ABSTRACT
Research studies of residential institutions for children in several countries are scrutinised to see if the size of establishment produces any distinct effects. All except one of the studies examined failed to find any effects due to size per se but there was evidence to support the benefits of smaller living units and teaching groups. However, the extent of any effects depends on the context in which groups operate. A model that considers size in the context of institutional aims and structures is proposed as a more fruitful approach to understanding the significance of size in service development.

Introduction

The aim of this paper is to assess the significance of institutional size for the operation of residential establishments for children and for child outcomes.

An initial scrutiny of the evidence suggests that size has some effects. For example, with regard to organisational structure, large establishments are likely to be more bureaucratic and decentralised than small ones. Size also seems related to the functions of establishments, so therapeutic communities tend to be relatively small and military academies large. But apart from these somewhat obvious relationships, comparative research studies of residential homes and schools have failed to find much correlation between the size of residential establishments, placement outputs and outcomes for children.

Even if such a relationship existed, it would be difficult to prove because there have been so few controlled studies in residential care and education. Practical constraints make it impossible to establish large and small establishments and then to allocate children randomly to them. The controlled trials that have been undertaken have mostly focused on the effects of contrasting regimes (Clarke and Cornish, 1975). The only example we have been able to find where size was the focus of concern is the 1960s Brooklands experiment in England. Here, a group of mentally retarded children were moved from a hospital ward with 40 beds and a high staff/child ratio (1:3) to small hostels accommodating 16 children cared for by seven staff. Beneficial effects were found in that the children’s language improved, they became less maladjusted and were able to play and relate better to other residents and adults (Tizard, 1964). However, because there was no observational evidence on interaction, it was difficult to attribute these effects exclusively to changes in institutional size. In fact, Tizard suspected that changes in management were equally important.

The research in which the effects of size have been explored comprises comparative studies of similar institutions of varied sizes. In these samples, it is possible to correlate outcomes with size to see if any patterns emerge. Studies of this kind have been conducted in England (Lambert et al., 1975; Millham et al., 1975; Kushlick 1972, 1974 and Tizard, B., 1975; Tizard, J., 1975), Germany (Planungsgruppe Petra, 1988) and Israel (Kashti, 1998). Between them, they have scrutinised a range of institutions, such as residential education (range 100-1200 places), training schools for young offenders (range 20-100 places) and homes for children in need of personalised care (range 5-15 places).

In all of these studies, no correlation was found between size and other variables, apart from the obvious relationships about organisational structures such as those cited in the opening of this paper. Lambert and colleagues (1975), for example, looked at boarding schools and found that in small private schools with strong ‘expressive’ aims, head teachers’ roles were different from those in other schools. They had much more responsibility for the pastoral care of pupils and staff leadership. This had consequences in that it was easier for head teachers to introduce radical changes as there were fewer established staff status positions, a less federal structure and greater consensus among staff about aims and methods. In fact, resistance to change was more likely to come from pupil cultures which in schools adopting a progressive, child centred ethos could be cohesive and powerful. In terms of resources, small schools also differed markedly from others. Staff-pupil ratios were often lower, buildings and facilities of poorer quality and the curriculum narrower and less imaginative.

In terms of outcomes, however, the results are even more barren, at least in terms of conventionally expected outcomes. There seem to be no distinct effects of size. The one exception is the survey by Sinclair and Gibbs (1998) of 48 residential homes (size 4-20 places) for children in the care of local authorities in England. Small homes were likely to do well on three criteria: the quality of the social environment of the home; levels of individual misery and children’s social adjustment. This effect was reinforced if the heads of the home felt that their roles were clear, mutually compatible, not disturbed by re-organisation and if they had autonomy. It also helped if staff agreed on how the home should be run.

The conclusion from all this evidence, with the exception of Sinclair and Gibbs’s research, is that all of these studies could find little influence of size on any aspect of children’s behaviour or development once other factors were taken into account. There was no obvious correlation between institutions ranked in order of size and of features of regime, young people’s behaviour or their social and academic achievements. There were small establishments where relationships between children and staff were rigid and impersonal and large establishments that effused pastoral care and individual attention.

Findings from research in schools
Findings from day schools echo these conclusions with regard to institutional size. Generally, there is no evidence to suggest that large mainstream schools are any more successful than smaller ones. Rutter and colleagues (1979) undertook a comparative study of 12 London secondary schools (age range 11-18) educating from 450 to 2,000 pupils. They found that the size of schools ‘proved to have no significant association with any of our outcome measures, which were attendance, behaviour, academic progress and delinquency… It was thought that this lack of association might be misleading if large schools were based on split or broken down into smaller sites. But, again, no significant relationship was found…It may well be that the size of a school does have an impact on its character and style but, at least within our sample, small schools were neither more nor less likely to have favourable outcomes, however they were measured’ (p.100).

The significance of size for sub-units
If the size of institutions produces no distinct effects, does the size of sub-units within them matter? The way the institution is divided up rather than the overall size may be the factor most affecting the daily lives of residents and staff. In 1830s England, Dr. Arnold of Rugby School sought to improve conditions in his large private boarding school by dividing it up into living houses of 60 children, a policy inspired by the twelve tribes of Israel. Similarly, residential provision for children with special needs is nowadays usually structured around much smaller living groups of about eight children.

There is evidence from organisational studies about the strengths and weaknesses of various sizes (Moos, 1968, 1974; Moos and Houts, 1968; Kushlick, 1972, 1974). This shows that although big institutions are no more successful than small ones in terms of child development outcomes, large size does produce its own problems of structure and communication. Research studies have noted the potential for difficulties of resistance among staff to change, an undue concern with questions of organisation and administration, potentially unhelpful staff and child sub-cultures and, as a corollary, a tendency to neglect emerging evidence about changing needs of children (Brown et al., 1998).

The literature also notes the potential strengths of large size placements. For example, effective large-scale residential settings capitalise on their size by showing flexibility in helping children who do not settle in one part of the establishment, providing a variety of services within the placement to achieve optimal child outcomes, allocating resources to an improved understanding of residents’ needs and displaying a preparedness to experiment with new methods to meet them.

The message from organisation theory is that size does not become important until the organisational demands of the institution displace its primary functions, such as when welfare or education aims are swamped by process, logistics or administration.

Findings about the size of sub-units within institutions
There are some rules of thumb when thinking about size in relation to the internal structure of residential placements but not sufficient knowledge to be categorical. There has been an understandable desire on the part of policy makers for numbers that can be used when planning residential provision and some indicators have emerged in response to this demand.

Tizard and Kushlick’s work in the 1960s, partly undertaken in residential placements for mentally handicapped children, showed that living groups operate most efficiently when they consist of six to eight people. For example, two or three people are not as creative as seven and a unit of 20 people requires organisational structures that detract from a focus on primary objectives. This work has been used to rationalise the size of units for children with special needs and by management theorists seeking to optimise group dynamics, such as McCullough and Ely (1975) who recommended the mystical number of seven. However, Barbara Tizard’s (1975) work on residential nurseries urged caution before accepting a set formula. She found that residential nurseries with two staff (a nurse and an assistant) to eight children produced less stimulation and staff-child interaction than those with one nurse to six children. This was because of the different staff dynamics in the two models and the disabling effects of pairings on staff feelings of empowerment and autonomy.

The largest body of research on unit size is provided by studies of class size in day schools. There is an extensive international literature on this but the results are by no means conclusive. Bennett (1998), in a major research review of British and North American studies, is sceptical about dramatic benefits of small classes although
he is more optimistic if small size is accompanied by other conditions. He writes, ‘little or nothing can be made of the findings of the British studies wither for policy formation or guidance on classroom processes. The American research base is stronger but lacks classroom process data and fails to present an unambiguous picture’. There are tentative suggestions for policy and practice but nothing concrete. Examples are found in the work of Howley (1994, 2001) who found that socially disadvantaged students do better in small secondary schools and by Friedkin and Necochea (1988) who found that students from higher social classes do better in large schools although the improvement was less marked than for the disadvantaged groups attending small schools.

Bennett’s review is important in providing a context in which to interpret research findings in this field, particularly as most of the results are correlations between factors. Even if there are reasonable grounds for believing that the introduction of class sizes of 20 or less will result in higher achievement in the first years of schooling, particularly among minority children, this is only likely if cognisance is taken of the conditions necessary for effective implementation. Important in this context is training for teachers to enable them to maximise the benefits of reduced pupil numbers

To answer the question conclusively, an experimental design in which changes are introduced under controlled conditions would be required and several examples are available (Shapson, 1980; Finn and Achilles, 1990; Prais, 1996). This helps overcome the difficulties of controlling for differences in children’s background.

Another method of determining the effects of class size is to re-analyse a national data cohort of children. A recent study by Iacovou (2001) does this using the British National Child Development Study cohort of children born in Britain between March 3rd and 9th 1958. She concluded that once all other influential variables had been eliminated from the analysis, reducing class size for infants (ages 5-8) from 30 to 22 produced better reading test scores, a 10 per cent improvement in the reading ages of the children. There were, however, problems in interpreting the evidence. For example, the same teaching methods may be used whatever the size of class and factors such as the external scrutiny of teachers or the inclusion of children with special needs in the teaching group may affect the results. There would also be difficulties of implementing such a policy. Reducing all classes to a maximum of 20 would require massive recruitment of new teachers and retraining of existing ones as opportunity on its own may not be enough to change practice. This is confirmed by the poor quality and unimaginative curricula in reform schools where class sizes were small (Millham et al., 1975). All of these factors could reduce the quality of teaching and negate any gains produced by the introduction of smaller classes. Nevertheless, teachers and politicians clearly feel that class size is important which may in itself produce a ‘Hawthorne’ effect.

Bennett’s conclusion is that ‘class size is but one contextual factor, alongside other factors such as location and intake, curriculum policy, school organisation etc. which interact with teacher and pupil characteristics to mediate classroom processes and, through them, educational outcomes’. This indicates the ‘complexity of adequately researching the issue of class size and shows the impossibility of divorcing the impact of size alone from the myriad of interacting variables that constitute the typical classroom’.

If size by itself does not seem to be important for outcomes, why do the sizes of different establishments vary and why has there been a reduction across the board in recent years?

Factors that affect institutional size
Historians argue that the main determinants of institutional size have been economic and ideological factors and the functions of residential establishments rather than considerations of optimal child development (Parker, 1988). They stress that although average sizes have mostly fallen in recent years, the rank order according to the average size of sectors, such as prisons, boarding schools, hostels etc., has remained consistent.

Economic factors have driven in both directions. On the one hand, public or semi-public bodies with access to capital loads and a responsibility for general populations tended to see large buildings, often constructed on inexpensive land and away from urban centres, as the best use of scarce resources. By contrast, many voluntary initiatives provided smaller placements that reflected restricted access to funds and a focus on specialist groups often neglected by the public sector. However, these small enterprises were always financially unstable and many were taken over by larger charities whose resident groups were much larger. It was only in the middle of the twentieth century when concern began to be expressed about the quality of care, the different needs of boys and girls, the plight of infants, the poor conditions of buildings and the lack of facilities that size became an issue. But, again, this was a passing concern in some countries. In England, for example, interest in questions of size declined once residential nurseries had been abolished in the 1970s.

Ideological factors are important, too. In the nineteenth century, although large institutional size was often justified in terms of supporting a chapel or band, an important aim was proselytisation. Much policy reflected arguments between religious sects, a wish to make public statements through a large building and a desire to socialise children into ways of thinking or behaving. Kashti (1986, 1988), for example, draws on evidence on the development of residential education in nineteenth century England, Translyvania in the 16th and 17th centuries and Israel in this century to show how the size of living groups for children has decreased as religious or national ethos and ethnic homogeneity has declined.

There is a perennial tendency for decisions about size to be ideologically driven with no obvious evidence on likely benefits for children and staff. Indeed, an obsession with size can distract from other qualities of regimes, such as treatment, which may rely on certain numbers to be viable. This can have the effect of removing a service option for ideological rather than scientific reasons. In New South Wales, Australia, for example, the state government has recommended that care homes should have no more than eight children and that there should be no more than three establishments on one site. This policy may be welcome on child care grounds but seems oblivious to the evidence that size is of little significance in itself.

From a sociological perspective, it is also clear that the function of the residential placement is a strong determinant of size of institutions and the sub-units within them (Tizard et al., 1975). There is a reasonable amount of consistency in differences of placement size between sectors, suggesting a strong functional influence. Youth correction facilities (range 40-500 in US), hospitals (range 1,000-2,000) and residential education (range 200-1,200) tend to be larger than provision for children with special educational, behavioural or psychological needs (range typically between 5-50).

There can, however, be a confusion between size and function. Large institutions may be difficult to run but can be stimulating for young people. Some literature on adolescence (Coleman and Hendry, 1999) stresses the benefits of a mixture of experiences in contexts where size varies, such as a supportive family, a peer group and a football crowd. Lambert’s (1971) idea for the development of Dartington Hall School into a constellation of centres may have sounded radical at the time but probably has some relevance. Indeed, with new technologies, Illich’s (1971) ‘the school is dead’ may yet come into its own.

Naturally, the above list of influences cannot be exhaustive and other unexpected factors can enter the calculation. In education, there has been a recent pre-occupation in the United States with questions of security in mainstream schools, primarily stemming from a series of incidents in which children have shot pupils and teachers. Although these are isolated events (only a tiny proportion of schools have experienced such an incident), there is a suggestion that large size may be a contributory factor. This has prompted the Gates Foundation to mount a study into the potential for capping school size at 400 places.

These powerful economic, ideological and sociological influences tend to produce stability within residential sectors but may also detract from an objective assessment of size. Generally speaking, there is a greater tendency to think about the potential to make placements more effective than to make them smaller. Thus, although size may be a factor taken into consideration, it is usually determined by extraneous issues. This is confirmed by the tendency for institutions to be satisfied with the size they are. There may be minor adjustments for economic or ideological reasons but staff in big establishments tend to emphasise the benefits of large size while those in smaller establishments argue that ‘small is beautiful’. In his study of 150 boarding schools in England, Lambert et al. (1975) found no examples of plans radically to alter a school’s size and few concerns among staff or pupils about size as an issue. This was despite the fact that the school populations varied from 60 to 1,400.

From the available evidence, it would seem reasonable to hypothesise that the size of placements for children and the sub-units within them decrease with (a) the age of children served; and (b) the amount of special needs (educational, behavioural, psychological) they present. Since many residential services in Western nations are increasingly (although not exclusively) reserved for children with some kind of special need, sizes of both institutions and living units have decreased in the last quarter of the last century.

Intellectual justifications for small living group
Two schools of thought, one philosophical the other psychological, have provided the intellectual justification for this trend to smaller living group size. The philosophical influence is the growing strength of individualism in Western societies and a move away from collective socialisation and group indoctrination. This is manifest in increasing sensitivity to the needs of individual children, the categories of need used to fashion plans for them and the obligation to try and meet needs effectively. It is also apparent in changes of desired aims from values such as encouraging religious belief or preparing for military or domestic service to ones of providing safety, allowing self expression, encouraging family links and improving social adjustment.

The psychological influence has been fuelled by evidence accumulated since the Second World War on the negative effects of separation of children from their birth families. This has arguably had the greatest impact on the organisation of size of living groups and residential placements for younger children.

Most of the early proponents of maternal deprivation were psychologists and psychiatrists attracted to psycho-analytical theories and especially concerned about the fate of illegitimate children in foundling homes. Their ideas gained influence after World War I and were further reinforced by public concern about the effects of separation brought about by the evacuation of children in World War II. Large numbers of children in Western Europe cities were moved to safe rural havens, which for some meant sailing to North America.

While there can be little doubt that theories of separation and maternal deprivation have had a major impact on child care thinking in the second half of the twentieth century, their proponents were unlikely to have perceived family group living as a ‘cure’ for maternal deprivation as it could not remedy the child’s fundamental deprivation of a mother figure who offered continuous care. The family focus was more a reaction against institutional practices rather than an attempt to provide expectations of a ‘normal’ family, such as permanence and stability of adults and children.

The effects of these arguments has been to confuse the issue of size with a reaction against institutional care. Small is seen as family like and caring but the reality can be different. For example, a home where all the residents are excluded from school will seem claustrophobic if the children are not stimulated by other experiences and an anti-social peer culture may flourish.

It is clearly difficult to disentangle all the relevant factors but the fact is that institutions across the board have created smaller, more family like environments within larger settings to meet the needs of their children more effectively. Variously called ‘cottages’, ‘living units’ or ‘family-group homes’, the tendency has been to establish units of 6-12 children supported by consistent care givers in a family like environment.

Approaching the issue of size
Some general conclusions about the effects of size can be drawn from this evidence but most are too general to be useful. There are simply insufficient proven effects of size as a factor in its own right to be categorical. What then offers something more fruitful?

A three stage approach is helpful. The first looks at the aims of the residential institution. This is the most useful starting point and so much emanates from them. Are the aims, therefore, to promote child development, to shelter, to treat behaviour problems or to educate? Favoured outcomes will naturally differ in particular historical periods (Parker, 1988) but it is possible to chart them, whether in terms of societal expectations or what staff believe they can achieve for children (Brown et al., 1998; Dartington Social Research Unit, 1998).

The second stage is to decide what practices are known from research to promote these aims. Here, there is much more information in some areas than others (Bullock et al.,1993; Department of Health, 1998). For example, approaches to reducing anti-social behaviour are better informed by research than managing family contacts. Nevertheless, there is sufficient practice wisdom for professionals to act decisively even if there are few validated service responses to the identified needs of children.

The third stage is to ask what organisational features facilitate these desired practices (Dartington Social Research Unit, 1999). What structures and training need to be put in place to achieve them? One aspect to be considered will be the size of the institution and the sub-units within it.

This approach has the advantage of moving away from seeking general effects of size, a perspective that has been found to be limited. It introduces size as an important variable but at the right stage of planning. Hopefully, this will lead to better decisions about the organisational aspects of residential institutions and better outcomes for children.

Conclusions
An initial scrutiny of residential establishments suggests some relationship between size, style of operation and outcomes but closer examination and a consideration of other variables show that this relationship is tenuous. All studies (except one on children’s homes) found no indication of the influence of size on any aspect of children’s behaviour or development once other factors were taken into account. The ranking of institutions by size produced no co-relations with rankings of other variables.

Although size may not be important for outcomes, it clearly produces its own problems of structure and communication. The message from organisation theory is that size does not become important until the organisational demands of the institution displace its primary functions.

If the significance of size is considered for different types of residential institution, some general conclusions emerge. For example, large size offers flexibility for placing young people in units, offers more choice of activities, more resources and more treatment opportunities. These can be especially effective if there are smaller living groups and good relationships between people who know one another well.

If aims of the establishment are associated with performance, a certain size might be necessary to achieve this. But, size brings its own problems, such as peer cultures in prisons or the admission of unsuitable children to keep up numbers. Size is a weapon in institutional management and has to be used intelligently to produce benefits.

A problem in discussion is that the issue of size has got caught up in a reaction against institutional care. There is also frequent confusion between size and function. Large institutions may be difficult to run but can be stimulating for young people.

Studies of the relationship between size and outcomes suggest that size in itself is less significant than the way it is managed. Evidence about classroom size, for example, shows that a reduction of class sizes to around 20 does help younger children, especially those from disadvantaged backgrounds, but only under conditions necessary for effective implementation.

A more fruitful approach to view size as a means to other ends. This requires initial clarification of the aims of residential establishments and the means of achieving them. It is in a consideration of those means that discussions of institutional size take their most useful place.

References

Bennett, N. (1998) ‘Class size and the quality of educational outcomes’, Journal of Child Psychology and Psychiatry, XXXIX, 797-804

Brown, E., Bullock, R., Hobson, C. and Little, M. (1998) Making Residential Care Work: Structure and Culture in Children’s Homes, Aldershot: Ashgate

Bullock, R., Little, M. and Millham, S. (1993) Residential Care for Children: A Review of the Research, London: HMSO

Coleman, J. and Hendry, L. (1999) The Nature of Adolescence, London: Routledge

Clarke, R. and Cornish, D. (1972) The Controlled Trial in Institutional Research – Paradigm or Pitfall for Penal Evaluators?, London: HMSO

Dartington Social Research Unit (1998 ) Structure, Culture and Outcome: How to Improve Residential Services for Children

Dartington Social Research Unit (1999) Matching Needs and Services

Department of Health (1998) Caring for Children away from Home: Messages from Research, Chichester: Wiley and Son

Finn, J. and Achilles, C. (1990) ‘Answers and questions about class size: a statewide experiment’, American Educational Research Journal, XXVII, 557-577

Friedkin, N. and Necochea, J. (1988) ‘School system size and performance: A contingency perspective’, Educational evaluation and Policy Analysis, X, 237-249

Howley, C. (1994) The Academic Effectiveness of Small Scale Schooling, Charleston: ERIC Digest, 372897

Howley, C. (2001) Research on Smaller Schools: What Education Leaders Need to Know to Make Better Decisions, Arlington: Educational Research Service

Iacovou, M. (2001) Does Class size Really Matter?, Institute for Social and Economic Research, University of Essex

Illich, I. (1971) Deschooling Society, New York: Calder and Boyars

Kashti, Y. and Arieli, M. (eds.) (1987) Residential Settings and the Community: Congruence and Conflict, London: Freund

Kashti, Y. and Arieli, M. (1986) People in Institutions: The Israeli Scene, London: Freund

Kashti, Y. (1998) Boarding Schools at the Crossroads of Change: The Influence of Residential Education on National and Societal Development, New York: Haworth Press

Kushlick, A. (1972) ‘Evaluating residential services for the mentally handicapped’ in McLachlan, G. (ed.), Approaches to Action – A Symposium on Services for the Mentally Ill and Handicapped, Oxford: University Press/Nuffield Hospitals Trust

Kushlick, A. (1974) ‘Evaluating residential services for mentally handicapped children’ in Wing, J. and Hafner, H. (eds.), Roots of Evaluation: The Epidemiological Basis for Planning Psychiatric Services: An International Symposium, Oxford: University Press

Lambert, R. (1971) Alternatives to School. Exeter: University Press

Lambert, R., Millham, S. and Bullock, R. (1975) The Chance of a Lifetime? A Study of Boarding Education, London: Weidenfeld and Nicolson

McCullough, M. and Ely, P. (1975) Social Work with Groups, London: Routledge and Kegan Paul

Millham, S., Bullock, R. and Cherrett, P. (1975) After-Grace-Teeth: A Comparative Study of the Residential Experiences of Boys in Approved Schools, London: Human Context Books

Moos, R. (1974) Evaluating Treatment Environments: A Social Ecological Approach in Health, Medicine and Society, A Wiley-Interscience Series, New York

Moos, R. (1968) ‘The assessment of social climate correlational institutions’, Journal of Research in Crime and Delinquency, V, 174-188

Moos, R. and Houts, P. (1968) ‘Assessment of the social atmospheres of psychiatric wards’, Journal of Abnormal Psychology, LXXIII, 595-604

Parker, R. (1988) ‘An historical background’ in National Institute for Social Work, Residential Care: The Research Reviewed, London: HMSO

Plannungsgruppe Petra (1988) Analyse von Leistungsfeldern der Heimerziehung: Ein empirische Beitrag zum Problem der Indikation, Frankfurt am Mainz

Prais, S. (1996) ‘Class size and learning: The Tennessee experiment - what follows?, Oxford Review of Education, XXII, 399-414

Rutter, M., Maughan, B., Mortimore, P. and Ouston, J. (1979) Fifteen Thousand Hours: Secondary Schools and their Effects on Children, London: Open Books

Shapson, S., Wright, E., Eason, G. and Fitzgerald, J. (1980) An experimental study of the effects of class size, American Educational Research Journal, XVII, 141 – 152

Sinclair, I. And Gibbs, I. (1998) Children’s Homes: A Study in Diversity, Chichester: Wiley and Son

Tizard, B. (1975) ‘Varieties of residential nursery experience’ in Tizard, J. et al., Varieties of Residential Experience, Op. Cit., 102-121

Tizard, J. (1964) Community Services for the Mentally Handicapped, Oxford: University Press

Tizard, J. (1975) ‘The quality of residential care for retarded children’ in Tizard, J. et al., Varieties of Residential Experience, Op. Cit., pp. 52-68

Tizard, J., Sinclair, I. And Clarke, R. (eds.) (1975) Varieties of Residential Experience, London: Routledge and Kegan Paul