal of Experimental Child Psychology, 88, 121151. Gelman, R. & Galistel, C.R. (1978). The childs understanding of number. Cambridge, MA: Harvard University Press. Gersten, R., Jordan, N.C. & Flojo, J.E. (2005). Early identification and intervention for students with mathematics difficulties. Journal of Learning Disabilities, 38, 293304. Ghesquire, P. & Ruijssenaars, A. (1994). Vlaamse normen voor studietoetsen Rekenen en technisch lezen lager onderwijs. [Flemisch norms for schooltests on Mathematics and Technical Reading in elementary school]. Leuven: K.U.L.C.S.B.O. Grgoire, J. (2005). Dveloppement logique et comptences arithmtiques. Le modle piagtien estil toujours actuel? In M. Crahay, L. Verschaffel, E. De Corte, & J. Grgoire (Eds.) Enseignement et Apprentissage des Mathmatiques (pp. 5777). Brussels: De Boeck. Grgoire, J., Noel, M. & Van Nieuwenhoven (2004). TEDI-MATH. TEMA (Flamish adaptation: A. Desoete, H. Roeyers & M. Schittekatte): Brussel/ Harcourt: Antwerpen. Kingma, J. (1983). Seriation, Correspondence, and transitivity. Journal of Educational Psychology, 75, 763771. Kingma, J. (1984). Traditional intelligence, Piagetian tasks, and initial arithemtic in kindergarten and primary-school grade-one. Jounal of Genetic Psychology, 145, 4960. Landerl, K., Bevan, A. & Butterworth, B. (2004). Developmental dyscalculia and basic numerical capacities: a study of 89-year-old students. Cognition, 93, 99125. Le Corre, M., Van de Walle, G., Brannon, E.M. & Carey, S. (2006). Re-visiting the competence/ performance debate in the acquisition of the counting principle. Cognitive Psychology, 52, 130169. LeFevre, J.A., Smith-Chant, B.L., Fast, L., Skwarchuk, S.L., Sargla, E., Arnup, J.S., Penner-Wilger, M., Bisanz, J. & Kamawar, D. (2006). What counts as knowing? The development of conceptual and procedural knowledge of counting from kindergarten through Grade 2. Journal of Experimental Child Psychology, 93, 285303. Loureno, O. & Machado, A. (1996). In defense of Piagets theory: A reply to 10 common criticisms. Psychological Review, 103, 143164. Jordan, N.C., Kaplan, D., Olah, L.N. & Locuniak, M.N. (2006). Numer sense growth in kindergarten: A longitudinal investigation of children at risk for mathematics difficulties. Child Development, 77, 153175. Piaget, J. & Szeminska, A. (1941). La gense du nombre chez lenfant. Neuchtel: Delachaux et Niestl. (7e dition, 1991). Rittle-Johnson, R., Siegler, R.S. & Wagner, R.S. (2001). Developing conceptual understanding and procedural skill in mathematics: An iterative process. Journal of Educational Psychology, 93, 346362. Stock, P., Desoete, A. & Roeyers, H. (2006). Focussing on mathematical disabilities: a search for definition, classisfication and assessment (pp. 2962). In Soren V. Randall (Ed.) Learning Disabilities New Research Nova Science: Hauppage, NY. Torbeyns, J., Verschaffel, L. & Ghesquire, P. (2004). Strategic aspects of simple addition and subtraction: the influence of mathematical ability. Learning and Instruction, 14, 177195.

Educational & Child Psychology Vol 24 No 2

37

Pieter Stock et al.

Appendix A: Subtsests and examples of test-items of the TEDI-MATH

Subtest 1. Procedural knowledge (13 items, max 14 points) Content and example of item Counting as far as possible (without help 2 points; with starting help 1 point) Counting forward to an upper bound (up to 9 ) Counting forward to an upper bound (up to 6 ) Counting forward from a lower bound (from 3 ) Counting forward from a lower bound (from 7 ) Counting forward from a lower bound to an upper bound (from 5 up to 9) Counting forward from a lower bound to an upper bound (from 4 up to 8) Counting forward (5 steps starting at 8 ) Counting forward (6 steps starting at 9 ) Count backward (from 7 ) Count backward (from 15 ) Count by step (by 2) Count by step (by 10) Counting linear pattern of items (how many rabbits are there? How many rabbits are there in total?, and how many rabbits are there if you start counting with this one. Why?). Max 3 points Counting linear pattern of items (how many lions are there? How many lions are there in total?, and how many lions have I hidden. Why? ). Max 3 point Counting random pattern of items (how many turtles are there? How many turtles are there in total ? ). Max 2 points Counting random pattern of items (how many sharks are there? How many sharks are there in total? ). Max 2 points Counting a heterogeneous set of items (how many animals are there in total? ). Max 1 point Understanding of the cardinal (Can you put as much counters as there are on this paper? ) Max 1 point Understanding of the cardinal (How many hat do I have in my hand, when all the snowman had a hat on this picture?) Max 1 point Seriation: Sort the cards form the one with fewer trees to the one with the most trees. forgot this card and Can you put this card in the correct order? and I give you carts with numbers now. Do the same as with the trees. Start with the cart with smallest number and go on with the other carts Max 3 points Classification: Make groups with the cards that go together. Can you put them together in another way? and Make groups with these cards that go together Max 3 points Conservation: 4 items: Do you have more counters than me? Do I have more counters than you? Or do we have the same number of counters? Why? Max 4 points

2. Conceptual knowledge (7 items, max 13 points)

3. Logical operations on numbers (9 items, max 9 points)

(Continued ) 38 Educational & Child Psychology Vol 24 No 2

Early markers for arithmetic difficulties Subtest 4. Estimation of the size (18 items, 18 points) Content and example of item Comparison of dot sets (subitising): in preschool and grade 1

Where do you have most dots? Here or here? Show me. 1 dot versus 3 dots 3 dots versus 2 dots 4 dots versus 6 dots 7 dots versus 2 dots 7 dots versus 12 dots 15 dots versus 8 dots (see example) max 6 points Estimation of size: Comparison of distance between numbers (in grade 1).

Target number is 4. What number is closed to this (5 or 9)? Target number is 2. What number is closed to this (7 or 4)? Target number is 8. What number is closed to this (7 or 3)? Target number is 9. What number is closed to this (5 or 7)? Target number is 7. What number is closed to this (3 or 9)? Target number is 3. What number is closed to this (8 or 2)? (see ex. above) Target number is 5. What number is closed to this (3 or 9)? Target number is 6. What number is closed to this (8 or 1)? Target number is 32. What number is closed to this (59 or 24)? Target number is 48. What number is closed to this (57 or 15)? Target number is 61. What number is closed to this (53 or 99)? Target number is 79. What number is closed to this (48 or 86)? max 12 points.

Educational & Child Psychology Vol 24 No 2

39

Assessing and teaching children who have difficulty learning arithmetic

Ann Gervasoni & Peter Sullivan
Abstract
Arithmetic difficulties have long captured the attention of teachers and researchers, but intervention programs for assisting children are seldom successful for all. Recent Australian research suggests that this is because we have failed to recognise the complexity of arithmetic difficulties. Analysis of some 30,000 oneon-one clinical interviews conducted over three years during the Early Numeracy Research Project provided rich data for charting the pathway of young childrens number learning in four domains (Counting, Place Value, Addition and Subtraction Strategies, and Multiplication and Division Strategies), and for identifying children who were having difficulty. We describe such children as being vulnerable or at risk of not being able to take advantage of everyday classroom experiences. The data show that the combinations of domains in which children were vulnerable were diverse, and suggest that there is no single formula for describing children who are vulnerable in number learning, or for describing the instructional needs of students. Indeed, children have learning needs that call for teachers to make individual decisions about the instructional approach for each child. Further, the diversity of childrens mathematical knowledge in the four domains suggests that knowledge in any one domain is not necessarily prerequisite for knowledge construction in another domain. This finding has implications for both intervention programs and for the way in which school mathematics is introduced to children. It seems likely that children may benefit from concurrent learning opportunities in all number domains, and that experiences in one domain should not be delayed until a level of mathematical knowledge is constructed in another domain.

SSISTING CHILDREN who have difficulty learning arithmetic is not straightforward. This conclusion became apparent to us after we analysed the number knowledge of several thousand Australian students during the Early Numeracy Research Project (ENRP, Clarke, Cheeseman, Gervasoni, Gronn, Horne, McDonough, Montgomery, Roche, Sullivan, Clarke & Rowley, 2002). What we discovered is that the number knowledge of children who are vulnerable is much more diverse that we had anticipated. This finding has important implications for the way classroom teachers and specialist teachers assess students and plan and implement curriculum and instruction that accelerates vulnerable childrens number learning. First, teachers need assessment tools that are responsive to the diversity of childrens number knowledge and that enable teachers to determine the extent of childrens current mathematical knowledge 40

in the number domains. Second, teachers need to provide effective learning experiences that can be easily customised to respond to childrens individual learning needs. This article explores these aspects with the aim of providing advice for teachers. In so doing, we draw on findings from the ENRP that researched mathematics learning in the first three years of schooling (ages 5, 6 and 7years). Arithmetic difficulties may not be straightforward, but once we recognise this fact, then the way forward to helping students is clearer.

Vulnerability in arithmetic learning

The argument presented in this paper is based on the assumption that it is important for school communities to identify children who, as emerging school mathematicians and after one year at school, have not thrived in the school environment, and to provide these children with the type of learning opportuniEducational & Child Psychology Vol 24 No 2 The British Psychological Society 2007

Assessing and teaching children who have difficulty learning arithmetic ties and experiences that will enable them to thrive and extend their mathematical understanding. Thus, the notion of intervention early in schooling is important. The perspective that underpinned our research was that those children who have not thrived have not yet received the type of experiences and opportunities necessary for them to construct the mathematical understandings needed to successfully engage with the school mathematics curriculum, or to make sense of this curriculum. As a result, these children are vulnerable and possibly at risk of poor learning outcomes. The term vulnerable is widely used in population studies (e.g. Hart, Brinkman & Blackmore, 2003), and refers to children whose environments include risk factors that may lead to poor developmental outcomes. The challenge remains for teachers and school communities to create learning environments and design mathematics instruction that enables vulnerable childrens mathematics learning to flourish. 1998; Bobis & Gould, 1999);
reflect, where possible, the structure of

mathematics;
allow the mathematical knowledge and

Using growth points to describe childrens number learning

The data presented in this article were collected from 1999 to 2001 as part of the ENRP (Clarke et al., 2002), the fundamental building block of which was a set of growth points (see Appendix A) that describe key stages in the learning of various domains of mathematics. The principles underlying the construction of the growth points were that they would: describe the development of childrens mathematical knowledge and understanding in the first three years of school, through highlighting important ideas in early mathematics understanding in a form and language that is useful for teachers; reflect the findings of relevant international and local research in mathematics education (e.g. Steffe, von Glasersfeld, Richards & Cobb, 1983; Steffe, Cobb & von Glasersfeld, 1988; Fuson, 1992; Boulton-Lewis, 1996; Mulligan & Mitchelmore, 1996; Wright, Educational & Child Psychology Vol 24 No 2

understanding of individuals and groups to be described; and enable identification of those students who may benefit from additional assistance. The growth points developed formed a framework for describing childrens development in four Number domains (Counting, Place Value, Addition and Subtraction Strategies, and Multiplication and Division Strategies), three Measurement domains (Length, Mass and Time), and two Space/Geometry domains (Properties of Shape, and Visualisation and Orientation). The processes for validating the growth points, the interview items and the comparative achievement of students in project and references schools are described in full in Clarke et al. (2002). To illustrate the nature of the growth points, the following are the points for the Addition and Subtraction Strategies domain: 1. Counts all to find the total of two collections. 2. Counts on from one number to find the total of two collections. 3. Given subtraction situations, chooses appropriately from strategies including count back, count down to & count up from. 4. Uses basic strategies for solving addition and subtraction problems (doubles, commutativity, adding 10, tens facts, other known facts). 5. Uses derived strategies for solving addition and subtraction problems (near doubles, adding 9, build to next ten, fact families, intuitive strategies). 6. Given a range of tasks (including multidigit numbers), can solve them mentally, using the appropriate strategies and a clear understanding of key concepts. Each growth point represents substantial expansion in mathematical knowledge. In discussions with teachers, growth points are described as key stepping stones along paths 41

Ann Gervasoni & Peter Sullivan to mathematical understanding (Clarke, 2001). It is not claimed that every student passes all growth points along the way, nor should the growth points be regarded as discrete. However, the order of the growth points provides a guide to the possible trajectory (Cobb & McClain, 1999) of childrens learning. In a similar way to that described by Owens & Gould (1999) in the Count Me In Too project: the order is more or less the order in which strategies are likely to emerge and be used by children (p. 4). In summary, the framework of growth points help teachers to: understand a possible trajectory for describing childrens learning; describe (following assessment) the mathematics achievements of each child; identify any children who may be vulnerable in a given domain; describe a zone of proximal development in each domain so as to customise planning and instruction; and identify the diversity of mathematical knowledge in a class. than 20,000 assessment interviews. In the ENRP, the assessment interview was conducted with every child in project schools at the beginning and end of each school year (March and November respectively). The March interviews enabled teachers to determine the growth points reached by children at the beginning of the school year, to customise curriculum planning and instruction accordingly, and to identify any children who were vulnerable in a particular domain. The November interview enabled growth over the year to be determined for each child. The Early Numeracy Interview takes between 30 to 40 minutes per student and during the ENRP was conducted by the regular classroom teacher. The full text of the interview involves around 60 tasks, although no child is presented with all of these. Given success with a particular task, the interviewer continues with the next task in the given mathematical domain (e.g. Addition and Subtraction Strategies) for as long as the child is successful. If the child cannot perform a particular task correctly, the interviewer moves on to the next domain or moves into a detour, designed to elaborate more clearly any difficulty a child might be having in a particular area. Figure 1 shows the first two questions from the Addition and Subtraction Strategies section of the interview. Words in italics are instructions for the interviewer, and the symbols and arrows indicate which question to ask next. Teachers gain insights into childrens mathematical knowledge from their responses to the interview tasks. For example, a childs response to Question 18 enables the teacher to determine whether a child can count-on or use reasoning strategies or known facts to solve the problem. Children who are successful move on to Question 19, while the others continue with 18(d) to determine whether they can use a count-all strategy to solve the problem when all items may be seen. Question 19 provides information about childrens arithmetic strategies in a simple subtraction context in which models are not initially provided. Children who are not sucEducational & Child Psychology Vol 24 No 2

The clinical assessment interview based on the growth points

The framework of growth points formed the structure for the creation of the assessment items used as the basis of a clinical interview, known as the Early Numeracy Interview (Department of Education Employment and Training, 2001). Clinical interviews enable the teacher to observe children as they solve problems to determine the strategies used and any misconceptions, and to probe childrens mathematical understanding through thoughtful questioning (Ginsburg, 1997; Wright, Martland & Stafford, 2000). The insights gained through this form of assessment inform teachers about the particular instructional needs of each student more powerfully than scores from traditional pencil and paper tests. Both the Early Numeracy Interview and the framework of growth points were refined throughout the first two years of the ENRP in response to data collected from more 42

Assessing and teaching children who have difficulty learning arithmetic

SECTION C: STRATEGIES FOR ADDITION & SUBTRACTION Equipment: green teddies, ice-cream lid, white, blue and yellow problem cards. 18) Counting On Place 9 green teddies on the table. a) Please get four green teddies for me. b) I have nine green teddies here (show the child the nine teddies, and then screen the nine teddies with the ice-cream lid). Thats nine teddies hiding here and four teddies here (point to the groups). c) Tell me how many teddies we have altogether. . . . Please explain how you worked it out. Q19 If the child does not say 13, please go to part (d) d) (Remove the lid). Please tell me how many there are altogether. 19) Counting Back For this question you need to listen to a story. a) Imagine you have 8 little biscuits in your play lunch and you eat 3. How many do you have left? . . . How did you work that out? Q20 part (b)

b) Could you use your fingers to help you to work it out? (its fine to repeat the question, but no further prompts please). Section D

Figure 1: Excerpt from the addition and subtraction strategies section of the ENRP assessment interview

cessful with this task are provided with a prompt to use their fingers as a model. If children are still unable to use a successful strategy, the interviewer moves to the questions on Multiplication and Division Strategies. Successful students move on to a series of more complex tasks in the Addition and Subtraction Strategies domain, with the final task involving the subtraction of 3-digit numbers. The Early Numeracy Interview provided teachers participating in the ENRP with insights about childrens mathematical knowledge that they reported might otherwise not have been forthcoming (Clarke, 2001). The project found that teachers were able to use this information to plan instruction that would provide students with the best possible opportunities to extend their mathematical understanding.

Using growth points to identify vulnerability in number learning for Grade 1 and 2 children (6- and 7 year-olds)
Participants in the Early Numeracy Research Project in the year 2000 included 1497 Grade 1 children (6-year-olds) and 1538 Grade 2 children (7-year-olds) from 34 ENRP trial schools. These schools included Government, Catholic and Independent schools from across the State of Victoria in Australia that were widely representative of the Victorian population. In order to identify children who are vulnerable in their number learning, sometimes a line is drawn across a distribution of test scores, and children below the line are deemed at risk (e.g. Ginsburg, 1997; Woodward & Baxter, 1997). In the ENRP the decision on where to draw the line was made on the basis of on the way growth points (Gervasoni, 2000). The on the way growth point in any domain is used to identify children who have constructed the mathematical knowledge 43

Educational & Child Psychology Vol 24 No 2

Ann Gervasoni & Peter Sullivan that underpins the initial mathematics curriculum in a particular domain and grade level, and who are likely to continue to learn successfully. Not yet reaching the on the way growth point in a particular domain is an indicator that children may be vulnerable in that domain and may benefit from support to help them reach the on the way growth point. The development of appropriate on the way growth points for Grade 1 (the second year of school) and Grade 2 children was guided by three data sources: the ENRP growth point distributions for Grade 1 and Grade 2 children in March 2000; the Victorian Curriculum and Standards Framework II (Board of Studies, 2000) for Grade 1 and Grade 2; and the opinions of ENRP Grade 1 and Grade 2 classroom teachers (see Gervasoni, 2004, for elaboration of this). The analyses and synthesis of these data resulted in the following on the way growth points being established for children at the beginning of Grade 1 in Counting, Place Value, Addition and Subtraction Strategies, and Multiplication and Division Strategies respectively: can count collections of at least 20 items; can read, write, order and interpret one digit numbers; counts all to find the total of two collections; counts group items individually in multiplicative tasks. In other words, being able to count a collection of 20 objects is one piece of evidence to suggest that children are ready for conventional Grade 1 experiences. If they cannot count 20 objects, they may be vulnerable and unable to take advantage of the activities they will experience during the year. The on the way growth points established for children at the start of Grade 2 were: counts forwards and backwards beyond 109 from any number; can read, write, order and interpret two digit numbers; counts on from one number to find the total of two collections; uses grouping to solve multiplicative tasks. In other words, a student who can read, write, order and interpret two digit numbers will benefit from the usual Grade 2 numeracy experiences in the Place Value domain. Students who cannot do so are vulnerable and may not be able to take advantage of those same experiences. Even though these respective on the way growth points are to some extend arbitrary, they are based on considered judgments and they form the basis of the data reported in the following section. The domains and combinations of domains for which children were vulnerable When considering the domains and combinations of domains for which children were vulnerable, there are three aspects of interest: the number of students who are vulnerable in the respective domains; the extent to which children are vulnerable in multiple domains; and the nature of vulnerable students achievements in domains in which they are not vulnerable. The analysis of ENRP participants growth point data found that there were 576 (out of 1497) Grade 1 children in 2000, or 38 per cent of the group, who were vulnerable in at least one number domain. The number of Grade 2 children in 2000 who were vulnerable in at least one number domain was 659 (n 1538), or 43 per cent of the Grade 2 children. The number of children who were vulnerable in each number domain is shown in Table 1. In Grade 1 (6-year-olds), the number of students vulnerable in any domain according to the earlier definition is small but large enough to require considered attention. The number of students vulnerable in Multiplication and Division Strategies is larger. These students were unable to solve a multiplicative task even by counting the items individually. While this might be a characteristic of a curriculum in the first year of school that avoided this domain, such students still require support. The number of vulnerable students in Grade 2 is larger, although it can be noted that this is perhaps a function of the definition. To illustrate the implications for teaching, these data suggest that in a typically Educational & Child Psychology Vol 24 No 2

44

Assessing and teaching children who have difficulty learning arithmetic Domain Counting Place Value Addition and Subtraction Strategies Multiplication and Division Strategies Grade 1 (n 1497) Grade 2 (n 1538)

163 (11%) 149 (10%) 176 (12%) 423 (28%)

371 (24%) 417 (27%) 294 (19%) 280 (18%)

Table 1: Number of children in year 2000 who were vulnerable in each number domain sized Grade 1 class of 24 students, about 14 children (62 per cent) will have reached the on the way points in all domains, while 10 children (38 per cent) will be vulnerable in one or more number domains. About three children will be vulnerable in each of Counting, Place Value, and Addition and Subtraction Strategies, and seven children will be vulnerable in the Multiplication and Division Strategies domain. It would be usually assumed that there was considerable overlap in these vulnerabilities. In a typical Grade 2 class (7-year-olds), the number of students who have reached the on the way points in all domains is similar, but the higher percentages in the respective domains indicate that more students are vulnerable in multiple domains. To explore the overlaps between the domains for Grade 1 children, Figure 2 provides a diagrammatic representation of the intersecting domains for which the Grade 1 children were vulnerable. There were 576 Grade 1 children who were vulnerable in at least one domain. By way of explanation, there are 33 students (top right hand side) who are vulnerable in Place Value but in no other domain, while the 36 students (nearby) are vulnerable in both Place Value and Multiplication and Division Strategies. The most striking inference is that there were only 23 children (out of a total of 1497) who were vulnerable in all four domains, and only another 57 who were vulnerable in any three domains. On one hand, this suggests that vulnerability is a function of absence of specific experiences. This may occur if children do not have enough experiences in a Educational & Child Psychology Vol 24 No 2 domain or are not able to take advantage of these so as to construct knowledge. On the other hand, it can be anticipated that specific interventions can address the respective vulnerabilities. We can expect that once teachers recognise that any child is vulnerable in a given domain, then specific interventions can be expected to redress this situation. A similar diagram indicating the intersecting domains for which Grade 2 children were vulnerable is shown in Figure 3 (N 1538). There were 659 Grade 2 children who were vulnerable in at least one domain. While there is still a remarkable spread of vulnerabilities in the respective domains, there are 219 (out of 1538) of the Grade 2s who were vulnerable in three or four domains. This is a higher proportion than for Grade 1 children. However, on the other hand, most Grade 2 children who were vulnerable were vulnerable in either one or two domains, and these domains varied. Overall, the diversity of domains and combinations of domains in which Grade 2 children were vulnerable is striking. There was a spread of vulnerability across all domains, and there were no combinations of domains that were common for children who were vulnerable. Making inferences about vulnerability The above figures indicate a surprising diversity of knowledge across the various number domains. To explore this further, we examined, for students who were vulnerable in one domain, their performance in the other domains. Indeed, while a shortcoming of the Early Numeracy Interview was the limited number of items associated with a particular 45

Ann Gervasoni & Peter Sullivan

37 14 15 36 18 231 48 23 16 12 8

33 36
19

65 67

83 24 44 54 93 8 20 10 58 32

21

43

60

Figure 2: Diagrammatic representation of the intersecting domains for which Grade 1 children (6-year-olds) were vulnerable (n 576)

Figure 3: Diagrammatic representation of the intersecting domains for which Grade 2 children (7-year-olds) were vulnerable (n 659)

growth point in any one domain, its strength was its capacity to facilitate comparisons of performance across domains. The following data refer to the students in Grades 1 and 2 who were vulnerable in Counting, and examines their achieved growth points in the other domains. Of the 163 Grade 1 children who were vulnerable in Counting (which meant that they did not count a collection of about 20 teddies accurately): 64 per cent were at the level of reading, writing, ordering and interpreting one digit numbers; 48 per cent could solve the addition problem (9 4 teddies) by counting-all the items, and a further 14 per cent could solve the solve that problem by countingon, that is without having to see the original nine teddies; 30 per cent could solve multiplicative tasks by counting objects one by one (actually making and quantifying 4 groups of 2 teddies, and sharing 12 teddies among 4 mats), and 14 per cent could use grouping strategies (such as skip counting) to solve those tasks; 61 per cent could accurately compare the 46

length of a string and a straw, and a further 26 per cent could use paper clips as a unit to quantify the length of the straw. Clearly, just because a student is vulnerable in Counting this does not mean that they need particular support in other domains. It is possible that some of these students made a mistake with the counting of the set of 20 teddies. However, in the analysis across six separate interviews for each child, we found extraordinary consistency in responses from the students implying there were few careless errors. In the project schools, the interviews were administered one by one by class teachers. This could account for the low number of aberrant results. Likewise the results are unlikely to be a result of fatigue, given that these items were asked at the start of the interview. Further the items were asked on the same day reducing variations due to external circumstances. A logical conclusion is that the capacity of students in particular domains is a function of their prior experiences with whatever are the necessary pre-requisite knowledge and skills for that domain, and that the respective domains (e.g. Counting, Place Value, Addition and Subtraction Educational & Child Psychology Vol 24 No 2

Assessing and teaching children who have difficulty learning arithmetic Strategies, and Multiplication and Division Strategies) are less connected to each other than we had anticipated. The main result is that teachers should be advised to evaluate students performance on particular tasks, and to take care when inferring from performance on one task to performance on another, or from one domain to another. There was less variability in the profiles of the Grade 2 children, but similar conclusions can be drawn nevertheless. Of the 371 Grade 2 children who were vulnerable in Counting (meaning that they could not continue a counting sequence): 26 per cent were at the level of reading, writing, ordering and interpreting two digit numbers, and a further 4 per cent were doing this for numbers up to 1000; 45 per cent could solve the addition problem (9 4) by counting-on, and a further 7 per cent could solve subtraction tasks like 8 3 and 12 9; 54 per cent could use grouping (evident by skip counting) to solve multiplicative tasks, and a further 2 per cent could abstract multiplicative thinking (meaning they could solve a task like 15 teddies seated in 3 equal rows, without using models); 51 per cent could accurately compare the length of a string and a straw and a further 40 per cent could use paper clips as a unit to quantify the length of the straw. In other words, close to half of the students who were vulnerable in the Counting domain were performing at least up to expectations in other number domains. We had anticipated that Counting would be fundamental to a capacity to respond to items in Place Value, Addition and Subtraction Strategies, and Multiplication and Division Strategies, but it seems that this is not the case. This has implications both for curriculum and for teaching. Grade 1 and Grade 2 children identified as vulnerable in aspects of learning school mathematics. This group is far from being a homogeneous one. Indeed, there were no patterns in the domains in which children were vulnerable, or in any combinations of domains for which children were vulnerable. Vulnerability was widely distributed across all four domains and combinations of domains in both grade levels. However, one feature of the findings for Grade 1 children is worth noting. Twice as many Grade 1 children were vulnerable in Multiplication and Division than for any other domain, but this level of vulnerability was not maintained for Grade 2 children. It is likely that this finding is an artefact of the mathematics curriculum in Victoria that does not recommend Multiplication and Division experiences for children in the first year of school. It is therefore likely that some teachers may not provide such learning opportunities for their students. The fact that so many Grade 1 children did reach the on the way growth point for Multiplication and Division indicates that children will benefit from opportunities to enhance their construction of knowledge in the Multiplication and Division domain throughout the first year of school. If this were to occur, then perhaps fewer Grade 1 children would be identified as vulnerable in the Multiplication and Division domain. A feature of the findings for Grade 2 children worth noting is that a higher proportion of students were vulnerable in three or four domains. It is likely that this finding is an artefact of the assessment interview and growth points. However, this situation also reflects the increasing complexity of the mathematics curriculum that Grade 2 children typically experience. An implication of this finding may be that it is important to provide intervention programs for children who are vulnerable as early in their schooling as possible, before their difficulties increase in complexity. The findings have several other implications for the instructional needs of children. 47

Discussion and implications

The findings presented in the previous sections highlight the diversity of mathematical understandings amongst the group of Educational & Child Psychology Vol 24 No 2

Ann Gervasoni & Peter Sullivan Most importantly, the results indicate that children who are vulnerable in aspects of learning school mathematics have diverse learning needs, and this calls for particular customised instructional responses from teachers. It is likely that teachers need to make individual decisions about the instructional approach for each child, and that there is no formula that will meet all childrens instructional needs. Further, the diversity of childrens mathematical knowledge in the four domains suggests that knowledge in any one domain is not necessarily prerequisite for knowledge construction in another domain. For example, some teachers may assume that children need to be on the way in Counting before they are ready for learning opportunities in the Addition and Subtraction Strategies domain. On the contrary, the findings presented in Figures 1 and 2 indicate that some children who are not on the way in Counting are already on the way in Addition and Subtraction, and this pattern is maintained for the other domains also. This finding has implications for the way in which the school mathematics curriculum is introduced to children. It seems likely that children will benefit from learning opportunities in all four number domains, provided in tandem with one another, and that learning opportunities in one domain should not be delayed until a level of mathematical knowledge is constructed in another domain. Responsive instruction and intervention for vulnerable learners of school mathematics The data presented earlier suggest that addressing arithmetic difficulties is not straight forward because of the diversity of the domains and combinations of domains in which children are vulnerable. A theme emerging in the literature is the need for instruction and learning experiences to closely match childrens individual learning needs (e.g. Ginsburg, 1997; Greaves, 2000; Wright et al., 2000). Rivera (1997) believes that instruction is a critical variable in effective programming for children with mathe48 matics learning difficulties, and that instruction must be tailored to address individual needs, modified accordingly, and evaluated to ensure that learning is occurring. The need for tailored instruction is because diversity among and within subgroups of children who have difficulty with learning mathematics is as great as for that across the general population (Shonkoff & Phillips, 2000). Certainly the number knowledge of the vulnerable learners explored in this paper was characterised by its diversity. Therefore, it is important to consider intervention for those with arithmetic difficulties as a concept rather than a program, and for programs to respond to diversity, as target groups are heterogeneous rather than homogeneous. Ginsburg (1997) articulated a process for using the zone of proximal development (Vygotsky, 1978) for enhancing childrens learning, believing that this is an important idea for assisting teachers develop suitable learning opportunities for children. The zone of proximal development is described as the distance between the actual developmental level, as determined by independent problem solving, and the level of potential development, as determined through problem-solving under adult guidance, or in collaboration with more capable peers (Vygotsky, 1978, p. 86), and defines those functions that have not yet matured but are in the process of maturation (Vygotsky, 1978, p. 86). Ginsburgs process is that first, the teacher analyses childrens current mathematical understandings and identifies their learning potential within the zone of proximal development. Next, the teacher presents a problem in an area with which the child had difficulty, and provides hints to assist the problem solving process. These hints range from general metacognitive hints to those specific to the mathematical demands of the task. The amount of help each child needs is an estimate of his or her learning efficiency within the domain. The teacher continues presenting problems to the student of a similar nature, providing as Educational & Child Psychology Vol 24 No 2

Assessing and teaching children who have difficulty learning arithmetic much help as necessary, until the student is able to solve the problems independently. Finally, the teacher presents near, far and very distant transfer problems and students are given assistance, as needed, to solve them. Apart from enhancing learning, Ginsburg (1997) argues that this type of technique is required to establish the extent to which cognitive difficulties persist despite constant efforts to remove them. Sullivan, Mousley and Zevenbergen (2006) also highlighted the importance of teachers adjusting tasks to enable children who experience difficulty to engage successfully in learning opportunities. For example, they claim that student difficulties with a particular task might be a result of: the number of steps; the number of variables; the modes of communicating responses; the number of elements in recording; the degree of abstraction or visualisation required; the size of the numbers to be manipulated; the language being used; or psychomotor considerations. They argued that a teacher could anticipate these difficulties, and prepare prompts and associated resources to: reduce the required number of steps; reduce the required number of variables; simplify the modes of representing results; reduce the written elements in recording; make the task more concrete; reduce the size of the numbers involved; simplify the language; or reduce the physical demand of any manipulatives. Thus, we are recommending that teachers be oriented towards responding to individual childrens learning needs by adjusting tasks to increase engagement, an important aspect of assisting vulnerable learners. Wright et al. (2000) also advocated that teachers should routinely make adjustments to planned activities on the basis of childrens responses. Further, they argue that tasks should be genuine problems for children, and that children should be challenged to bring about reorganisations in their thinking (Wright et al., 2000). Each of the strategies outlined above is important to consider when teaching children with arithmetic difficulties. However, Educational & Child Psychology Vol 24 No 2 fundamental to meeting childrens individual learning needs is the notion of a framework of growth points or stages of development, such as that developed for the ENRP. Growth points help teachers to identify childrens zone of proximal development in mathematics so as to create appropriate learning opportunities, and in order to adjust activities to increase engagement and remove features that are creating barriers to learning. Thus, reference to a framework of growth points helps to ensure that instruction for vulnerable children is closely aligned to childrens initial and ongoing assessment, and is at the cutting edge of each childs knowledge (Wright et al., 2000). In summary, we believe the following instructional practices are important for enhancing mathematics learning for children who have arithmetic difficulties: 1. targeting instruction within a childs zone of proximal development in each domain, based on current assessment information about the childs mathematical understandings and the probable course of the childs learning; 2. making adjustments to planned activities on the basis of childrens responses; 3. presenting rich, challenging problems that promote hard thinking within a childs zone of proximal development and in an area with which the child had difficulty; providing hints to assist the problem solving process, ranging from general metacognitive hints to those specific to the mathematical demands of the task; continually presenting problems of a similar nature, providing as much help as necessary, until the student is able to solve the problems independently. These approaches are important to consider when providing intervention programs for children with arithmetic difficulties, and when examining the effectiveness of an intervention program.

Conclusion
The findings presented in this paper suggest that there is no single formula for describing 49

Ann Gervasoni & Peter Sullivan children who have difficulty learning arithmetic or for describing the instructional needs of this diverse group of students. Further, it is not possible to assume that because a child is vulnerable in one aspect of number learning, then he or she will be vulnerable in another. This finding may surprise some teachers and highlights why assisting children with arithmetic difficulties is not straight forward. Meeting the diverse learning needs of children is a challenge, and requires teachers to be knowledgeable about how to identify each childs learning needs and customise instruction accordingly. This calls for rich assessment tools capable of mapping the extent of childrens knowledge in a range of domains, and an associated framework of growth points capable of guiding teachers curriculum and instructional decision-making. It follows that intervention programs for children who are vulnerable need to be flexible in structure in order to meet the diverse learning needs of each participating child. Intervention teachers need to provide instruction and feedback that is customised for the particular learning needs of each child, and based on knowledge of childrens current mathematical knowledge. Further, teachers need to ensure that children take advantage of learning experiences by drawing childrens attention to the salient features to facilitate the construction of knowledge and understanding. Another issue for Intervention programs is that the diversity we discovered in childrens number knowledge and abilities in a range of number domains suggests that programs need to focus on all number domains in tandem. It is not appropriate to wait until children reach a certain level of knowledge in one domain before experiences in another domain are introduced. Childrens construction of number knowledge in a specific domain is not dependent on prerequisite knowledge in another domain, but is dependent on being able to take advantage of a range of experiences in a given domain. Assisting children with arithmetic difficulties is complex, but teachers who are equipped with the tools necessary for responding to the diverse needs of individuals who have not previously thrived when learning arithmetic, are able to provide children with the type of learning opportunities and experiences that will enable them to thrive and extend their mathematical understanding further.

Address for Correspondence

Dr Ann Gervasoni, Australian Catholic University Ballarat Campus, 1200 Mair St, Ballarat, VIC 3350, Australia E-mail: Ann.Gervasoni@acu.edu.au Professor Peter Sullivan, Monash University, Clayton VIC 3800, Australia E-mail: Peter.Sullivan@education.monash. edu.au

References
Board of Studies (2000). Curriculum and Standards Framework II: Mathematics. Carlton, Victoria: Author. Bobis, J. & Gould, P. (1999). The mathematical achievement of children in the Count Me In Too program. In J.M. Truran & K.M. Truran (Eds.), Making the difference (Proceedings of the 22nd Annual Conference of the Mathematics Research Group of Australasia (pp. 8490). Adelaide: MERGA. Boulton-Lewis, G. (1996). Representations of place value knowledge and implications for teaching addition and subtraction. In J. Mulligan & M. Mitchelmore (Eds.), Childrens number learning: A research monograph of MERGA/AAMT (pp. 7588). Adelaide: Australian Association of Mathematics Teachers. Clarke, D. (2001). Understanding, assessing and developing young childrens mathematical thinking: Research as powerful tool for professional growth. In J. Bobis B. Perry & M. Mitchelmore (Eds.), Numeracy and beyond: Proceedings of the 24th Annual Conference of the Mathematics Education Research Group of Australasia (Vol. 1, pp. 926). Sydney: MERGA. Clarke, D., Cheeseman, J., Gervasoni, A., Gronn, D., Horne, M., McDonough, A., Montgomery, P., Roche, A., Sullivan, P., Clarke, B. & Rowley, G.

50

Educational & Child Psychology Vol 24 No 2

Assessing and teaching children who have difficulty learning arithmetic

(2002). Early Numeracy Research Project Final Report. Melbourne: Australian Catholic University. Cobb, P. & McClain, K . (1999). Supporting teachers learning in social and institutional contexts. Paper presented at the International Conference on Mathematics Teacher Education, Taipei, Taiwan. Department of Education Employment and Training (2001). Early numeracy interview booklet. Melbourne: Department of Education, Employment and Training. Gervasoni, A. (2000). Using growth-points profiles to identify Grade 1 students who are at risk of not learning school mathematics successfully. In J. Bana & A. Chapman (Eds.), Mathematics education beyond 2000. Proceedings of the 23rd Annual Conference of the Mathematics Education Research Group of Australasia. (pp. 275282). Fremantle, WA: MERGA. Gervasoni, A. (2004). Exploring an intervention strategy for six ans seven year old children who are vulnerable in learning school mathematics. Unpublished PhD thesis, La Trobe University, Bundoora. Ginsburg, H.P. (1997). Mathematical learning disabilities: A view from developmental psychology. Journal of Learning Disabilities, 30(1), 2033. Greaves, D. (2000). Private provider services for students with learning difficulties. In W. Louden, L.K.S. Chan, J. Elkins, D. Greaves, H. House, M. Milton, S. Nichols, J. Rivalland, M. Rohl & C. Van Kraayenoord (Eds.), Mapping the territory: Primary Students with Learning Difficulties: Literacy and Numeracy (Vol. 1, pp. 135156). Canberra: Commonwealth of Australia. Fuson, K. (1992). Research on whole number addition and subtraction. In D.A. Grouws (Ed.), Handbook of Research on Mathematics Teaching and Learning (pp. 243275). New York: Macmillan. Hart, B., Brinkman, S. & Blackmore, S. (2003). How well are we raising our children in the North Metropolitan Region? Perth: Population Health Program, North Metropolitan Heath Service. Mulligan, J. & Mitchelmore, M. (1996). Childrens representations of multiplication and division word problems. In J. Mulligan & M. Mitchelmore (Eds.), Childrens Number Learning: A Research Monograph of MERGA/AAMT (pp. 163184). Adelaide: Australian Association of Mathematics Teachers. Owens, K. & Gould, P. (1999). Framework for elementary school space mathematics (discussion paper). Rivera, D.P. (1997). Mathematics education and students with learning disabilities: Introduction to the special series. Journal of Learning Disabilities, 30(1), 219. Shonkoff, J. & Phillips, D. (2000). From neurons to neighborhoods: The science of early childhood development. Washington: National Academy Press. Steffe, L., Cobb, P., & von Glasersfeld, E. (1988). Construction of arithmetical meanings and strategies. New York: Springer-Verlag. Steffe, L., von Glasersfeld, E., Richards, J. & Cobb, P. (1983). Childrens counting types: Philosophy, theory, and application. New York: Praeger. Sullivan, P., Mousley, J. & Zevenbergen, R. (2006). Developing Guidelines for Teachers Helping Students Experiencing Difficulty in Learning Mathematics. In P. Grootenboer, R. Zevenbergen & M. Chinnappan, (Eds.), Proceedings of the29th Annual Conference of the Mathematics Education Research Group of Australasia (pp. 496503), Canberra, July. Vygotsky, L.S. (1978). Mind in society: The development of higher mental processes. Cambridge, MA: Harvard University Press. Woodward, J. & Baxter, J. (1997). The effects of an innovative approach to mathematics on academically low-achieving students in mainstreamed settings. Exceptional Children, 63, 373388. Wright, R. (1998). An overview of a research-based framework for assessing and teaching early number learning. In C. Kanes, M. Goos & E. Warren (Eds.), Teaching Mathematics in New Times: Proceedings of the 21st Annual Conference of the Mathematics Education Research Group of Australasia (pp. 701708). Brisbane, Queensland: Mathematics Education Research Group of Australasia. Wright, R., Martland, J., & Stafford, A. (2000). Early Numeracy: Assessment for teaching and intervention. London: Paul Chapman Publishing

Educational & Child Psychology Vol 24 No 2

51

Ann Gervasoni & Peter Sullivan

Appendix A: Early numeracy research project assessment framework

Growth Points for the number domains Notes
Growth points are not necessarily hierarchical,

but involve increasingly complex reasoning and understanding. It must be emphasised that conclusions drawn in relation to placing students at levels within this framework are based on a 30-minute (approx.) interview only. Ongoing assessment by the teacher during class will provide important further information for this purpose. Student understanding may be reported as a 0. This should not be taken as an indication of no knowledge or no understanding, but rather as an indication of a lack of evidence of 1. Counting Not apparent. Not yet able to say sequence of number names to 20. Rote counting. Rote counts the number sequence to at least 20, but is not yet able to reliably count a collection of that size. Counting collections. Confidently counts a collection of around 20 objects. Counting by 1s (forward/backward, including variable starting points; before/after). A count forward and backwards from various starting points between 1 and 100; knows numbers before and after a given number. Counting from 0 by 2s, 5s, and 10s. Can count from 0 by 2s, 5s, and 10s to a given target. Counting from x (where x 0) by 2s, 5s, and 10s. Given non-zero starting points, counts by 2s, 5s,10s to given target. Extending and applying counting skills. Counts from non-zero starting points by any 1-digit number, and applies counting skills in practical tasks. Place Value Not apparent. Not yet able to read, write, interpret and order single digit numbers. Reading, writing, interpreting, and ordering single digit numbers. Can read, write, interpret and order single digit numbers.

Reading, writing, interpreting, and ordering twodigit numbers. Can read, write, interpret and order two-digit numbers. Reading, writing, interpreting, and ordering threedigit numbers. Can read, write, interpret and order 3-digit numbers. Reading, writing, interpreting, and ordering numbers beyond 1000. Can read, write, interpret & order numbers beyond 1000. Extending and applying place value knowledge. Can extend and apply knowledge of place value in solving problems. Strategies for addition and subtraction Not apparent. Not yet able to combine & count 2 collections of objects. Count all (two collections). Counts all to find the total of two collections. Count on. Counts on from one number to find the total of two collections. Count back/count down to/count up from. In subtraction contexts, chooses suitably from count-back, count-down-to & count-up-from strategies. Basic strategies (doubles, commutativity, adding 10, tens facts, other known facts). Given an addition or subtraction problem, strategies such as doubles, commutativity, adding 10, tens facts, and other known facts are evident. Derived strategies (near doubles, adding 9, build to next ten, fact families, intuitive strategies). Given an addition or subtraction problem, strategies such as near doubles, adding 9, build to next ten, fact families and intuitive strategies are evident. Extending and applying addition and subtraction using basic, derived and intuitive strategies. Given a range of tasks (including multi-digit numbers), can solve them mentally, using the appropriate strategies and a clear understanding of key concepts. Strategies for Multiplication and Division Not apparent. Not yet able to create and count the total of several small groups. Counting group items as ones. To find the total in a multiple group situation, refers to individual items only.

52

Educational & Child Psychology Vol 24 No 2

Assessing and teaching children who have difficulty learning arithmetic Modelling multiplication and division (all objects perceived). Models all objects to solve multiplicative and sharing situations. Abstracting multiplication and division. Solves multiplication and division problems where objects are not all modelled or perceived. Basic, derived and intuitive strategies for multiplication. Can solve a range of multiplication problems using strategies such as commutativity, skip counting and building up from known facts. Basic, derived and intuitive strategies for division. Can solve a range of division problems using strategies such as fact families and building up from known facts. Extending and applying multiplication and division. Can solve a range of multiplication and division problems (including multidigit numbers) in practical contexts.

Educational & Child Psychology Vol 24 No 2

53

Assessing pupil knowledge of the sequential structure of numbers

David Ellemor-Collins & Robert Wright
Abstract
Research on childrens mental strategies for multidigit addition and subtraction identifies two categories of strategy. Collections-based strategies involve partitioning numbers into tens and ones, and can be modeled with base-ten materials. Sequence-based strategies involve keeping one number whole, and using the sequential structure of numbers. They can be modelled as jumps on an empty number line. Studies have found sequence-based strategies to be more successful, and to correlate with more robust arithmetic knowledge, particularly among low-attaining pupils. Studies also suggest that sequence-based strategies and sequential structure are not explicitly developed in many primary mathematics classrooms. This report draws on results from a three-year project which has the goal of developing pedagogical tools for intervention in the number learning of low-attaining third- and fourth- graders (8- to 10-year-olds). These tools include assessment tasks to inform intervention. The report focuses on four groups of assessment tasks that collectively enable detailed documenting of pupils knowledge of the sequential structure of numbers. Tasks and pupils responses are described in detail. Some examples follow. When asked to count back from 52, pupils said, 52, 51, 40, 49, 48, and so on. When asked to count back by tens from 336, pupils had difficulty continuing after 326. Thus teen numbers in the hundreds (316) presented particular difficulties. Pupils had difficulty saying the number that is ten less than 306. Pupils had difficulty with locating the numbers 50, 25, 62 and 98 on a number line on which zero and 100 were marked. The report provides insight into assessing knowledge of sequential structure and argues that this is important basic number knowledge.

ULTI-DIGIT ADDITION and subtraction is an important aspect of number sense and mental computation (Anghileri, 2000; Thompson & Smith, 1999). As well, this topic is important because it provides a basis for more advanced arithmetic. This article draws from a current three-year project focusing on the development of pedagogical tools to support intervention in the number learning of low-attaining third- and fourthgraders (8- to 10-year-olds). These tools include schedules of diagnostic assessment tasks and instructional procedures. This paper focuses on some of the assessment tasks which enable assessment of knowledge of the sequential structure of numbers. Developing significant knowledge of the sequential structure of numbers provides an important basis for multi-digit addition and subtraction (Beishuizen & Anghileri, 1998). The paper will:

1. elaborate the term sequential structure of numbers; 2. review literature relevant to the sequential structure of numbers; 3. set out relevant diagnostic assessment tasks; and 4. describe the range of low-attaining pupils responses to those tasks.

Sequential structure of numbers

This paper discusses low-attaining pupils knowledge and use of what we call the sequential structure of numbers. By sequential structure of numbers we refer to the decade-based structures in the linear sequence 1, 2, 3, . . . 99, 100, 101. . . . Specifically, this number sequence consists of a sequence of decades, which can be further organised in a sequence of hundreds. The decade numbers (10, 20, 30, . . .) are reference points in the sequence, at even intervals of Educational & Child Psychology Vol 24 No 2 The British Psychological Society 2007

54

Assessing pupil knowledge of the sequential structure of numbers ten. Each decade follows the same pattern as, for example, 20, 21, 22, . . . 28, 29, 30. By the neat symmetry in this sequence, a pair of numbers such as 18 and 28, or 71 and 81, is always ten steps apart. Referring to the sequential structure of numbers, 57 can be regarded as one after 56, seven after 50, three before 60, or 10 after 47. We can also describe collections-based structures in multidigit numbers. These involve thinking of numbers in terms of collections of ones, tens, hundreds and so on. For example, 57 can be constructed as fifty and seven, or as 7 ones and 5 tens. Anghileri, 1998; Cobb et al., 1997; Foxman & Beishuizen, 2002; Thompson & Smith, 1999). The standard example of a sequencebased strategy is the jump strategy. Jump involves keeping the first number whole and adding (or subtracting) the second via a series of jumps. For example, a pupil might add 57 and 26 using jump by reasoning as follows: 57 and ten is 67, and ten more is 77; three more is 78, 79, 80; and three more make 83. Researchers note that such sequence-based strategies depend on knowledge of sequential structures to jump by ten, and to make steps and hops in the number sequence (Fuson et al., 1997; Treffers & Buys, 2001; Yackel, 2001). Classroom use of settings such as a number line that highlights the decades or a bead string with the decades demarked by colour (110 is blue, 1120 is red, 2130 is blue etc.) are linked to pupil use of sequential structure and sequence-based strategies (Klein, Beishuizen & Treffers, 1998). The standard example of a collectionsbased strategy is the split strategy. Split involves partitioning both numbers into tens and ones, adding (or subtracting) separately with the tens and the ones, and finally recombining the tens and ones subtotals. A pupil might add 57 and 26 using split by reasoning as follows: 50 and 20 are 70, 7 and 6 are 13, 70 and 13 make 83. Collections-based strategies use collections-based structures (Fuson et al., 1997; Treffers & Buys, 2001; Yackel, 2001). Classroom settings such as base-ten blocks are linked to the use of collections-based structures and strategies (Beishuizen, 1993). Fuson et al. (1997) suggest that an advanced understanding of multi-digit addition and subtraction requires an integration of sequence-based and collections-based strategies. For example, an advanced pupil asked to add 5 doughnuts to 58 doughnuts might use a sequence-based strategy, jumping through 60 to 63, which is more efficient than split in this case; but when then asked how many boxes of ten she could fill, use her knowledge of collections-based structure to recognize 6 tens in 63. 55

Literature review
Emphasis on mental computation In the last 15 years, research and curriculum reforms in a range of countries highlight a renewed emphasis on mental computation with multidigit numbers (Beishuizen & Anghileri, 1998; Thompson, 1997). An early emphasis on mental strategies, rather than formal written algorithms, may better support number sense and conceptual understanding of multidigit numbers, and support development of important connections to related knowledge (Askew, Brown, Rhodes, Wiliam & Johnson, 1997; Hiebert & Wearne, 1996; McIntosh, Reys & Reys, 1992; Sowder, 1992; Yackel, 2001). Mental computation can also stimulate the development of numerical reasoning and flexible, efficient computation (Anghileri, 2001; Treffers, 1991). Mental strategies: sequence-based and collections-based In response to the emphasis on mental computation, research projects in several countries focused on pupils informal mental strategies for multi-digit addition and subtraction (Beishuizen, Van Putten & Van Mulken, 1997; Cobb et al., 1997; Cooper, Heirdsfield, & Irons, 1995; Foxman & Beishuizen, 1999; Fuson et al., 1997; Ruthven, 1998; Thompson & Smith, 1999). Several studies described two main categories of strategies sequence-based and collections-based (e.g. Beishuizen & Educational & Child Psychology Vol 24 No 2

David Ellemor-Collins & Robert Wright Infrequency of sequence-based strategies among low-attaining pupils Researchers have found that low-attaining pupils tend to use split strategies, indicating the development of knowledge of collectionsbased structure (Beishuizen, 1993; Foxman & Beishuizen, 2002). Research also suggests that many low-attaining pupils do not develop the strategy of jumping by tens and thus may not develop sequence-based structures (Beishuizen, 1993; Beishuizen et al., 1997; Menne, 2001). Thus it is unlikely that these pupils can advance to integrated sequence-collections-based strategies which, we would argue, is important for number sense and mental computation. Advantages of sequence-based strategies Jump strategies can develop as abbreviations of pupils informal counting strategies (Beishuizen & Anghileri, 1998; Olive, 2001). Following the view that pupils knowledge should build on their informal strategies (Anghileri, 2001; Resnick, 1989), some researchers recommend teaching jump strategies (Klein et al., 1998). A common difficulty with multi-digit addition and subtraction arises for pupils when they separate the digits in the tens place from the digits in the ones place and do not adequately regroup. For example, 57 26 is found to be 73 or even 713. These difficulties arise in the case of split strategies but do not arise in the case of jump strategies (Beishuizen & Anghileri, 1998; Cobb, 1991; Fuson et al., 1997). Beishuizen and colleagues found that pupils made significantly more errors when using split strategies than when using jump strategies. Importantly, even within a group of pupils identified as low-attaining, jump strategies were much more successful (Klein et al., 1998). These results were confirmed by Foxman (2002). Studies comparing the use of split and jump strategies found that split led to more difficulty developing independence from concrete materials (Beishuizen, 1993), more procedural and conceptual confusion (Klein et al., 1998) and slower response times, suggesting a heavier load on working memory (Wolters, 56 Beishuizen, Broers & Knoppert, 1990). Subtraction tasks are a source of particular difficulties in multidigit arithmetic, and the potential confusions of subtraction using a split strategy are well documented. Confused responses using split suggest the collectionsbased structure offers a problematic representation of subtraction tasks (Fuson et al., 1997). Success with split requires strong number sense and subtle insight into the procedure itself, whereas success with jump mainly requires knowing how to jump ten from any number (Beishuizen, 1993). Developing flexibility with strategies An important goal in improving multidigit number sense is flexibility with strategies, including recognising efficient short-cuts and making adaptations for unfamiliar problems (McIntosh et al., 1992). Studies indicate that pupils more readily adapt the jump strategy to make efficient computation choices. According to Beishuizen et al., this is due to the underlying mental representation of the number row up to 100 (1997). That is, using the sequential structure of number readily supports strategic insight into computation tasks. In summary, pupils with arithmetic difficulties tend not to develop sequential structure and sequence-based strategies such as jump. It is likely that this denies them an integrated approach to multi-digit addition and subtraction, and access to the preferred strategies of arithmetically successful pupils. Further, development of sequential structure and strategies might resolve a number of typical multi-digit difficulties prevalent with collections-based strategies.

Assessment task groups and responses

As part of the project (referred to earlier in this article), 204 low-attaining pupils were interviewed twice during the school year to assess their number knowledge. The pupils were in third and fourth grades (8- to 10year-olds) from a broad demographic range across the state of Victoria, and were selected for the study based on low results in screenEducational & Child Psychology Vol 24 No 2

Assessing pupil knowledge of the sequential structure of numbers ing tests. A method involving one to one, dynamic interview was used, in which the pupil is posed number tasks, and the interviewer pays close attention to the pupils thinking process (Wright, Martland & Stafford, 2006). Interview assessments were recorded on videotape for later analysis. We use the term task group to refer to a group of closely related tasks used to investigate pupils knowledge of a specific topic. In this paper, we discuss four task groups we found particularly valuable in assessing pupil knowledge of sequential structure: 1. Number word sequences by ones. 2. Number word sequences by tens. 3. Incrementing and decrementing by ten. 4. Locating numbers. For each, we describe the range of lowattaining pupils responses and difficulties, evident from analysis of the videotaped interviews. where pupils responded correctly to these tasks, their responses indicated a lack of certitude, particularly when bridging decade or hundred numbers. All of our low-attaining pupils made errors with number word sequences bridging 1000. Younger childrens difficulties in establishing the number word sequence are well documented (e.g. Fuson, Richards & Briars, 1982; Wright, 1994). We have found the persistent errors and uncertainties of these older children striking. Our conclusion is that the assessment tasks described above are indicative of areas of knowledge that should be explicitly taught, at least in the case of low-attaining pupils.

Task group 2: Number word sequences by tens

Focus Number word sequences by tens, forwards and backwards, on and off the decade. Examples Count by tens. Stop at 120. Count by tens from 24. Ill tell you when to stop. Stop at 104. Bridging 50 or 40 backwards 52, 51, 40, 49, 48 . . . 52, 51, ^ 49, 48 . . . 52, 51, 50, 89, 88 . . . 42, 41, 40, 49, 48 . . . Bridging 100 98, 99, 100, ten hundred 102, 101, ^ 99, 98 . . . Bridging 110 forwards 108, 109, 1000, 1001 . . . 108, 109, 200, 201, 202 . . . Number word after 109: 1000 Bridging 200 198, 199. Thats all I know. 198, 199, 1000, 1001 . . . 198, 199, ^ 201, 202 . . . 202, 201, ^ 199, 198 . . . Table 1: Pupils errors in oral number word sequences. Note: Errors are marked in bold, omissions are marked with . 57

Task group 1: Number word sequences by ones

Focus Short sequences of number words, backwards and forwards; number word before or after; and bridging decades and hundreds. Examples Count from 97. Ill tell you when to stop. Stop at 113. Count backwards from 103. Stop at 95. Say the number that comes just after 109. Say the number that comes just before 100. Similarly for bridging 40, 210, 300, 990, 1000, 1100 forwards and backwards. Low-attaining pupils difficulties Table 1 sets out examples of pupils errors with number word sequences. Errors bridging 50 backwards indicate that the pupils have not fully constructed the number word sequence. Rather, they are aware of separated chains such as 4149 and 5159, and link these chains incorrectly when going backwards (Skwarchuk & Anglin, 2002). In the range 100 to 1000, number word sequence errors were common (Table 1). In many of the cases Educational & Child Psychology Vol 24 No 2

David Ellemor-Collins & Robert Wright Count by tens back from 52. Stop at 2. Count by tens from 167. Stop at 237. Low-attaining pupils responses and difficulties The patterns of number word sequences by tens are inherent in the sequential structure of the base-ten number system. Jump strategies are derived from these patterns. Researchers have suggested that pupils can have difficulty producing number word sequences by tens off the decade, and hence be unable to develop a jump strategy (e.g. Beishuizen, 1993). Skip counting by tens on the decade. All pupils interviewed could produce the sequence of decade numbers 10, 20, 30, . . ., although some had difficulty in continuing beyond 90. Cannot count by tens from 24. Some pupils could not count by tens from 24. Responses included: (a) 24, 25, 20 and again 24, 25, 20?; and (b) 24, 30, 34, 40. One pupil could not count by tens from 24, but could count by tens from 25 25, 35, 45. It seems that these pupils inability to make sense of the task arises from an unfamiliarity with sequences of tens off the decade compared with sequences of fives and of tens on the decade. Indeed, some of these pupils could count by tens on the decade up to 1000. Counting by ones. When asked to count by tens from 24, some pupils counted each ten by ones. This could be laborious and sometimes unsuccessful. Sometimes, a pupil would seemed to become aware of the pattern they were producing, and their sequence would become more fluent perhaps curtailing the counting by ones. Difficulties with teen numbers. Many pupils could not coordinate the teen numbers with a larger number sequence. When counting by tens back from 52 (52, 42, 32, . . .), some pupils had difficulty after 22. Responses included: (a) 22, 2; (b) 22, 14, 4; and (c) 22, 10, 1. Some were successful but their response involved counting back by 58 ones after 22. Many pupils had difficulties with teen numbers in the hundreds, for example saying 336, 326, 316, 314, 306, 304. A few pupils had difficulties with teens when counting back by tens on the decade: (a) 70, 60, . . . 30, 20, 15, 10.; (b) 70, 60, . . . 30, 12, 10.; and (c) 70, 60, 50, 40, 12, no, 20, 0? or 10?. Irregularities in the names of teens mask their ten-structure (Fuson et al., 1997) and this results in significant difficulties in saying sequences by tens. Difficulties in the range 100 to 1000. Pupils who could skip count by ten off the decade in the range 1 to 100 experienced difficulties with bridging one hundred or higher hundred numbers. One pupil said 177, 187, 197, one hundred and- then 297, then 207, 217. . . . When skip counting back by ten on the decade some pupils produced a sequence such as 430, 420, 410, 300, 390, 380. . .. This is analogous to a common error among younger pupils at decade numbers when counting backward by ones, for example, when counting back from 45, the pupil says, 45, 44, 43, 42, 41, 30, 39, 38 . . . . Pupils who erred when skip counting by saying 300 as the number ten less than 410, did not make a corresponding error when skip counting back by ten off the decade but there were difficulties at hundred numbers such as 336, 326, 316, 297 corrected to 296, and thus omitting 306. Another difficulty was discriminating the new hundreds number from the tens number, for example, an attempt to skip count by ten from 167 was 267, 367, 467. When counting back by tens from 336, one response was: 326, 316, 312 (pause), 326, 226, 206, two hundred and zero, 196, 186, 176. This sequence illustrates a persistent difficulty with the teen numbers within the hundreds, and confusion when bridging 200. It is clear that knowledge of sequences of tens beyond 100 is a significant extension of knowledge of sequences of tens up to 100. The responses described above are indicative of weaknesses in pupils knowledge of the sequential structure of numbers.

Educational & Child Psychology Vol 24 No 2

Assessing pupil knowledge of the sequential structure of numbers We believe it is important to address these weaknesses through intervention. number. Alternatively, we could say that a pupil who can increment by ten to bridge a hundred number has constructed a sequence-based strategy for the operation of adding ten. Progressions in incrementing and decrementing by ten Pupils success with tasks involving incrementing or decrementing by ten tended to progress as follows: 2-digit off-the-decade: ten more/less than 79, 3-digit off-the-decade: ten more/less than 356, forward across a hundred number: ten more than 195, backward across a hundred number: ten less than 306, forward across 1000: ten more than 999, backward across 1000: ten less than 1005. Thus a pupil who was successful at the third progression (starting from the uppermost progression) was likely to succeed with the tasks at the first two progressions and not succeed with the tasks from the fourth progression onward. Some pupils could not increment by ten off the decade at all. Difficulties with the teen number sequence were also evident in these tasks, for example, a pupil could find ten more than 356, but not find ten more than 306. Tasks involving 1000, that is ten more than 999 and ten less than 1005, were especially difficult for virtually all of the pupils. Which number is ten less than 306? The task of finding ten less than 306 was particularly difficult for many pupils. Pupils responses included: I dont know. Incorrect counting: 210?, 299, 300. Counting back by ones with an incorrect sequence, and using fingers to keep track of the ten counts: 305, 304, 303, 302, 301, 330, 329, 328, 327, 326!. Counting back by ones and answering 295. Counting back by ones successfully. Jumping back ten successfully. 59

Task group 3: Incrementing and decrementing by 10

Focus Ten more and ten less, on and off the decade, and bridging decades and hundreds. Examples Show 20 on a card. Ask Which number is ten more than this? Similarly for 79, 356, 195, 999. Show 30 on a card. Ask Which number is ten less than this? Similarly for 79, 356, 306, 1005. Relationship to knowledge of the tens structure of the number sequence We were interested in whether pupils could solve the tasks in Task Group 3 without counting by ones or trying to use an algorithm for addition or subtraction. We would regard such pupils as having knowledge of the tens structure of the number sequence. The tasks of (a) incrementing and decrementing by ten and (b) skip counting by ten seemed to be linked in the sense that pupils showed similar levels of advancement in their responses to these task groups. Nevertheless, the tasks involving incrementing or decrementing by ten are distinct from tasks of skip counting by ten. There was incongruence in pupils responses on these two task groups. For example, one pupil skip counted by tens from 167 successfully: 177, 187, 197, 297 self-corrected to 207, 217. . ., but could not solve ten more than 195: one hundred and . . . changed to 220? changed to 225. Pupils might construe the incrementing task as an addition task rather than a task based on a number sequence with increments of ten. Thus they are unable to regard ten more as one increment of a sequence with increments of ten. Even if they can regard ten more as one increment in a sequence with increments of ten, they are apparently unable to increment the sequence from a standing start, when the increment involves bridging a hundred Educational & Child Psychology Vol 24 No 2

David Ellemor-Collins & Robert Wright To solve this task by jumping back ten requires knowing the decade before 301 is the 290s. Many of these pupils did not know this or could not apply this knowledge to solve the task, that is, they lacked knowledge of the sequential structure of numbers. Further, many of the pupils did not count back by ones to solve this task. Apparently these pupils could not construct a representation of this problem that was embedded in the number sequence. Those pupils who attempted to count back on this task were consistently more successful on other tasks involving incrementing by ten than pupils who did not attempt to count back. sequential structure and also a lack of knowledge of linear measure. We have found pupils responses to be interesting, and revealing of their number sequence knowledge. Four examples are discussed below. Renees response, shown in Figure 2, is typical. It would seem there is some sense of global location: 50 is placed at half-way; 25 placed perhaps from a sense of decades, or from half of 50; 98 is probably located to be near 100, but with a weak sense of the measure of the 2-step gap. A weaker response can be very revealing. Helen (see Figure 2) does find 50 as half way. But to locate 25, she marks all the ones from 0 to 25. She does not count in tens, though she does emphasise her 20 point. Her 25 ends up almost at 50, and it is not clear whether she regards this as problematic. She locates 98 two steps back from 100, but the steps are too big. Helen is using an aspect of the number structure, but is not checking against another aspect, that is, 98 as 8 more than 90. She does not seem to have a global or embedded sense of the structures of the sequence. To locate 62, Helen again counts by ones. She does count on from 50, and she emphasises the 60 point along the way, showing some appreciation of how the number structure can support her solution. But she does not curtail counting by ones. Nate finds each of his numbers by counting and marking fives (see Figure 2). He locates 98 after counting to 95, and locates 62 just past 60. He does not count by ones, but doesnt regard the decades as reference points. Rather, he counts by fives. Perhaps more striking is that, in contrast to Helen, Nate finds both 62 and 98 by counting by fives from five. He marks 98 just short of 100 but does not use 100 as the reference to locate 98. He does seem concerned to line up the successive counts to 50 at the same 50 mark, but he does not simply count-on from 50he begins afresh from five each time. His approach is analogous to counting-all rather than counting-on to solve an addition task. Both Helen and Nate frequently counted by ones on their fingers to solve addition Educational & Child Psychology Vol 24 No 2

Task group 4: Locating numbers on a number line

Focus Locating numbers on a linear representation of the number sequence from 0 to 100. Example Pupil is given a pen, and a line on paper with only the endpoints 0 and 100 labeled (see Figure 1). Pupil is asked to Mark where 50 is on the line, and then Label that as 50. Similarly mark and label 25; 98; 62. Low-attaining pupils responses and difficulties A locating number task requires knowledge of the number sequence. To locate the numbers efficiently requires using ideas such as: 50 at half-way; 25 at half-way to 50; 98 at two steps before 100; and 62 just after 60, which is ten after 50. To do the task well also requires knowledge of linear measure and proportion which may be somewhat distinct from number sequence knowledge. Pupils difficulties with this task indicate a lack of knowledge of

100

Figure 1: Blank number line for locating numbers task 60

Assessing pupil knowledge of the sequential structure of numbers

Renee

Helen, counting by ones

Nate, counting by fives from zero

Nate on post-assessment
Figure 2: Locating numbers on a number line

tasks, and neither could skip count by tens off the decade. Their responses on this Task Group indicate a lack of knowledge of structures in the number sequence. Both of these pupils received intensive individual intervention instruction after these assessments. The instruction did not focus on tasks of locating numbers but included a significant focus on counting by ones and tens, and on recording on an empty number line, additions using a jump strategy. When assessed with the locating number task after the intervention period, Helen was no longer marking ones, and she located 25 appropriately. Nates post-assessment response is shown in Figure 2. His pen moves in jumps of ten and one. This learning that Nate demonstrated in his post-assessment can be attributed to the instruction that focused on recording jump strategy additions on a number line. He shows a clear use of decade structure, and he is no longer working from one. Interestingly, his knowledge of linear measure and proportion has also advanced. Educational & Child Psychology Vol 24 No 2

Thus, without explicit instruction on locating numbers, his broad development of number sequence knowledge has made significant differences to his responses on the task group of locating numbers. This Task Group is very useful because it can reveal knowledge of number sequence structure, can differentiate levels of understanding, and can enable learning over time to be documented.

Conclusion
We claim that the sequential structure of numbers is important basic number knowledge. We advocate that pupils number learning should include a focus on number word sequences up to 1000, skip counting and incrementing by tens off the decade, and locating numbers in the range 1 to 100. It is striking that many third and fourth grade pupils (aged 8 to 10 years) are not successful on the assessment tasks described in this report. In our view, a focus on sequential structure exemplifies an informed approach 61

David Ellemor-Collins & Robert Wright to tackling numeracy difficulties (Dowker, 2005). Studies suggest that weakness in these sequence-based tasks is characteristic of lowattaining pupils (Beishuizen et al., 1997; Menne, 2001). Our study accords with this. We recommend that low-attaining pupils be assessed for knowledge of sequential structure, and that intervention include explicit attention to development of this knowledge. The four assessment task groups discussed in this report can inform detailed assessment of pupils number sequence knowledge. We are developing instructional activities for this topic in our current research project with lowattaining pupils, trialling, for example, flexible incrementing and decrementing by tens and ones (Wright, Martland, Stafford & Stanger, 2002) and jumping on an empty number line (Menne, 2001). We are also developing activities targeting the pupils development of the related sequence-based mental strategies for addition and subtraction. There has been considerable discussion of pupil and curriculum choices between collections-based and sequence-based strategies for addition and subtraction (Beishuizen, 2001). Studies suggest that low-attaining pupils can have more success with sequencebased addition strategies, such as jump, than with collection-based strategies, such as split (Beishuizen, 1993). Importantly, if teachers choose to emphasise jump, pupils will require a co-development of knowledge of sequential structure (Menne, 2001). Further, regardless of choice of arithmetic strategy (jump or split) our curriculum should recognise the importance of sequential structure as a basic aspect of number.

Acknowledgements
The authors gratefully acknowledge the support for this project from the Australian Research Council under grant LP0348932 and from the Catholic Education Commission of Victoria. As well, the authors acknowledge the contributions of Gerard Lewis and Cath Pearn (partner investigators) to this project. Finally, the authors express their sincere thanks to the teachers, pupils and schools participating in the project.

Address for correspondence

School of Education, Southern Cross University, PO Box 157, Lismore NSW 2480, Australia E-mail: Bob.Wright@scu.edu.au

References
Anghileri, J. (2000). Teaching number sense. London: Continuum. Anghileri, J. (2001). Intuitive approaches, mental strategies and standard algorithms. In J. Anghileri (Ed.), Principles and practices in arithmetic teaching Innovative approaches for the primary classroom. Buckingham: Open University Press. Askew, M., Brown, M., Rhodes, V., Wiliam, D. & Johnson, D. (1997). Effective teachers of numeracy: Report of a study carried out for the Teacher Training Agency. London: Kings College, University of London. Beishuizen, M. (1993). Mental strategies and materials or models for addition and subtraction up to 100 in Dutch Second Grades. Journal for Research in Mathematics Education, 24(4), 294323. Beishuizen, M. (2001). Different approaches to mastering mental calculation strategies. In J. Anghileri (Ed.), Principles and practices in arithmetic teaching (pp. 119130). Buckingham: Open University Press. Beishuizen, M. & Anghileri, J. (1998). Which mental strategies in the early number curriculum? A comparison of British ideas and Dutch views. British Educational Research Journal, 24(3), 519538. Beishuizen, M., Van Putten, C. M. & Van Mulken, F. (1997). Mental arithmetic and strategy use with indirect number problems up to one hundred. Learning and Instruction, 7(1), 87106. Cobb, P. (1991). Reconstructing elementary school mathematics. Focus on Learning Problems in Mathematics, 13(2), 322. Cobb, P., Gravemeijer, K.P.E., Yackel, E., McClain, K. & Whitenack, J. (1997). Mathematizing and symbolizing: the emergence of chains of signification in one first-grade classroom. In D. Kirshner & J.A. Whitson (Eds.), Situated Cognition Theory:

62

Educational & Child Psychology Vol 24 No 2

Assessing pupil knowledge of the sequential structure of numbers

Social, semiotic and neurobiological perspectives (pp. 151233). Mahwah, NJ: Lawrence Erlbaum. Cooper, T.J., Heirdsfield, A. & Irons, C.J. (1995). Years 2 and 3 childrens strategies for mental addition and subtraction. In B. Atweh & S. Flavel (Eds.), Proceedings of the eighteenth annual conference of the Mathematics Education Research Group of Australasia (pp. 195202). Darwin: Mathematics Education Research Group of Australasia. Dowker, A. (2005). Individual differences in arithmetic: Implications for psychology, neuroscience and education. Hove, UK: Psychology Press. Foxman, D. & Beishuizen, M. (1999). Untaught mental calculation methods used by 11-year-olds. Mathematics in School, 57. Foxman, D. & Beishuizen, M. (2002). Mental calculation methods used by 11-year-olds in different attainment bands: A reanalysis of data from the 1987 APU Survey in the UK. Educational Studies in Mathematics, 51(12), 4169. Fuson, K.C., Richards, J. & Briars, D. (1982). The acquisition and elaboration of the number word sequence. In C.J. Brainerd (Ed.), Progress in Cognitive development: Vol. 1 Childrens logical and mathematical cognition (pp. 3392). New York: Springer-Verlag. Fuson, K.C., Wearne, D., Hiebert, J., Murray, H., Human, P., Olivier, A., Carpenter, T.P. & Fennema, E. (1997). Childrens conceptual structures for multidigit numbers and methods of multidigit addition and subtraction. Journal for Research in Mathematics Education, 28, 130162. Hiebert, J. & Wearne, D. (1996). Instruction, understanding, and skill in multidigit addition and subtraction. Cognition and Instruction, 14(3), 251283. Klein, A.S., Beishuizen, M. & Treffers, A. (1998). The empty number line in Dutch second grades: Realistic versus gradual program design. Journal for Research in Mathematics Education, 29(4), 443464. McIntosh, A., Reys, B.J. & Reys, R.E. (1992). A proposed framework for examining basic number sense. For the Learning of Mathematics, 12, 244. Menne, J. (2001). Jumping ahead: an innovative teaching program. In J. Anghileri (Ed.), Principles and practices in arithmetic teaching Innovative approaches for the primary classroom (pp. 95106). Buckingham: Open University Press. Olive, J. (2001). Childrens number sequences: An explanation of Steffes constructs and an extrapolation to rational number of arithmetic. The Mathematics Educator, 11(1), 49. Resnick, L.B. (1989). Developing mathematical knowledge. American Psychologist, 44, 162169. Ruthven, K. (1998). The use of mental, written, and calculator strategies of numerical computation by upper primary students within a calculator aware number curriculum. British Educational Research Journal, 24(1), 2142. Skwarchuk, S. & Anglin, J.M. (2002). Childrens aquisition of the English cardinal number words: A special case of vocabulary development. Journal of Educational Psychology, 94(1), 107125. Sowder, J.T. (1992). Making sense of numbers. In G. Leinhardt, R. Putnam, & R.A. Hattrup (Eds.), Analysis of arithmetic for mathematics teaching (pp. 151). Hillsdale, NJ: Lawrence Erlbaum. Thompson, I. (Ed.) (1997). Teaching and learning early number. Buckingham: Open University Press. Thompson, I. & Smith, F. (1999). Mental calculation strategies for the addition and subtraction of 2-digit numbers (Report for the Nuffield Foundation). Newcastle upon Tyne: University of Newcastle upon Tyne. Treffers, A. (1991). Didactical background of a mathematics program for primary education. In L. Streefland (Ed.), Realistic mathematics education in primary school (pp. 2156). Utrecht: Freudenthal Institute. Treffers, A. & Buys, K. (2001). Grade 2 and 3 Calculation up to 100. In M. van den Heuvel-Panhuizen (Ed.), Children learn mathematics. The Netherlands: Freudenthal Institute, Utrecht University. Wolters, G., Beishuizen, M., Broers, G. & Knoppert, W. (1990). Mental arithmetic: Effects of calculation procedure and problem difficulty on solution latency. Journal of Experimental Child Psychology, 49, 2030. Wright, R.J. (1994). A study of the numerical development of 5-year-olds and 6-year-olds. Educational Studies in Mathematics, 26, 2544. Wright, R.J., Martland, J. & Stafford, A.K. (2006). Early numeracy: Assessment for teaching and intervention (2nd ed.). London: Paul Chapman Publishing. Wright, R.J., Martland, J. Stafford, A.K. & Stanger, G. (2002). Teaching number: Advancing childrens skills and strategies. London: Paul Chapman Publishing. Yackel, E. (2001). Perspectives on arithmetic from classroom-based research in the United States of America. In J. Anghileri (Ed.), Principles and practices in arithmetic teaching Innovative approaches for the primary classroom (pp. 1531). Buckingham: Open University Press.

Educational & Child Psychology Vol 24 No 2

63

What can intervention tell us about arithmetical difficulties?

Ann Dowker
Abstract
146 children (mean age 6 years 10 months) were included in the Numeracy Recovery intervention programme. The programme involved working with children who have been identified by their teachers as having problems with arithmetic. These children were assessed on nine components of early numeracy, and received weekly individual intervention (half an hour a week for approximately 30 weeks) in the particular components with which they have been found to have difficulty. Six months after the start of intervention, the children showed significant improvement on three standardised tests. Scores on three components (estimation; derived fact strategies; translation between concrete, numerical and verbal presentations of arithmetic problems) were analysed in relation to one another; to general level of addition performance; to standardised test scores; and to improvements in these scores. Regressions showed few relationships between the components, addition level, and standardised test scores. However, addition level predicted both the initial standardised test scores and improvement in these scores; and derived fact strategy use showed a negative relationship to test improvement. Implications of these findings are discussed.

HERE IS a much smaller research base on mathematical development and difficulties than on some other areas of development, such as language and literacy. However, there has recently been an increased emphasis on mathematics in cognitive developmental research (e.g. Baroody & Dowker, 2003; Campbell, 2005; Royer, 2003); in neuroscience (Ansari, Garcia, Lucas, Hamon & Dhilil, 2005; Butterworth, 1999; Dehaene, 1997); and in educational policy and practice in the UK and abroad (Askew & Brown, 2001; Kilpatrick, Swafford & Fundell, 2001). In particular, there is by now overwhelming evidence from experimental, educational and factor analytic studies of typically developing children and adults (e.g. Dowker, 1998, 2005; Geary & Widaman, 1992; Ginsburg, 1977; Lefevre & Kulak, 1994; Siegler, 1988); studies of children with arithmetical deficits (Butterworth, 2005; Dowker, 2005; Geary & Hoard, 2005; Ginsburg, 1977; Jordan & Hanich, 2000; Russell & Ginsburg, 1984; Shalev, Gross-Tsur & Manor, 1997); studies of patients (Butterworth, 1999; Dehaene, 1997; Delazer, 2003); and functional brain imaging studies 64

(Castelli, Glaser & Butterworth, 2006; Dehaene, Spelke, Pinel, Stanescu & Tsivkin, 1999; Gruber, Indefrey, Steinmetz & Kleinschmidt. 2001; Rickard, Romero, Basso, Wharton, Flitman & Grafman, 2000) that arithmetical ability is not unitary. Its broad components include counting, memory for arithmetical facts, the understanding of concepts, and the ability to follow procedures. Each of these broad components has, in turn, a number of narrower components: for example, counting includes knowledge of the counting sequence, ability to follow counting procedures in counting sets of objects, and understanding of the principles of counting: for example, that the last number in a count sequence represents the number of objects in the set, and that counting a set of objects in different orders will give the same answer (Greeno, Riley & Gelman, 1984; Munn, 1997). Moreover, though the different components often correlate with one another, weaknesses in any one of them can occur relatively independently of weaknesses in the others. Weakness in even one component can ultimately take its toll on performance in Educational & Child Psychology Vol 24 No 2 The British Psychological Society 2007

What can intervention tell us about arithmetical difficulties? other components, partly because difficulty with one component may increase the risk of the child relying exclusively on another component, and failing to perceive and use relationships between different arithmetical processes and problems. In addition, when children fail at certain tasks, they may come to perceive themselves as no good at maths and develop a negative attitude to the subject. However, the components described here are not seen as a hierarchy. A child may perform well at an apparently difficult task (e.g. word problem solving) while performing poorly at an apparently easier component (e.g. remembering the counting word sequence). Though certain components may frequently form the basis for learning other components, they need not always be prerequisites. Several studies (e.g. Denvir & Brown, 1986) have suggested that it is not possible to establish a strict hierarchy whereby any one component invariably precedes another component. Many children have difficulties with some or most aspects of arithmetic. It is hard to estimate the proportion who have difficulties, since this depends on the criteria that are used to define difficulty. Moreover, as arithmetical thinking involves such a wide variety of components, there are many forms and causes of arithmetical difficulty, which may assume different degrees of importance in different tasks and situations. It is likely that at least 1520 per cent of the population have difficulties with certain aspects of arithmetic, sufficient to cause significant practical and educational problems for the individual (Bynner & Parsons, 1997) though the proportion that might be described as dyscalculic is much lower than this. The present paper describes a study of the relationships between components of arithmetic in a group of children selected for having arithmetical difficulties. The research is based on the Numeracy Recovery Programme (Dowker, 2001, 2005), an intervention that takes a multiple components view of arithmetic. Educational & Child Psychology Vol 24 No 2 The components were selected on the basis of earlier research (Dowker, 1998) and discussions with teachers about what they considered to be important components of arithmetic, which were sources of difficulty for children. Since childrens performance in these components was assessed in depth before the interventions took place, the study can be used to analyse relationships between the components, and to investigate whether they should indeed be seen as separate multiple components, they are just specific tests of an overarching general arithmetical ability. This can be studied by investigating the initial relationships between components; by investigating whether they predict standardised test scores similarly or differently; and by investigating whether they have a differential role in predicting test improvements. Thus, the first aim of this study is to investigate whether there is a relationship between scores on the different components of the Numeracy Recovery Programme, or whether they are relatively independent. Previous research, as summarised above, has suggested some functional independence between these components. It appears possible, however, that different components can be more closely linked in children with mathematical difficulties, due to difficulties in one having an adverse effect on development of others, or due to the difficulties sharing a common cause. The second aim of the study is to investigate the extent to which different components might contribute independently not only to standardised test scores, but to the level of improvement in these scores following intervention. This is an important question for theoretical reasons, in as much as it can provide information about the specific components of arithmetic that have the greatest effect on overall growth, and for practical reasons, in helping us to predict which children are most likely to benefit from a relatively non-intensive intervention, and which may need more or different help. 65

Ann Dowker

Method
Sample 146 children who had been included in the Numeracy Recovery programme. They included 64 boys and 82 girls. Ages ranged from five years six months to eight years three months (mean: six years ten months; standard deviation 6.1 months). Measures of arithmetical components Arithmetical components were measured using assesments of three selected components of arithmetical ability (derived fact strategy use, estimation and translation) derived from the programme.The three components (described more fully in Appendix 2) were selected because they had already been studied in some detail with unselected groups of children (Dowker, 1997, 1998, 2005).

Outcome measures
The subsample of 145 were assessed before and after intervention on the British Abilities Scales Basic Number Skills subtest (Elliott, 1997), the WOND Numerical Operations test (Wechsler & Rust, 1996), and the WISC Arithmetic subtest (Weschler, 1991). The first two place greatest emphasis on computation abilities and the latter on arithmetical reasoning, and so the BAS and WOND were taken as measures of computation and the WISC as a measure of arithmetical reasoning. Analyses To establish whether the arithmetical components were interrelated, three separate multiple regression analyses were run with each of the three selected components as outcome measures in turn. To establish whether the arithmetical components contributed to the improvement level, multiple regression analyses were run using the improvement scores on BAS, WOND and WISC as outcome measures. The numeracy recovery programme Numeracy Recovery is an intervention based on a multiple components view of arithmetic 66

and has been described previously (Dowker, 2001, 2005). The programme has involved working with children who have been identified by their teachers as having problems with arithmetic. These children are assessed on nine components of early numeracy, which are summarised in Appendices 1 and 2. The children receive weekly individual intervention (half an hour a week) in the particular components with which they have been found to have difficulty. The interventions are carried out by the classroom teachers, using techniques proposed by the researcher. The teachers are released (each teacher for half a day weekly) for the intervention. Each child remains in the program for 30 weeks, or until their teachers feel they no longer need intervention, whichever is shorter. New children join the project periodically. The initial scores on the standardised tests and retest scores after six months of the first 146 children to take part in the project are described here. The median standard scores on the BAS Basic Number Skills subtest were 96 initially and 100 after approximately six months. The median standard scores on the WOND Numerical Operations test were 91 initially and 94 after six months. The median standard scores on the WISC Arithmetic subtest were 7 initially, and 8 after six months (the means were 6.8 initially and 8.45 after six months). Wilcoxon tests showed that all these improvements were significant at the 0.01 level. 101 of the 146 children have been retested over periods of at least a year and have maintained their improvement.

Using the intervention to examine the structure of arithmetic

As outlined in the introduction, there were two aims. The first was to investigate whether there were relationships between scores on the different components or whether they were relatively independent. The second was to examine the extent to which different components might have contributed inde-

Educational & Child Psychology Vol 24 No 2

What can intervention tell us about arithmetical difficulties? pendently to the level of improvement in standardised scores following intervention. The performance of a subsample of children in the intervention group was examined on three selected components of arithmetical ability (derived fact strategy use, estimation and translation) against outcome measures of the standardised tests. The three components (described more fully in Appendix 2) were selected because they had already been studied in some detail with unselected groups of children (Dowker, 1997, 1998, 2005). The children were divided into five levels according to their performance on a mental calculation pre-test. Table 1 gives brief descriptions of the levels, and examples of the problems that could and could not be solved at these levels. In practice, only the first three levels were represented in the present group. and estimate strategies from an unselected sample at the same levels (Dowker, 1998). The figures show that these strategies are used more frequently by children at addition levels 2 and 3 than by children at Addition level 1. Comparison shows that children from the 1998 unselected sample at all three levels used derived fact strategies more often. The mean translation score of the intervention group as a whole was 22.58 (s.d. 9.81). The unselected children studied by Dowker, Gent & Tate (2000) were not divided by performance levels in this way; but the overall average score obtained by 6-year-olds was 32, so on he whole these scores too were higher in the unselected sample.

Relationships among different measures

In order to investigate the independent contributions that different factors made to target measures, entry level multiple regressions were carried out on each of the measures as dependent variable, with the other measures as the predictors. Table 4 shows the results of the regression analyses. Addition Level was included in the analyses because it was a determinant of the precise content of the derived fact strategy and estimation tasks, and it seemed desirable to control for it. As it is an ordinal rather than cardinal score, there could however be doubts as to the appropriateness of its inclusion in a regression. To check for this, the same analyses were carried out with this variable omitted, and results were identical as regards the significance of the other predictors, except that Estimation became a definitely rather than borderline significant predictor of Translation, (beta 0.25; t 2.23; p 0.05) and of course vice versa (beta 0.28; t 2.23; p 0.05). Factors contributing to standardised test scores, and to levels of improvement in these scores. Entry method linear multiple regressions were carried out [1] with each standardised score at t1 as the dependent variable and Age, Addition Level, Addition Derived Fact 67

Results
Table 2 gives the descriptive statistics for all the variables for 146 children in the subsample. Before the intervention, 37 of these children were at Addition Performance Level 1 (Beginning Arithmetic); 86 at Level 2 (Facts to 10) and 23 at Level 3 (Facts to 25). Table 3 shows the mean number of the three kinds of additive strategies used by children from the intervention group at level 1, 2 and 3 respectively. The table also shows, for comparison purposes, data on derived fact

Level

Problem just within range 2 5 8 23 52 2 3 6 44 39

Problem outside range 5 8 23 52 523 3 6 44 39 168

Beginning arithmetic Facts to 10 Simple facts 2-Digit (no carry) 2-Digit (carry)

Table 1: Levels of arithmetical performance in addition Educational & Child Psychology Vol 24 No 2

Ann Dowker n WOND (first score) WOND (second score) WOND improvement BAS (first score) BAS (second score) BAS improvement WISC (first score) WISC (second score) WISC improvement Age at start (months) Addition Level Addition derived fact strategies Estimation Translation 175 146 146 175 146 146 175 146 146 175 175 175 175 175 Mean 90.32 92.79 2.45 95.19 100.38 5.11 6.86 8.33 1.51 80.7 1.94 1.05 4.08 21.9 Standard deviation 10.99 12.34 10.9 11.76 12.45 11.87 2.87 2.5 3.13 6.05 0.67 1.13 2.21 9.93 Range 62 to 123 68 to126 23 to 33 67 to 123 62 to 133 25 to 49 2 to 17 2 to 15 8 to 11 66 to 97 1 to 4 0 to 5 0 to 9 2 to 47

Table 2: Means and standard deviations of test scores

Level 1 Intervention: Number of Derived fact strategies Number of reasonable estimates (out of 9) Translation score Unselected: Number of derived fact strategies Number of reasonable estimates 0.7 1.8 Mean 0.3 Sd 0.6 Mean 1.3

Level 2 Sd 1.2 Mean 1.2

Level 3 Sd 1.1

1.7

4.5

2.3

4.3

2.2

15.3

7.2

25.0

9.5

25.9

8.4

3.0

2.4

5.4

5.0

Table 3: Means and standard deviations of additive strategies by addition performance level for intervention and previous (unselected) samples 68 Educational & Child Psychology Vol 24 No 2

What can intervention tell us about arithmetical difficulties? Predictor Derived facts Addition level Estimation Translation Age in months Estimation Addition level Derived facts Translation Age in months Translation Addition level Derived facts Estimation Age in months Table 4: Results of regressions on task scores 0.25 0.05 0.21 0.27 2.2 0.43 1.94 2.47 P n.s. P P 0.06 (*) 0.05* 0.05* 0.09 0.09 0.26 0.04 0.67 0.78 1.93 0.32 n.s. n.s. P n.s. 0.06(*) 0.17 0.1 0.06 0.8 1.32 0.78 0.43 0.6 n.s. n.s. n.s n.s. Beta t Significance

Strategies, Estimation, and Translation as the predictors and [2] the improvement in score (t2 t1) as the dependent variable and Age, t1 standardised scores, Addition Level, Addition Derived Fact Strategies, Estimation, and Translation as the predictors.

Discussion
The children with arithmetical difficulties appeared in general to show some weaknesses in the components investigated: derived fact strategies, estimation and translation, as compared with unselected children in other studies. The groups may not be directly comparable, due to the time lapse and changes in the educational system since the studies of the unselected children; but the figures suggest that the children in the intervention group used on average somewhat fewer derived fact strategies and make fewer reasonable estimates than the unselected children, and that this was especially true of those at the higher addition performance levels. Perhaps Educational & Child Psychology Vol 24 No 2

children at the higher addition performance levels were only regarded by their teachers as arithmetically weak and needing intervention if they did have additional weaknesses in aspects of arithmetical reasoning. The study of unselected childrens translation (Dowker, 2005) did not assess the childrens addition performance level; but their translation performance as a group appeared to be somewhat better than that of the children with arithmetical difficulties. It should be noted, however, that not all children in the latter group performed poorly in the components investigated; that derived fact strategy use and estimation were often quite good; and that their translation performance in particular seemed better than that which would have been predicted by Hughes (1986), who found extreme translation difficulties even in unselected 9-year-olds. Standardised test scores are more related to some specific components of arithmetic 69

Ann Dowker Predictor WOND (first score) Addition level Derived facts Estimation Translation Age in months WOND (improvement) Addition level Derived facts Estimation Translation Age in months WOND (first score) BAS (first score) Addition level Derived facts Estimation Translation Age in months BAS (improvement) Addition level Derived facts Estimation Translation Age in months BAS (first score) WISC (first score) Addition level Derived facts Estimation Translation Age in months Table 5: Results of regressions on these scores (Continued on next page) 70 standardised 0.39 0.07 0.11 0.27 0.21 test 3.3 0.6 1.0 2.24 1.81 scores and P P 0.01** n.s. n.s. 0.5* n.s. improvements in 0.25 0.33 0.08 0.05 0.25 0.66 2.2 3 0.78 0.44 2.22 5.6 P P P P 0.05* 0.01** n.s. n.s. 0.05* 0.01** 0.47 0.09 0.16 0.05 0.14 3.94 0.78 1.24 0.43 1.19 P 0.01** n.s. n.s. n.s. n.s. 0.43 0.27 0.21 0.03 0.4 0.68 2.86 2.33 1.85 0.22 2.9 4.81 P P P P P 0.01** 0.01** 0.07 (*) n.s. 0.05* 0.01** 0.6 0.13 0.01 0.08 0.41 5.71 1.31 0.13 0.74 4.03 P P 0.01** n.s. n.s. n.s. 0.01** Beta t Significance

Educational & Child Psychology Vol 24 No 2

What can intervention tell us about arithmetical difficulties? Predictor WISC (improvement) Addition level Derived facts Estimation Translation Age in months WISC (first score) 0.1 0.05 0 0.02 0.04 0.82 0.96 0.43 0 0.15 0.43 8.0 P n.s. n.s. n.s. n.s. n.s. 0.01** Beta t Significance

Table 5: (continued) Results of regressions on standardised test scores and improvements in these scores than to others. Addition Level was a significant independent predictor of initial scores in all tests: not surprisingly, as all the standardised tests emphasised competence at calculation. Neither Estimation nor Derived Fact Strategies was an independent predictor of any test scores. Translation predicted performance in the WISC Arithmetic test, but not in the other tasks. This may be due to the fact that the WISC Arithmetic test places an emphasis on word problem solving, whereas the other tests place greater emphasis on calculation and on reading and writing numbers. The pattern was more complex with regard to the factors affecting improvements in performance. Initial scores were negative predictors of improvement in all three tests. This is not surprising, as the lower the initial score, the more room for improvement; and the initial scores had indeed only been included in the regressions so as to control for them, while investigating the effects of the scores on different components of arithmetic. None of these scores did in fact predict improvement in Arithmetic. With regard to improvements in the WOND and BAS tests, there was no significant effect of either Estimation or Translation. Addition Level was, however a positive independent predictor of improvement (even after partialling out the initial score on the standardised test) and Derived Fact Strategy score was a negative independent predictor. It is intriguing that different components of arithmetic appear to play different roles in predicting improvement; but a lot more research needs to be done before we can draw strong conclusions about the nature of these roles. The finding that derived fact strategy scores appear to act as a negative predictor of improvement is puzzling, especially as most studies (Dowker, 1998, 2005) show a positive relationship between derived fact strategy use and other aspects of arithmetical performance. There are at least three possible explanations. One is that, in the case of children with arithmetical difficulties, good derived fact strategy use actually does make them less likely to show overall improvement in arithmetic at least in the short term, perhaps because the existence of useful compensatory strategies reduces the need and motivation to acquire conventional strategies. A second possible explanation is that the association is secondary to some other characteristics that are associated with relative strengths in derived fact strategy use. Dowker (1998) found that children who are significantly better at derived fact strategy use than at exact calculation (whether or not they have actual calculation difficulties) are

Educational & Child Psychology Vol 24 No 2

71

Ann Dowker more likely than others to show large discrepancies, in either direction, between Verbal and Performance IQ. It may be that such discrepancies are associated with lower improvements, at least in the short term, and especially in a group selected for mathematical difficulties, which may include some children with co-morbid disorders associated with Verbal/Performance IQ discrepancies. This hypothesis should be investigated in the future by giving the children IQ and perhaps other cognitive tests, and investigating relationships to improvements in arithmetic. A third possible explanation is that the effects on improvement in performance are not linked to the childs arithmetical or cognitive characteristics, but to the form of intervention that was given. Children who were weak at derived fact strategies were given interventions that involved derived fact strategy training; those who already performed well at derived fact strategies, did not receive such intervention. Perhaps derived fact strategy training has a particularly beneficial effect. It may be desirable to investigate the effect of giving derived fact strategy training to all children in an intervention program, in addition to the more specific interventions for components in which particular children demonstrate weaknesses. Investigations revealed some general correlations between specific components of arithmetic, but there were few significant independent relationships between these components. It is possible that more such relationships would be found if a larger sample were studied. The results are, however, consistent with the view that arithmetic is made up of many components; that there is no such thing as arithmetical ability; only arithmetical abilities (Dowker, 2005). Indeed, they show rather less relationship between different components than was found in Dowkers (1998) study of an unselected group of children. That study (which did not investigate translation) showed significant independent effects of addition level on both estimation and derived fact strategies, and a particularly strong independent relationship between 72 derived fact strategy use and estimation. By contrast, in the present study, derived fact strategy use and estimation not only did not show an independent relationship; they were not even correlated before other factors were partialled out. It may be that in a group of children with arithmetical difficulties, there is even less relationship between different arithmetical components than in a typical sample. Perhaps in completely typical mathematical development, different components, though perhaps functionally separable, do inform and reinforce one another in the course of development (as Baroody & Ginsburg (1986) propose for the development of principles and procedures in younger children, in their mutual development theory). In children with arithmetical difficulties, this integration may not occur to the same extent, either because it is impeded by marked weaknesses in individual components, or because of a failure in the integrative process itself. However, another possible explanation for any differences found between Dowkers (1998) study and the present one is that the findings are linked to educational changes. There were crucial changes in British mathematics education in 19981999, with the introduction of the National Numeracy Strategy (Department for Education and Employment, 1998). There is no transparently obvious reason why the changes in mathematics education at that time should have led to greater dissociation between different components of arithmetic. If anything, one might have expected that the more explicit structure of the mathematics curriculum, and the inclusion of derived fact strategies and estimation in primary mathematics instruction, might have led to the components becoming more integrated with one another. However, curriculum changes sometimes have effects other than the predictable or intended ones. In any case, it is risky to assume that differences in findings between groups are entirely the result of group characteristics, when there are also differences in the instruction that they have received. It would be desirable to Educational & Child Psychology Vol 24 No 2

What can intervention tell us about arithmetical difficulties? compare the children in this group with unselected children undergoing similar mathematics instruction. Such a study is currently underway. provide children with individual attention: for example, interventions in literacy. It is also desirable to investigate whether different approaches to such intervention (e.g. age when intervention starts; degree of intensiveness; degree of individualization; the particular components emphasised) may differ in general effectiveness and/or differentially appropriate to different groups of children. The present study also demonstrates the possibilities for bidirectional relationships between research and intervention. The project integrates the implementation and evaluation of the intervention scheme with the investigation of individual differences in, and relationships between, certain selected components of arithmetic. Thus, the intervention project, which was inspired by my earlier research and conclusions about the components of arithmetic, also serves to test theories about these components.

Conclusion
The findings discussed in this paper strongly support the view that arithmetic is made up of multiple components rather than being unitary; though further research is necessary to establish how the relationships between components vary with ability level and with educational factors.They are also consistent with the view that arithmetical difficulties can be significantly ameliorated by interventions targeting specific weaknesses, though more research comparing different forms of intervention would be needed to confirm this view. There is more research to be done on exactly how such interventions lead to improvement. Indeed, the project is undergoing further development and evaluation. Further investigations are of course necessary to show whether and to what extent specific interventions in mathematics are more effective in improving childrens mathematics than other interventions which

Address for correspondence

Ann Dowker, Department of Experimental Psychology, University of Oxford, South Parks Road, Oxford OX1 3UD. E-mail: ann.dowker@psy.oc.ac.uk

References
Askew, M. & Brown, M. (2001, eds.). Teaching and Learning Primary Numeracy: Policy, Practice and Effectiveness. Notts: British Educational Research Association. Ansari, D., Garcia, N., Lucas, E., Hamon, K. & Dhilil, B. (2005). Neural correlates of symbolic number processing in children and adults. Neuroreport, 16, 17691775. Baroody, A.J. & Ginsburg, H.P. (1986). The relationship between initial meaningful and mechanical knowledge of arithmetic. In J. Hiebert (Ed.), Conceptual and Procedural Knowledge: The Case of Mathematics (pp. 75112). Hillsdale, NJ: Erlbaum. Baroody, A.J., Ginsburg, H.P. & Waxman, B. (1983). Childrens use of mathematical structure. Journal for Research in Mathematics Education, 14, 156168. Baroody, A. and Dowker, A. (2003). The Development of Arithmetical Concepts and Skills. Mahwah, N.J.: Erlbaum. Butterworth, B. (1999). The Mathematical Brain. London: Macmillan. Butterworth, B. (2005). Developmental dyscalculia. In J.I.D. Campbell (Ed.), Handbook of Mathematical Cognition (pp. 455467). Hove: Psychology Press. Bynner, J. and Parsons, S. (1997). Does Numeracy Matter? London: Basic Skills Agency. Campbell. J.I.D. (Ed.), (2005) Handbook of Mathematical Cognition. Hove: Psychology Press. Canobi, K.H., Reeve, R.A. & Pattison, P.E. (1998). The role of conceptual understanding in childrens addition problem solving. Developmental Psychology, 34, 882891. Canobi, K.H., Reeve, R.A. & Pattison, P.E. (2003). Young childrens understanding of addition concepts. Educational Psychology, 22, 513532. Castelli, F., Glaser, D.E. & Butterworth, B. (2006). Discrete and analogue quantity processing in the parietal lobe: a functional MRI study. Proceedings of the National Academy of Sciences, 103, 46934698. Cowan, R., Bailey, S., Christakis, A. & Dowker, A.D. (1996). Even more precisely understanding the order irrelevance principle. Journal of Experimental Child Psychology, 61, 84101.

Educational & Child Psychology Vol 24 No 2

73

Ann Dowker
Decorte, E. & Verschaffel, L. (1987). The effect of semantic structure on first-graders strategies for solving addition and subtraction word problems. Journal for Research in Mathematics Education, 18, 363381. Dehaene, S. (1997). The Number Sense. London: Macmillan. Dehaene, S., Spelke, E., Pinel, P., Stanescu, R. & Tsivkin, S. (1999). Sources of mathematical thinking: behavioural and brain-imaging evidence. Science, 5416, 970974. Delazer, M. (2003). Neuropsychological findings on conceptual knowledge of arithmetic. In A. Baroody & A. Dowker (Eds.), The development of arithmetical concepts and skills (pp. 385407). Mahwah, N.J. Erlbaum. Denvir, B. & Brown, M. (1986). Understanding number concepts in low-attaining 7-to-9-yearolds. Part 1: Development of descriptive frameworks and diagnostic instruments. Educational Studies in Mathematics, 17, 1536. DfEE (1999). The National Numeracy Strategy: Framework for Teaching Mathematics, Reception to Year 6. London: Department for Education and Employment. Dowker, A.D. (1997). Young childrens addition estimates. Mathematical Cognition, 3, 141154. Dowker, A.D. (1998). Individual differences in normal arithmetical development. In C. Donlan (Ed.), The Development of Mathematical Skills (pp. 275302). Hove: Psychology Press. Dowker, A.D. (2001). Numeracy recovery: a pilot scheme for early intervention with young children with numeracy difficulties. Support for Learning, 16, 610. Dowker, A.D. (2005). Individual Differences in Arithmetic: Implications for Psychology, Neuroscience and Education. Hove: Psychology Press. Dowker, A., Gent, M. & Tate, L. (2000). Is arithmetic a foreign language? Young childrens translations between arithmetical problems in numerical, concrete and word problem formats. Paper presented at the European Congress of Psychology, London. Elliott, C.D., Smith, P. & McCullouch, K. (1997). British Ability Scales, 2nd Edition. Windsor: NFER. Fuson, K. (1992). Research on learning and teaching addition and subtraction of whole numbers. In G. Leinhardt, R. Putnam & R.A. Hattrop (Eds.), Analysis of Arithmetic for Mathematics Teaching. Hillsdale, N.J.: Erlbaum. Geary, D.C. & Hoard, M.K. (2005). Learning disabilities in arithmetic and mathematics: theoretical and empirical perspectives. In J.I.D. Campbell (Ed.), Handbook of Mathematical Cognition (pp. 253267). Hove: Psychology Press. Geary, D.C. & Widaman, K.F. (1992). On the convergence of componential and psychometric models. Intelligence, 16, 4780. Ginsburg, H.P. (1977). Childrens Arithmetic: How They Learn It and How You Teach It. New York, N.Y.: Van Norstrand. Gray, E. & Tall, D. (1994). Duality, ambiguity and flexibility: a proceptual view of simple arithmetic. Journal for Research in Mathematics Education, 25, 115144. Greeno, J., Riley, M. & Gelman, R. (1984). Conceptual competence and childrens counting. Cognitive Psychology, 16, 94134. Griffin, S., Case, R. & Siegler, R. (1994). Rightstart: providing the central conceptual prerequisites for first formal learning of arithmetic to students at risk for school failure. In K. McGilly (Ed.), Classroom Learning: Integrating Cognitive Theory and Classroom Practice (pp. 2550). Boston: M.I.T. Press. Gruber, O., Indefrey, P., Steinmetz, H. & Kleinschmidt, A. (2001). Dissociating neural correlates of cognitive components in mental calculation, Cerebral Cortex, 11, 350369. Holt, J. (1966). How Children Fail. New York: Pitman. Hughes, M. (1986). Children and Number: Difficulties in Learning Mathematics. Oxford; Blackwell. Jordan, N. & Hanich, L.B. (2000). Mathematical thinking in second grade children with different forms of LD. Journal of Learning Disabilities, 33, 567578. Kilpatrick, J., Swafford, J. & Fundell, B. Adding It Up: Helping Children Learn Mathematics (Eds.) (2001). Washington, D.C.: National Academy Press. Lefevre, J.A. & Kulak, A.G. (1994). Individual differences in the obligatory activation of addition facts. Memory and Cognition, 22, 188200. Lefevre, J.A., Greenham, S.L. & Waheed, N. (1993). The development of procedural and conceptual knowledge in computational estimation. Cognition and Instruction, 11, 95132. Mayer, R. (2003). Mathematical problem solving. In J.M. Royer (Ed.), Mathematical Cognition. (pp. 6992). Greenwich, CT: Information Age Publishing. Mazzocco, M. & Myers, G.F. (2003). Complexities in identifying and defining mathematics learning disability in the primary school years. Annals of Dyslexia, 53, 218253. Munn, P. (1997). Childrens beliefs about counting. In I. Thompson (Ed.), Teaching and Learning Early Number (pp. 919). Buckingham: Open University Press. Ostad, S. (1998). Developmental differences in solving simple arithmetic word problems and simple number-fact problems: a comparison of mathematically normal and mathematically disabled children. Mathematical Cognition, 4, 119. Rickard, T.C., Romero, S.G., Basso, G., Wharton, C., Flitman, S. & Grafman, J. (2000) The calculating brain: an fMRI study. Neuropsychologia, 38, 325335.

74

Educational & Child Psychology Vol 24 No 2

What can intervention tell us about arithmetical difficulties?

Riley, M.S., Greeno, J.G. & Heller, J.I. (1983). Development of childrens problem-solving ability in arithmetic. In H.P. Ginsburg (Ed.). The Development of Mathematical Thinking (pp. 153196). New York: Academic Press. Royer, J.M. (Ed.), (2003) Mathematical Cognition. Greenwich, CT: Information Age Publishing. Russell, R. & Ginsburg, H.P. (1984). Cognitive analysis of childrens mathematical difficulties. Cognition and Instruction, 1, 217244. Shalev, R.S., Gross-Tsur, V. & Manor, O. (1997). Neuropsychological aspects of developmental dyscalculia. Mathematical Cognition, 3, 105170. Siegler, R.S. (1988). Individual differences in strategy choice: good students, not-so-good students and perfectionists. Child Development, 59, 833851. Siegler, R.S. & Booth, J.L. (2005). Development of numerical estimation: a review. In J.I.D. Campbell (Ed.), Handbook of Mathematical Cognition (pp. 197212). Hove: Psychology Press. Sowder, J.T. & Wheeler, M.M. (1989). The development of concepts and strategies use in computational estimation. Journal for Research in Mathematics Education, 20, 130146. Straker, A. (1996). Mental Maths for Ages 5 to 7: Teachers Book. Cambridge: Cambridge University Press. Straker. A. (1999). The National Numeracy Project, 19961999. In I. Thompson (Ed.), Issues in Teaching Numeracy in Primary Schools (pp. 3948). Buckingham: Open University Press. Temple. C.M. (1994). The cognitive neuropsychology of the developmental dyscalculias. Cahiers de Psychologie Cognitive/Current Psychology of Cognition, 13, 351370. Tronsky, L.T. & Royer, J.M. (2003). Relationships among basic computational automaticity, working memory and complex problem solving: what we know and what we need to know. In J.M. Royer (Ed.), Mathematical Cognition (pp. 117145). Greenwich, CT: Information Age Publishing. Warrington, E.K. (1982). The fractionation of arithmetical skills: a single case study. Quarterly Journal of Experimental Psychology, 34A, 3151. Wynn, K. (1990). Childrens understanding of counting. Cognition, 36, 155193. Yeo, D. (2003). Dyslexia, Dyspraxia and Mathematics. London: Whurr. Wechsler, D. (1991). Wechsler Intelligence Scale for Children, 3rd Edition. London: Harcourt. Wechsler, D. & Rust, J. (1996). Wechsler Objective Numerical Dimensions. London: Harcourt.

Educational & Child Psychology Vol 24 No 2

75

Ann Dowker

Appendix 1
Aspect Counting procedures Assessment (i) accuracy of counting sets of 5, 8, 10, 12 and 21 objects; (ii) rote verbal counting to 10 and to 20 The children watch an adult count a set of objects, and are then asked to predict the result of further counts: (a) in the reverse order (b) after the addition of an object and (c) after the subtraction of an object. Children are shown a set of 5 items, and then shown one more item being added, and asked to say, without counting, how many there are now. This is repeated up to 15. Children are shown a set of 10 items, and then shown one item being subtracted, and asked to say, without counting, how many there are now. This is repeated down to zero. Intervention Children are given practice in counting sets of objects, ranging in number from 5 to 25. Children practice counting and answering order-irrelevance questions about very small numbers of counters (up to 4). He adult makes statements such as, Its four this way, and four that way its four whichever way you count it! The child is given practice with increasingly large sets. Practice in observing and predicting the results of such repeated additions and subtractions with counters (up to 20). Verbal number after and number before problems: What is the number before 8?, What is the number after 14?, etc. Worksheets devised for the project, including repeated addition and subtraction by 1 from a set of circles. Number After Dominoes and Number Before Dominoes which are played like dominoes except that the added domino must be the number after (or before) the end item, rather than the same number. Practice in reading and writing numbers, sorting objects into groups of ten, and recording them as 20, 30, etc. and sorting and recording tasks where there are extra units as well as the groups of ten.

Counting principles (i) Order irrelevance (ii) Repeated addition by 1 and repeated subtraction by 1

Written symbolism for numbers

Children are asked to read aloud a set of single-digit and two-digit numbers. A similar set of numbers is dictated to them for writing.

(Continued ) 76 Educational & Child Psychology Vol 24 No 2

What can intervention tell us about arithmetical difficulties? Aspect Place value Assessment Children are asked to add 10s to units (20 3), to add 10s to 10s (20 30) and to combine the two into one operation (20 33). They are also asked to point to the larger number in pairs of 2-digit numbers, that vary in the units (23 vs 26); in the 10s (41 vs 51); or in both tens and units in conflicting directions (27 vs 31; 52 vs 48). Intervention Addition of tens to units and the tens to tens in several different forms: Written numerals, Number line or block, Hands and fingers in pictures, 10-pence pieces and pennies or Any apparatus (Multilink, Unifix) with which the child is familiar. The fact that these give the same answers is emphasised. Practice with arithmetical patterns such as: 20 10; 20 11; 20 12, etc; being encouraged to use apparatus when necessary. Short addition and subtraction word problems of Change, Compare and Combine types are discussed with them: What are the numbers that we have to work with? What do we have to do with the numbers? Do you think that we have to do an adding sum or a takingaway sum? Do you think that John has more sweets or fewer sweets than he used to have?, etc. They are encouraged to use counters to represent the operations in the word problems, as well as writing the sums numerically. Children practice reading and writing numbers. Children with difficulties in reading or writing two-digit numbers (tens and units) are given practice in sorting objects into groups of ten, and recording them as 20, 30, etc. They are then given such sorting and recording tasks where there are extra units as well as the groups of ten. The children are shown the same problems in different forms and are shown that they give the same results

Word problem solving

Word problems test (Griffin et al., 1995)

Translation between concrete, verbal and written number

Children are asked to read aloud a set of single-digit and two-digit numbers. A similar set of numbers is dictated to them for writing. Children are presented with sums and are invited to show how to do this sum with the counters in a rage of translation contexts: numerical to concrete, concrete to numerical, verbal to concrete, verbal to numerical, numerical to verbal and concrete to verbal. The childrens performance on this pretest is looked at in the context of their performance on the Written Symbolism and Word Problem pretests.

(Continued ) Educational & Child Psychology Vol 24 No 2 77

Ann Dowker Aspect Derived fact strategies Assessment Children are given the Addition and Subtraction Principles Test (Dowker, 1998). They are given the answer to a problem and then asked them to solve another problem that could be solved quickly by the appropriate use of an arithmetical principle. Children are presented with a series of problems with estimates made by imaginary characters (Tom and Mary). The children are asked to evaluate the estimates on a five-point smiley faces scale and to suggest good guesses for these problems themselves. Intervention Training in the use and application of derived fact strategies (specifically commutativity, the n 1 principle, and the inverse principle).

Arithmetical estimation

Children are given additional Tom and Mary evaluation tasks, and are asked to give reasons for their answers; and further practice in producing their own estimates. They also play Twenty Questions-type number-guessing games (cf. Holt, 1966), which involve focussing on the range within which a number lies. The child is asked to do the same sums repeatedly in the hope that the repetition will lead to retention of the facts involved. They also play number games (e.g. Straker, 1996) that reinforce number fact knowledge.

Number fact retrieval

Russell and Ginsburgs (1984) Number Facts Test has been expanded to include some subtraction facts.

78

Educational & Child Psychology Vol 24 No 2

What can intervention tell us about arithmetical difficulties?

Appendix 2: The nine components of the numeracy recovery programme

1. Counting procedures: Arguably the most basic component of arithmetic is the ability to make appropriate use of counting. While most six-year-olds have achieved relatively effortless counting, a significant number have not (Griffin, Case & Siegler, 1994; Yeo, 2003). This may seriously impede their development of arithmetic, both because of the intrinsic logical relationships between counting and arithmetic, and because the effort of counting may distract attention from other aspect of arithmetic (Gray & Tall, 1994; Yeo, 2003). 2. Counting-related principles and their application: Most counting principles are acquired before the age of five or six, even in children with some mathematical difficulties. However, the order irrelevance principle (that counting the same set of items in different orders will result in the same number) is usually the latest of the main counting principles to be acquired (Cowan, Dowker, Christakis & Bailey, 1996); and is sometimes weak even in six-year-olds. Evidence suggests that understanding the order irrelevance principle is closely related to the ability to predict the result of adding or subtracting an item from a set (Cowan, Dowker, Christakis, & Bailey, 1996; Dowker, 2005). 3. Written symbolism for numbers: Here is much evidence that children often experience difficulties with written arithmetical symbolism of all sorts, and in particular with repre-senting quantities as numerals (Ginsburg, 1977; Fuson, 1992). With regard to this component, children are asked to read aloud a set of single-digit and two-digit numbers. A similar set of numbers is dictated to them for writing. 4. Understanding the role of place value in number operations and arithmetic : This involves the ability to add 10s to units (20 3 23); the ability to add 10s to 10s (20 30 50); and the ability to combine the two into one operation (20 33 53). A related task involves Educational & Child Psychology Vol 24 No 2 pointing to the larger number in pairs of 2-digit numbers, that vary either just with regard to the units (e.g. 23 versus 26); just with regard to the 10s (e.g. 41 versus 51); or where both tens and units vary in conflicting directions (e.g. 27 versus 31; 52 versus 48). 5. Word problem solving: There is a considerable body of evidence (Hughes, 1986; Mayer, 2003; Riley, Greeno & Heller, 1983) that young children often experience difficulty with word problems in arithmetic, even when they are capable of performing the necessary calculations. Indeed, some studies have suggested (e.g. Russell & Ginsburg, 1984) that performance on word problems is one of the tasks that most strongly defines the difference between mathematically normal and mathematically disabled schoolchildren. It is important to take into account the nature of the problems, as their semantic nature has a strong influence on how easily they are solved. For example children tend to find problems involving changes in quantity (Change problems) easier than those involving comparisons between quantities (Compare problems) (Riley et al., 1983; DeCorte & Verschaffel, 1987). 6. Translation between arithmetical problems presented in concrete, verbal and numerical formats: Several authors have suggested that translation between concrete, verbal and numerical formats is a crucial area of difficulty in childrens arithmetical development. For example, Hughes (1986) reported that many primary school children demonstrate difficulty in translating between concrete and numerical formats (in either direction), even when they are reasonably proficient at doing sums in either one of these formats and has suggested that this difficulty in translation may be an important hindrance to childrens understanding of arithmetic. The childrens performance on the assessment described in the table is 79

Ann Dowker looked at in the context of their performance on the Written Symbolism and Word Problem pretests. If the children perform particularly badly on translations that involve numerical material, and also perform poorly on the Written Symbolism pretest, then it is likely that their main problem is with written symbolism. If the children perform particularly badly on translations that involve verbal material, and also perform poorly on the Word Problems pretest, then it is likely that their main problem is with word problem comprehension. However, if they perform uniformly poorly on all parts of the translation pretest, and/or if their performance on the translations involving numerical material is disproportionately worse than their performance on the Written Symbolism pretest, and/or if their performance on the translations involving the verbal material is disproportionately worse than their performance on the Word Problem pretest, then it is likely that the problem is with translation as such. 7. Derived fact strategies in addition and subtraction: One crucial aspect of arithmetical reasoning is the ability to derive and predict unknown arithmetical facts from known facts, for example by using arithmetical principles such as commutativity, associativity, the addition/subtraction inverse principle (Baroody, Ginsburg & Waxman, 1983; Canobi, Reeve & Pattison, 1998, 2004; Dowker, 1998). For example, if we know that 29 13 42, we can use the commutativity principle to derive the fact that 13 29 is also 42. 8. Arithmetical estimation: The ability to estimate an approximate answer to an arithmetic problem, and to evaluate the reasonableness of an arithmetical estimate, are important aspects of arithmetical reasoning (LeFevre et al., 1993; Siegler & Booth, 2005; Sowder & Wheeler, 1989). 9. Number fact retrieval : Although most psychologists, educators and mathematicians agree that memorisation of facts is not the essence of arithmetic, knowledge of number facts does contribute to efficiency in calculation (Tronsky & Royer, 2003), and is a significant factor in distinguishing between mathematically normal and mathematically disabled children (Geary & Hoard, 2005; Ostad, 1998; Jordan & Hanich, 2000; Russell & Ginsburg, 2004).

80

Educational & Child Psychology Vol 24 No 2

What can intervention tell us about arithmetical difficulties?

Appendix 3: A description of the three tasks used in the study Derived fact strategy task (use of arithmetical principles)
Children are given the answer to a problem and then asked to solve another problem that can be solved quickly by using this answer, together with the principle under consideration. The exact arithmetic problems vary according to the previously assessed calculation ability of the child and are selected to be just a little too difficult for the child to solve unaided. The principles investigated are as follows, in order of their difficulty for the children: 1. The identity principle (e.g. if one is told that 8 6 14, then one can automatically give the answer 14, without calculating, if asked What is 8 6?). 2. The commutativity principle (e.g. if 9 4 13, 4 9 must also be 13). 3. The n 1 principle (e.g. if 23 44 67, 23 45 must be 68). 4. The n 1 principle (e.g. if 9 8 17, 9 7 must be 17 1 or 16). 5. The addition/subtraction inverse principle (e.g. if 46 27 73, then 73 27 must be 46). A child is deemed to be able to use a principle if (s)he can explain it and/or used it to derive at least 2 out of 3 unknown arithmetical facts, while being unable to calculate any sums of similar difficulty when there is no opportunity to use the principle. The Derived Facts score is the total number of derived facts used. good estimates (e.g. 7 2 10, 71 18 90); three that are too small; and three that are too large. The children are asked to evaluate each guess on a fivepoint scale from very good to very silly, represented by a set of schematic faces ranging from very smiling to very frowning, and were themselves asked to suggest good guesses to the sums. The Estimation score is the number of reasonable estimates, out of a maximum score of 9. Reasonable estimates are defined as those that are within 30 per cent of the correct answer, and are also larger than each of the addends.

Translation task
The tasks involve translations between word problem, concrete and numerical formats for additions and subtractions. The concrete formats involve the use of counters. Word problems included Change, Compare and Combine problems for addition and subtraction. All six combinations of presentation and response domain are given, as demonstrated below. No sum in any of these translation tasks includes a number greater than 10. 1. Translation from numerical to concrete (2 items): Children are presented with written sums (2 5 7; 6 4); and invited to show me how to do this sum with the counters. 2. Translation from concrete to numerical (2 items): Children watch the researcher perform arithmetical operations with counters (adding 7 counters to 2 counters; subtracting 6 counters from 9 counters) and are then asked to write down the sum that goes with what I did. 3. Translation from verbal to concrete (5 items): Children are presented with word problems and asked to show me this story with the counters e.g. Paul had 4 sweets; his mother gave him 3 more; so now he has 7 sweets. (Addition: Change semantic category). 81

Estimation task
The addition estimation task has been used in previous studies (Dowker, 1997, 1998, 2003). In the present study, each child is presented with an set of addition problems within their base correspondence as defined above. Each set includes a group of nine sums to which a pair of imaginary characters (Tom & Mary) estimate answers. Each set of Tom & Marys estimates includes three Educational & Child Psychology Vol 24 No 2

Ann Dowker 4. Translation from verbal to numerical (5 items): Children were presented with word problems (similar but not identical to those above), and asked to write down the sum that goes with the story. 5. Translation from numerical to verbal (2 items): Children are presented with written sums (3 6 9, 8 6 2), and invited to tell me a story that goes with this sum. 6. Translation from concrete to verbal (2 items): Children watch the researcher perform arithmetical operations with counters (e.g. adding 5 counters to 3 counters; subtracting 6 counters from 9 counters) and then asked to tell a story to go with what I just did with the counters. The Translation score is calculated by giving 3 for every fully complete response, 2 for every complete response, which involves inverting the operation (e.g. representing the story Peter had 5 buns; he ate 3 buns; so now he has 2 chocolates, as 3 2 5 rather than 5 2 3); 1 for every incomplete response and 0 for every incorrect response. As the total number of items is 18, the maximum possible score is 54.

82

Educational & Child Psychology Vol 24 No 2

Investigating variability in classroom performance amongst children exhibiting difficulties with early arithmetic
Jenny Houssart
Abstract
Researchers in both psychology and mathematics education acknowledge that childrens mathematical performance can vary inexplicably from day to day, though there has been little detailed investigation of the form of variability discussed in this paper. The paper builds on research suggesting this might be a particular issue for children considered to have learning difficulties in mathematics. The children concerned were seven- to nine-year-olds taught together for mathematics in a small group with high levels of adult help in assessment, planning and teaching. Observational research was conducted, with the researcher making weekly visits over the course of a year. Findings synthesise a range of evidence for each child gathered both during planned assessment tasks and as part of routine classroom activity. The data are used to chart the performance of individuals over this period. Findings suggest that arithmetical capacities were not fixed and easily assessed, but varied from day to day. This variability is considered in some detail with the aim of offering explanations for perceived differences. Elements such as task presentation and subtle mathematical differences between tasks provide partial explanations. Many differences remain unexplained, and it is argued that variability is in fact a feature of learning. Finally, implications for practice in assessing children and planning for their mathematical development, are considered.

HE ARTICLE uses detailed observational data gathered in a natural classroom situation, to compare childrens responses to similar tasks over a period of time. While some variability may be expected, observation shows major variability, with children apparently demonstrating specific capacities only to lose them later. The research questions are whether there is evidence for variability of the performance of individuals and whether it is possible to explain any variability. Variability can take many forms, for example between children or for individual children between different aspects of mathematics, different aspects of number or tasks presented in different ways. Another form of variability is between similar or identical tasks carried out by the same child on different occasions. This final type of variability is the Educational & Child Psychology Vol 24 No 2 The British Psychological Society 2007

main focus of this research although other types are relevant. The research arises from a wider ethnographic study carried out in various settings. This particular research question arose in this setting from the concern of the adults who worked there. They expressed concern at the difficulty of assessing children during classroom activities due to perceived variability of their responses. The purpose of my investigation, therefore, was firstly to interrogate my existing data to see if there was evidence for this variability. A further question was whether variability really occurred over identical or similar tasks or whether it could in fact be explained by task differences. The children concerned were all considered to have learning difficulties in mathematics and the article ends by considering whether variability is a particular issue for 83

Jenny Houssart such children or is in fact a wider phenomenon. The papers contribution is to show that classroom research confirms clinical psychological experiments on variability, suggesting that this may be integral to the very process of learning itself rather than an aberration requiring alternative explanations. A recent detailed discussion of individual differences in arithmetic (Dowker, 2005) brings together findings confirming that many children show variability across different aspects of number. A central theme of Dowkers book is that it is not appropriate to talk about arithmetical ability, but rather arithmetical abilities which can be grouped in to several categories. It is suggested that there can be strong discrepancies in either direction between almost any two components. Although the focus is mainly on variability between aspects of arithmetic, some points are also made relevant to variability between occasions. For example, in relation to her studies of estimation, Dowker suggests that the know or not know dichotomy in relation to particular types of arithmetic is inadequate, and she refers to a zone of partial knowledge and understanding. She provides many examples of individuals demonstrating uneven performance across different aspects of mathematics and uses the phrase cognitively uneven for those who have verbal reasoning which is either much better or much worse than their spatial reasoning. Despite detailed discussion of the performance of individuals on particular aspects of arithmetic, the author stresses the difficulty in trying to break down arithmetic into components for the purposes of assessment and intervention. A major contributor to discussions of variability in arithmetic is the noted psychologist Siegler (1996) who focuses on childrens strategies and how changes occur in their strategies and ways of thinking. He suggests that evidence for variability is present in the detail of much research but that, for several reasons, this remains peripheral and underemphasised. A key reason he advances is that within Developmental Psychology variability between age groups is the main focus of attention; hence variability within groups is minimised. Although there is little work aimed directly at studying variability, clues to variability can be found by looking at the findings of apparently contradictory studies. In particular, work challenging Piaget (e.g. 1952, 1953) highlighted Educational & Child Psychology Vol 24 No 2

Background
Both psychologists and educational researchers note, frequently as an aside, that childrens mathematical capacities are not fixed and easy to assess but vary markedly from day to day or even between similar tasks on the same day. The issue is not currently foregrounded in English schools, possibly because it conflicts with contemporary initiatives. For example, the National Numeracy Project (DfEE 1999) focuses on detailed learning objectives, often to be shared with children with the hope that they will be achieved within the lesson. It is stated explicitly (page 33) that assessment during every lesson should check that children have grasped the main teaching points and determine whether they can move on, or whether misunderstandings need to be addressed. The implication is that such decisions can reasonably be made in a lesson and that, in the medium term, records can be kept confirming which key objectives have been met. The current mathematics curriculum in England puts an emphasis on teaching number, though other aspects of mathematics are also included. Much of the detailed discussion about children with apparent difficulties in mathematics focuses on number (e.g. Dowker, 2004, Wright et al., 2002). Other writers, (e.g. Gabb, 2005) assert that pupils with special educational needs should have an appropriate diet of mathematics, not just restricted to basic number work. My own research in other settings suggests that some pupils who are apparently low attainers in mathematics can respond well to non-number tasks such as measuring or shape (Houssart, 2004). However, most detailed studies of children considered to be low attainers concentrate on aspects of number. 84

Investigating variability in classroom performance amongst children young children able to cope with concepts such as class inclusion and conservation at an earlier age than previously thought (e.g. McGarrigle & Donaldson, 1974, McGarrigle, Grieve & Hughes, 1978). Such findings are discussed by Siegler (1996) who suggests that recognising variability of thinking is important in trying to reconcile evidence of young childrens competence with evidence of their incompetence. Within Education some writers touch on possible variability of performance, often in an assessment context. Black (1998) considers whether a pupil might perform differently on different days when discussing the reliability and validity of formal tests. He argues that this has received far less attention from researchers than other issues of reliability and validity. It could be argued that if variability was widely recognised this would undermine the system of formal assessment currently in use, and it is therefore perhaps not surprising that this is not a high profile issue. Discussions held with teachers (Watson, 2000) indicate their strong belief that pupil performance varies from day to day. These teachers taught mathematics to children aged between ten and twelve years. Asked about how they reached judgments about their pupils mathematics, roughly half of the thirty teachers said it was possible for pupils to be able to do some mathematics on one day but not on the next. Related work by the same author (Watson, 2001, 2006) suggests that the whole issue of teacher assessment is problematic and that it is not possible to say for certain what a pupil knows. Work aimed specifically at assessing pupils with learning difficulties in arithmetic also accepts that there is some variability (Wright, Martland & Stafford, 2000). In this case the emphasis is again on variability of strategy, with the authors saying that children frequently use strategies that are less sophisticated than those of which they are capable. They give possible reasons for this, including the facts that a less sophisticated strategy may be easier or that some feature of the childs thinking prior to solving Educational & Child Psychology Vol 24 No 2 the current task may focus them on a less sophisticated strategy. A related, currently more prominent issue in education is that pupils may perform differently according to how a task is presented. Clausen-May (2005) discusses pupils with different mathematics learning styles. She uses the outline VAK model, incorporating Visual, Auditory and Kinesthetic styles in a discussion of how pupils may exhibit learning differences in mathematics with preferences for tasks involving seeing, hearing or doing. She also suggests that classroom mathematics tasks may have a strong literacy base, disadvantaging some pupils. Different thinking styles in mathematics are also discussed by Chinn (2004) who uses the idea in the context of pupils considered to have learning difficulties in mathematics. Both writers work is relevant to the current study because the data are drawn from a range of classroom tasks and it is possible that variability might be explained by pupils being asked to work on tasks presented in different ways. There is therefore some support amongst both writers and teachers for the idea of variable performance. However, research on the issue is relatively sparse for methodological reasons; such studies are time-consuming and problematic. Most of the research evidence is provided by Siegler and his colleagues (Siegler & Jenkins, 1989; Siegler, 1996; Siegler & Stern, 1998). These studies use micro genetic methods where individuals undertake the same task on several occasions, methods described in detail by Siegler & Crowley (1991) who argue that the concept of micro genetic methods and the rationale for using them go back for over eighty years. This paper uses data gathered in a classroom whilst children undertake their normal activities. It is similar in some ways to micro genetic studies since it is longitudinal, with children frequently returning to similar tasks. It differs from them in that the researcher can not control the type or number of tasks carried out and because it is conducted in a classroom context rather 85

Jenny Houssart than as a clinical experiment. This carries limitations in the number of times a child may work on each calculation but has the advantage of providing an opportunity to see whether Siegler and his colleagues results in experimental situations can be replicated in a natural classroom context. Context and method Data are drawn from a long-term research project carried out with four groups of children using ethnographic methods. The data considered were gathered from one group of children aged seven to nine years old. The twelve children in the group, who were drawn from different classes, were all considered to have learning difficulties as far as mathematics was concerned. The group was similar to other groups considered to be low attainers in mathematics (Denvir, Stolz & Brown, 1982; Haylock, 1991; Robbins, 2000) in that they exhibited a range of apparent difficulties with a corresponding range of possible reasons. The teacher was joined in mathematics lessons by two classroom assistants. Whilst researching, I visited the group for one mathematics lesson each week for a year, adopting a role similar to that of the classroom assistants and observed the childrens responses to activities carried out with the whole group. While children worked alone or in smaller groups, I worked alongside them as requested by the teacher. I also occasionally conducted assessment activities with individual children as requested by the teacher. Detailed notes of childrens responses to tasks were kept. The data was analysed by examining field notes and extracting all those incidents which named individuals. These were reorganised to obtain a personal record for each child providing information drawn over a year for each individual, arranged chronologically. Incidents were coded on each personal record according to the aspect of mathematics concerned. The next step was to focus on examples where there were a large number of incidents for an individual featuring the 86 same aspect of mathematics. These incidents were extracted and compared in order to examine variations across the year. This process is exemplified in the sections below with particular reference to one child, Claire, and with examples drawn from other children also shown for comparison. Findings Overview of activities. Initial examination of the personal records for each child gave some indication of the curriculum covered by this group of children and in particular which aspects of mathematics were revisited several times. To illustrate this, the information from Claires personal record has been tabulated to show which aspects of mathematics she was observed studying across each of the twenty-five weeks for which there were observations for her (see Table 1). Each mark on the table indicates aspects of mathematics covered in the notes, though these vary from brief mentions to records of whole lessons containing several activities on the same aspect. It is clear from the chart that the majority of time was spent on number rather than on other aspects of mathematics. There are only six entries covering data handling, shape and space and measures and most of these are very brief. This means it is impossible to draw any conclusions about whether Claires performance on number tasks varied from her performance on other aspects. The shape and space observation for Week 11 related to Claires use of a computer program concerned with tables facts. It had been noted that Claire negotiated the maze with apparent ease. This raises a slight possibility that Claire may be happier working with spatial tasks but there is insufficient data to examine this possibility. This was the same for all children, with the vast majority of observations being related to number. For this reason, the focus for the rest of this article will be on variability within number. It is worth noting, however, that any children in this group with strengths in other aspects of mathematics had few opportunities to demonstrate those strengths. Educational & Child Psychology Vol 24 No 2

Data handling * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

Shape and space

Measures

Money

Counting and ordering

Place value

Counting in 2s, 5s and 10s

Multiples of 2, 5 and 10

Odd and even numbers

Addition

Subtraction

Week 1

Week 2

Week 3

Week 4

Week 5

Week 6

Week 7

Week 8

Week 9

Educational & Child Psychology Vol 24 No 2

Week 10

Week 11

Week 12

Week 13

Week 14

Week 15

Week 16

Week 17

Week 18

Week 19

Week 20

Week 21

Week 22

Week 23

Week 24

Investigating variability in classroom performance amongst children

Week 25

87

Table 1: Summary of entries for Claire

Jenny Houssart Focusing on number. Table 1 indicates a large amount of data concerning Claires responses to aspects of mathematics which were revisited frequently throughout the year, such as addition, subtraction and counting in twos, fives and tens. The next step in analysis was to look separately at aspects of mathematics for which there was sufficient data. Observations for chosen aspects were tabulated with brief details included of the task involved. At this point tentative coding was used to indicate whether the task was completed correctly ( ), incorrectly (X), or whether the response was mixed ( ). Because this represents a fairly crude categorisation, some comments were added to give details of the outcome. Where responses were almost entirely correct a tick was used, but any small errors were noted. Similarly, if responses were almost entirely incorrect, a cross was used and any correct response noted. A cross was also used when no response was given, but this was noted. The symbol for mixed response was used for a mixture of correct and incorrect answers and also for cases where work was completed with adult help. Table 2 shows Claires response to addition tasks tabulated in this way and Table 3 shows her response to tasks involving counting in twos, fives and tens and multiples of two, five and ten. Some observations can be made from these two tables. Firstly, it is clear that on several occasions, Claire did not provide answers to questions or did not participate in games or joint counting activities. It is not possible to say for certain whether Claire could have answered the questions or not and this presented a difficulty for the staff working with her. Towards the beginning of the year, for example, when she did not answer questions about the total number of spots on a domino in Week 4, the staff were concerned that she did not understand these very basic ideas and was unable to answer. Later in the year, they felt that Claire sometimes chose not to participate in tasks that she might be able to complete. There were 88 also a few occasions when Claire arrived at lessons distressed about matters outside of mathematics and this may have been a factor in her lack of participation. The tables also show some general tendencies. For example, Table 2 suggests that Claire achieved more success in written addition tasks than in similar tasks presented in games or practical formats. Table 3 suggests that her counting and recitation of tables was better when she was picked to recite in front of the class or to an adult as an assessment activity. However, neither of these patterns applies entirely. Generally, both charts show a mixed performance across the weeks with Claire often failing to answer or answering incorrectly, sometimes on tasks similar to those she had completed previously. This will be considered in more detail later. A similar process was carried out for other children and information is shown for Seth in Tables 4 and 5. Seth was one of the older children in the group, and one of the most successful at mathematics, although the adults remarked that he was not consistently successful. The charts for Seth differ from those for Claire in several ways. Firstly, it was very rare for Seth not to provide answers or participate in activities, so, when he did not answer correctly, incorrect answers were given which sometimes shed light on his difficulties. The chart for addition also indicates that Seth usually completed tasks correctly and sometimes did more than was expected, for example with his comments about addition of zero in Week 4 and his systematic recording of possibilities in Week 7. Although Seths difficulties in Weeks 4, 14 and 17 have no obvious explanation, it is not surprising that he needed help with the second task in Week 20 as it was harder than those carried out previously. Seth also differs from Claire in that the two tables show a different picture. The table for place value suggests that he was less confident with this aspect and had some difficulty with tens and units. This observation is supported by other entries on Seths personal record. For example, Seth had difficulEducational & Child Psychology Vol 24 No 2

Investigating variability in classroom performance amongst children Week Week 2 Week 4 Week 4 Week 4 Week 5 Task details Addition to 10 Adding numerals from cards Adding spots on dominoes Totals to 10 Number walls (written addition) Outcome X X X X Outcome details Does not answer Does not answer Does not answer Does not answer Had adult help with first few examples, then worked alone Mixture of correct and incorrect answers 19 out of 20 calculations correct with no adult help Correct answers, no help X X X X Not participating Initially incorrect but corrected after adult help Not participating Mostly incorrect, heavy rubbing out Mostly incorrect, apparently tried to make use of a pattern Incorrect answer Not participating Completed with intensive adult help Mixture of instant correct responses and correct responses after counting on fingers Task interrupted All answers correct Mixed

Week 6 Week 6 Week 8 Week 15 Week 15 Week 15 Week 16 Week 16

Pairs to 10 (oral activity) Making 10 (written activity) Adding money (oral activity) Addition to 20 (game format) Number card addition to 20 Addition dominoes Make 10 worksheet (no help) Make 20 worksheet (no help)

Week 16 Week 16 Week 17 Week 20

Pairs to 20 (mental, then check with calculator) Addition to 20, dice game Make 10 and make 20 worksheet Addition to 20 using number cards

X X

Week 24 Week 25

Dice addition Addition of money (oral activity)

Table 2: Claires response to addition tasks. Key to outcomes: Mostly or entirely correct, response, X Mostly or entirely incorrect or answers not given ties in identifying multiples of two, five and ten in Week 18. He ringed the number 205, suggesting it was a multiple of two, and when challenged about this said that it did end in two. When asked to ring multiples of ten, he ringed almost everything. Apparently, Seths performance across aspects of number Educational & Child Psychology Vol 24 No 2

was somewhat uneven, with his confidence in addition not being matched with his understanding of place value. Care needs to be taken here as it is acknowledged that place value is often more problematic than addition (Cockburn, 1999). However, Seths performance on place value tasks was also 89

Jenny Houssart Week Week 5 Week 5 Week 7 Task details Counting in 5s (game) Counting in 5s (missing number game) Tape for 2x table Outcome X Outcome details Incorrect answers Correct answer with adult help No response initially, joins in later, numbers are correct but out of step with the questions Claire counts correctly up to 30s then misses out 36 Intermittent participation Joins in enthusiastically and correctly volunteers to recite it alone (not chosen) Joins in enthusiastically and correctly volunteers to say it alone Needs extensive adult help to start with but later completes correctly by counting in 2s, 5s and 10s Recited correctly alone as assessment task Intermittent participation Mostly correct, a few small errors X X Intermittent participation Incorrect answers Incorrect answers Correctly recites alone as assessment activity Does not participate to start with, then joins in correctly Intermittent participation Completed correctly Does not stay on task Completed correctly as individual assessment activity X Initially incorrect, completes with adult help Incorrect answers, lots of adult encouragement

Week 7 Week 8 Week 8

Counting in 2s (joint oral activity) Counting in 5s (joint activity) 5x table tape

Week 9 Week 11

10x table tape 2x, 5x and 10x tables as part of computer game 2x table tape Counting in 2s (joint oral activity) Counting in 2s worksheet Counting in 5s (oral activity) Counting down in 10s Counting in 10s, missing number game 10x table Counting in 5s (joint oral activity) Multiples of 5, card activity Multiples of 5 (writing on board) Multiples of 2, 5 and 10 (worksheet) Counting in 2s, 5s and 10s Counting in 2s oral activity (high-starting numbers) Counting in 2s using coins

Week 12 Week 13 Week 13 Week 14 Week 14 Week 14 Week 15 Week 18 Week 18 Week 18 Week 18 Week 19 Week 19 Week 23

Table 3: Claires response to tasks involving counting in 2s, 5s and 10s and multiples of 2s, 5s and 10s 90 Educational & Child Psychology Vol 24 No 2

Investigating variability in classroom performance amongst children Week Week 4 Week 4 Week 5 Week 7 Week 8 Week 14 Week 15 Task details Adding spots on dominoes Domino addition Oral addition Addition to twelve (written task) Finding coins for given totals Addition to 100 Make twenty (number cards) Outcome X Outcome details Incorrect answers Discusses addition of zero Answers quickly and correctly in plenary All possibilities recorded systematically Correctly answered for totals of 10p, 1, 50p Some errors initially, corrected with adult help Completed correctly, comments on connection between 5 15 and 15 5 Completed without mistakes Completed without mistakes Appeared to complete easily Completed correctly Make ten part corrected easily, slows down and makes some mistakes on make twenty part Completed correctly and helped another child Correct on 16 15, needed help with 32 33 Completed correctly Completed correctly, number lines were available but Seth appeared to make use of known facts sometimes Seth appeared to find these questions (e.g. 3 2) easy

Week 16 Week 16 Week 16 Week 16 Week 17

Make ten worksheet (no help) Make twenty worksheet (no help) Make ten and make twenty worksheet Adding three numbers Make ten and make twenty worksheet

Week 20 Week 20 Week 24 Week 24

Addition program on computer Two-digit addition using hundred square Adding four numbers Dice addition

Week 25

Oral addition

Table 4: Seths response to addition tasks compared with that of others in this group who completed many of them correctly yet had greater difficulties than Seth with addition tasks. Analysis of similar tasks. Further analysis was carried out by extracting activities from each aspect of mathematics which were as similar as possible. This led to a series of shorter Educational & Child Psychology Vol 24 No 2 tables which tended to contain between five and ten activities. Carrying out this process for Claire revealed a series of tables which showed mixed progress. Two of these tables are shown as Table 6 and 7. Table 6 shows those addition tasks which required Claire to add up numbers rather than the harder tasks which involved finding numbers that added to a given total. This chart suggests that 91

Jenny Houssart Week Week 2 Task details Putting tiles on hundred square Outcome Outcome details Initially made mistakes but completed correctly after adult explanation Confused tens and units Made initial mistakes, corrected after adult help Mixture of correct and incorrect answers Completed correctly Completed quickly and correctly X Answered questions incorrectly Could make and read three-digit numbers

Week 2 Week 3 Week 3 Week 7 Week 7 Week 8 Week 16

Making numbers with arrow cards Putting tiles on hundred square Questions about hundred square Representing numbers with tens and units pieces Hundred square jigsaw Counting stick Making numbers with cards

Table 5: Seths response to place value tasks

Claires performance improved over the year, though with some variation across the weeks. It also shows, as mentioned before, that she was more likely to leave questions unanswered than to answer incorrectly. She did also sometimes need adult help to start with. For example, in the number walls activity in Week 5 she used a number line to add, and although she was secure about starting at the first number rather than zero, she needed reminding to count one on the first jump rather than on the starting number. After this initial help, Claire completed this

written activity correctly. The tables for Claire suggest that she tended to participate less enthusiastically in practical and game activities. The detailed observations for Claire contained comments that support this. For example, in Week 4, Claire and some other children struggled with the introduction to addition via spots on dominoes. For this reason, the adults decided to abandon the planned worksheet and carry on with the practical introduction. At this point, Claire said Are we doing games all

Week 4 Week 4 Week 5 Week 8 Week 15 Week 24 Week 25

Adding numerals from cards Adding spots on dominoes Number walls (written addition) Adding money (oral activity) Addition dominoes Dice addition Addition of money (oral activity)

X X

Does not answer Does not answer Had adult help with first few examples, then worked alone Correct answers, no help

Not participating Task interrupted All answers correct

Table 6: Claire adds single-digit numbers 92 Educational & Child Psychology Vol 24 No 2

Investigating variability in classroom performance amongst children Week 7 Tape for 2x table No response initially, joins in later, numbers are correct but out of step with the questions Claire counts correctly up to 30s then misses out 36 Needs extensive adult help to start with but later completes correctly by counting in 2s, 5s and 10s Recited correctly alone as assessment task Intermittent participation Mostly correct, a few small errors Completed correctly as individual assessment activity X Initially incorrect, completes with adult help Incorrect answers, lots of adult encouragement

Week 7 Week 11

Counting in 2s (joint oral activity) 2x, 5x and 10x tables as part of computer game

Week 12 Week 13 Week 13 Week 19 Week 19 Week 23

2x table tape Counting in 2s (joint oral activity) Counting in 2s worksheet Counting in 2s, 5s and 10s Counting in 2s oral activity (high-starting numbers) Counting in 2s using coins

Table 7: Claire counting in twos and multiples of two

day? Much later in the year, when work was being discussed at the beginning of a lesson, Claire made the comment I love sums, I love writing in my book. It appears that Claire considered calculations written in books or on the whiteboard as real maths. Although Claires attitude was extreme, it is echoed in a more moderate form in the findings of Gregory, Snell & Dowker, (1999) who carried out an international study about attitudes to mathematics and suggested that children may see written sums as a core aspect of the subject. Table 7 deals with activities in which Claire was asked to count in twos or identify multiples of two. It is not surprising that Claire found the activity in Week 11 harder, as this computer game required her to multiply given numbers by two, whereas the normal activity was to chant multiplication facts in order only. The second task carried out in Week 19 was also harder than some of the others, as Claire was asked to count in twos Educational & Child Psychology Vol 24 No 2

starting from 74. It is less easy to explain Claires difficulties in Week 23, when asked to count in twos as the teacher dropped 2p coins in a tin. 12p was dropped in the tin and Claire said the amount was 50p. The activity was repeated with intensive adult help. Looking at these activities for Claire suggests variability between occasions which can sometimes but not always be explained by looking at differences in the way tasks were presented or in the mathematics involved. In many cases, Claires incorrect answers could not be explained by considering common errors or misconceptions and it is possible that she sometimes gave answers based on the idea of arithmetic as an arbitrary game as outlined by Ginsburg (1977). For some children, very similar activities were identified, often over consecutive weeks, to see if there was still variability. There were several such cases where children were recorded successfully completing a mathematical task one week and then 93

Jenny Houssart experiencing difficulty with a very similar task the following week. It was also sometimes recorded that children who did not appear to understand a piece of mathematics however it was presented in one lesson were able to cope with it in the next lesson. The following example concerns the performance of one child, Douglas, across similar tasks on three consecutive weeks. Douglas was amongst those in the group having greatest difficulty with mathematics and his record reflected this. In an activity concerning addition pairs which made 20, Douglas did not answer when asked what should be added to 15. In a later activity in the same lesson, each child was given a number tile and asked to find another number tile to go with their tile to make 20. Douglas was given 19 but didnt answer. The teacher asked How many do you need to make 20? There was no answer, so the teacher said Count on. Douglas said, 20. The teacher asked, How many? and he said 20. A number line was found and used to demonstrate. Eventually, Douglas gave the required answer: 1. This and similar incidents led me to conclude that Douglas could not provide the missing number in addition pairs. However, the following week I was proved wrong. The children were given a number and asked What has to be added to it to make 20? They had calculators which they were allowed to check the addition with after the numbers had been suggested. 15 was given as the first number in the activity and Douglas correctly suggested 5 for the second number. The calculator was not used to give Douglas the answer, merely to check, though this incident was in keeping with others which suggested he did better in an activity involving technology even when the technology did not actually do the mathematics for him. This activity suggested to me that contrary to the evidence of the previous week, Douglas was able to understand the idea of pairs of numbers with a given total. The following week the idea of pairs to 20 was introduced in a different way. The children had a worksheet on which they had 94 to ring and join pairs of numbers to total 20. Douglas made little progress with the sheet so I explained what he had to do and picked some individual numbers asking him for the pairs. The sheet was eventually completed with a very high level of help. Although it is not possible to be certain why this variability occurred, it seems likely that part of the explanation lies in the way the task was presented. Perhaps Douglas was motivated by calculator use and presenting the task in that context maximised his potential. Perhaps Douglas understood the underlying idea but declined to cooperate when the task was presented in other ways, or maybe he was genuinely unable to understand the format of the worksheet. Douglas was similar to Claire in responding differently when tasks were presented in different ways. This was true for others in the group. Some, in contrast to Claire, succeeded in mental calculation but had more difficulty when the same calculations were presented in written form. However, task presentation did not explain all or even most of the cases of variability. There were many examples where children were successful on a task one week but failed to carry out a similar task presented in a similar way the following week. Variability also occurred within extended tasks carried out on one occasion as shown in the examples which follow. Analysis of single tasks. Activities in this classroom commonly involved counting in twos, fives or tens. Often this counting was carried out as a group but occasionally children were asked to count alone with the adults and other children listening. On one occasion Neil was asked to count alone in fives to 100s. He started slowly and deliberately. He started 5, 10, 15, 20, 13. Asked to try again, he said 20, 25, 40 and was then correct to 70. He was unsure whether 74 or 75 was next and was helped by the adults. He ended with 75, 80, 85, 100. It was common for the teacher to complete a checklist concerning counting when children were asked to count alone in this Educational & Child Psychology Vol 24 No 2

Investigating variability in classroom performance amongst children way. However, it was not clear whether Neil could be recorded as able to count in fives to 100. There are several possible reasons for his slight difficulties. His slow start suggested that he may have been mentally adding on five every time. The 13 is harder to explain, but could have been a combination of skipping the 25 and confusing 13 with 30. Since Neil managed to count from 20 to 70 correctly, he may have been aware of the number pattern involved. However, he considered 74 as a possibility suggesting that he had not seen the pattern. Omitting 90 and 95 may have been a mathematical error, an accidental slip or a desire to get to 100 as requested. On another occasion, the group was working on multiples of five and children had been invited to write multiples of five on the board. The numbers 55 and 80 were written and the teacher said one of those was also a multiple of ten. Michael quickly put his hand up and answered 80. The teacher praised him and asked him to write another multiple of 10 on the board. He wrote 56. Michaels enthusiasm for answering the teachers initial question and his correct answer of 80 had suggested to the adults that he could recognise multiples of ten. The teachers intervention was important and was presumably designed to confirm Michael understanding. However, it had the opposite effect, leaving us wondering if he understood multiples at all. His correct answer could have been a guess, especially since he had two numbers to choose from, though he seemed confident and was not obliged to answer. The 56 is harder to explain, though children occasionally did activities related to hundred squares where they were asked to identify the next number and writing 56 on the board near to 55 would have been correct if that had been the activity. Summary of findings. The above incidents are selected to show examples of different types of variability. Variability across lessons, as illustrated by Douglas, was common, and could sometimes be explained by factors such Educational & Child Psychology Vol 24 No 2 as task presentation. In many cases, more than one explanation was possible. Variability within individual incidents, as shown by Neil & Michael is harder to explain, suggesting that real difficulty exists in trying to determine from such an incident whether or not a child can do a piece of mathematics or even whether they did it successfully on that occasion. In other words, in general there was no explanation available in terms of task presentation or indeed any other obvious factor for most cases of variability. The adults who worked in the classroom were well aware of the issue of variability, especially for children such as Douglas & Claire, for whom this was a marked issue. My analysis of personal records indicates that variability was actually present for all children, though in some cases it was much less frequent and was not necessarily evident as part of normal classroom assessment. Discussions amongst the adults often attempted to explain variability in terms of factors such as task preference, lack of concentration and other personal or social factors. Such discussions often led to the crucial issue of what should be done next as far as planning with this group of children was concerned; a tension existed between reinforcement and repetition or moving on, with perceived pressure to move on to harder aspects of mathematics. Issues of mastery and progression are related to this dilemma. Some believe it desirable to master a piece of mathematics, that is, by performing consistently. For some, mastery is a pre-requisite for moving on. The idea of progression in mathematics is based for many on the belief that any new piece of mathematics can only successfully be learned when those preceding it have been understood thoroughly. However, my findings suggest that this may be an unrealistic aspiration. Implications These findings have clear implications for practitioners by casting doubt on the usefulness of assessments conducted on single occasions, especially if they are based on only one item only of each type. There are 95

Jenny Houssart dangers in extrapolating such assessments to make statements about what a child can or cannot do. It could rather be argued that contradictory assessment can be more useful in diagnosing difficulties and in planning by indicating that children respond better to certain types of activity, or have difficulties with specific aspects of arithmetic or with particular methods. This information can be used to structure appropriate activities, and to inform adults about those activities they are likely to need particular help with. However, my findings suggest that it is unrealistic to expect busy classroom practitioners to compile detailed pictures of all children and reach appropriate conclusions. This is an extremely time-consuming task, and interpretation is problematic. Thus, although detailed assessment can inform teaching it can not be relied upon. A further issue is whether the variability observed applies particularly to groups of children considered to have learning difficulties or whether it applies more widely. It is not possible to answer this question from this data, but it is useful to speculate about why variability in performance was apparently so marked in this classroom. One possible reason is that variability is more common amongst children who experience difficulties in learning arithmetic. Another is that the adults concerned were in the unusual position of being able to observe children closely as they worked on similar tasks over a long period of time and therefore more able to observe variability, which could be present but less noticeable in other situations. It is interesting to note that my analysis of personal records detected variability amongst all children, even those for whom it was not evident from normal classroom observation. It appears possible that variability is a natural part of learning and is present in classrooms, for all children. Perhaps the key issue for practitioners is how to proceed in situations similar to the one described. Teachers need to make decisions about when to move on to harder aspects of mathematics and when to repeat or reinforce ideas. Some views of mathematics and learning point to the conclusion that individual aspects should be mastered to the point where performance is accurate and automatic before moving on. Perhaps the fact that the children concerned are considered to be low attainers makes this option more tempting. However, my findings suggest that aiming for mastery before moving on is unrealistic and likely to be demoralising for all concerned. In moving forward, however, teachers need to be aware that reminders are often required. This could be seen as a positive step with reinforcement taking place as required in order to enable progress to be made rather than being seen as an end in itself. Frequent repetition of tasks can also produce a reaction in some children. Using micro genetic methods with a group of ten children, Siegler & Jenkins (1989) had to stop working with two of the ten children because of the way they reacted to the repetition of tasks. One child became over-anxious about succeeding on the tasks, whereas the other apparently became bored and gave evidence of not trying. It is possible that the situation in our classrooms in which work is frequently repeated brings about a similar reaction in children. Ironically, this repetition is often carried out in a search for mastery and automaticity.

Address for correspondence

Dr J. Houssart, Institute of Education, 20 Bedford Way, London WC1H 0Al. E-mail: j.houssart@ioe.ac.uk

96

Educational & Child Psychology Vol 24 No 2

Investigating variability in classroom performance amongst children

References
Black, P. (1998). Testing: Friend or foe? Theory and practice of assessment and testing. London: Falmer Press. Chinn, S. (2004). The trouble with maths. London: Routledge-Falmer. Clausen-May, T. (2005). Teaching maths to pupils with different learning styles. London: Paul Chapman Publishing. Cockburn, A. (1999). Teaching mathematics with insight: The identification, diagnosis and remediation of young childrens mathematical errors. London: Falmer Press. Denvir, B., Stolz, C. & Brown, M. (1982). Low attainers in mathematics, 516: Policies and practices in schools. London: Methuen Educational. DfEE (1999). The National Numeracy Strategy, framework for teaching mathematics from reception to Year 6. Sudbury: DfEE Publications. Dowker, A. (2004). What works for children with mathematical difficulties? (DfES Research Report 554). Nottingham: DfEE Publications. Dowker, A. (2005). Individual differences in arithmetic: Implications for psychology, neuroscience and education. Hove and New York: Psychology Press. Gabb, J. (2005). Equals at BCME 6. Equals, 11(2), 2223. Ginsburg, H. (1977). Childrens arithmetic: The learning process. New York: D. Van Nostrand Company. Gregory, A., Snell, J. & Dowker, A. (1999). Young childrens attitudes to mathematics: A cross-cultural study. Paper presented at the Conference on Language, Reasoning and Early Mathematical Development, University College London. Haylock, D. (1991). Teaching mathematics to low attainers, 812. London: Paul Chapman Publishing. Houssart, J. (2004). Low attainers in primary mathematics: The whisperers and the maths fairy. London: Routledge Falmer. McGarrigle, J. & Donaldson, M. (1974). Conservation accidents. Cognition, 3, 341350. McGarrigle, J., Grieve, R. & Hughes, M. (1978). Interpreting inclusion: a contribution to the study of the childs cognitive and linguistic development. Journal of Experimental Child Psychology, 26, 528550. Piaget, J. (1952). The childs conception of number. London: Routledge and Kegan Paul. Piaget, J. (1953). How children form mathematical concepts. Scientific American, 189(5), 7479. Robbins, B. (2000). Inclusive mathematics 511. London: Continuum. Siegler, R. (1996). Emerging minds: The process of change in childrens thinking. New York: Oxford University Press. Siegler, R. & Crowley, K. (1991). The microgenetic method: A direct means for studying cognitive development. American Psychologist, 46(6), 606620. Siegler, R. & Jenkins, E. (1989). How children discover new strategies. Hillsdale, New Jersey: Lawrence Erlbaum Associates. Siegler, R. & Stern, E. (1998). Conscious and unconscious strategy discoveries: A microgenetic analysis. Journal of Experimental Psychology, 127(4), 377397. Watson, A. (2000). Mathematics teachers acting as informal assessors: Practices, problems and recommendations. Educational Studies in Mathematics, 41, 6991. Watson, A. (2001). Making judgments about pupils mathematics. In P. Gates (Ed.), Issues in mathematics teaching. London: Routledge Falmer. Watson, A. (2006). Raising achievement in secondary mathematics. Maidenhead: Open University Press. Wright, R., Martland, J. & Stafford, A. (2000). Early numeracy: Assessment for teaching and intervention. London: Paul Chapman Publishing. Wright, R., Martland, J., Stafford, A. & Stanger, G. (2002). Teaching number: Advancing childrens skills and Strategies. London: Paul Chapman Publishing.

Educational & Child Psychology Vol 24 No 2

97

Language-based retrieval difficulties in arithmetic: A single case intervention study comparing two children with SLI
Tuire Koponen, Tuija Aro, Pekka Rsnen & Timo Ahonen
Abstract
The aim of this single-case intervention study was to examine whether difficulties in fluent language-based retrieval are related to learning to retrieve arithmetical facts from long-term memory. Two 10-year-old Finnish-speaking children considered to have Specific Language Impairment (SLI) were trained individually twice a week for two months using computerised game-like addition tasks. The participants were matched for non-verbal reasoning and non-verbal numerical skills as well as linguistic skills (verbal shortterm memory, comprehension and vocabulary). The key cognitive difference between the participants was naming fluency. Child A had difficulties in fluent language-based retrieval while child Bs performance on the same task was close to age-mean. A multiple baseline across-subjects design was used, with three baseline assessments and three follow-up assessments. Before the intervention both children used finger-counting strategies only. During the intervention child B progressed from finger-counting strategies to fact retrieval, while child A continued to use finger counting only. The results suggest that the benefit of an intervention programme, focusing on teaching fluent calculation skills with simple additions, is related to the specific features of the childs language competencies. It is proposed that an inability to shift from a finger counting to a fact retrieval strategy is connected to difficulties in fluent language-based retrieval from long-term memory.

LUENT CALCULATIONS are important cornerstones in mathematical learning and are also essential in many everyday situations. The learning develops gradually. First, when learning to solve simple arithmetic problems, children use calculation strategies based on counting such as finger counting or verbal counting (Siegler & Shrager, 1984). Initially, children count both numbers presented in the addition countingall procedure (Fuson, 1982; Geary, Hamson & Hoard, 2000). Later, children shift to counting on from the cardinal value of the first (counting-on max) or larger number (counting-on min) presented, which is a more efficient strategy (Geary et al., 2000). Frequent successful use of counting strategies increases memory representations of arithmetical facts and leads to a strategy of retrieving arithmetical facts from long-term 98

memory (Barrouillet & Fayol, 1998; Siegler & Shrager, 1984). In age-appropriate development of arithmetical skill children usually start using fact retrieval as the main strategy by the age of nine years. However, some children have difficulties in acquiring the skill of fluent calculation. For example, children with specific language impairment (SLI) have been found to have difficulties in shifting from counting-based strategies to a fact retrieval strategy (Fazio, 1999; Koponen, Mononen, Rsnen & Ahonen, 2006). These findings suggest that learning arithmetical facts relies on linguistic processing. This idea has also been presented in models of number processing (e.g. Dehaene & Cohen, 1995). Models of number processing state that arithmetical facts are stored in an associative network (e.g. Ashcraft, 1992; Campbell & Clark, 1988; Siegler & Shrager, 1984). Ashcrafts Educational & Child Psychology Vol 24 No 2 The British Psychological Society 2007

Laguage-based retrieval difficulties in arithmetic (1992) network retrieval model proposes that arithmetical facts are represented in memory in an organized network from which access and retrieval occur via a process of spread activation. The strength with which the nodes are stored and interconnected is a function of the frequency of occurrence and practice. How these nodes are represented remains an open question. However, there is theoretical justification as well as empirical evidence for both modular (e.g. McCloskey, Caramazza & Basil, 1985; Butterworth, 1999; Temple & Sherwood, 2002; Landerl, Bevan & Butterworth, 2004) and non-modular (Campbell & Clark, 1988; Dehaene & Cohen, 1995) representations of arithmetical facts. Butterworth & colleagues (Butterworth, 1999; Landerl et al., 2004) propose that the numerical, rather than other cognitive domains, is central in arithmetical fact retrieval deficit. In contrast, Campbell & Clark (1988) take the view that the associative network of arithmetical facts includes multiple numerical codes, (e.g. phonological, semantic and visual) which are interconnected. Dehaene and Cohen (1995) go even further by proposing that arithmetical facts are stored in verbal form. The goal of the present study was to test the theory that simple calculation is a language-connected skill. Of particular interest was whether problems in fluent languagebased retrieval are related to learning to retrieve arithmetical facts from long-term memory. This question was explored in a single case intervention study with two participants considered to have specific language impairment (SLI). Only a few studies have focused on the mathematical skills of children with specific language impairment (SLI). These studies show that children with SLI lag significantly behind their age peers in arithmetical skills (e.g. Arvedson, 2002; Fazio, 1999; Tieche Christinat, Conne & Gaillard, 1995). Fluent calculation skills, in particular, are one of those basic numerical skills that seem to be hard to acquire if the child has a language impairment (Fazio, 1999; Koponen, et al., 2006). Fazio (1999) reported that compared Educational & Child Psychology Vol 24 No 2 to their age peers, 9- to 10-year-old children with SLI had more problems when fast arithmetical fact retrieval was demanded. Koponen and colleagues (2006) found that only 31 per cent of 9- to 11-year-old children with SLI used fact retrieval as the main strategy in single-digit calculations. Most of the children with SLI used counting-based strategies, despite having practised simple calculation for several years. Rapid serial naming of objects and colours was the only one of the variables that explained the differences between the children who mainly used retrieval and those who used counting-based strategies in single-digit calculations. The children who calculated simple additions and subtractions slowly also named colours and objects more slowly than the children who calculated fluently. Also, Temple and Sherwood (2002) found that a group of children with arithmetical difficulties were slower at colour and object naming than control children, although the authors did not claim a causal relationship between rapid naming and difficulties in fact retrieval. These findings of language-based retrieval difficulties are in line with the theory of Geary (1993), who has suggested that representing and retrieving phonological information from long-term memory may underlie problems in learning arithmetical facts as well as coexistence with reading difficulty. The results of his later study (Geary, Hamson & Hoard, 2000) support this theory while suggesting that poor inhibition of irrelevant associations might also contribute to fact retrieval difficulties (see Barrouillet & Fayol, 1998). The present study was concerned with the impact of language-based retrieval difficulties on arithmetical acquisition. To better understand this connection, which so far has mainly been investigated via group studies, an intervention study design was used. Two SLI children with different abilities in languagebased retrieval received an arithmetical intervention in order to examine whether the capacity to learn arithmetical facts is related to language-based retrieving ability. 99

Tuire Koponen et al.

Method
Participants The participants were two 10-year-old Finnish speaking children (1 boy, 10; 10 years and 1 girl, 10; 5 years) with specific language impairment comprising both receptive and expressive language. They had been diagnosed by a phoniatrist using the ICD-10 criteria (WHO, 1993). They had participated in an extended compulsory education programme and were attending a state school for children with SLI. They were studying mathematics according to their individual educational plan (IEP), using special education textbooks (concise versions of the standard books) for second graders (aged 8 to 9 in Finland). The children were drawn from a larger sample collected for a group assessment study conducted in spring 2002 (see Koponen et al., 2006). These two participants were chosen on the grounds that, while they both had calculation difficulties (calculation fluency was at the level of 6-year-old preschoolers without any formal education in arithmetic), only one of them had difficulty in fluent language-based retrieval. They were matched in non-verbal reasoning skill, linguistic skills (digit span, sentence comprehension and vocabulary), and non-verbal numerical skills (comprehension of numbers presented in Arabic numerals and as play-money and estimation). In addition, the Paired-Associate Word Learning Test (WMSR, Wechsler, 1995) was administered in order to determine the participants ability to learn verbal associations. Their raw and standard scores are presented in Table1. Design The intervention study was run in three phases: pretests, intervention and posttests (see Figure 1). A multiple baseline acrosssubjects design was used (Kazdin, 1982) to examine the effects of this computer-aided intervention on the calculation skill acquisition of two children with SLI. Child A (male) had a one-week interval between the first and second and a three-week interval between the second and third pre-intervention assess100

ments, while Child B (female) had three baseline pre-intervention assessment sessions in eight days, after which the intervention was applied. The differences in the intervals between the baseline assessments were designed to enhance the validity of interpretation of the intervention effect, thus enabling it to be more powerfully argued that the improvements found in the childrens calculation performances were due mainly to the intervention and not, for example, to teaching at school or maturation during the intervention. The effects of the intervention were assessed at three follow-up assessments. Both participants had their first post-intervention session three days after their last intervention and their second and third assessments one week and one-month, respectively, after the first post-intervention assessment. Measures At each of the six assessments the children were presented with 31 simple additions to be done using a computer. The children heard instructions through headphones and viewed examples of the task on the screen. The answers and time taken for each trial of every task were recorded on the computer. The computer-administered tasks were constructed and carried out using the NEURE program, a graphical experiment generation tool (see http://www.edu.fi/ oppimateriaalit/neure). Each testing session was also recorded on video. Addition task. Single-digit additions using permutations from one to nine were presented one at a time in a horizontal format on a computer screen. The additions were in random order. The children were instructed to answer as quickly as possible by pressing any key on the keyboard. After pressing a key the addition disappeared and a box for the answer appeared with the cursor in it. The children were instructed to write the answer and to mouse-click an on-screen OK button. After giving the answer the children were asked how they calculated this addition. The different strategies (fact retrieval, counting, Educational & Child Psychology Vol 24 No 2

Laguage-based retrieval difficulties in arithmetic Children Child A Raw score Standard score Raw score (M 10, Sd 3) 6 13 38 57 72 9/22 24 3 4 NA 5 1 N/A 74 6 14 33 43 52 14/36 25

Task Language skill Digit Spana Comprehensionb Vocabulary c Naming colors d Naming objects d Paired-Associate Word Learning Teste RCPM (IQ)f Non-verbal number skill Comparison of numbersg Pre Post Comprehension of numbersh Estimationi

Child B Standard score (M 10, Sd 3) 3 6 NA 10 9 N/A 78

11 12 10 5

6 7 10 6

Table 1: Background measures. Note : a Digit Span task (Wisc-III; Wechsler, 1991), raw score. b Sentence comprehension (NEPSY; Korkman, Kirk & Kemp, 1997), raw score, max. 21. c Word finding vocabulary test (Renfrew, 2001), raw score, max. 50. d Rapid automatised naming (colors and objects, RAN; Denkla & Rudel, 1974; the Finnish version Ahonen, Tuovinen & Leppsaari, 1999), time in seconds. e Paired-Associate Word Learning Test (WMS-R), raw score of 3 and 6 trials, max. 24/48. f Ravens Coloured Progressive Matrices (Raven, 1993), raw score. g Number comparison, raw score, max. 15. h Moneybag, raw score, max. 20. i Estimation, raw score, max. 20.

Figure 1: Multiple baseline design counting on the fingers, decomposition) were introduced and gone through with the children before the testing. The testing sesEducational & Child Psychology Vol 24 No 2 sion was also recorded on video and the childrens strategies were confirmed afterwards. The task consisted of two training 101

Tuire Koponen et al. items and 31 assessment items. There were three parallel versions, each with 16 identical and 15 different additions. The 15 different additions were selected so that each was of the same average level of difficulty according to Wheeler ranking (see Wheeler, 1939). s Three scores were derived: calculation speed, accuracy and fact retrieval strategy. Intervention. The intervention was conducted twice weekly for a period of about two months, each session lasting 45 minutes. The first author conducted all the intervention sessions during class time in a secluded room at the school. Previous studies (e.g. Tournaki, 2003) have shown that children with mathematical learning difficulties benefit more from strategy instruction than from instruction through drill and practice. In the present intervention study the children received instruction and training in the use of more efficient calculation strategies, in accordance with the typical development of calculation strategies such as learning to use the counting-on min instead of counting-on max strategy. A faster counting strategy increases the possibility of having all the terms present at the same time in working memory in order to form and store the associations between the terms. Faster counting may also increase calculation accuracy by shortening the counting process a rather error-prone procedure. The counting-on min strategy was introduced by using two kinds of practices: first, the order-irrelevance principle in addition was introduced. This was done with concrete material by demonstrating that when two sets containing items x and y are combined, the final amount is always the same, despite the arrangement or order of the objects (e.g. 2 red and 3 blue or 2 blue and 3 red balls; you have 2 balls and you get 3 more or you have 3 balls and you get two more). After that the same idea was presented in the addition context with numbers and the symbol for the operation: the sum of a certain two numbers is always the same, despite the order in which they are pre102 sented (e.g. 2 3 3 2 5). Second, the child was asked to calculate the addition 2 9 and write down the answer. Next, the child was asked to calculate 9 2 and write the answer. After that it was discussed with the child whether the answer was the same or different, why it was the same (relating this practice with the earlier one), which way it was faster and easier to calculate and how they could use this information when calculating other additions like 3 9 or 4 7. During the computer games the children were observed and reminded to use the counting-on min strategy if they used the counting-on max strategy. Direct retrieval was trained both with computer tasks and board games. The tasks were of three kinds. In the first, mathematical problems were presented in visual quantities without any arithmetical symbols. The children were encouraged to identify the quantities without counting, name them and say the sum aloud. After the children had solved the problem correctly, positive feedback was given and the same problem was presented with arithmetical symbols and numbers. In the second, mathematical problems were presented both in visual quantities and in arithmetical symbols. The children had to solve the problem and then write the correct answer by pressing the number key. In the third, mathematical problems were presented in arithmetical and number symbols without the corresponding visual quantities. Moreover, there was practice in such tasks as memorization of pairs of numbers which equal a given numerical outcome (e.g. try to find all number pairs which equal 4). In these exercises counting-on strategies did not adapt well; hence the aim of this task was to strengthen the memory-associations between the addends and the answers with the support of the relevant conceptual knowledge. All the training tasks were single-digit additions using permutations from one to nine. Throughout the intervention the children were encouraged to retrieve Educational & Child Psychology Vol 24 No 2

Laguage-based retrieval difficulties in arithmetic answers from their memory. The researcher gave additional positive feedback when the children retrieved a fact from memory when playing the computer and board games. However, strong pressure or prohibition from using counting-based strategies was avoided. Both of the children were well motivated and co-operative and participated willingly in each intervention session. Data Analysis The analysis was performed in three steps. The first examined whether there are any significant differences on the childrens calculation performances (on calculation speed, accuracy and fact retrieval strategy) between the six assessment sessions. The second examined whether the childrens calculation performances within the baseline and follow-up were stable. The third compared the childrens pre and post intervention performances. For this purpose the baseline data were combined as were the follow-up data. The baseline variable thus included all three baseline assessments (93 additions) and the follow-up variable all three follow-up assessments (93 additions). Raw scores for calculation speed were subjected to a non-parametric Friedman Test, which is a nonparametric equivalent of a one-sample repeated measures design. Both the number of assessments (k) and sample size (N) exceeded five and so the value of statistic (denote as Fr) was distributed approximately as 2 with df k 1 (Siegel & Castellan, 1988). The raw scores for calculation accuracy and fact retrieval strategy were subjected to the non-parametric Cochran test, which is identical to the Friedman test but is applicable where all the variables are binary. The sample size (N) exceeded four and the product Nk was greater than 24 and so the value of statistic (denote as Q) was distributed approximately as 2 with df k 1 (Siegel & Castellan, 1988). The nonparametric tests were used because the variables were not normally distributed and the database was small. The medians for Educational & Child Psychology Vol 24 No 2 calculation speed as well as percentage of accuracy and use of the fact retrieval strategy at each baseline and follow-up assessment session are presented in Figure 2.

Results
The test for the possible effect of the assessment sessions on calculation speed showed a significant Fr value for both children (child 23, A Fr 36,27, p .001 and child B Fr 72, p .001), while for the effects on calculation accuracy and fact retrieval strategy significant differences were found for child B (Q 29,48 exact p .001 and Q 44,79, exact p .001), but not for child A (exact p .05). Calculation performance during the baseline and follow-up In calculation speed both children had a stable baseline (exact p .05), with no significant differences in the childrens calculation speed between the three pre assessments. Both children also had a stable follow-up, with no significant differences in their calculation speed between the three post assessments. In calculation accuracy child B had both a stable baseline (exact p .05) and stable follow-up (exact p .05), with no significant differences in her calculation accuracy between either the three pre assessments or between the three post assessments. During the baseline assessments child B did not use the fact retrieval strategy at all, and thus her baseline score was not subjected to analysis. During the follow-up B showed significant differences in the use of the fact retrieval strategy in the first and second compared to the third post assessment session (exact p .05 and exact p .01), retrieving fewer facts from her memory in the first (23 per cent) and second (10 per cent) than in the third (42 per cent) assessment. At the time of the second post assessment session, child Bs teacher reported that she had had a bad day, and been upset and tearful, which most likely influenced her performance during the session. Nevertheless, 103

Tuire Koponen et al.

Calculation speed, accuracy and fact retrieval strategy of child A
100,00 % 10000

80,00 %

8000

60,00 %

6000

Accuracy Fact retrieval Speed

40,00 %

4000

20,00 %

2000

0,00 % BL1 BL2 BL3 FO1 FO2 FO3

Calculation speed, accuracy and fact retrieval strategy of child B

100,00 % 10000

80,00 %

8000

60,00 %

6000

Accuracy Fact retrieval Speed

40,00 %

4000

20,00 %

2000

0,00 % BL1 BL2 BL3 FO1 FO2 FO3

Figure 2: Calculation performance of child A and child B

the assessment was done, as rescheduling was not possible. Comparison of calculation performance before and after the intervention. Child A showed significant differences in calculation speed between the baseline assessment and followup (Fr 25,04, exact p .001), being faster after the intervention was applied. As before, there were no significant differences in accuracy or the use of the fact retrieval strategy. Child B showed significant differences in calculation speed, accuracy and usage of the fact retrieval strategy (Fr 21,77, exact p .001, Q 24,14, exact p .001, Q 23,00, exact p .001), being faster, more accurate and using significantly more fact retrieval after than before the intervention. 104

At the baseline both children used counting-based strategies, particularly the counting-on max strategy. They took the first addend by lifting an equivalent number of fingers without counting and then joined it to the second addend by lifting one finger at a time while counting (e.g. 3 5 3, 1, 2, 3, 4, 5) until they got the answer (8). The intervention did not help child A to shift from counting to fact retrieval. He did not use the fact retrieval strategy in the first or second post assessment at all. At the third post assessment he remembered only one answer (8 1 9); using a following number rule rather than fact retrieval to solve it. However, he was able to learn the counting-on min strategy (instead of counting-on max strategy), although he used it Educational & Child Psychology Vol 24 No 2

Laguage-based retrieval difficulties in arithmetic only in the first and second but not in the third post assessment. During the intervention child B was able to shift from finger-counting strategies to fact retrieval. During the three follow-up sessions she used the fact retrieval strategy to solve 23 per cent, 10 per cent and 42 per cent of all items. In addition she was able to learn the counting-on min strategy (instead of counting-on max strategy) and used it in all the post assessment sessions. fluently retrieving the names of familiar objects presented serially in the RAN. Likewise, he was unable to retrieve eight associative pairs of words during the six trials in the WMS-R. In contrast, both children performed alike in the other linguistic and cognitive tasks, such as the digit span, in which both were able to recall three digits forward (short-term memory) and two digits backward (working memory). Because the participants were matched in all other linguistic and non-verbal reasoning skills, the difference in their ability to benefit from the intervention suggests that a specific language difficulty in retrieving could be one aspect causing the fact retrieval difficulty. In addition, it should be considered whether child As weak performance in learning associative pairs of words is an indication of a storing difficulty. At least, it can be said that he did not form representations of arithmetical facts which were retrievable later. These propositions are in line with the theory of Geary (1993), who has suggested that fact retrieval problems are a reflection of a more general problem in the ability to represent and retrieve phonetic information from long-term memory. This in turn could explain the coexistence of arithmetical and reading difficulties. Some of the more recent theories propose that a number-specific module could be responsible for calculation deficit (Butterworth, 1999). In order to see whether these theories offered a potential explanation of the present findings, the number skills of the participants were analysed. These analyses indicated that child B had severe difficulties in understanding the structure of Arabic numbers and their connections to magnitude. For instance, she had difficulties in selecting the largest 2-digit number out of three items. In contrast, child As performance in number comparison was closer to that of educational age controls and he did not commit errors with 2-digit numbers. Their ability at matching Arabic numbers with play money as well as estimating the dis105

Discussion
Previous studies have shown that children with SLI have difficulties in retrieving arithmetical facts from memory (Fazio, 1999; Koponen et al., 2006). Koponen and others (2006) found that rapid serial naming of objects and colours differentiated those children with SLI who after several years of practice still depended on slow calculation strategies in single-digit additions and subtractions from those who had learnt to retrieve those same calculations from memory (Koponen et al., 2006). The aim of this study was to explore further the question whether the development of calculation fluency shares some of the underlying processing abilities required in fluent retrieval of the names of objects or colours. The primary focus of the study was to examine the effect of an intensive intervention on the calculation skill of two participants: one had difficulties in both fluent calculation and language-based retrieval (child A) while the other had difficulties only in calculation (child B). The question to be considered was whether difficulties in fluent language-based retrieval had an impact on learning to retrieve arithmetical facts from long-term memory. Before the intervention both participants used a finger-counting strategy (counting-on max) only. Although both of them calculated significantly faster after the intervention, only child B was able to progress partially from using finger counting to fact retrieval. It was of interest that child A, who was unable to retrieve facts from memory, had difficulties in Educational & Child Psychology Vol 24 No 2

Tuire Koponen et al. tance between numbers was very similar. Thus, it can be concluded that a difference in number skills at the outset did not explain the differences found in the effects of the intervention. Another possible explanation for the results could be that the motivation to change strategies may not have been same for both children. Child A was rather accurate from the start and thus the rewards obtained by changing his strategy may not have been as high for him as for Child B, who made more errors in the beginning but showed a good increase in accuracy during the intervention. However, this is an unlikely explanation in view of the fact that Child A was not resistant to the intervention, but started to use the counting-on min strategy during the intervention. Moreover, both of the children were co-operative, motivated and participated willingly in each intervention session. The results also raise questions as to how it is possible that child B had not learned simple arithmetical facts during her 4-years schooling, and yet learned them during the two months intervention. There are several possible explanations. First, before the intervention child B committed many errors of calculation owing to her use of an errorprone counting strategy. During the intervention her counting strategy progressed considerably, enhancing her accuracy and speed, which in turn enabled her to form correct associations between the terms of the problem and the answer. Second, compared to the teaching in the classroom, the intervention was restricted, focusing only on simple addition through very intensive practice with immediate feedback and a large amounts of repetition. Third, despite her SLI, child B did not present language difficulties of a type, which, according to our interpretation, might have an impact on the learning of arithmetical facts. The results suggest that the particular nature of the language impairment has a major impact. That is, a specific difficulty in verbal retrieval and in forming verbal associations seems to be connected to the ability to learn arithmetical facts. In addition to lending support to earlier findings, indicating that children with language impairment have a specific problem in fact retrieval, the present study shows the importance of defining in detail the particular SLI phenotype when trying to understand the role played by language in arithmetical difficulties.

References
Tuire Koponen, Nilo Maki Jyvaskyla, Finland. E-mail: tuire.koponen@nmi.fi Institute,

References
Ahonen, T., Tuovinen, S. & Leppsaari, T. (1999). Nopean sarjallisen nimemisen testi [The test of rapid serial naming. Research reports.]. Jyvskyl: Haukkarannan koulun julkaisusarjat. Tutkimusraportit. Ashcraft, M.H. (1992). Cognitive arithmetic: A review of data and theory. Cognition, 44, 75106. Arvedson, P.J. (2002). Young children with specific language impairment and their numerical cognition. Journal of Speech, Language, and Hearing Research, 45, 970982. Barrouillet, P. & Fayol, M. (1998). From algorithmic computing to direct retrieval: Evidence from number and alphabetic arithmetic in children and adults. Memory & Cognition, 26, 355368. Butterworth, B. (1999). What counts: How every brain is hardwired for math. New York. Campbell, J.I.D. & Clark, J.M. (1988). An encodingcomplex view of cognitive number processing: Comment on McCloskey, Sokol & Goodman (1986). Journal of Experimental Psychology: General, 117, 204214. Dehaene, S. & Cohen, L. (1995). Towards an anatomical and functional model of number processing. Mathematical Cognition, 1, 83120. Denckla, M.B. & Rudel, R. (1974). Rapid automatized naming of pictured objects, colors, letters and numbers by normal children. Cortex, 10, 186202. Fazio, B.B. (1999). Arithmetic calculation, short-term memory and language performance in children

106

Educational & Child Psychology Vol 24 No 2

Laguage-based retrieval difficulties in arithmetic

with specific language impairment: A 5-year follow-up. Journal of Speech, Language and Hearing Research, 42, 420421. Fuson, K.C. (1982). An analysis of the counting on solution procedure in addition. In T.P. Carpenter, J.M. Moser & T.A. Romberg (Eds.), Addition and subtraction: A cognitive perspective (pp. 6781). Hillsdale NJ: Lawrence Erlbaum Associates, Inc. Geary, D.C. (1993). Mathematical disabilities: Cognitive, neuropsychological, and genetic components. Psychological Bulletin, 114, 345362. Geary, D.C., Hamson, C.O. & Hoard, M.K. (2000). Numerical and arithmetical cognition: A longitudinal study of process and concept deficits in children with learning disability. Journal of Experimental Child Psychology, 77, 236263. Kazdin, A.E. (1982). Single-case research designs: Methods for clinical and applied settings. New York: Oxford University Press. Koponen, T., Mononen, R., Rsnen, P. & Ahonen, T. (2006). Basic numeracy in children with specific language impairment: Heterogeneity and connections to language. Journal of Speech, Language, and Hearing Research, 46, 116. Korkman, M., Kirk., U. & Kemp, S.L. (1997). NEPSY. Lasten neuropsykologinen tutkimus [NEPSY: A developmental neuropsychological assessment]. Helsinki, Finland: Psykologien kustannus. Landerl, K., Bevan, A. & Butterworth, B. (2004). Developmental dyscalculia and basic numerical capacities: A study of 89-year--old students. Cognition, 93, 99125. McCloskey, M., Caramazza, A. & Basili, A. (1985). Cognitive mechanisms in number processing and calculation: Evidence from dyscalculia. Brain and Cognition, 4, 171196. Niilo Mki Instituutti 2002. NEURE-programme: http://www.nmi.jyu.fi/neure/.1.2.2003. Raven, J.C. (1993). Coloured progressive matrices. Revised and reprinted. Oxford: Oxford Psychologists Press Ltd. Renfrew, C. (2001). Word finding vocabulary test (4th ed). Bicester: Speechmark Publishing Ltd. Siegel, S. & Castellan N.J (1988). Nonparametric statistics: for the Behavioral Sciences (2nd edn.) (pp. 174183). Singapore: McGraw-Hill, Inc. Siegler, R.S. & Shrager, J. (1984). Strategy choices in addition and subtraction: How do children know what to do? In C. Sophian (Ed.), Origins of cognitive skills (pp. 229293). Hillsdale, NJ:Lawrence Erlbaum Associates, Inc. Temple, C.M. & Sherwood, S. (2002). Representation and retrieval of arithmetic facts: Developmental difficulties. Quarterly Journal of Experimental Psychology, 55, 733752. Tieche Christinat, C., Conne, F. & Gaillard, F. (1995). Number processing in language impaired schoolchildren. Approche Neuropsychologique des Apprentissages chez lenfant, hors srie, 5257. Tournaki, N. (2003). The differential effects of teaching addition through strategy instruction versus drill and practice to students with and without learning disabilities. Journal of Learning Disabilities, 36, 449458. Wechsler, D. (1991). WISC-III. Wechslerin lasten lykkyysasteikko. [Wechsler Intelligence Scale For Children]. Helsinki, Finland: Psykologien Kustannus Oy. Wechsler, D. (1995). WMS-R. Wechsler Memory Scale revised. Helsinki, Finland: Psykologien Kustannus Oy. Wheeler, L.R. (1939). A comparative study of the difficulty of the 100 addition combinations. Journal of Genetic Psychology, 54, 295312. World Health Organization (1993). The ICD-10 classification of mental and behavioural Disorders: Diagnostic criteria for research (pp. 142143). Geneva.

Educational & Child Psychology Vol 24 No 2

107

Achieving new heights in Cumbria: Raising standards in early numeracy through mathematics recovery
Ruth Willey, Amanda Holliday & Jim Martland
Abstract
This article describes how standards in early numeracy were raised within Cumbria by the application of the Mathematics Recovery Programme. It reports data showing how childrens numeracy improved as a result of the programme, and describes effective elements of the in-service teacher training programme which was implemented. This work is an example of how teachers and educational psychologists can work together to develop and disseminate good practice in teaching, which is based on a sound theoretical and evidence base. Mathematics Recovery (MR) is an evolving, research-based programme which was first developed in the 1990s, in order to meet the needs of children who were not reaching age-related expectations for numeracy skills. There is an underlying model of how children acquire strategies and numerical knowledge, and an explicit set of principles of good teaching. The MR materials include short-term, intensive, individual teaching programmes, as well as group and class teaching. The paper reflects on the nature of best practice in assessment for learning (DFES 2005b), which we argue is dynamic in character, in that the assessment is embedded in the teaching, with the assessor playing a mediating role, supporting the learner to construct and elaborate their own model of number. We show how Mathematics Recovery implements this approach to assessment and teaching, through the design, implementation and evaluation of programmes of shortterm intervention. The paper concludes with some evidence of how the above has impacted on teachers professional development and changed classroom practice in Cumbria.

Support for numeracy difficulties

Within the British educational system, there has traditionally been more emphasis placed upon addressing childrens literacy difficulties, than on addressing their numeracy difficulties. This is reflected in the resources which Local Authorities have allocated to the two areas: in many Authorities there will be a team of specialist literacy teachers to address Special Needs in literacy, but no corresponding team for such needs in numeracy. However, we do have evidence both of the existence of childrens difficulties with numeracy, and of the disadvantage which people with poor numeracy skills suffer in adult life. Since the introduction of the National Numeracy Strategys Framework for teaching mathematics in 1999, the number of children achieving the target level for their age in mathematics at the end of Key Stage 2 has increased by 16 per cent 108

(DfES, 2004 and 2005a). But there is still a long tail of children who do not achieve the target level, with a steady proportion of children achieving below level 3, between 2001 and 2004 (DfES, 2005b). The longterm consequences of numeracy difficulties are serious: research suggests that, amongst adults, poor numeracy is more disadvantageous in the labour market than is poor literacy (Basic Skills Agency, 1997). There is, then, a pressing need to address this area. The Primary National Strategy has responded through its model of waves of intervention: Wave 1 being high quality learning and teaching for all in daily lessons; Wave 2 being targeted, short term small group interventions; and Wave 3 being a more individualised, short term intervention to address fundamental errors and misconceptions (DfES, 2005b). Wave 3 is intended for pupils in Key Stage 2. Thus, whilst Wave 2 Educational & Child Psychology Vol 24 No 2 The British Psychological Society 2007

Achieving new heights in Cumbria might be regarded as being early intervention to prevent children from falling behind, Wave 3 is targeted at children who have made unsatisfactory progress across the three years of infant-aged schooling. In 2002 the Cumbria LEA piloted and then implemented the Mathematics Recovery programme which complements all three waves of intervention. In its original form, it was a short term, intensive individual programme for pupils in Year 1 of Key Stage 1, and thus constituted intervention to prevent failure, earlier than Wave 3 and more intensive than Wave 2. There are not many such documented intervention programmes available for the Key Stage 1 age group. In her recent review What Works for Children with Mathematical Difficulties, Ann Dowker reviews MR very positively, and cites it as one of the two available large-scale, individualised, componential programmes based on cognitive theories of arithmetic (Dowker, 2004). As will be seen below, the MR programme has also been developed so that it can be applied more widely than just as an individual programme, and it can be used as part of Wave 1, Wave 2 and Wave 3. entire programme provides an extensive professional development course to prepare the specialist teachers, and ongoing collegial and leader support for these teachers. MR was originally developed in New South Wales, between 1992 and 1995. It emerged from detailed research studies of how childrens number knowledge develops (Wright, 1991; Aubrey, 1993; YoungLoveridge, 1989, 1991). From this, a model of the usual course of this learning was constructed, and assessment tools and techniques were developed, to enable individual childrens knowledge to be described in the terms of the model. Wright and his colleagues went on to design an individual teaching approach and materials, intended to move children on through the model, by working in a very detailed way within the childs Zone of Proximal Development (ZPD), that is, planning instruction which is focussed just beyond the childs current levels of knowledge. These individual teaching programmes were evaluated, and shown to be very successful in moving children on through the stages and levels of the model. (Wright et al., 1994; Wright et al., 1998). The approach has been further developed into its current, published form (Wright, Martland & Stafford, 2006; Wright, Stanger, Stafford & Martland, 2006; Wright, Martland, Stafford & Stanger, 2002). Materials now include assessment tools, teaching programmes for individual children and a book on using the approach in classroom teaching. MR is now in wide, international use, in Australia, the USA, New Zealand, Canada, the UK and Ireland.

The key features and origins of the mathematics recovery programme

The key features of MR can be summed up under four headings Early Intervention, Assessment, Teaching and Professional Development. The assessment and teaching strands use a strong underpinning theory of young childrens mathematical learning which leads to a comprehensive and integrated framework for both assessment and teaching. The programme has a detailed approach to, and specific diagnostic tools for, the assessment of childrens early number strategies and knowledge. Following the assessment, teachers can employ an especially developed instructional approach and distinctive instructional activities which can be applied to individuals in small-group or class situations. The programme also has an intensive, short-term teaching intervention for low-attaining 58year-old children by specialist teachers. The Educational & Child Psychology Vol 24 No 2

Assessment in the mathematics recovery programme

Mathematics Recovery involves a distinctive approach to assessing young childrens numerical knowledge. The origins of this method are in research projects conducted in the 1980s and 1990s that focused on understanding childrens numerical strategies for addition and subtraction and the modifications children make to their strategies 109

Ruth Willey et al. over time (e.g. Cobb & Steffe, 1982; Steffe & Cobb, 1988; Steffe et al., 1983; Wright, 1989, 1991a). In 1998 the approach was extended to include a focus on childrens early multiplication and division knowledge. This work drew on an extensive range of research (Steffe, 1992b; Steffe & Cobb, 1988; Steffe, 1994; Mulligan, 1998) and the Count Me In Too project (NSW Department of Education and Training, 1998). The assessment in MR is distinctive on two counts. First, it is interview based and second, the assessment interview is videotaped so that the teacher does not need to record the childs responses during the course of the interview. The benefit of not having to make notes is that the assessor is free to observe, listen and engage in questions with the child in order to detect the most sophisticated strategy the child uses. Underlying the development of MR is a belief that in early number learning it is very important to understand, observe and take account of childrens knowledge and strategies when solving tasks. Childrens early numerical knowledge varies greatly and their strategies are multifarious. Thus, across children, early numerical knowledge is characterized by both commonalities and diversity. As indicated by the research of Denvir & Brown (1986a, 1986b), it is insufficient to think that every childs early numerical knowledge develops along a common developmental path. For example, one important factor in a particular childs developmental path, it is believed, relates to the nature of the settings in which the childs prior learning has occurred. Also, children who may appear to an observer to be in the same setting, or learning situation, will construct the situation idiosyncratically and thus different kinds of learning are likely to occur. The childs process of constructing numerical knowledge can be thought of in terms of progression or advancement. Children reconstruct or modify their current strategies and doing so is nothing more or less than progression, advancement or learning. Given this, it is useful to consider the notion 110 of the relative sophistication of childrens strategies. For example, the child who has no means of working out nine plus three other than counting out nine counters from one, counting out three counters from one, and then counting all of the counters from 1 to 12, is using a far less sophisticated strategy than the child who ignores the counters and says nine plus three is the same as ten plus two, and I know that is 12 without counting. Understanding the progression of the strategies which children use in early number situations is the key to advancing teaching staffs professional knowledge and learning. We refer to the progression as SEAL (Stages of Early Arithmetical Learning). They are: Stage Stage 0: Stage 1: Stage 2: Stage 3: Stage 4: Stage 5: Significant tasks Emergent Counting Perceptual Counting Figurative Counting Initial Number Sequence Intermediate Number Sequence Facile Number Sequence

Assessment tasks as a source of instructional activities Virtually all of the assessment tasks are ideally suited for adaptation to instructional activities. Further, because the assessment tasks are organised into task groups, the tasks within a task group or across several groups typically constitute an implied, instructional sequence. Again, although the tasks are presented in a format for one-toone interaction, they are easily adapted to situations involving small or large group instruction.

Implementation of mathematics recovery in Cumbria

Cumbria began its involvement with MR on a small scale in 2002 with a group of eight teachers, two Numeracy Consultants and one Educational Psychologist. Subsequently, a MR Team consisting of a Numeracy Consultant (two days per week), an Educational Educational & Child Psychology Vol 24 No 2

Achieving new heights in Cumbria Psychologist (half a day per week) and three teachers (one day per week each) has worked to support and develop the use of MR within the County. A major focus is the running of an annual course to train teachers and teaching assistants. The course takes place over two terms, with assessment being covered in the first term, and the teaching programme in the second term. There is a total of seven centrebased training days, with two or more tutor visits to participants schools. During the course, participants engage in video-taped practice assessments, and design and run a teaching programme with an individual child. So far, 97 schools have undertaken the training, which represents almost one third of the primary schools in Cumbria. (The training has also been found useful by some special school and secondary school teaching staff). Schools have been encouraged to send a teacher and teaching assistant on the course together, in order to promote the use of the programme later in school at both the classroom and individual child level. The teachers and assistants have worked closely together on the programmes for children, and have found this particularly helpful in the development of their skills. Staff who successfully complete the MR training are able to apply for funding to run individual MR programmes, for pupils whom they have assessed as functioning well below the expected levels on the MR assessments. The effectiveness of these programmes is evaluated, through analysis of pupils results on the MR assessments before and after the programme. Most pupils make gains of two SEAL stages (e.g. they move from having to see and count concrete objects in order to add two sets (Stage 1 on SEAL) to being able to work without visible objects and to countup-from and count-down-from to solve addition and subtraction problems, including missing addends and missing subtrahends (Stage 3 on SEAL)). They also increase their ability in other aspects of number: saying forward and backward number word sequences, to identifying numerals and recognising spatial patterns. Indeed, so far the small number of pupils who have not made a gain of at least one SEAL stage during their MR individual programme have all made measurable gains in these other aspects. See Table 1 below, for a summary of the gains in SEAL stages made by 2 1 0 Cumbrian pupils who received individual programmes between April 04 and March 06. These increases are similar to those reported in the Australian MR research, although the Cumbrian programmes are shorter in duration (about 20 sessions, half an hour each, taking place three or four times a week whereas the Australian programmes were more than twice this length). As Ann Dowker says, Relatively small amounts of individual intervention may make it possible for a child to benefit far more fully from whole class teaching (Dowker, 2004). Evaluating how well the children have generalised and adapted the learning is more difficult.

No stages gained April 04 March 05 100 pupils April 05 March 06 110 pupils % of total 9 4 6.2

1 stage gained 26 31 27.1

2 stages gained 44 57 48.1

3 stages gained 19 12 14.8

4 stages gained 2 6 0.04

Table 1: Gains in SEAL stages made by 210 Cumbrian pupils who received individual programmes (Holliday, 2005; Holliday, 2006) Educational & Child Psychology Vol 24 No 2 111

Ruth Willey et al. Informal teacher reports, collected on tutorial and other visits to schools, consistently say that the pupils are performing better in class, and that they have gained in confidence and independence. It will soon be possible to analyse the mathematics SATs results from the end of Key Stage 1, tracking those pupils who received a funded programme, and comparing results with comparable pupils who did not. In addition to staff training and monitoring of individual programmes, the MR team are developing other ways of supporting the use of MR in schools. These include the publication of guidance for schools on the use of MR for group work and for work in the Foundation Stage; work to develop ICT materials for whole class use; work with teaching staff on using MR in the daily mathematics lesson; establishing regular support meetings to update staff and for them to share developments. The development of the MR work within Cumbria would not have been possible without dedicated funding for the project. This was not initially available. However, following the evaluation of the success of the first cohort of training, it was possible to argue the case for some of the existing Special Needs resources to be directed towards numeracy. Currently, there is an annual budget which pays for the salaries of the MR Team, and for the delivery of some individual programmes to pupils in schools. with pupils who have some history of mathematics difficulty? A large part of the answer to this, we believe, lies in the way in which the assessment and teaching are used together, within a framework which is constructivist in its nature. The assessment is not seen just as a measure of what has been learned, but as an integral and ongoing facet of the teaching, which will inform both what is taught next, and the approach and materials which are used in the teaching. Although the initial assessment does give summative information about the levels and stage at which the child is functioning, its central purpose is to allow a qualitative, detailed analysis of the strategies the child is using. The assessor presents the child with a problem, observes the child working, and explores the childs responses (through questioning and judicious presentation of new problems), to find out how the child thought whilst solving the problem. This information is recorded (after the assessment interview) in a Pupil Profile, which highlights the childs present strategies, strengths and weaknesses, and possible next steps for development. This profile is then used to design the teaching programme for the pupil, drawing on the range of available teaching activities within the MR materials. This approach to assessment continues throughout the teaching programme, as the childs responses during teaching are observed closely, and used to guide the next teaching steps. The aim is always to be working within the childs Zone of Proximal Development (Lunt, 1993; Lidz, 1995), so that the child succeeds, with small but welltargeted prompts from the MR teacher. The MR teacher role is critical here: it is to select appropriate problems for the child, present them in a suitable setting, support the child successfully to find their own solution to the problem and help the child to reflect on what they are doing. This is not didactic teaching: modelling of solutions rarely happens, and when it does is usually associated with the learning of basic facts (such as the words in the forward number word Educational & Child Psychology Vol 24 No 2

Why does MR work so well?

Our ongoing evaluation of the work suggests that MR is highly effective, including with children who have already received Wave 2 interventions, in the form of small group work based on National Numeracy Strategy Springboards, or on National Numeracy Strategy objectives which had been tracked back to earlier curriculum stages. Yet many of the MR teaching activities resemble those in other programmes, and cover ground which is also in the Numeracy Strategy. So why is it so successful, especially when used

112

Achieving new heights in Cumbria sequence). This is a mediating role, and is similar to the teacher role in the Dutch Realistic Mathematics work, which refers to pupils as engaging in guided reinvention, and stresses the importance of knowledge being constructed by the child. (Gravemeijer, 1994; Milo, Ruijssnaars & Seegers, 2005). The problem-centred approach is used, within MR, as a very important tool for ensuring that the teaching remains constructivist in its orientation. A central aim of the MR programme is for children to develop their own, increasingly powerful concepts of number, which they will be able to use as a basis for subsequent learning (Cobb & Merkel, 1989). Because they have been developed by the children elaborating their concepts in the course of their own problem-solving, these constructs will be fully understood by the children, in a relational, rather than only an instrumental, manner (Skemp, 1976). Thus, the children will not merely be following a learned recipe for solving a particular, familiar type of problem (showing instrumental understanding), but will be able to devise their own strategies and algorithms for solving novel types of problem (showing relational understanding). Such a shift from procedures to reasoning (Wheatly & Reynolds, 1999) is essential, if children are to become confident and independent learners who will be able to generalise and extend their knowledge in new contexts. The problem-centred approach fosters this relational understanding, through developing a setting in which children can invent and discuss their own strategies (Cobb & Merkel, op cit). In an individual MR session, this will be done by presenting the child with a problem which is slightly more difficult than those which the child has previously solved, and allowing as much time as is necessary for the child to work on the problem. The teacher will observe closely, and use their knowledge of what strategies and concepts that child already possesses, to offer prompts that will lead the child towards developing more sophisticated strategies. Importantly, the child will be encouraged to Educational & Child Psychology Vol 24 No 2 check their solutions, by using less sophisticated strategies. This will enable them to build links between their developing concepts, so that they are continuously elaborating their mental model of the number system, through solving the problems. As Wright expresses it, for the constructivist teacher, advances in the childrens knowledge occur when the children modify their current ways of operating in response to a problematic situation. (Wright, 1990). During the teaching sessions, the MR teacher continues to observe and assess the childs responses, with a strong focus on how the child makes use of the support which the teacher offers. The teacher is continuously making and testing hypotheses about what experiences will now help the child to develop further their models of number. Thus, throughout the programme, the teacher can be regarded as engaging in Dynamic Assessment (Lidz, 1995; Elliott, 2003). The role of the teacher is to mediate the childs learning experience, by locating the childs Zone of Proximal Development (ZPD) (i.e., the region where the child can only succeed with some support), and working with the child in this Zone. An example from a teaching programme may serve to illustrate the style of the teaching, showing how the teacher mediates the childs learning, choosing different prompts and settings, in order to help the child to construct her responses. The section of dialogue and commentary below comes from an early session in a programme with Gertie, a girl aged 6 years and 5 months.

Programme extract: Session 2: work on forward number word sequence to 20

Gertie has been doing FNWS successfully, although she hesitated at 12. The teacher presents the Number Word After task, and Gertie succeeds with 5, 15, 24, 29 and 7. The teacher then presents 12, and the following dialogue happens:

T: What number comes straight after 12? G: 11 T: were going forwards, so its the number that comes just after 12. 113

Ruth Willey et al. G: 14 T: 12. . . . . .? (re-presents task, and waits) G: . . . . . .13 T: Good. How did you do that? G: I just thought in my head it was. T: Did you count? (G shakes her head.) G then succeeds with NWA 3 and 12 T: 19 G: 18 T: 19, what comes next, the number just after 19? (re-presents task) G: . . . . . . . . . . . . . . .17
At this point, Gs body language is interesting. She is slumped in her chair, with her hands up around her mouth, beginning to squirm. She seems bored or tired, and uncomfortable. She seems to be signalling that she has had enough of this hard task, but T persists : In this extract, Gertie eventually succeeds with NWA for 12 and 19, but with a lot of difficulty, and with external prompting. Future sessions will show that her counting skills vary a lot, from day to day. She is also worse in the afternoons, when she seems tired, and has more difficulty in concentrating. Several interesting points emerge, from this extract. Although Gertie can produce the Forward Number Word Sequence beyond 20, she has not connected this knowledge to the Number Word After task. The teacher attempts to help her make that connection. First, this is done verbally, by encouraging her to count. (She uses a series of graduated prompts: Did you count? . If you were counting forwards, what would come just after 19?. . .do some counting forwards. . . .) The teacher observes carefully, and decides to fine tune the teaching, through the choice of problems. (Having discovered a difficulty with 12, she works on this, then asks a different question, before revisiting 12 to check. The difficulty with 19 is then worked on, until this is resolved. At this point, Gertie has had enough of this task, and the teacher moves on to a different Key Topic.) This teaching is in Gerties ZPD: she cannot do it independently, but eventually gets there, with teacher mediation.

T: Youre going backwards, youre doing the numbers before. If you were counting forwards, what would come just after 19?
No reply from G, after a long pause. T decides to try to get G to count forwards, and listen to her own voice saying the number after 19 :

T: Can we do some counting forwards . . . well start at 18. 18 . . . G: 19 T: 20 G: 21 T: OK, so what comes just after 19? G: 21 T: 19. . .? G: 19 . .
This has not worked. T brings out a numeral track, from 11 to 20. T points to each number, and G reads them out correctly.

T: Wheres 19? G points to 19 T: What comes just after it? G: 20 (Points to it.) T puts numeral track away. T: What comes just after 19? G: 20 T: Good. What comes just before 20? G: . . T: If you were counting, what would come just before 20? G: . .19 T: Well done! 114

Perhaps the most powerful and positively evaluated session in the teaching course is one which is run as a class tutorial, where teaching staff bring scenarios from their ongoing teaching programmes and ask the group to help them with teaching ideas (i.e., ideas for mediating the childs experience). This leads to exchange of ideas and rich discussion, focussed on how to set up an experience which will support that particular child to solve the problem in question. At this point in the course, it becomes evident that not only are the teaching staff working within the childrens ZPD, but that they are also developing their teaching skills through working within their own ZPD, using the tutors and each other for support.

Beyond individual programmes

In its original form, MR was delivered to children as short term individual programmes, targeted at Year 1 aged children who were beginning to fall behind their peers in Educational & Child Psychology Vol 24 No 2

Achieving new heights in Cumbria numeracy. However, it has developed considerably beyond this. For example, in New South Wales and in New Zealand the approach is used across the first three years of schooling, as a framework for the teaching of numeracy to all pupils. This implementation is called Count Me In Too (CMIT), and does not deliver individual programmes to children. It has been evaluated (through pre and post assessments of pupils, as well as questionnaires and case studies with teaching staff and facilitators involved in the programme) as highly successful, both in promoting pupil progress and in increasing teacher knowledge and understanding (Thomas & Ward, 2001). Key elements in the success of this group approach were found to be: increased teacher understanding of how children learn number; increased focus on the strategies which individual children actually use; the availability of assessment tools which can be used to group children appropriately for working on particular learning objectives (Thomas & Ward, op cit). The Cumbrian experience also shows that the effectiveness of MR goes well beyond the individual programmes. Examples of this include: Teaching staff who attend the training often respond to the assessment course by spontaneously implementing changes in the way they deliver their class teaching. They come to the second and subsequent training sessions keen to talk about changes they have already made. Many focus initially on the mental and oral starter part of their lessons, noticing that most children are working in their ZPD for only a small part of this activity. They then find different ways to organise and present the activity, so that it is better differentiated to match the childrens needs. Teaching staff raise their expectations of what children can achieve. About half of the teaching staff who complete the course apply for funding to run individual programmes. (In the financial year 20052006, funding was approved Educational & Child Psychology Vol 24 No 2 for 111 programmes, and 93 per cent of those programmes have so far been completed.) However, many of the remaining teaching staff use the approach to make changes in their classroom teaching, to organise groups of children for teaching, or to deliver teaching to small groups. Some teaching staff are using the assessment materials to track childrens progress throughout the infant school, and to pinpoint the need for specific interventions with particular children, or for staff training in particular areas of numeracy teaching. Teaching staff feel that their expertise in early numeracy is recognised within their school and feel that their skills are used to advantage, as colleagues consult them regarding childrens progress and the construction of teaching strategies. The teaching staff report that they can readily use existing classroom materials but now in a more effective way: they try to promote mental strategies through providing a set of integrated activities in multiple settings. Although MR can be very effectively applied in whole class and small group contexts, the MR Team take the view that, in initially learning to apply the principles of MR, it is extremely effective to work with an individual child. This allows the teacher to work continuously at finding where the child is, and developing ways of supporting them to move on. This cannot be done so precisely when working with a group, where there is often the need to make a compromise, or move on before one child is really ready. The MR Team have formed the impression that many teaching staff develop their knowledge, understanding and practice of numeracy teaching considerably, through using MR. A research project is attempting to explore these changes, through in-depth Personal Construct Psychology interviews with staff (Willey, ongoing). Results so far suggest that staff constructs about teaching of number change markedly, following 115

Ruth Willey et al. training and use of MR. Changes to constructs are in line with the underlying principles of MR, and include the following: Trying to take pupils back to first principles so they can build understanding, rather than trying to plug gaps in their procedural knowledge Having an understanding of how children develop number knowledge A growing commitment to promoting pupils independent learning A belief that good teaching will be enjoyable and motivating for pupils Willingness to wait whilst pupils think, and to observe closely what they do A belief that children will learn effectively, if they are given tasks within their ZPD and a small amount of support A view of the teacher as facilitator and guide, rather than transmitter of knowledge. The staff who were interviewed were asked to rate themselves on the construct teaches numeracy very well, for both before and after the MR training. All but one of them felt that they had improved. Even staff with many years of experience felt that engaging with MR had moved their teaching on significantly. It seems, then, that teaching staff do develop skills and knowledge through implementing MR, and that they do put this into practice in their subsequent work with pupils. This is because MR provides a structure within which staff feel safe to experiment with a more constructivist approach. The teaching activities given within the programme function as examples. Although it would be possible to run individual programmes using only the teaching activities given, this does not generally happen: the teaching staff adapt, tailor and extend the activities, to address more exactly the needs and interests of each child. The teaching staff are able to do this because they have learned to use the principles of MR, to generate their own solutions to new situations. The teaching staff have gained confidence in pupils abilities to learn in a constructivist way, and in their own abilities to guide such learning.

Conclusion
We have shown how one Local authority has implemented Mathematics Recovery and evaluated its impact. Individual pupils who receive MR programmes make good progress in basic numeracy skills. Teachers and teaching assistants develop their knowledge, skills and confidence to teach numeracy. The Maths Recovery principles, assessment tools and activities work well at a number of levels: in individual programmes, in group work and in informing good classroom teaching. The greatest power of Mathematics Recovery lies in its use as a tool for the professional development of teaching staff. Staff who engage with Mathematics Recovery develop an enhanced faith in pupils ability to learn and to solve problems for themselves. Alongside this, they become more confident in their own ability to assess where pupils are, and to offer appropriate support to help pupils learn. For many staff, this results in a significant shift in their teaching style, away from the didactic and towards a more pupil-focused, constructivist outlook.

Address for correspondence

Ruth Willey, Cumbria County Council, Childrens Services, Newbeck Centre, Wigton Rd, Carlisle, CA2 6LB E-mail: ruth.willey@cumbriacc.gov.uk A. Holliday, Cumbria County Council, Childrens Services, Kendal Area Office, Busher Walk, Kendal, LA9 4RQ E-mail: amanda.holliday@cumbriacc.gov.uk J. Martland, Mathematics Recovery programme (UK) Ltd, 11 Station Lane, Mickle Trafford, Chester, Cheshire, CH2 4EH E-mail: jrmartland@btopenworld.com

116

Educational & Child Psychology Vol 24 No 2

Achieving new heights in Cumbria

References
Aubrey, C. (1993). An investigation of the mathematical knowledge and competencies which young children bring into school. British Educational Research Journal, 19(1), 2741. Basic Skills Agency (1997). Does numeracy matter? Evidence from the National Child Development Study on the impact of poor numeracy on adult life. Basic Skills Agency. Cobb, P. & Merkel, G. (1989) Thinking strategies: teaching arithmetic through problem-solving NCTM Yearbook. Denvir, B. & Brown, M. (1986a) Understanding of number concepts in low attaining 79 year olds: Part 1. Development of descriptive framework and diagnostic instrument. Educational Studies in Mathematics, 17, 1536. Denvir, B. & Brown, M. (1986b) Understanding of number concepts in low attaining 79 year olds: Part II. The teaching studies. Educational Studies in Mathematics, 17, 143164. Department for Education and Skills (DfES) (2005a). Reviewing the frameworks for teaching literacy and mathematics. Primary National Strategy, 17862005 DOC-EN. London: DfES. Department for Education and Skills (DfES) (2005b). Targeting support: implementing interventions for children with significant difficulties in mathematics. Primary National Strategy, 1083-2005. London: DfES. Department for Education and Skills (DfES) (2004). Raising standards in mathematics Achieving childrens targets: Primary National Strategy, 10752004. London: DfES. Dowker, A. (2004). What works for children with mathematical difficulties? Nottingham: DfES Publications. Elliott, J. (2003). Dynamic assessment in educational settings: Realising potential. Educational Review, 55(1)1532. Gravemeijer, K.P.E. (1994). Developing realistic mathematics Education. Utrecht, The Netherlands: CD-B Press. Holliday, A. (2005). Mathematics recovery evaluation of funding April 2004 to March 2005. Unpublished report of Cumbria Childrens Services. Holliday, A. (2006). Mathematics recovery evaluation April 2005 to March 2006. Unpublished report of Cumbria Childrens Services. Lidz, C.S. (1991). Practitioners guide to dynamic assessment. New York: Guilford. Lidz, C.S. (1995). Dynamic assessment and the legacy of L. S. Vygotsky. School Psychology International, 16, 143153. Lunt, I. (1993). The practice of assessment. In H. Daniels (Ed.) Charting the agenda: educational activity after Vygotsky. London: Routledge. Milo, B.F., Ruijssenaars, A.J.J.M. & Seeger, G. (2005). Math instruction for students with special educational needs: effects of guiding versus directing instruction. Educational and Child Psychology, 22(4), 6880. Mulligan, J.T. (1998). A research-based framework for assessing early multiplication and division. In C. Kanes, M. Goos & E. Warren (Eds.), Proceedings of the 21st Annual Conference of the Mathematics Education Research Group of Australasia (Vol. 2, pp. 404411). Brisbane: Griffith University. NSW Department of Education and Training (1998). Count me in too: A professional development package. Sydney: NSWDET. Skemp, R. (1976). Relational understanding and instrumental understanding. Mathematics Teaching, 77, 2026. Steffe, L.P., von Glasersfeld, E., Richards, J. & Cobb, P. (1983). Childrens counting types: Philosophy, theory, and application. New York: Praeger. Steffe, L.R. & Cobb, P. (with E. von Glasersfeld) (1988). Construction of arithmetic meanings and strategies. New York: Springer-Verlag. Steffe, L.P. (1992b). Schemes of action and operation involving composite units. Learning and Individual Differences, 4, 259309. Steffe, L.R. (1994). Childrens multiplying schemes, In G. Harel & J. Confrey (Eds.), The Development of Multiplicative Reasoning in the Learning of Mathematics (pp. 341). Albany, NY: State University of New York Press. Thomas, G. & Ward, J. (2001). An evaluation of the Count Me In Too pilot project. Exploring issues in mathematics education. Wellington: New Zealand Ministry of Education. Wheatley, G. & Reynolds, A. M. (1999). Part 1: Helping children learn mathematics. Coming to know numbers. Mathematics Learning, 625. Wright, R.J. (1989). Numerical development in the kindergarten year: A teaching experiment. Doctoral Dissertation, University of Georgia. Wright, R. (1991). What number knowledge is possessed by children entering the kindergarten year of school? Mathematics Education Research Journal, 3(1), 116. Wright, R.J. (1990). Interactive communication: Constraints and possibilities, In L.P. Steffe & T.Wood (Eds.), Transforming childrens mathematics education: International perspectives. Hillsdale NJ: Lawrence Erlbaum. Wright, R.J. (1991a). An application of the epistemology of radical constructivism to the study of learning. Australian Educational Researcher, 18(1), 7595. Wright, R., Cowper, M., Stafford, A., Stanger, G. & Stewart, R. (1994). The mathematics recovery

Educational & Child Psychology Vol 24 No 2

117

Ruth Willey et al.

project A progress report: Specialist teachers working with low-attaining first graders. Proceedings of the 17th annual conference of the Mathematics Education Research Group of Australia, 2, 709716. Wright, R., Stewart, R., Stafford, A. & Cain, R. (1998). Early mathematics. Proceedings of the 20th annual meeting of the North American chapter of the International Group for the Psychology of Mathematics Education, 1, 211216. Wright, R., Martland, J., Stafford, A. & Stanger, G. (2002). Teaching number: advancing childrens skills and strategies. London: Paul Chapman Publishing/Sage. Wright, R., Martland, J. & Stafford, A. (2006a). Early numeracy: Assessment for teaching and intervention. London: Paul Chapman Publishing/Sage. Wright, R., Stanger, G., Stafford, A. & Martland, J. (2006b). Teaching number: Advancing childrens skills and strategies. London: Paul Chapman Publishing/Sage. Young-Loveridge, J. (1989). The development of childrens number concepts: The first year of school. New Zealand Journal of Educational Studies, 24(1), 4764. Young-Loveridge, J. (1991). The development of childrens number concepts from ages five to nine, Volumes 1 and 2. Hamilton, NZ: University of Waikato.

118

Educational & Child Psychology Vol 24 No 2

Educational psychologists assessment of childrens arithmetic skills

Submission to special issue educational and child psychology september 2006
Susie Mackenzie
Abstract
This paper explores a range of tools and approaches available to Educational Psychologists asked to assess mathematical skills and understanding. The main assessment tools and techniques commonly used by EPs are standardised tests, curriculum based assessment, observation and dynamic assessment. Compared with assessment of reading skills there are fewer standardised tests available for the assessment of mathematical skills and the validity of these is questioned. It is suggested that when an EP has a good understanding of how mathematical skills develop and of the key cognitive skills, social experiences, attitudes and strategies that underpin progress then an interactive assessment style can emerge. I describe how observation combined with interactive questioning using a detailed guide developed in Leeds provides a useful and flexible adjunct to standardised tests and curriculum based assessment. This more wide-ranging approach may be more time consuming but can give a better understanding of the gaps in the pupils understanding, the strategies they bring to mathematical tasks and how they may, or may not, be able to use these skills in everyday life.

HIS ARTICLE seeks to address the needs of educational psychologists currently working in the UK. There are a number of reasons for this article at this time: increasing understanding of the nature of mathematical skills, changes in the delivery of the maths curriculum and the introduction of particular interventions to address childrens difficulties with mathematical skills. The purpose of this article is to consider practical aspects of educational psychology work in the context of the other contributions to this special issue providing information about areas of theory, research and practice. This article will focus on assessment of individual children as this continues to be a significant part of the work of educational psychologists, though at times there will be reference to needs of specific groups or whole school issues. Assessment is frequently discussed in professional and academic Educational & Child Psychology Vol 24 No 2 The British Psychological Society 2007

literature (Fredrickson, Webster & Wright, 1991; Miller & Leyden, 1999; Freeman & Miller, 2001; Elliott, 2003: Tymms & Elliott, 2006). This article does not seek to revisit arguments about relative merits of specific forms of assessment, or the question of labels, which frequently arises in conversations with parents, and others, when we are asked for an assessment. The aim here is to provide a relevant summary of research relating to maths learning, and assessment tools, to guide applied psychologists in their thinking when asked to do an assessment in relation to problems with numbers. It is intended to promote assessment that can lead to a good understanding of the childs difficulties, strategies and misconceptions and so that educational psychologists can provide appropriate advice and intervention.

119

Susie Mackenzie

The purpose of assessment of maths skills

Freeman & Miller (2001) provided a taxonomy of assessment with three main paradigms: norm-referenced (psychometrics), criterion-referenced (curriculum based) and dynamic assessment or analysis of the childs responses in a learning situation. Freeman & Miller (2001) repeated an assertion made originally by Fredrickson, Webster & Wright (1991) that assessment by a psychologist needs to go beyond a description of what a child can or cant do, and suggest that the aim of an assessment should be to help those around the child, or teaching the child, to have a better understanding of WHY the child has this pattern of strengths or difficulties. In the context of maths teaching and learning Wiliam (2006) provided a thorough analysis of the role of assessment of maths learning for teachers, and asked some questions that are equally relevant to psychologists involved in the assessment of maths skills. Do we want an assessment that tells us the childs existing state of knowledge, or one that enables us to play a role in improving or supporting learning? Do we want to do an assessment that gives us clues as to why the child is failing or one which gives us a score and tells us that the child is performing below average for their age? For educational psychologists one main purpose of an assessment of maths skills would be to formulate and test hypotheses about why a particular child is failing to progress or falling behind peers. Another issue raised by Wiliam (2006) is that most maths assessment tools measure the childrens ability to compute, which is important, but this is not the same as the childrens numeracy: their ability to solve problems using mathematical knowledge in the real world. While it is important to assess what computations a child can and cannot do, we may also want to assess childrens ability to say whether an answer is correct or reasonable or to know when they have been short changed in a shop. At its most basic we 120

might find that a child can give correct answers to test questions such as 3 2 5, or even 50 20 30, but a more important assessment question might be Can this child go into a shop with a 50 pence piece, buy a packet of crisps and come out with the correct change? Therefore another key element or purpose behind an assessment of maths skills could be to find out how the child manages number problems in everyday life (Hughes, Desforges & Mitchell, 2002).

Typical maths development and key skills

Educational psychologists need to have some idea of what it would be reasonable to expect a child to know, or do, at a particular age or stage of development. What maths tasks should an average 3, 5, 7, 10 or 15 year old be able to do? When the National Numeracy Strategy was introduced in England and Wales (DfES, 1999) specific targets were set for each year group, though these did not necessarily relate to the available evidence relating to maths attainment (Hughes, 2000), or take account of the evidence that there are wide variations in childrens levels and rates of development of maths understanding pre-school (Young-Loveridge, 1989) and in first years of formal maths teaching (Hughes, 2000; Munn, 2004). Research on the development of childrens mathematical skills shows that maths development is far from a simple linear progression (Nunes and Bryant, 1996; Donlan, 1998; Munn, 2004; Dowker, 2005), it may be better to think in term of a set of skills rather than a unitary skill (Munn, 2004; Dowker, 2005). Recent reviews by Munn (2004) and Dowker (2005), looking at teaching mathematics and development of arithmetic and mathematical skills, conclude that there are multiple components that make up each persons ability, and that there are individual differences in the development of each of these components (see Dowker and Gervasoni in this issue). Therefore our assessment will need to consider as many of these components as possible, and acknowledge that it is Educational & Child Psychology Vol 24 No 2

Educational psychologists assessment of childrens arithmetic skills not necessarily helpful to make simplistic statements about what maths knowledge or skills an individual should have at a particular age. When thinking about factors underlying development of maths skills we can start with cognitive factors such as perception of quantity, number sense and subitising the ability to recognise the number of objects in a visual display, for example, * * * * is a set of four (Deheane, 1997; Butterworth, 2003), working memory (Baddeley & Hitch, 1974; Adams & Hitch, 1998), knowledge about number (Dowker, 2005), and how this is different to knowledge about how to carry out specific procedures with numbers (RittleJohnson & Seigler, 1998), fluency of counting and speed of processing (Bull & Johnston, 1997), strategies, application of rules, procedures and holding numbers in their head while calculating (Wright, Martland & Stafford, 2000; Dowker, 2005). Then there are another set of factors that could be described as interactive or environmental such as early everyday experiences relating to number, classroom experience and willingness to solve real life problems using mathematical concepts and computation strategies. For example, experiences with number and maths concepts (such as size, quantity, shape) in the home (YoungLoveridge, 1989; Sammons, Sylva et al., 2002), the quality of teaching, class size, setting policies (Askew, 2001; Boaler & Wiliam, 2001), nature of school curriculum and whether it connects with the childs home experiences with number and maths concepts (Hughes, 2000), myths and emotions about maths learning and how children develop self-perception as a learner of maths (Walkerdine, 1998). Anxiety about learning and doing maths (Buxton, 1981; Ashcroft & Faust, 1994) is a particularly important area. Applied psychologists will be aware of the role emotions play in all aspects of learning and assessment and there is an extensive research literature documenting negative emotions and problems with childrens self-perception as a learner of Educational & Child Psychology Vol 24 No 2 maths, from UK children as young as eight years old (Newstead, 1998), to secondary age pupils (Walden & Walkerdine, 1985; Walkerdine, 1998) and even among students in Further and Higher sector of Education in USA (Betz, 1978; Hembree, 1990) and in the UK (Mackenzie, 2002). There is evidence of the effects of anxiety on maths test performance (Ashcraft & Faust, 1994) and advice on how to cope with or overcome fears that interfere with performance of maths skills (Tobias, 1980; Buxton, 1981; Hembree, 1990). Mathematical skills and their development are complex, therefore a range of key skills or components of maths understanding need to be considered when doing an assessment. To make an analogy with the assessment of reading difficulties most practitioners would agree that at least basic phonics, word decoding and understanding of texts are key ingredients of a reading assessment. An assessment of maths and numeracy skills may need to include the following (depending on the age of the child and the description of the problem provided by their teacher or parent): understanding of basic concepts of size and sets (at perceptual and symbolic levels); knowledge of the number system and counting skills; understanding of the language of maths; memory for specific number facts such as times tables or number bonds; computation strategies and procedures; attitudes to learning maths and selfperception as a learner of maths; and finally, ability to solve problems using mathematical concepts and numbers in everyday life.

Problems that can occur in the development of arithmetic and maths skills
Gifford (2005) provided a comprehensive review of the literature and research evidence relating to maths difficulties in children from a range of countries (including USA, Europe and the Middle East). Her summary of the evidence states that about five percent of children may experience difficulties learning number skills, with boys and girls in equal 121

Susie Mackenzie numbers. Within this context, the only agreed defining characteristic for dyscalculia is poor arithmetic skills (in particular poor number fact recall) that persists despite appropriate teaching. She also noted another finding, common across a range of studies and cultures, that difficulties with maths learning can co-occur with a range of other difficulties to do with reading, language, spatial awareness, coordination, attention and memory. In a very small number of children, such as those with global cognitive deficits and complex learning difficulties, there may be a failure to develop even a perceptual understanding of number (Dehearne, 1997; Staves, 2001), or inability to develop from initial iconic numerosity to understanding a symbolic and language based number system (Gifford, 2005). For other children, without complex difficulties, there are a range of problems. Some can rote count without understanding the principles underlying counting: count each object once and only once, fixed order, the final number defines the set, the number for the set is independent of other qualities such as the size of the objects and that the number of the set is the same no matter which order you count individual items (Gelman & Galistel, 1978). Some children have good situated mathematics for dealing with number in the real world, but cannot connect this to the maths taught in the classroom (Nunes & Bryant, 1997), for example their understanding of number relating to real objects may not connect with the mathematical language and symbolic number system used in classroom learning (Sammons, Sylva et al., 2002; Munn, 2004). For some children language deficits may contribute to difficulties understanding maths (Donlan, 1998). Other difficulties contributing to problems with maths include use of inefficient computation strategies or misapplication of a learned strategy (Wright et al., 2000), inappropriate targets for year groups driven by the National Numeracy Strategy and gaps in knowledge not picked up by teachers (Hughes, 2000; Munn, 2004), 122 slow speed of processing (Bull & Johnston, 1997) or working memory problems (Adams & Hitch, 1998; Henry & MacLean, 2003; McKenzie, Bull & Gray, 2003). We should not be looking only at cognitive processing difficulties but also considering emotional and environmental/situational factors (Walkerdine, 1998; Gifford, 2005). We also need to be aware of groups that are known to have specific difficulties learning maths skills, for example, specific conditions such as Downs syndrome (Bird & Buckley, 2000) specific language disorders (Donlan, 1998) and hearing impairments (Nunes & Moreno, 1998). Equally we need to be wary of unhelpful myths and generalisations such as girls cant do maths, you need to be brainy to do maths, Asian kids are good at maths or you wont need maths if you are going to be a professional footballer, mechanic, rock star or mother that may be affecting assumptions about children learning (Walkerdine, 1998).

Psychometric assessment
Psychometric or norm-referenced tests compare an individual with a large group of children, and children of a range of ages. Performance on these tests can be an indication of capacities, but is not an explanation of those capacities (Howe, 1997), and interpretations of test scores will generally need to take notice of contextual factors, particularly where the child being tested has limited communication skills or where spoken English is not their first language. Table 1 compares the content of some UK standardised tests commonly used by educational psychologists to look at mathematical understanding, understanding of numbers and computation skills. When using these assessment tools practitioners need to be aware of, and consider the implications for test performance of, a number of issues. Where a test includes written calculations it is important to check that the format and symbols used accord with those familiar to the child. For example, children at Key Stage 1 (5 to 7 years old) who are following the National Educational & Child Psychology Vol 24 No 2

British Ability Scales BAS II (Elliott, Smith & McCulloch, 1996) Content 2:67:11 Age range Comment Standardised prior to implementation of the NNS. Mathematical language used in questioning even for very first items. Standardised prior to implementation of the NNS. Early items all visual rather than needing understanding of mathematical language. Later items require child to apply knowledge about computation and number system requiring holding interim results in WM. Standardised prior to implementation of the NNS. Vertical layout of computation and symbols used for division such as 3)96 not compatible with NNS. No word problems or conceptual content at early levels.

Sub test

Early number concepts

Basic concepts, size, more than; Basic number (zero to 2 digits, tens and units) and counting. 5:017:11

Educational & Child Psychology Vol 24 No 2 5:017:11

UK

Quantitative reasoning

Pattern, matching and basic number concepts; more than, less than, double, increasing to more complex rules such as divide by 2 then take away 1.

Number attainment

Recognition of numbers, symbolic computation ( , , x and ); fractions, percentages, decimal notation, powers, some application of number such as discounted prices and fuel consumption. (Wechsler, 2004) Age range 6:016:11

Wechsler Intelligence Scale for Children WISC IV Content

Sub test

Comment Standardised post implementation of the NNS. First 5 items have visual cues, all other items rely on understanding of a verbal question and on holding information in WM. (Continued)

Arithmetic Test

Basic counting, simple computation ( , , x and ) and application of number and computation in word problems

Educational psychologists assessment of childrens arithmetic skills

123

124
UK

Susie Mackenzie

Wechsler Individual Achievement Test WIAT II Content 4:016:11 Age range Comment

(Wechsler, 2005)

Sub test

Mathematical Reasoning

Ranges from basic counting, more than, size, shape, pattern, knowledge of the number system, money, clocks and calendars, to geometry, fractions and probability theory

Standardised post implementation of the NNS. Complex mathematical language is used in questioning, but every test item has visual information that reduces load on WM. Response booklet tells examiner what skills are being tested in each item and includes a grid for noting qualitative observations. Standardised post implementation of the NNS. Horizontal layout of computations and notation compatible with NNS. Response booklet tells examiner what skills are being tested in each item and includes a grid for noting qualitative observations. No word problems.

Number Operations

Recognition and writing of number symbols, counting, and computation ( , , x and ), fractions, percentages, square roots and higher powers, geometry

4:016:11

Educational & Child Psychology Vol 24 No 2

Table 1: (Continued ) Cognitive batteries used by educational psychologists in the assessment of arithmetic skills

Educational psychologists assessment of childrens arithmetic skills Numeracy Strategy will not be familiar with the vertical format used on the BAS II number skills attainment test, as they have only been introduced to calculation with a horizontal format (e.g. 3 2 5 and 4 6 24). Similarly some test forms will use the old fashioned) rather than the symbol for division. Large steps between test items make it hard to determine actual competence or specific gaps or misconceptions and the content of the tests tends to have a very narrow focus on symbolic computation and not to include application of number (Wiliam, 2006), or to test ability to manage computation with visual or concrete support (number lines, structured apparatus etc). Children may fail some test items due to any one of many problems: lack of knowledge of the number system or maths concepts, limited language understanding, the demands made on working memory or heightened anxiety due to performing in a test situation with an unfamiliar adult. The focus when using these tools needs to be not only on the actual scores and what an individual got right or wrong, but also on how the child approached the task, what strategies the child used to attempt to answer test items and careful scrutiny of wrong answers to get a feel for an underlying gap in knowledge, misconception or mis-applied strategy. Teachers will be familiar with test materials for assessment relating to the National Curriculum such as Standard Attainment Tests (SATs) for each Key Stage of Education, NFER tests and such tests as the Basic Number Screening Test (Gillham, 2001). The Basic Number Screening Test (BNS) has the advantage of being easy to administer and child friendly and has two parallel forms allowing for test/re-test comparison (e.g. before and after an intervention). Standardisation of the BNS is from 7 to 12 years but there are not enough test items to allow discrimination at the bottom end of the scale to get a clear picture of specific gaps in understanding for many primary age children with serious weaknesses in maths. One additional assessment tool on the market is the Dyscalculia Screener (Butterworth, 2003). It is standardised for children aged 6 to 14 years, and as its name suggests, it is supposed to identify children at risk by providing information to teachers about skill levels on a set of specific tasks presented on computer, with information about response times as well as numbers of correct and incorrect answers. Though response times may be of interest, the computer tasks have little ecological validity (how the child uses number skills in everyday life), and would not be advisable for any child experiencing maths anxiety. The screener is a mixture of tests based on assumptions about underlying processes such as subitising, and straight forward knowledge about the number system and computation. Profiles produced for individuals on the screener would not necessarily provide clues as to the nature of childs difficulties, or gaps in their understanding, and do not lead to specific advice for teachers about appropriate interventions. The Dyscalulia Screener does not test language understanding, visual memory, working memory or what strategies are being used in computation. At present we do not have a specific standardised test designed to highlight strengths and weaknesses of underlying skills when a child fails to make progress with maths, or to give us detailed information that could be used for the design of an individual intervention programme. The Working Memory Test Battery (Gathercole & Pickering, 2001) could be used to explore components of working memory that might be affecting performance of computation, but this would not identify gaps in basic knowledge, problems with the language of maths or inappropriate procedures used in calculations. If assessing a child that has been experiencing failure with maths learning for some time evidence relating to maths anxiety is worth noting and bearing in mind when interpreting test results. Ashcroft & Faust (1994) took groups of young adults who had experienced failure with maths at school and 125

Educational & Child Psychology Vol 24 No 2

Susie Mackenzie gave them computations to do in test and informal conditions. Performance in test conditions was markedly worse than that in informal conditions. Performance on tests where participants knew their performance was being monitored did not reflect their actual ability to perform computations in less stressful conditions or in everyday life. Standardised tests provide scores that teachers or parents find useful as a guide to the extent of the problem, or confirmation by an independent expert of their assertion that this child has significant difficulties (Freeman & Miller, 2001). Test performance may give some indication as to whether or not there is cause for concern or a need for further investigation or intervention. However, scores on these tests on their own do little in terms of furthering understanding of the nature of the difficulties of an individual child, tell us nothing of how the child can use number skills in real life and do not necessarily provide clues for designing an appropriate intervention. Keeping detailed notes relating to the childs performance while administering a standardised test will be helpful when formulating hypotheses about the nature of their difficulties: such as strategies being used, need for visual aids or writing down workings, misapplying procedures or gaps in knowledge, time taken to retrieve information, what skills are fluent and which require attentional resources. The WIAT (see Table 1) does provide a useful guide in the record booklet for recording some observations. Curriculum-based assessment At its most basic a curriculum based assessment is a descriptive assessment giving a clear picture of what the child can do, without reference to what other children can or cant do, or what is expected at a given age. In terms of maths skills it might be statements such as can add two single digits or knows number bonds to 10. As with test scores such statements may be useful descriptors, or indications that there is a problem, but do little to answer why questions or allow for 126 testing of a hypothesis about the nature of the problem. When an educational psychologist is asked to look at progress of an individual or small number of children falling significantly behind peers National Curriculum (NC) levels routinely used by schools to monitor progress are not detailed enough, or the child may not have reached NC level 1 (i.e., the child does not have the skills expected of a five year old). In these cases we can suggest use of Performance Indicators for Value Added Target Setting or PIVATS (Lancashire, 2002). PIVATS cover NC levels 1 to 5, but in addition they provide indicators for levels P1 to P8 leading up to NC level 1. PIVATS categorise maths learning in three components: Number, Application of Number and Shape, Space and Measures with detailed descriptive statements about what the child can do. In addition each level (P1 to P8) is broken down into five particular skills. For example P5 level for Number has five indicators if the child has achieved one of these indicators then they could be said to be at NC level P5i, when they achieve another of the indicators they are at level P5ii and so on. These detailed descriptors can be particularly useful when holding a conversation with a teacher regarding targets for an Individual Education Plan (IEP), personalisation of learning and for monitoring learning over a term or year. Informal, qualitative and interactive assessment Educational psychologists can adopt approaches known as dynamic assessment to look at the skills the child brings to learning situations in general (Elliott, 1993; Elliott, 2003; Resing, 2006) such as adaptive thinking skills, mediation of learning, ability to learn from demonstration of a rule. This approach can be applied to understanding of maths constructs, ability to carry out particular mental maths tasks or use maths in real life. Given the lack of diagnostic psychometric tools discussed earlier, these may be more helpful in making sure an assessment can test a range of hypotheses or proEducational & Child Psychology Vol 24 No 2

Educational psychologists assessment of childrens arithmetic skills vide evidence for recommending a particular intervention, but it will take considerably more time and by its nature it is not standardised and could be seen by some as biased or subjective (Elliott, 1993; Dowker, 2005). In an informal assessment of maths skills the educational psychologist can engage the child in maths activities and games in a relaxed way (minimising test anxiety, Ashcraft & Faust, 1994). We can use observation, questioning and mediated learning to look for zones of proximal development and to gather information relating to the childs understanding of quantity, numbers, language of maths, counting fluency, strategies, recall of number facts and ability to apply knowledge in real, practical ways. The assessment can become a conversation: asking the child to do a range of maths related activities, taking into account the age of the child, expected targets for their age and information from school about what the child can and cant do, or how the child approaches activities relating to maths and arithmetic. As prompts for questions and concepts to use to identify specific maths learning problems Wright et al. (2000) provide a very useful assessment technique (p3962) and Denvir & Bibby (2001) have produced a very clear hierarchy of skills in their Diagnostic Interviews in Number Sense. Education Leeds School Support Service has devised their own assessment sheets based on principles set out by Denvir & Bibby (2001). A copy of the assessment guide is included as an appendix. The focus is on key skills and both asking questions of the child and asking the child to set challenges for you, with hints as to what to recommend if a child cannot complete a particular task. It is very flexible, particular skills to check in detail can be related to the age and general developmental level of the child and what the class teacher or parent has already told you. This might include: Does this child count with one to one correspondence? Does this child understand specific mathematical language? Educational & Child Psychology Vol 24 No 2
Is this child operating at an iconic or sym-

bolic level of understanding of number? (Can the child count or compute using symbols or do they need objects, fingers etc.) How sound is their long-term memory for the number system/number bonds/ multiplication facts/place value? What skills are fluent (retrieving number facts from long term memory) and which require conscious effort (and working memory resources)? What strategies is the child using for computation ( , , and )? Items I have in my maths assessment kit, used alongside this guide, include a range of beads for sorting, counting and estimating, large dice, playing cards, a small clock, real coins (1 to 1p), Numicon plates, place value (PV) arrow cards, a 100 square and white board and marker. With children in the last year of Primary School and at Secondary level I am interested in not only number but also application of knowledge especially relating to clocks and money (hence the small clock and set of coins in my kit), and their experiences as a maths learner to check beliefs about maths, motivation and attitudes to learning about maths. Most EPs would include in an assessment questions about likes and dislikes, favourite activities etc., and these can be easily applied to an assessment of affect, motivation and attitudes to learning about numbers asking not only obvious questions such as do you like learning about numbers? And do you feel confident in maths lessons, but also exception questions such as which bits of numeracy lessons do you like best? or which bits do you find easy? or how does Mrs X help you?.

Results of assessments: advice and intervention

If the purpose of an assessment were to identify what the nature of the problem is, then one of the results would be to give advice tailored to help the individual. Educational psychologists may also be asked for their advice if a whole class appears not to be making progress or when tracking for the school 127

Susie Mackenzie shows that value added for maths is not as good as that for literacy or science. A useful place to start for both individual solutions and class or school level interventions would be to discuss with staff their whole approach to maths teaching and suggest they look at publications from the DfES such as Supporting Pupils with gaps in their mathematical understanding (DfES, 2005; see also Gross in this issue), and encourage teachers to implement specific teaching to address gaps in understanding (Dowker, 2005; Gifford, 2005; see also Dowker in this issue). Sometimes a child struggling with maths may be expected to practice skills they find difficult, rather than have specific teaching to address gaps in their understanding, but this should be avoided as it could be counter-productive, leading to frustration, boredom and lack of motivation (Boaler & Wiliam, 2001). If assessment of the difficulties of an individual or group identifies a specific problem, then approaches can be recommended to address that problem. For example, if it is a problem with working memory then methods can be introduced to reduce memory load (such as use of visual and concrete aids); if it is a problem of misapplication of a set of strategies then this can be addressed through specific teaching and explanation. There are now intensive structured programmes available such as the Mathematics Recovery Programme (Wright et al., 2000; see also Willey et al., this issue), the Numeracy Recovery Programme (Dowker, 2001; Dowker, 2004 see also Dowker this issue) and approaches that make use of visual support for maths (Wing and Tacon, 1999; DfES, 2005; Brighton & Hove, 2006). Older pupils will benefit from developing strategy flexibility or improving knowledge of the sequential structure of numbers (see Verschaffel et al; Ellemor-Collins and Wright, this issue). Schools might also want to consider peer tutoring and group collaboration, re-thinking setting arrangements and other in house initiatives (Topping & Bamford, 1998; Boaler & Wiliam, 2001; Topping, Campbell, Douglas 128 & Smith, 2003; Dowker, 2005). Intervention for older pupils and young adults might also need to address self-perceptions and constructs relating to learning maths (Hembree, 1990). Structured recovery and catch up programmes for pupils with problems need to be delivered by staff with a good understanding of mathematics, how skills develop and the nature of the childs specific problems. Teaching assistants (TAs) are increasingly likely to be the main providers of support for individual children who need individual learning plans, and good quality training is essential for TAs (Farrell, Balshaw & Polat, 1999). Teaching assistants supporting children struggling with maths may need particular assistance with understanding how to teach number concepts, or particular skills and strategies, in which case the best intervention might be to provide specific training for TAs regarding the nature of mathematical development: key concepts, language, processing required and so on. Alternatively we might advise the class teacher, who we would expect to know more about teaching mathematics, to let the TA oversee more capable students while she teaches the child with difficulties or devises an individual programme to be delivered by a TA. Sure Start and other community initiatives provide another potential intervention through pre-school community based schemes, such as the BigMath programme (Ginsburg, Balfanz & Greenes, 1999) which was designed to raise maths understanding in young children through games and other fun activities. Workshops can be run for families with ideas for games and activities to do at home: everyday activities from the mundane such as setting the table for a meal; counting songs with classics such as five speckled frogs; card, dice and board games. These activities provide practice for a range of concepts, counting and computation skills in fun and engaging ways, and can build confidence for parents as well as children (Warren & Westmoreland, 2000).

Educational & Child Psychology Vol 24 No 2

Educational psychologists assessment of childrens arithmetic skills

Conclusion
This article aims to lead practitioners to a more thorough understanding and approach to the assessment of basic arithmetic (ability to compute with numbers), numeracy (ability to use numbers in real life) and identifying gaps in mathematical understanding and knowledge. Educational psychologists will be aware of the five outcomes of education as set out in the Childrens Act (2004): to be happy, safe, enjoying learning, become economically secure and able to make a positive contribution to society. Basic numeracy and maths skills contribute to all these, especially economic security, and helping children to become mathematically literate citizens in society (Wiliam, 2006). Standardised tests and curriculum-based assessments have their place in the psychologists toolkit. However, if we want to answer why? questions, to consider different hypotheses about the nature of the problem and to be in a position to recommend welltargeted interventions, then these approaches are best used with caution and in conjunc-

tion with more searching observation and interactive questioning along the lines suggested in this article. What is really important is that practitioners doing an assessment of maths difficulties have a good understanding of the complexity of maths learning and what different factors might be contributing to the difficulties of an individual. Educational psychologists are in an important position to inform and support those working directly with the child, so we need to be well informed. Where children have problems learning maths our efforts to make a difference will be depend on the quality of our own knowledge and understanding regarding maths learning in general and maths problems in particular.

Address for correspondence

Susie Mackenzie, Psychology and Assessment Service, Education Leeds Blenheim Centre Crowther Place Leeds LS6 2ST E-mail: susie.mackenzie@educationleeds.co.uk

References
Adams, W. & Hitch, G.J. (1998). Childrens mental arithmetic and working memory. Chapter 7 in C. Donlan (Ed.), The development of mathematical skills. Hove: Psychology Press. Ashcraft, M.H. & Faust, M.W. (1994). Mathematics anxiety and mental arithmetic performance: an exploratory investigation. Cognition and Emotion, 8, 97125. Askew, M. (2001). Policy, practices and principles in teaching numeracy: What makes a difference? In P. Gates (Ed.), Issues in mathematics teaching (chapter 8). London: Routledge. Baddeley, A.D. & Hitch, G.J. (1974). Working Memory. In G.H. Bower (Ed.), The psychology of learning and motivation, 8, 4789. Betz, N.E. (1978). Prevalence, distribution and correlates of maths anxiety in college students, Journal of Counselling Psychology, 25, 441448. Boaler, J. & Wiliam, D. (2001). Weve still got to learn! Students perspectives on ability grouping and mathematics achievement. In P. Gates (Ed.), Issues in mathematics teaching. London: Routledge Press. Bird, G. & Buckley, S. (2001). Number skills of individuals with Down Syndrome an overview. Portsmouth: Down Syndrome Educational Trust. Brighton & Hove (2006). Visual Models and Images supported by Signs and Symbols: A Wave 3 programme designed by Brighton and Hove LEA. Bull, R & Johnston, R.S. (1997). Childrens arithmetic difficulties: Contributions from processing speed, item identification and short-term memory. Journal of Experimental and Children Psychology, 64, 124. Butterworth, B. (2003). Dyscalculia screener: Highlighting pupils with specific learning difficulties in maths. Windsor: nferNelson. Buxton, L. (1981). Do you panic about maths? Coping with maths anxiety. London: Heinemann. Dehaene, S. (1997). The number sense: How the mind creates mathematics. London: Allen Lane. Denvir, H. & Bibby, T. (2001). Diagnostic interviews in number sense. London: Kings College. DfEE (1999). The national numeracy strategy. Suffolk: Cambridge University Press. DfES (2005). Primary national strategy: Using models and images to support mathematics teaching and learning in Years 1 to 3. Norwich: HMSO.

Educational & Child Psychology Vol 24 No 2

129

Susie Mackenzie
DfES (2005). Primary national strategy: Supporting children with gaps in their maths understanding: Wave 3 mathematics. Norwich: HMSO. Donlan, C. (Ed.) (1998). The development of mathematical skills. Hove: Psychology Press. Dowker, A. (2001). Numeracy recovery: a pilot scheme for early intervention with young children with numeracy difficulties. Support for Learning, 16(1), 610. Dowker, A, (2004) What works for children with mathematical difficulties. London: DfES Research Report 554. Dowker, A. (2005). Individual Differences in Arithmetic: implications for psychology, neuroscience and education. Hove: Psychology Press. Elliott, C., Smith, P. & McCulloch, K. (1996). British ability scales second sdition (BASII). Windsor: nferNelson. Elliott, J. (1993). Assisted Assessment: if it is dynamic why is it so rarely employed? Educational and Child Psychology, 10(4), 4858. Elliott, J. (2003). Dynamic Assessment in educational settings: Realising potential. Educational Review, 55(1), 1532. Farrell, P., Balshaw, M. & Polat, F. (1999). The management, role and training of learning support assistants. London: Department for Education and Science. Fredrickson, N., Webster, A. & Wright, A. (1991). Psychological assessment: A change of emphasis. Educational Psychology in Practice, 7, 2029. Freeman, L. & Miller, A. (2001). Norm-referenced, Criterion-referenced and dynamic assessment: What exactly is the point? Educational Psychology in Practice, 17(1), 316. Gathercole, S. & Pickering, S. (2001). The working memory test battery for children. London: Harcourt Assessment. Gelman, R. & Gallistel, C.R. (1978). The childs understanding of number. Cambridge MA: Harvard University Press. Gifford, S. (2005). Young childrens difficulties in learning mathematics: Review of research in relation to dyscalculia. London: Qualifications and Curriculum Authority. Gillham, B. (2001). Basic number screening test. London: Hodder & Stoughton. Ginsburg, H.P., Balfanz, R. & Greene, C. (1999). Challenging mathematics for young children. In A. Costa (Ed.), Teaching for intelligence, II: A collection of articles. Arlington Heights: Skylight. Hembree, R. (1990). The nature, effects and relief of maths anxiety. Journal for Research in Maths Education, 21, 3346. Henry, L.A. & MacLean, M. (2003). Relationships between working memory, expressive vocabulary and arithmetical reasoning in children with and without intellectual disabilities. Educational and Child Psychology, 20(3), 5164. Howe, M.J.A. (1997). IQ in question: The truth about intelligence. London: Sage. Hughes, M. (2000). The national numeracy strategy: Are we getting it right? The Psychology of Education Review, 24, 311. Hughes, M., Desforges, C. & Mitchell, C. (2002) Numeracy and beyond. Buckingham: Open University Press. Lancashire County Council (2002). Performance indicators for value added target setting: Assessment of learning, performance monitoring and effective target setting for all pupils revised edition. Preston: Lancashire Professional Development Service. Mackenzie, S. (2002). Can we make maths count in HE? Journal of Further and Higher Education, 26(2), 159172. McKenzie, B., Bull, R. & Gray, C. (2003). The effects of phonological and visuospatial interference on childrens arithmetical performance. Educational and Child Psychology, 20(3), 93108. Miller, A. & Leyden, G. (1999). A coherent framework for the application of psychology in schools. British Educational Research Journal, 25, 389400. Munn, P. (2004). The psychology of maths education in the early primary years. The Psychology of Education Review, 28(1), 48. Newstead, K. (1998). Aspects of childrens mathematics anxiety. Educational Studies in Mathematics, 36, 5371. Nunes, T. & Bryant, P. (1996). Children doing mathematics. Oxford: Blackwell. Nunes, T. & Bryant, P. (Eds.) (1997), Learning and teaching mathematics: An international perspective. Hove: Psychology Press. Nunes, T. & Moreno, C. (1998). Is hearing impairment a cause of difficulties in learning mathematics? In C. Donlan (Ed.), The development of mathematical skills (chapter 10). Hove: Psychology Press. Rittle-Johnson, B. & Seigler, R.S. (1998). The relationship between conceptual and procedural knowledge in learning mathematics: A review. In C. Donlan (Ed.) The development of mathematical skills (chapter 4). Hove: Psychology Press. Resing, W.C.M. (2006). Using childrens ability to learn in diagnosis and assessment. Educational and Child Psychology, 23(3), 611. Sammons, P., Sylva, K., Melhuish, E. C., Siraj-Blatchford, I., Taggart, B., & Elliot, K. (2002). The effective provision of pre-school education project. Technical Paper 8a: Measuring the impact on childrens cognitive development over the pre-school years. London: Institute of Education/DfES. Staves, L. (2001). Mathematics for children with severe and profound learning difficulties. London: David Fulton.

130

Educational & Child Psychology Vol 24 No 2

Educational psychologists assessment of childrens arithmetic skills

Tobias, S. (1980). Maths anxiety: What you can do about it. Todays Education, 69(3), 2629. Topping, K. & Bamford, J. (1998). Parental involvement and peer tutoring in mathematics and science. London: David Fulton. Topping, K.J., Campbell, J., Douglas, W. & Smith, A. (2003). Cross-age peer-tutoring in mathematics with seven and eleven year olds: Influence on mathematical vocabulary, strategic dialogue and self-concept. Educational Research, 45(3), 287308. Tymms, P. & Elliott, J. (2006). Difficulties and dilemmas in the assessment of special educational needs. Educational and Child Psychology, 23(3), 2534. Walden, R. & Walkerdine, V. (1985). Girls and mathematics: From primary to secondary schooling. Bedford Way Papers 24 London: Heinemann. Walkerdine, V. (1998). Counting girls out: Girls and mathematics (2nd edn). Hove: Falmer Press. Warren, V. & Westmoreland, S. (2000). Number and play in everyday life. In Drury, R., Miller, L. & Campbell, R. (Eds.), Looking at early years education and care (chapter 16). London: David Fulton. Wechsler, D. (2004). Wechsler Intelligence Scale for children (UK 4th edn). London: Harcourt Assessment. Wechsler, D. (2005). Wechsler Individual Achievement Test (UK 2nd edn)London: Harcourt Assessment. Wiliam, D. (2006). Keeping learning on track: Classroom assessment and the regulation of learning. In F.K. Lester (Ed.), NCTM handbook of research on mathematics education. Wing, T. & Tacon, R. (1999). A visual approach to mental arithmetic. Paper presented at the BCME-4 University College, Northampton. Wright, R.J., Martland, J. & Stafford, A.K. (2000). Early numeracy: Assessment for teaching and intervention. London: Paul Chapman. Young-Loveridge, J. (1989). The relationship between childrens home experiences and their mathematical skills on entry to school. Early Child Development and Care, 43, 4359.

Educational & Child Psychology Vol 24 No 2

131

132 Questions Observations Recommendations

Appendix: Education Leeds maths assessment guide

Susie Mackenzie

Task

Count forwards in 1s from zero

Count backwards in 1s

Id like you to start at 1 and count as Fluency/pace far as you can go (stop when appropriate). confuses ty/teen babbles/clarity of words This time start at 10 and count backwards. hesitancy/sequencing across decades beyond 100 when Now start at 20 and count backwards. appropriate understanding term count back can count back when prompted
hesitancy/sequencing across

practise laying out numeral cards practise with number line/own 100 square move finger/pointer whilst counting develop photo of 100 square /order use teen/ty prompt 100 square jig-saws place missing numbers back in to 100 square

Count forwards in 1s from a given number

More counting on. This time start at * and count on until I say stop.

as above use flash-cards, e.g. forwards Move forwards/backwards along

colour line
practise putting finger/pointer on

Count backwards in 1s from a given number

understands the key vocabulary can do this when demonstrated can point to number before/after

Now counting backwards again. Start at * and count back until I say stop.

decades says given number/unable to just think given number and continue counting beyond 100 when appropriate

number thinking it and then counting

point to next number after */number

Give next number after/next 2 numbers

What number comes next after * Tell me the next 2 numbers after *

just before* (or object/picture in a line)

adult says number sequence, stops

Give number before/ 2 numbers before

What number comes just before * Tell me the 2 numbers that come just before *

on a 100 square/number line

and pupil gives next number e.g. 14 15 16 _ practise with 100 square .Use flash cards/vocab book

Add 1 to a given number/add 2

understands the key vocabulary knows to take-away when asked

to subtract
does this mentally, i.e. relates to

work practically/real life problems build in time to visualise use own 100sq and link add to

Educational & Child Psychology Vol 24 No 2

Subtract 1 from a given number/subtract 2

Im going to say a number and I want you to add 1 3 1 etc Then adding 2 Tell me what subtract means Im going to say a number and I want you to subtract 1. 5 1

next number and subtract to above number before uses fingers or requires apparatus for adding/subtracting 2 use whisper/shout use flash-cards/vocabulary book

Task
understands key vocabulary can do when given a real life

Questions

Observations

Recommendations

Say the number that is 1 more example/apparatus

Tell me the number that is 1 more than *

Say the number that is 1 less

can do this randomly/quick recall needs to give answers in a learnt Use visual prompts/RAP

Tell me the number that is 1 less than *

work practically/real life problems build in time to visualise practise with own 100 square/photo use flash cards/vocabulary book

Knows pairs of numbers which make 10

You did really well with the adding up What does this number say? (10) Tell me 2 numbers that add up to 10, etc.

Educational & Child Psychology Vol 24 No 2 (school and home) pattern/sequence Domino activity/ele flips needs to be given a number Allow processing time prompt does not recognise commutativity all above apply to higher numbers
work practically/real life problem teach to count all

Knows pairs of numbers which make 20 and relate to higher numbers

As above

Add 2 numbers under 10/20 counts all counts on from first number counts on from largest number count back only uses known facts to derive

Tell me what * add * is/makes/equals

Subtract 2 numbers under 10/20

Show card, e.g. 9 4. Read this to me. Give me the answer. Tell me what * subtract /take away * is/ makes/equals Show card, e.g. 10 4. Read this to me. Give me the answer

count on/back count on from largest number use number bonds use doubles

Count forwards in 10s to 100

Count on in 10s for me

practise with Place Value (PV)

Arrow Cards
use ty visual prompt relate to 100 square

Count backwards in 10s from 100

Count backwards in 10s from 100

understands instruction gets to 90 then stops or confused gets to 20 then reverts to teens gets to 10 then stops does not say zero when counting back

Educational psychologists assessment of childrens arithmetic skills

133

134 Questions
does not relate this to counting

Task

Observations

Recommendations

Susie Mackenzie

Add/subtract 10 (then 20/30, etc. ) on/back in 10s reverts to fingers can read/write multiples of 10 will say that 40 is 4 tens will say that 40 is 40 tens 3 digit numbers if appropriate

Say how many 10s in 40, etc.

Tell me what 10 add 10 is/makes What about 20 add 10, etc. What is 20 subtract/take away 10, etc. What does this number say? (e.g. 40) How many 10s make/are there in 40?

use PV/Arrow cards write out sequences develop visualisation of moving cards use flash-cards and relate adding to counting on/subtracting to counting back lay out PV cards on a base board if appropriate and count down, e.g. 1, 2, 3 tens equals 30, etc. relate to 100 square use ty visual prompt
practise with 100 square and PV cards develop visualisation of moving cards write out sequences

Count forwards in 10 from a given number demonstration/written prompt/ using PV cards/100 sq beyond 100 if appropriate

This time I want you to count in 10s starting at the number * (e.g. 27)

unable to do this can do this following

Count backwards in 10 from given number

This time I want you to count backwards in 10s starting at the number * (e.g. 52)
as above not relating counting on to add/ as above use flash cards and relate to counting

Add 10 to a given number

Lets start with the number * (e.g. 4,6,8,9) Now add 10 (and 10 and 10, etc.)

counting back to subtracting

on/back

Educational & Child Psychology Vol 24 No 2

Subtract 10 from a given number Task

Lets start with the number * (e.g. 96,98,99) Now subtract/take away 10 (take away 10 take away 10, etc)

Task
can read/write 2 digit numbers practise with PV cards:

Questions

Observations

Recommendations

Place value Say how many tens/units in a 2 digit number

Point to the number **/Write the number ** for me (or ***) Show me the tens? Show me the units? (T Us ) How many tens are there?(Hs Ts) Lets do some more adding. What is 20 3 etc (100 40 3, etc.) (3 digit if appropriate ) reverses the order of the digits can identify the tens digit and say that 4 tens are 40, etc. can identify tens digit but unable to give the value of the 4 i.e., says that it is 40 tens/or that it is 4 as above with 3 digit numbers when asked 20 3 reverts to counting on when asked 46 6 reverts to counting back

Educational & Child Psychology Vol 24 No 2 put the 10 card on the table say 10 fit the 1 card on top say 10 and 1 equals 11 fit the 2 card on top say 10 and 2 equals 12, etc. use the teen visual prompt write 10 and ( ) 1 10 etc build in time to visualise cards partition 2/3 digit numbers using PV cards use PV cards to make 2/3 digit numbers
when counting in 2s stops at 10/20 use apparatus/count in 2s and colour when counting in 5s stops in a 100 square. Point out the pattern

Can partition, i.e. 23 20 3 146 100 40 6, etc.

Lets use PV Arrow Cards. Add 20 and 3. What number is it, etc (200 70 5 etc) What number is this? (**/***) Which two/three cards do you need?

Knows that 46 6 40 Count in 2s/5s before 100

Subtraction/take aways . What is 46 6, etc. Lets use the cards. Count in 2s ( say stop if appropriate )

Count in 5s (say stop if appropriate)

Keep as visual prompt

use whisper/shout to count

can count in 2s/5s but unable to

apply this when counting grouped objects/coins

in 2s
as above for counting in 5s count grouped objects

e.g. 2 4 6 8 & 1 more

understands vocabulary relates to counting in 2s recognises significance

9
make a poster (for home and school).

Count in 2s from zero/recognise as even

What is this number? (within 10 ) Is it odd or even? How did you know that?

of last digit

Point out the importance of last digit. Have fun with car number plates/ large numbers 2 4 6 8 0 even

Count in 2s from 1/ recognise as odd

If appropriate show numbers beyond 10

Educational psychologists assessment of childrens arithmetic skills

135

136 Questions
unable to give answer to double 2 identify which are double dominoes

Task but knows what 2

unable to identify a double domino use card dominoes. Fold to teach

Observations 2 equals, etc. halves.

Observe strategies used Listen to explanations as they

Recommendations dots/pictures/numbers

Give doubles

Whats double 2, 3, 4, 5, 10, 20, 50, 100

Susie Mackenzie

Give halves

Whats half of 4, 6, 8, 10, 20, 40, 100, 200

Observe pupil using pencil and paper methods

on appropriate tasks. work

Make note of any misconceptions

Ask pupil to talk about what they are doing.

Table compiled by Jane Sellers and Susie Mackenzie

Educational & Child Psychology Vol 24 No 2

Linking childrens home and school mathematics

Martin Hughes, Pamela Greenhough, Wan Ching Yee, Jane Andrews, Jan Winter & Leida Salway
Abstract
Current approaches to teaching mathematics in English primary schools pay little attention to the kind of mathematics which children engage in outside of school. This paper attempts to redress the balance by describing the nature and characteristics of childrens out-of-school mathematics, and looking at how connections might be made between in-school and out-of-school mathematics. At home, mathematics is frequently encountered during play and games, and in authentic household activities such as cooking and shopping. There are also more school-like mathematical activities such as homework and commercially available maths schemes. The paper argues that it is important for connections to be made between home and school mathematics, but this is often impaired by teachers lack of knowledge about home mathematics and by parents lack of knowledge about school mathematics. One solution to this problem lies in knowledge exchange activities, and examples are provided of activities which operate in both the school-to-home and home-to-school directions. The main implications for teachers and educational psychologists are to pay much greater attention to childrens out-of-school mathematics, and to develop further ways of linking home and school mathematics.

HE NATIONAL Numeracy Strategy (DfEE, 1999) was introduced into English primary schools in 1999. Its overall aim was to raise standards of attainment in mathematics, and in particular to meet the governments target that by 2002 threequarters of 11-year-olds should reach level 4 in their mathematics SATs. Yet by 2006 it was being accepted that although the Strategy had been widely welcomed by teachers, it had not fully achieved its targets. A new approach was required, one which was less prescriptive than what had gone before and which allowed teachers to be more creative and innovative in their practice (DfES, 2006). While the 2006 Framework is less prescriptive in its approach, it shares with the original National Numeracy Strategy a number of key assumptions about the teaching and learning of mathematics. In particular, both documents appear to regard mathematics as a set of skills and competences which are acquired and assessed within the classroom setting, under the Educational & Child Psychology Vol 24 No 2 The British Psychological Society 2007

supervision of a trained professional teacher. Despite general encouragement within the 2006 Framework to build good home-school links, there is little or no recognition that mathematics may be acquired and used outside the classroom, or that parents and other family members might play an important role in supporting childrens mathematical learning. Such an approach within the primary mathematics curriculum is surprising, given that much recent research on mathematics learning is taking a very different perspective. Many psychologists and mathematics educators are paying increasing attention to the ways in which mathematics is actually acquired and used in a wide range of real-life contexts. In a classic study by Nunes and her colleagues (Carraher et al., 1985), the authors looked at the ways in which young Brazilian street traders carried out calculations while conducting their businesses. Carraher et al., found that the traders not only used methods which were very different 137

Martin Hughes et al. from those they had been taught in school, but they were also more successful with their own methods than when the same problems were presented as school maths. More recently, a number of studies have shown that diverse groups such as farmers (Abreu, 1999), nurses (Hoyles et al., 2001) and fishermen (Nunes et al., 1993) typically develop and use methods for mathematical calculation which are different from those conventionally taught in schools. Such findings provide support for Laves (1988) claim that mathematics like other cognitive processes is essentially situated, and that different cultural practices may generate and support different ways of doing mathematics. A similar picture is starting to emerge from the more limited research on home mathematics. Street et al. (2005) argued that the numeracy practices in which children participate at home and at school differ on a number of dimensions. For example, home numeracy practices usually centre on solving a specific local problem, while school numeracy practices are about learning skills or concepts determined by the teacher. Street et al. also argued that there is considerable variation across home numeracy practices. While they were at pains to point out that they are not operating from a deficit position, they claimed that a key component in childrens mathematical attainment at school lies in the degree of consonance or dissonance between childrens home and school numeracy practices. In this paper we will look more closely at the nature of home mathematics, and its relation to school mathematics. In particular, we will argue that the possibility of making meaningful connections between home and school mathematics may be limited by the fact that teachers and parents often know very little about what is going on at home and at school respectively. We will further argue that this problem can be addressed by knowledge exchange activities between home and school.

The home school knowledge exchange project

This paper draws on the work of the Home School Knowledge Exchange Project, which took place between 2001 and 2005. The overall aim of the project was to develop and implement programmes of home-school knowledge exchange activities and look at their impact on children, teachers and parents. The project consisted of three main strands, focusing in turn on literacy at key stage 1, numeracy at key stage 2, and primary/secondary transfer. The data to be presented in this paper comes from the numeracy strand of the project (see Winter et al., 2004; Winter et al., 2007 for more details of this strand). The numeracy strand involved children in Years four and five from four contrasting primary schools in Bristol and Cardiff. In each city one school had a relatively high proportion of pupils eligible for free school meals, and one school had a relatively low proportion of such pupils. In addition, the ethnic composition of the sample attempted to match the ethnic diversity to be found in the two cities. Within each school, a teacherresearcher seconded part-time to the project worked with teachers, parents and children to develop and implement a range of home-school knowledge exchange activities, designed to communicate knowledge between home and school. The impact of the HSKE activities was evaluated using a range of measures. These included standardised assessments of attainment and attitude undertaken by all the children in the cohort (whose performance was then compared with children from similar schools where HSKE activities did not take place). In addition, in each school six children were chosen for more intensive study, on the basis of gender and attainment, and in-depth interviews were carried out with these children, their teachers and their parents. The data presented in this paper comes primarily from these target children and their families. Findings related to other aspects of this strand will be reported in subsequent papers. Educational & Child Psychology Vol 24 No 2

138

Linking childrens home and school mathematics As part of the involvement in the project, each target family was loaned a video camera and asked to record mathematics events which took place in the home. This request was made after a long interview in which the kinds of mathematics taking place at home had been explored. Where detailed interactions are reported below, they are based on the transcriptions of these videos. UK. For example, Dhanu and his older brother made a video of themselves playing Carrom. This game, which has been described as a combination of pool, marbles and air hockey, is extremely popular on the Indian subcontinent and in other countries with a substantial south-Asian population. It is essentially a board game, and sinking ones pieces down various holes around the board scores points. Dhanu and his brother were clearly experienced and skilful players, with a strong shared understanding of the game which meant that conversation was kept to a minimum. It should be noted that mathematics was often present in the childrens play, including their fantasy or role-play. In the following extract from an interview with Chloe, she talks about how she likes to play at estate agents. She describes how in her play she recreates some of the mathematics that adults would need to engage in if they were really selling houses: Interviewer: So when youre playing mortgages, what does that involve, what do you do there? Chloe: Like I pretend to talk to people and I like say how much money do you want to spend and .. whats the total money you want to spend, and theyll say like 300,000, something like that, and then I have to try and find them a house, like how many rooms they want if they want like a three-bedroomed house I have to try and look in pretend to look in books for a three-bedroomed house. And if they like if they find if they want one in like say [name] (local area where she lives), but then they want one in say Newport, Ill pretend to go . Ill go on like the laptop and Ill look and see what ones the best quality and they have to choose and something like that. Then I write it all down, like where theyre moving and how much they really want to spend, and then how much it costs, and then they have to write me a cheque out, and then Ill pretend to fax the cheques off, and then Ill do other stuff. 139

The nature of home mathematics

The children in our project were engaged in a substantial number of activities which involved mathematics in some way. These activities fell into three main groups play and games, authentic household activities, and school-like activities. Many of the games which the children played involved some kind of mathematics, and some of the games involved a great deal. In Monopoly, for example, the participants need to add together the numbers shown on two dice, move their counters the appropriate number of spaces around the board, and give and receive change as they buy property and collect rent from other players. They also have to make more strategic financial decisions about whether or not to buy a particular property or to pay a fine to get out of jail. Some of the children also played games which they had designed or developed themselves. Ryan, for example, was filmed by his mother playing outside on the street with his younger brother and some friends. The street was in a quiet housing estate and there was little traffic. The game which the children called Kerbs involved taking turns to throw a football from one side of the street to the other. The aim was to hit the opposing kerb as near to the edge as possible, and this counted as 20 points A near hit meant the thrower could get a second throw, this time taken from the middle of the street, and a successful throw here scored 10 points. The game proceeded at a high pace and it was clear that Ryan and his friends were very familiar with it. Several of the games involving mathematics had clear cultural roots outside the Educational & Child Psychology Vol 24 No 2

Martin Hughes et al.

Authentic household activities

Many everyday activities which take place in and around the home involve mathematics. Cooking, for example, can involve weighing quantities, using multiplication or division to amend a recipe, calculating the time needed for different parts of the process, and switching between different units of measurement. Planning a journey or holiday can involve consulting bus or train timetables, setting a budget, and working out the most cost effective means of travel. Programming the video can involve calculating how long a programme will last, working out whether there is enough blank tape on the video, and grappling with the 24-hour clock. Most of the children in our research were regularly engaged in these kind of household activities. We termed them authentic (cf Brown et al., 1989) because the purposes underlying the activities are an integral part of household life. Amongst other things, this means that it is important for all those involved that the mathematics is carried out correctly. If a mistake is made, then someones favourite TV programme will not be recorded, or their dinner will be ruined. At the same time, it should be noted that these activities are not fundamentally about mathematics or learning mathematics, despite the important role which mathematics plays in their successful completion. Many of these authentic household activities involve money. Nadia, for example, accompanied her mother on shopping trips as her mothers English was relatively modest. Nadias role was to read out the prices on the goods in the shops, check what was bought against the shopping list, and make sure her mother received the correct change. Other children encountered money through assisting in their parents work. Olivias mother was the part-time manager of a local centre for learning disabled people, and Olivia frequently helped her at the centre, running the coffee-bar and taking responsibility for the takings. Aaqils father was a taxi-driver, and Aaqil would collect spare change from his father which he would then give to charity. 140

One girl in the study, Ellie, made a video describing how she used mathematics to solve an authentic household problem. Ellies family were going away on holiday for two weeks, and during this time their cat would be looked after by neighbours. Ellie had to work out how much cat food to leave them. The cat food was in the form of granules and the daily amount depended on the weight of the cat. So, Ellie needed to weigh her cat. She placed the bathroom scales in front of the camera but the cat did not want to stay on the scales by itself. Ellie told the camera how she was going to solve the problem:
Im going to weigh my cat. First of all Im going to stand on the scales and tell you my weight, then Im going to stand on the scales with the cat in my arms and take that weight and tell you. Then Im going to take away the first weight I tell you from the second weight, and the weight left will be my cats weight. From that Ill be able to work out how much cat food my cats going to need.

In the next scene Ellie is holding up a piece of card on which she has written: My weight cat 53kg / My weight 461 2kg 6.5 kg

School maths at home

In the two kinds of activities which we have described above, the mathematics is embedded in play and games or authentic household activities. Both these kinds of activities share certain characteristics they are based around purposes and intentions which are independent of the mathematics involved. Indeed, they are not primarily about learning mathematics at all. In contrast, the third category of home mathematics activity is rather different. Here the activity is essentially focused on learning mathematics, and its main purpose is to develop new mathematical knowledge and skills or to practice and rehearse existing knowledge and skills. In this respect, these activities closely resemble mathematics activities undertaken in school, and we will refer to them here as school maths at home. In terms of the distinction made by Street et al . Educational & Child Psychology Vol 24 No 2

Linking childrens home and school mathematics (2005) they are activities belonging to the school domain but taking place on the home site. School maths took various forms at home. Perhaps the most common was homework which had actually been set by the childs teacher. In addition, some of the children attended maths classes given by private tutors, and had to carry out homework given by their tutor. Other children were set maths problems or calculations by their parents or siblings. Sometimes these were set in the context of a game in which the children would play schools, and this would involve one child in the role of teacher setting some maths problems for the other one in the role of pupil. For example, the video made by Nadias family shows her working through a sheet of maths problems drawn up by her older sister. In addition, some of the parents had bought commercially available maths schemes or software packages, and the children worked through these at home. While the distinctions between these different types of home maths activities are easy to make in theory, in practice the boundaries between them can become blurred. The following conversation took place while Molly was playing a game of darts with her younger brother Stephen. However their mothers interventions gave the game a strong schoolmaths feel. Mother: Stephen, whos going first? Stephen? (Stephen has picked up the darts and starts to throw) Mother: OK then (Stephen throws 15, then treble 3, then hits the board outside the ring) Mother: Right, Molly, how much did he get? (Molly looks at the board, says nothing) Mother: Right well count them up. One is fifteen Molly: Fifteen Mother: and treble three, whats three threes? Molly: Six Mother: No, treble three, three threes? Molly: Nine Educational & Child Psychology Vol 24 No 2 Yes, nine plus fifteen (pauses while she works it out) Twenty-four Mother: Yes. Right put Stephens score down as twenty-four then The conversation continues in this vein throughout the game. Mother: Molly:

Difficulties in linking home and school mathematics

As we have seen, children are engaged in a range of different activities at home which involve mathematics in some way. It is clearly important that ways should be found of linking this home mathematics to the mathematics taking place in school indeed, we would argue that childrens potential to learn mathematics will be substantially limited unless this is done. Yet there seem to be a number of reasons why such links are not made as frequently as they might be. First, there is a general lack of awareness about the existence and nature of home mathematics. Teachers frequently assume that there are few mathematical activities taking place at home, and that those activities which are going on are of little value for school learning or indeed, may actually serve to undermine it. Even if teachers are convinced of the value of home mathematics and the desirability of making links with it, they may not necessarily know how to go about it. Few teachers are trained in techniques for eliciting information from parents or other family members, and attempting to do so with a class of 30 or more children from contrasting backgrounds may seem a daunting task. It is also possible that parents and children may not wish to disclose information about family life, for whatever reason. Nadia, for example, was growing up in a family of Bengali origin who still maintained strong links with their extended family in Bangladesh. Like many families of South Asian origin, Nadias family use a system of finger counting in which the three sections on each finger are counted rather than just a single finger. This system allows counting up to 30 on the two hands compared with up to 10 using the Western 141

Martin Hughes et al. system. Nadia was good at mathematics and was strongly supported in this by her family. However, she did not use her finger counting method at school, and did not want her teacher to know about it. In this case, the barrier between home and school was being erected by the child herself (Incidentally, Abreu, 1999, reports that children often do not report on or value their home mathematics practices, which she discusses in terms of valorisation). Just as teachers may not know very much about home mathematics, so parents and other family members often know very little or think they know very little about what is happening in school. This has been particularly evident in the last few years in English primary schools, with the introduction of the National Numeracy and Literacy Strategies. Some parents may have noticed that work which their children do in school or bring home as homework employs different terminologies and procedures from ones they may be familiar with. Other parents have simply have a vague awareness that things are different now. If parents feel that there is a difference between their approach to mathematics and those used by the school, then they may feel inhibited from providing help when this is needed, possibly on the grounds that this might cause confusion for their children. Alternatively, they may go ahead and offer help anyway, even if the outcome is unsatisfactory. As Lucys mother pointed out:
What confuses me is that they do their calculations slightly different to how we were taught to do them . . . I try and show her my way and she says oh you dont know what youre doing (laughs) I can read it out to him but he always says Im wrong because Im not doing it properly . . . so . . .and we end up at loggerheads and I just think. . .I think well you need to just take it back to your teacher and say you cant do it, oh she says, Ive showed him and Ive showed him and Ive showed him, but he just doesnt seem to take it in.

On the video which Ryans mother made for the project there is a long section in which Ryan is attempting a homework sheet of subtraction calculations. Much of this section shows Ryans mother attempting to help him and Ryan resisting her help. At one point the following interchange takes place: Mother: To take em to be able to take five away frae three you have to put one unit off the four and put it onto the three, do you not? Ryan: No Mother: Well why You have to Ryan: You dont. Not in my school we dont. We do it a different way. Many parents were clearly feeling deskilled by the changes in mathematics teaching and their lack of knowledge of current methods being used at school. This feeling was often compounded with their lack of confidence in their own mathematical ability, as a result of their own experiences at school and afterwards.

Exchanging knowledge between home and school

In the previous section we argued that both teachers and parents may lack sufficient knowledge about the mathematics taking place at home and school respectively, and that this may severely limit the opportunities for making links between the mathematics taking place in the two contexts. In this section we describe what we have termed home-school knowledge exchange activities, which are designed to increase the exchange of knowledge about mathematics between home and school. The activities developed and implemented on the Home School Knowledge Exchange Project fell into two main groups. First, there were activities which we called Educational & Child Psychology Vol 24 No 2

While another mother who had been educated in India commented that I wish I went to school here but I didnt . . . A further example comes from Ryans mother, who grew up in Scotland and was taught to use methods which were different from those which Ryan was currently being taught. Ryans mother described how her attempts to help him often ended in conflict and lack of communication: 142

Linking childrens home and school mathematics school-to-home, where the primary aim was to inform parents and/or other family members about the school mathematics curriculum and the teaching methods being used. Examples of such school-to-home activities included regular newsletters, home-school folders which travelled backwards and forwards between home and school, allowing space for both teachers and parents to make comments, and videos made about school maths lessons which were made available for parents to see. At one school parents were invited into the classroom in small groups to be shown the procedures being used by the teachers. One mother commented as follows:
Theyre obviously teaching maths a lot different to the way I learnt maths, which was the main problem, communication. Her form teacher, they had an afternoon where some of the mums went in and they actually taught us for an hour how they teach children. And it helped so much, we got all these sheets and we came home and once I had it in my head, this is how shes got to do it . . . I mean, the answer came out the same, whether I did it my way or her way, but it was nice to know how theyre being taught, how they break it all down.

outside-of-school. The children returned the cameras to the school where the photographs were developed and put on displays, together with pieces of writing produced by the children which explained what mathematics was involved in the activity. The everyday maths activities fell mostly into the categories of play and games and authentic household activities described above, although there were some more idiosyncratic ones such as working out how long their grandparents had been married! While some children did this activity on their own, others involved their parents or other family members. In the following quote Adams mother talks about how they worked on the activity together:
He enjoyed doing it. And it made him think about what we actually do in the house that involves numbers. Like the clock and telling the time, and going to the shops, and change, and money. Because thats another thing, him and (his sister) were changing their English money into euros so that was more maths trying to work that out. And also trying to work out when we were on holiday, if something was 8 euros how much that was in English money . . . when he went to the shops Id say we can take a picture of this and we laid it all out and the change and the money, and then he took the picture.

This mother added that this had been probably the best hour Ive spent at the school, actually. The second main type of knowledge exchange activity were those we termed hometo-school. Here the overall aim was to bring knowledge about the childrens out-of-school lives into the classroom so that connections could be made with the childrens school mathematics. These activities included pupil profile sheets, which provided information about the childrens out-of-school interests and aptitudes; bringing maths games from home into school and playing them during lessons; and developing home maths trails, where the children sought out and tabulated information about their home lives, such as the average age of the family members. In one particularly successful activity all the children were supplied with disposable cameras and asked to take photographs of everyday maths events taking place Educational & Child Psychology Vol 24 No 2

Further details of the knowledge exchange activities and their impact on children, teachers and parents can be found in Winter et al. (2007).

Discussion and implications

In this paper we have argued that current approaches to teaching mathematics in English primary schools do not pay adequate attention to the kinds of mathematics in which children are engaging outside of school. We have attempted to redress this imbalance by providing an account of some of the main features of home mathematics, based on data recently collected with the help of children and their families. Our account makes clear that mathematics can be found in a wide range of home activities, but that it usually looks rather different from school mathematics. For example, the kind 143

Martin Hughes et al. of mathematics involved in play and games and authentic household activities is characterised by being heavily embedded (to use Donaldsons, 1978 term) in the immediate intentions and purposes of everyday life. Indeed, these purposes are usually not primarily about learning mathematics, but are about winning the game, or cooking dinner, or ensuring that the shopkeeper gives you the right change. In this respect they show a marked contrast to the mathematics activities found in school (and the schoollike activities which we also found at home), where the primary purpose of the activities is the learning of mathematics. We have also argued in this paper that it is important that children are able to make meaningful connections between the two kinds of mathematics they encounter. These connections might take different forms. For example, teachers might treat childrens outof-school mathematical knowledge as a resource which they could draw on in the classroom to help with the introduction of a new concept or procedure. Alternatively, they might treat the contexts in which children use mathematics as sites for the application of the more abstract (or disembedded) knowledge acquired in school (see Hughes et al., 2000, for further development of these ideas). There are also potential roles for parents as mediators or brokers (see Wenger, 1998) between school mathematics and childrens individual out-of-school understandings, on the grounds that parents are exceptionally well placed for playing this role. One factor which seems to be working against such connections being made is the lack of knowledge by teachers and parents of what is happening at home and school respectively. In this paper we have described a number of knowledge exchange activities which have been successfully used in schools on the Home School Knowledge Exchange Project. What is more important than the specific activities, however, is the recognition that such activities are desirable and can enhance childrens mathematics learning. The exact nature of the activities is less 144 important, and they can readily be modified to suit local circumstances. One issue which arose in the paper was that of different calculation methods being used by parents and children, and the conflict and frustration or withdrawal of support this may lead to. While it is important that parents are made aware of the methods being taught at their childrens schools, it is also important that they come to recognise that there is no single correct method. Indeed, one of the strengths of the 2006 Framework is that it emphasises that children should leave primary school having a range of methods both written and mental for carrying out calculations, and that some may be more appropriate for some kinds of problems than others. If this could be successfully communicated to parents, then many of the problems we noted above might be eased. At the same time, a home-to-school activity in which children brought in examples of different methods used by different family members, which were then openly discussed and compared within the classroom, could well help the children to appreciate that, in mathematics, there are many ways to accomplish the same purpose. The observations reported here are also relevant for all those including teachers and educational psychologists who need to make assessments of childrens ongoing mathematical progress. The present paper suggests that traditional school-based assessment procedures are likely to engage with only one part of childrens mathematical experiences. If we want to obtain a fuller picture of childrens competence in mathematics then it might be valuable to widen the range of contexts, tasks and materials used to assess children, and to make these closer to the kinds of meaningful and authentic mathematical activities in which they engage at home. This in turn further emphasises and reinforces the need to draw on the knowledge of those who know the children in environments other than school primarily but not exclusively parents. Educational & Child Psychology Vol 24 No 2

Linking childrens home and school mathematics To conclude, one of the main implications of this work is the need for all those concerned with childrens mathematics learning in school whether they be policy-makers, teachers or educational psychologists to be aware of and take account of the mathematics which children are engaged in outside of school, and to look for ways of making meaningful links between in-school and out-ofschool mathematics. The activities described above suggest some ways in which this might be done, but they represent only the start of what needs to be a major shift in perspective about the nature of mathematics learning and how it can be enhanced. 1078) as part of the Teaching and Learning Research Programme (TLRP). More information about the HSKE project and TLRP can be found at http://www.home-school-learning. org.uk and http://www.tlrp.org. We are very grateful to the children, parents and teachers who participated in the project and to the LEAs of Cardiff and Bristol for their support. The HSKE project team consisted of: Martin Hughes (project director), Andrew Pollard (who is also director of TLRP), Jane Andrews, Anthony Feiler, Pamela Greenhough, David Johnson, Elizabeth McNess, Marilyn Osborn, Mary Scanlan, Leida Salway, Vicki Stinchcombe, Jan Winter and Wan Ching Yee.

Acknowledgements
This paper draws on the work of the Home School Knowledge Exchange Project (HSKE), which was funded by the Economic and Social Research Council (reference number L139 25

Address for correspondence

Graduate School of Education, University of Bristol, 35 Berkeley Square, Bristol, BS8 1 JA E-mail: martin.hughes@bristol.ac.uk

References
Abreu, G. (1999). Learning mathematics in and outside school: Two views on situated learning. In J. Bliss, R. Saljo & P. Light (Eds.), Learning Sites: Social and Technological Resources for Learning. Oxford: Pergamon, 1731. Brown, J.S., Collins, A. & Duguid, P. (1989). Situated cognition and the culture of learning. Educational Researcher, 18(1), 3242. Carraher, T. N., Carraher, D.W. & Schliemann, A.D. (1985). Mathematics in the streets and in schools, British Journal of Developmental Psychology, 3, 2129. Department for Education & Employment (1999). The National Numeracy Strategy: Framework for teaching mathematics from Reception to Year 6 London: DfEE. Department for Education and Skills (2006). Primary framework for literacy and mathematics London: DfES. Donaldson, M. (1978). Childrens minds London: Fontana. Hoyles, C., Noss, R. & Pozzi, S. (2001). Proportional reasoning in nursing practice. Journal for Research in Mathematics Education 32(1), 427. Hughes, M., Desforges, C. & Mitchell, C. (2000). Numeracy and beyond. Buckingham: Open University Press. Lave, J. (1988). Cognition in practice: Mind, mathematics and culture in everyday life. Cambridge: Cambridge University Press. Nunes, T., Schliemann, A. & Carraher, D. (1993). Street mathematics and school mathematics. Cambridge: Cambridge University Press. Street, B., Baker, D. & Tomlin, A. (2005). Navigating numeracies: Home/school numeracy practices Dordrecht: Springer. Wenger, E. (1998). Communities of practice: Learning, meaning and identity. Cambridge: Cambridge University Press. Winter, J., Andrews, J., Greenhough, P., Hughes, M., Salway, L. & Yee, W. C. (in press). Improving primary mathematics: Linking home and school. London: Routledge. Winter, J., Salway, L., Yee, W. & Hughes, M. (2004). Linking home and school mathematics: The Home School Knowledge Exchange Project. Research In Mathematics Education, 6, 5975.

Educational & Child Psychology Vol 24 No 2

145

Supporting children with gaps in their mathematical understanding: The impact of the National Numeracy Strategy (NNS) on children who find mathematics difficult
Jean Gross
Abstract
This paper outlines and evaluates developments within the NNS that have addressed the needs of lower attaining pupils. Data is presented to show that these pupils have not benefited from the NNS as much as other pupils. The rationale and form of the additional Wave 3 intervention materials are then described. Evaluation of impact has so far been qualitative, involving comments made by teachers and the children themselves. Quantitative information is still awaited and will include statistics of on the number of pupils continuing to attain below National Curriculum Level 3 at the end of Key Stage 2 (at the age of 11 years). Implications for the term dyscalculia are discussed. It is concluded that there is an opportunity, in the relatively new field of mathematical difficulties, to develop from the start effective teaching strategies that work for all rather than distinct routes based on diagnostic categories.

N THE context of this special journal issue, where other papers consider research and interventions addressing childrens difficulties with mathematics or arithmetic (see for example Dowker; Gervasoni & Sullivan; Willey, Holliday & Martland, this issue), it is important also to outline and evaluate developments that have taken place within the English educational system. This paper sets out to do that. It starts with the rationale of the National Numeracy Strategy (NNS) and provides information about the impact of the strategy on lower attaining pupils. It then considers responses by the NNS to concerns about children with significant difficulties in mathematics and the rationale, development and evaluation of additional teaching materials described as Wave 3 intervention. Finally, the paper takes a pragmatic stance in relation to the term dyscalculia and suggests how this stance might inform educational psychology practice. 146

From its inception in 1999, the NNS has had the explicit aim of narrowing the gap between higher and lower attaining children, identified in international studies as a particular feature of the English system (Reynolds & Farrell, 1996). Resources have deliberately been targeted according to levels of disadvantage, and teachers have been encouraged to develop whole-class interactive teaching followed by carefully differentiated group work as an alternative to the lonely journey through graded textbooks that formed childrens main mathematical experience before the introduction of the NNS. The potential benefits for children who find mathematics difficult have included the opportunity for them to learn from others rather than being thrown back on their own sometimes limited resources, the opportunity for teachers to pinpoint appropriate learning objectives from a clearly defined progression, the focus on oral work, the use of visual models to develop mental imagery, explicit teaching of key mathematical vocabEducational & Child Psychology Vol 24 No 2 The British Psychological Society 2007

Supporting children with gaps in their mathematical understanding ulary, and the emphasis on metacognition children identifying and explaining the strategies they use to solve problems. TIMSS (Trends in International Mathematics and Science Study) found that the progress made in mathematics since 1995 was larger in England than in any other country and that these improvements applied equally across the range of attainment, including the most and least able. However, findings from the Leverhulme numeracy research programme (Brown & Millett, 2003), suggested that the range of variation in results between lower and higher attainers between 1998 and 2002 had actually increased. Evidence from national datasets shows that, while the numbers of children achieving nationally expected levels at the end of Key Stage 2 have grown significantly (Figure 1), the reduction in numbers achieving very low levels (below National Curriculum Level 3) has been much less dramatic (Table 2). The profile of children attaining below Level 3 in mathematics is of interest. It does suggest some degree of specificity in mathematical difficulties: 5.9 per cent of the 2005 Year 6 cohort were below Level 3 in mathematics only, compared to 6.3 per cent in English only and 3.9 per cent in both subjects. And while 68 per cent of those below Level 3 in English are boys, the corresponding figure for mathematics is only 55 per cent. In mathematics as in English, children experiencing Interpretation in schools Questioning used to funnel pupils responses towards a required answer, rather than promote discussion and cognitive engagement Increased use of setting as a response to difficulty in coping with diversity Lower attaining groups generally work with teaching assitants rather than the teacher Teachers and TAs have little liaison time and TAs may lack subject knowledge. Their role is often to support children in completing inadequately differentiated tasks

To what extent have NNS goals been achieved?

Overt goals, as stated above, often translate imperfectly into classroom practice and there is some evidence (Hardman, 2003; McSherry & Ollerton, 2002; Blatchford et al., 2004), summarised in Table 1, to suggest that the interpretation and implementation of NNS guidance often falls short of the ideal. The gap between aspiration and reality may help to explain a number of findings that raise concerns about the impact of the NNS on lower attainers. For example, several researchers (Landerl et al., 2004; Anderson, 2000) have noted that public questioning during whole-class teaching can lead to raised anxiety levels for children who find mathematics difficult. The Gatsby numeracy project (Muijs, 2003) found that low attaining children in Years 1 and 2 (age range five to seven years), who were supported by a teaching assistant, did not make greater progress than control children who were not supported. Evidence on the impact of NNS on the standards achieved by lower attaining pupils has often been contradictory. The 2003 Stated goals of NNS Whole class interactive strategies rather than didactic teaching Differentiation by ability grouping within a class Teachers work with all ability groups Well-trained Teaching Assistants (TAs) work with groups on appropriately differentiated tasks under the direction of the teacher and with frequent opportunities for liaison Table 1: Interpretations of NNS goals Educational & Child Psychology Vol 24 No 2

147

Jean Gross
90 85 79 85

% pupi l s achievi ng l evel 4+

80 75 75 70 65 60 55 50 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

English
Figure 1: The percentage of children achieving Level 4 a 2008 target

Maths
at the end of Key Stage 2, projected towards

1998 7

1999 6

2000 6

2001 5

2002 5

2003 6.3

2004 6.1

2005 5.8

Table 2: Percentage of pupils achieving below Level 3 in mathematics at the end of Key Stage 2 (age around 11 years). Note: Figures before 2003 rounded to the nearest whole number

social disadvantage are overrepresented in the low attaining group (by a factor of two), as are summer born children (three summer born for every two autumn born children achieving below Level 3).

Response by the NNS to concerns about children with significant difficulties in mathematics
Concern about the largely static numbers of very low attainers in mathematics led the NNS to develop a number of initiatives to raise standards for these children. Professional development materials were produced aimed at improving the quality of inclusive classroom teaching (DfES, 2002; DfES, 2004) and helping teachers work more effectively with teaching assistants (DfES, 2005a). The Strategies also developed a model based on three Waves of intervention, with Wave 1 represent148

ing high quality everyday inclusive classroom teaching, Wave 2 the provision of short periods of assistance in small groups for children who are not making satisfactory progress (about 20 per cent), and Wave 3 additional targeted interventions for the much smaller proportion (approximately 5 per cent) for whom the additional small group teaching is insufficient. The first step in developing a strategy for Wave 3 intervention was to map existing provision in schools. In the spring of 2002 all local authorities (LAs) were asked to provide information on literacy and mathematics interventions used in their area. Fifty per cent responded. All named at least one Wave 3 literacy intervention and 21 different literacy interventions were identified as in fairly widespread use.

Educational & Child Psychology Vol 24 No 2

Supporting children with gaps in their mathematical understanding Only seven of the LAs mentioned any Wave 3 maths interventions. The most widely used programme, mentioned by three respondents, was Mathematics Recovery (Wright et al., 2000), a one to one intensive (daily) teaching system for children in Year 1. The remaining responses described locally devised interventions, including the use of Numicon materials in a structured teaching programme (Horner, 2002). Other programmes known to be in use at the time but not mentioned in survey returns were Family Numeracy (Brooks & Hutchison, 2002), Barking and Dagenhams Group Education Plans (Whitburn, 1997), a Numeracy Recovery scheme for six year olds (Dowker, 2001) and Paired Maths(Topping et al., 2003), in which peers work together on a tutoring programme. The conclusions from the survey were that addressing mathematical difficulties was a low priority in many schools and local authorities. In contrast to literacy, few programmes were available. Those that existed were confined to particular geographical areas. Consequently, the National Strategies took the decision not to develop or publish any new literacy programmes, since a plethora was already available, but rather to guide schools choice of programmes by providing information on their evidence base. In mathematics the position was clearly different. There was a need to stimulate greater activity and there was a particular gap in materials for Key Stage 2 (children aged 7 to 11 years). For Key Stage 1 (age five to seven years) the NNS developed support sessions built into published teaching plans (DfES, 2003). These describe twenty-minute sessions for use with a group of children who are struggling with a key concept in the unit of work, to be taught in addition to the main lesson. To inform the development of new Key Stage 2 materials, the DfES commissioned a review of research to identify what works for children with mathematical difficulties. The review (Dowker, 2004) suggested that: mathematical difficulties are common and often quite specific; Educational & Child Psychology Vol 24 No 2
they are equally common in boys and

girls, in contrast to language and literacy difficulties which are more common in boys; they are varied and heterogeneous, as arithmetical ability is not a single entity, but is made up of many components; they represent one end of a continuum rather than a discrete disorder; their causes are varied and include, for example, individual characteristics, inadequate or inappropriate teaching, absence from school resulting in gaps in mathematics learning, lack of preschool home experience with mathematical activities and language; children with mathematical difficulties typically combine significant strengths with specific weaknesses; some children have particular difficulties with the language of mathematics; difficulty in remembering number facts is a very common component of arithmetical difficulties, often associated with dyslexia; other common areas of difficulty include word problem solving, representation of place value and the ability to solve multistep arithmetic problems. The review concluded that mathematical difficulties can be addressed through appropriate intervention, which can take place successfully at any time and can make an impact . . . it is not the case that a large number of children are simply bad at maths, and that nothing can be done about it (Dowker, 2004, p 42). The review drew out some general principles for effective intervention: It should be individualised, but the amount of time given to such individualised work does not, in many cases, need to be very large to be effective. It should be provided as early as possible, partly because mathematical difficulties can affect performance in other areas of the curriculum, and partly to prevent the development of negative attitudes to and anxiety about mathematics. Interventions that focus on the particular components of mathematics with which 149

Jean Gross the child has difficulty are more likely to be successful than those which follow a set programme. Distinguishing between specific mathematical difficulties and those associated with generally low cognitive ability does not seem to be helpful in planning intervention.
Experiences of success to build childrens

confidence in themselves as learners

A clear link to whole-class activities Involvement of parents and carers in

Developing the materials

The principles set out in Dowkers report were used to develop Supporting children with gaps in their mathematical understanding (DfES, 2005b), a set of materials for children in Key Stage 2. The materials take the form of assessment tools that allow class teachers to identify the particular errors and misconceptions that are limiting a childs ability to progress. A series of booklets each address a particular error or misconception, and describe teaching activities to undertake with the child on a one-to-one basis. The design of the materials draws on key psychological understandings about how all children learn, and on recent investigations of the pedagogies that are appropriate for children with special educational needs the intensification of what is needed by all children, and the common pedagogies delivered under special conditions described by Lewis and Norwich (2005). These include: Distributed practice short, focused periods of one-to-one work Overlearning continuing to practise new learning beyond the stage of apparent mastery Review and practice at progressively increasing intervals, to support recall An emphasis on metacognition helping the child make explicit the strategies they are using so that they can achieve independence in their learning Multisensory learning, with maximum use of structured equipment and everyday materials to model mathematical concepts Linking learning to familiar and relevant contexts Attention to language highlighting and modelling key vocabulary throughout 150

their childrens learning The materials reflect best practice in assessment for learning (Black & Wiliam, 2001) through the use of questions to elicit information about childrens understanding, sharing the purpose of the activity with the learners and encouraging childrens reflection on their learning so that they identify for themselves possible next steps.

Learning from the pilot

Twenty-seven local authorities volunteered to pilot the materials in 20032005. As a result of feedback, improvements were made to the assessment for learning opportunities throughout the materials, further links to whole class work were added and activities in the form of games both for use during teaching sessions and to involve parents and carers incorporated. The resource proved popular with school staff and children. Children in the pilot noted with pleasure the lack of worksheets. Schools reported that the pilot had helped them move on from a sole focus on literacy in their special needs support. Some schools, however, found it difficult to resource and implement the suggested model of detailed assessment and tailored one-to-one intervention. Some simply dovetailed the activities to objectives in the medium term planning for each year group and used the activities with lower sets/less able pupils as and when the Wave 3 activities and objectives coincided (Hurt, 2005). The more effective local authorities responded to issues like these by targeting support from central services. In the London Borough of Bromley, for example, specialist teachers familiarised senior managers with the aims of the intervention, worked alongside class teachers to identify gaps in childrens knowledge, modelled use of the teaching materials for teaching assistants, then gradually handed over Educational & Child Psychology Vol 24 No 2

Supporting children with gaps in their mathematical understanding more and more of the teaching to them. The outcome, as reported anecdotally, was greatly increased confidence amongst teaching assistants (many of whom had been reluctant to be involved in a mathematics intervention) and a sustained and more strategic use of the intervention materials in schools. Local findings like these are helpful, but wait to be replicated nationally. Certainly the qualitative findings from the pilot need to be treated with caution, given the history of adherence to unproven Wave 3 literacy programmes based only on perceptions that children love doing them and grow in confidence. Warm feelings like these need to translate into hard results if we are to take them seriously. There is, then, need for further research into the impact of the NNS materials, but also much to be learned from the impact of a messy but real activity in the field. Controlled experimental studies, robust as they might be, are often hard to replicate when they move beyond the conditions in which they were originally devised. The National Strategies traditional means of evaluation (combining feedback from users with impact on national attainment patterns), whilst less rigorous, at least has the advantage of a firm grounding in an applied context.

Evaluation of the impact of the materials

Evaluation of the Wave 3 intervention at national level has so far been mainly in the form of qualitative comments with quantitative information on numbers of children attaining below Level 3 still awaited. There seems to have been much change at the attitudinal level. Schools have mentioned childrens raised self esteem and more active participation in the daily mathematics lesson. For example: We often hear I can do it by myself whereas before it was more of I cant do it. Anything that gets kids this excited about maths has to be worth putting lots of time and effort into. Typical comments from pupils have been: I dont feel sick any more because I can do this work; It gave me a warm feeling in my tummy because I could do it; I now feel confident to do my SATs next summer. Some local authorities have undertaken their own evaluations, generally initiated by educational psychologists. In Norfolk, for example, educational psychologists tested children in fourteen schools at the beginning and end of a six month intervention period using BEAM diagnostic interviews in number sense (Denvir and Bibby, 2001) along with tests of cognitive ability. The results showed that almost all of the pupils had made greater progress in the time the Wave 3 materials had been used than in previous years. Many had increased their National Curriculum levels by at least one sub-level. Not surprisingly the greatest gains were in schools that had used the materials in a planned, systematic way and had effective support from senior management ( Johnson, 2005). Educational & Child Psychology Vol 24 No 2

Pragmatic stance on the emerging dyscalculia label

Early guidance from the NNS (DfES, 2001) made explicit use of the term dyscalculia, defining it as a condition that affects the ability to acquire arithmetical skills. As much as anything this represented a wish to give childrens mathematical difficulties a prominence in teachers minds similar to that engendered by the dyslexia label. Later NNS publications, however, including the Wave 3 mathematics intervention materials, are more cautious in the use of the term. The reasons are those that will be familiar to educational psychologists in relation to dyslexia: the evidence of a continuum of difficulties, rather than a discrete cut-off point, and the consequent risks of medical models that suggest children might either have, or not have a particular learning difficulty, just as they have or dont have measles. There is also a perceived risk of inadvertently encouraging varying levels of provision for children with different cognitive abilities, 151

Jean Gross based on the inappropriate but pervasive use of a discrepancy model. Finally, note has been taken of the Dowker research overview (op.cit) that failed to give support to the idea of a discrete teaching methodology for any one group of learners with arithmetic difficulties. Another risk is an over-investment of time and energy in diagnosis and labelling in local authority support services. It is salutary to reflect on the response of one special needs support service to the National Strategies Wave 3 materials, which was to use NFERNelson dyscalculia tests to identify children requiring intervention. The tests revealed three children who were dyscalculic, but there were another twenty who could not do maths and whose support needs were, as far as anyone could see, identical. In educational terms, the support services time might better have been spent working with teachers to help them overcome perceived barriers to appropriate, teacher-led use of the Wave 3 intervention. This is an example of how a medicaldiagnostic model based on assumed pathology can obstruct the development of appropriate curricular responses in the classroom. It is not unique. We know that teachers are identifying ever more children with special needs (DfES, 2006) and that the rate of so-called special educational needs (SEN) amongst boys of primary age is reaching epidemic proportions (almost one third of eight year old boys are now identified by their teachers as having special educational needs). It would seem that the concept has become little more than an excuse: a social construction that identifies a large number of children as not teachable unless they are provided with additional resources and support. Yet the provision of such additional resources may not be helpful: the number of teaching assistants has doubled in England since 1998 (Blatchford et al., 2004), most of them working with children who find learning difficult, but the percentage of very low attainers has hardly shifted over the same period. 152 It is important to avoid this pattern repeating as we develop our knowledge about mathematical difficulties. The focus needs to be on holding such children in the teachers sphere of responsibility, rather than placing them outside it. A medical-diagnostic model is unlikely to achieve this. Diagnosis and labelling can have negative effects on childrens attainment as illustrated by Tymms & Merrell (2004). According to their research, information for teachers on strategies to help inattentive and hyperactive children boosted the pupils attainments but screening and identifying specific pupils as having ADHD had the opposite effect. The explanation suggested by the researchers was that, once labelled, teachers focused on keeping the pupils happy and calm rather than encouraging them to achieve. Yet diagnosis and labelling can also have benefits. For some conditions such as autistic spectrum disorders (ASD) there is evidence that early diagnosis is important in reducing family stress and for the later mental health of the children involved, because it leads adults to respond in ways that minimise the potentially disabling effects of autism (Gross, 1994; Papps & Dyson, 2004). Many children and adults would equally attest to the importance of having a label such as dyslexia, because of its power in eliminating the disabling effects of how others construe them most usually as stupid or lazy. A social model of disability requires us to listen carefully to the perspectives of the learner, perhaps more carefully than to sterile debates about whether certain conditions exist or do not exist. For the moment, then, educational psychologists might want to adopt the scientifically less interesting but educationally more useful approach of taking dyscalculia by its literal meaning (an inability to calculate). They can then start from the assumption that all children who struggle with numbers and the number system are to some extent dyscalculic and proceed, in their work with teachers, to the much more important question So what are we going to do to about it in the classroom? Educational & Child Psychology Vol 24 No 2

Supporting children with gaps in their mathematical understanding

Conclusion
In their analysis of pedagogies for low attaining children, Dyson & Hick (2005) concluded that successful literacy interventions have a basis in a universally applicable model of reading development that leads them to play down an aetiological approach to understanding childrens difficulties in favour of a functional one. Put simply, whatever the underlying causes of childrens falling behind, the reading task remains the same (p.195). The parallels between mathematics and literacy are evident, and the history of dyslexia research and debate illuminating. We have the opportunity, in the relatively new field of mathematical difficulties, to develop from the start effective teaching strategies that work for all rather than distinct routes based on diagnostic categories. Implications for educational psychology practice are clear. In their consultations with individual teachers, they need to encourage a teaching and learning response rather than one based on the probing of cognitive causation. The diagnosis should be about what the children know and what they need to learn: teachers still need support with this, and educational psychologists can help by modelling the use of assessment tools such as

those provided in the NNS materials. The skills of educational psychologists, moreover, can be applied at systems level, helping the school develop effective strategic management of additional provision (Gross & White, 2003). Effective provision for children with mathematical difficulties involves a coherent whole-school approach. It requires systematic, targeted and time limited support informed by data and evidence on what works. There is a need for good systems for tracking and regular review of pupil progress, close connections between the intervention and the work of the class as a whole, the positive engagement of parents and carers, and rigorous evaluation. Educational psychologists have traditionally had the skills to help schools develop such systems. It is to be hoped that their background and training in the future will continue to provide them with the expertise they need to provide this type of consultative support.

Address for Correspondence

Jean Gross, Every Child a Reader, KPMG Foundation, Salisbury Square, Londn EC4Y 8BB. E-mail: jean.gross@kpmg.co.uk.
DfES (2001). Guidance to support pupils with dyslexia and dyscalculia. London: DfES. DfES (2002). Including all children in the literacy hour and daily mathematics lesson. London: DfES. DfES (2003). Models and images. London: DfES. DfES (2004). Learning and teaching for children with SEN in the primary years. London: DfES. DfES (2005a). The effective management of teaching assistants to raise standards in literacy and mathematics. London: DfES. DfES (2005b). Supporting children with gaps in their mathematical understanding. London: DfES. DfES (2006). Special educational needs in England, January 2006. London: DfES. Dowker, A. (2001). Interventions in numeracy: the development of a numeracy recovery project for young children with arithmetical difficulties. Support for Learning, 16, 610. Dowker, A. (2004). What works for children with mathematical difficulties. London: DfES.

References
Anderson, J. (2000). Teacher questioning and pupil anxiety in the primary classroom. Paper presented to the British Educational Research Association conference, Research Student Symposium, Cardiff University, September 2000. Black, P. & Wiliam, D. (2001). Inside the Black Box: raising standards through classroom assessment. London: Kings College London School of Education. Blatchford, P., Russell, A., Bassett, P., Brown, P. & Martin, C. (2004). The role and effects of teaching assistants in English primary schools (Years 4 to 6). London: DfES. Brooks, G. & Hutchison, D. (2002). Family numeracy adds on. London: Basic Skills Agency. Brown, M. & Millett, A. (2003). Has the National Numeracy Strategy raised standards? In I. Thompson (Ed.), Enhancing primary mathematics teaching. Buckingham: Open University Press. Denvir, H. & Bibby, T. (2001). Diagnostic interviews in number sense. London: BEAM Education.

Educational & Child Psychology Vol 24 No 2

153

Jean Gross
Dyson, A. & Hick, P. (2005). Low attainment,. In A. Lewis & B. Norwich (Eds.), Special teaching for special children? Buckingham: Open University Press. Gross, J. (1994). Asperger syndrome: A label worth having. Educational Psychology in Practice, 19(2), 10410. Gross, J. & White, A. (2003). Special needs and school improvement. London: David Fulton. Hardman, F. (2003). Interactive whole class teaching in the national literacy and numeracy strategies. Paper presented at the Learning Conference 2003: What Learning Means. University of London, Institute of Education. Horner, V. (2002). Counting on it. Special, Autumn 2002. Hurt, L. (2005). Personal communication. Johnson, J. (2005). Norfolk Wave 3. Personal communication. Landerl, K., Bevan, A. & Butterworth, B. (2004). Developmental dyscalculia and basic numerical capacities: A study of 89 year old students. Cognition, 93, 99125. Lewis, A. & Norwich, B. (Eds.) (2005). Special teaching for special children? Buckingham: Open University Press. McSherry, K. & Ollerton, M. (2002). Grouping patterns in primary schools. Mathematics in school, 31, 27. Muijs, D. (2003). The effectiveness of the use of learning support assistants in improving the mathematics achievement of low achieving pupils in primary school. Educational Research, 45(3), 219230. Papps, I. & Dyson, A. (2004). The costs and benefits of earlier identification and effective intervention. London: DfES. Reynolds, D. & Farrell, S. (1996). Worlds apart: A review of the international surveys of educational achievement involving England. London: HMSO. Topping, K., Campbell, J., Douglas, W. & Smith, A. (2003). Cross-age peer tutoring in mathematics with seven and 11 yearolds. Educational Research, 45(3), 231240. Tymms, P. & Merrell, C. (2004). Screening and interventions for inattentive, hyperactive and impulsive children. University of Durham: http://pips. cem.dur.uk/PDFs/ESRCReport.pdf. Whitburn, J. (1997). Improving mathematics attainment: lessons from abroad? Paper presented to Scottish Educational Research Association annual conference, September 18th20th, University of Dundee. Wright, R., Martland, J. & Stafford, A. (2000). Early numeracy: Assessment for teaching and intervention. London: Chapman.

154

Educational & Child Psychology Vol 24 No 2

Supporting children with gaps in their mathematical understanding

Appendix
Supporting children with gaps in their mathematical understanding focuses on addition/subtraction and multiplication/division objectives in the English National Numeracy Strategy mathematics framework. The materials consist of a series of teaching activities, each addressing one of the following common errors and misconceptions. Difficulty in partitioning numbers with zero place holders and/or numbers less than one Difficulty in choosing suitable methods for calculations that cross boundaries.

Multiplication and division

Confusing numbers when counting in twos Difficulty with identifying doubles and adding a small number to itself Making unequal groups when grouping, and being unable to compare the groups Difficulty when sharing in dealing with any left over after making equal groups Difficulty in counting reliably in tens from a multiple of ten Not understand vocabulary such as groups of, multiplied by Failing to link counting up in equal steps to the operation of multiplication Seeing an array as a collection of ones rather than focusing on rows of or columns of Failure to use partitioning when doubling Difficulty in relating multiplying by two to known facts about doubles Failure to use knowledge of doubles to find half of a number Not being systematic when sharing into equal groups or using the language of division to describe the process Not understanding that sets of or groups of need to be subtracted to solve a problem Lack of confident recall of multiplication facts Not understanding the relationship between multiplication and division facts Not understanding the operation of multiplying by ten Failure to apply partitioning and recombining when multiplying Assuming that the commutative law holds for division Writing a remainder that is larger than the divisor 155

Addition and subtraction

Difficulty in counting can only begin counting at one, lacks systematic approaches Misunderstanding the meaning of one more and one less, and not being able to identify the number before or after a given number Failing to relate the combining of groups of objects to addition Not being confident about when to stop counting when subtracting in answer to the question how many are left? Making counting mistakes when using teen numbers and/or crossing boundaries Difficulty in remembering number pairs totalling between 10 and 20 Counting up unreliably still counting the smaller number to get one too many in the answer Failing to relate finding a difference and complementary addition to the operation of subtraction Not making links between addition and subtraction and/or recognising inverses Not readily using number patterns to support calculating Insecure understanding of the structure of the number system Difficulty in partitioning Difficulty in deciding when to use calculations laid out in columns Difficulty in adding three numbers in a column Inefficient counting strategies for large numbers Rounding inaccurately, particularly with decimals, and having little sense of the size of the numbers involved Educational & Child Psychology Vol 24 No 2

Jean Gross Not understanding the significance of a remainder Not understanding rules about multiplying and dividing by powers of ten and the associative law Difficulty in interpreting a remainder as a fraction Interpreting division as sharing but not as grouping (repeated subtraction) Difficulty in making reasonable estimates.

156

Educational & Child Psychology Vol 24 No 2