Radical
Statistics
Poor areas and the 'ecological fallacy'
Paul Spicker
Introduction
This paper argues that there is such a thing as a 'poor area'. The idea has been attacked as an 'ecological fallacy', but this criticism is based on a misapplication of the principle of the ecological fallacy and a misunderstanding of the nature of poverty. Poverty is a constellation of different kinds of deprivation, and the relationship between these patterns of deprivation is complex. The forms of deprivation are patterned spatially by a series of urban processes, which lead to greater concentrations of problems in particular places. The area affects poor people, because the experience of living in a bad area exacerbates poverty, it makes people more vulnerable to poverty, and it is itself part of the experience of poverty. Moreover, the area affects more than the people who can be identified as 'poor', both because deprivations are more extensive, and because those who live in the area are also disadvantaged in their experiences and command over resources.
Understanding poverty
The concept of poverty is a contested one, and many different constructions have been placed on it (Spicker, 1999). Poverty implies, for some, a lack of resources; for others, a constellation of needs (Baratz and Grigsby, 1971); while for others, poverty is a social position related to the ability to participate in society (Townsend, 1979). The range of problems extends across a wide area of concern. The issues identified in the World Bank's participative poverty assessments include not only needs and resources, but problems of social relationships, including gender relations, precarious economic status, lack of security and abuse by those in power; limitations on the ability to participate in society, and on the capabilities of the poor; and issues relating to collective disadvantage, including disempowering institutions, weak community organisations and (of course) excluded locations (Narayna et al, 2000). Poverty is a complex, multi-dimensional phenomenon.
If poverty cannot be identified in terms of any single problem (like lack of money), or even a single set of problems, there is no effective distinction to be made between the factors which are associated with poverty and those which are constitutive of it. The idea of poverty describes, rather, a complex cluster of deprivations, sometimes described as a 'web' (Coffield and Sarsby, 1980; Kolvin et al, 1990; Narayna et al, 2000); the more deprivations that someone experiences, the more difficult it is to extricate oneself. (A similar position is implicit in the work of Mack and Lansley (1985), and of Townsend (1979), who both identify poverty in terms of permutations from a range of deprivations rather than specific defining problems.)
Poverty and poor areas
For over a century, social scientists have operationalised the concept of poverty principally as a property of individuals, families or households. Poverty is not usually referred to as an attribute of groups, communities, or regions. The concept of poverty has been individualised; the 'poverty' of poor areas is simply an aggregate of the conditions of the people who live in them.
On this basis, the proposition that areas can be poor is questionable. There are higher concentrations in the poorest areas of different kinds of social problem, but poverty is widely dispersed; most poor people do not live in poorer areas, and most of the people who live in such areas are not poor. This was the central insight of a seminal comment on the 1971 census (Holtermann, 1975), and it is still true. The Social Exclusion Unit's figures on the 'worst' areas in Britain show some concentration of problems, with 44 districts having 85% of the most deprived electoral wards. The report does not, however, show that most people in the worst areas were deprived. Less than half the children live in families on low incomes; only 20% of the population was unemployed, and less than 10% were single parents (Howarth et al., 1998, Table 3). The basic objection to describing the 'areas' as poor is, simply, that it misrepresents the situation of the people who live there.
The claim that such areas are poor nevertheless can be understood in two senses. The first is the argument that there is something about these areas which helps to explain the greater vulnerability to poverty of the people who live in them. The second, stronger sense is that the area itself is poor, even if many of the people in it are not.
Poor areas and the ecological fallacy
The idea that poverty can be understood in terms of areas has been attacked as an 'ecological fallacy'. I first heard this from Stein Ringen, but the fullest discussion in relation to area-based policy is made by Martin Bulmer (Bulmer, 1986, ch 11). The burden of Bulmer's position is that associations do not necessarily present evidence of relationships, and that aggregated data do not reveal information about the relationships which affect individuals. The evidence on the spatial distribution of problems is that although there may be greater concentrations in certain places, most people in the areas do not have the problems, and most of the problems are found elsewhere. The belief that there is an association is, therefore, based on a fallacious inference about the characteristics of individuals within the aggregate population.
The kind of study which Bulmer is criticising is fairly common in the literature (see e.g. Evans et al, 1992; Jencks and Peterson, 1991). Crane, for example, examines 'neighbourhood effects' on school drop-out rates and teenage pregnancies. He reviews several studies, controlling for family incomes, benefit receipt, religion, race, educational attainment and so forth, before presenting his own data (Crane, 1991). Neighbourhood effects seem to be large, but without an explanation of the process this might be seen as coincidental - the 'neighbourhood' acting as a residual figure for a series of factors which remain to be accounted for.
The ecological fallacy refers to the principle in empirically based social research that it may be illegitimate to make generalisations from data obtained between different settings, whether by aggregating data or by disaggregating it. This does not mean that identifying associations between aggregate figures is intrinsically defective, which would invalidate much economic analysis; nor does it mean that no inferences can be drawn about associations between the characteristics of an aggregate population and the characteristics of groups or smaller units within the population, which would make sampling impossible. What it does say is that the process of aggregating or disaggregating data may conceal the influence of disparate variables in different settings.
An ecological fallacy may be committed when the characteristics of groups are held to convey information about individuals within groups, or conversely the characteristics of individuals are taken as equivalent to those of groups. The basis for describing this as a fallacy, Schwartz (1994) points out, is really a matter of logic rather than of empirical observation. She gives the example of a jury which cannot reach a decision, which cannot be taken to mean that individuals within the jury cannot make a decision - usually, the opposite is true, and the jury is hung because some individuals are too decided. The principle is, at root, a simple one: groups have to be seen as different from the individuals who make them up. However, in the process of translating this principle into practice, the nature of the argument has shifted, so that it has come to focus on associations between factors rather than the characteristics of groups or individuals.
Bulmer's argument is based in a mis-application of the theory. The proposition that areas can be poor depends not on the aggregation of the circumstances of individuals but on a series of processes in the area as a whole. If poverty is seen not as an individualised concept, but as a description of a series of problems associated with deprivation, it is possible to refer to the problems of areas as a whole, rather than as an aggregate of the conditions of individuals. Bulmer takes it to be self-evident that we are dealing with aggregate figures. But a village is not simply an aggregate of households, and a city is not an aggregate of villages; they are spatial entities of different kinds. In geographical studies, 'correlations in which the units of observation are areal units are by no means always computed merely as an inferior substitute for the theoretically preferable individual correlations' (cited in Przeworski and Teune, 1970, p.60). That is one reason why the ecological fallacy holds true: the effect of aggregation or disaggregation can be to ignore important differences in the pattern of relationships which arise at different levels.
Bulmer also rejects any relationship between spatial factors and the circumstances of individuals. The 'ecological fallacy' is a statement about 'ecological correlation', not about ecology in itself. There is clearly an association between space and the prevalence of deprivation, though the association is weaker than many people supposed in the 1960s and 1970s; the question is whether the association means anything. Bulmer assumes that there is no basis for believing that the association stems from a relationship, which must depend on the analysis which is made.
Schwartz argues, in relation to public health, that:
the ecological fallacy, as often used, encourages three interrelated, fallacious notions: (1) that individual-level models are more perfectly specified than ecological-level models; (2) that ecological correlations are always substitutes for individual-level correlations, and (3) that group level variables do not cause disease. (Schwartz, 1994, p.819)
The same flaws occur in Bulmer's argument. The supposition that we are dealing with aggregates, rather than collective problems, and the rejection of a relationship between space and individual conditions, are based on profoundly individualistic assumptions. The ecological fallacy has become a bugbear to frighten people away from interpreting social observations in collective terms.
The concentration of problems
The reasons why an area is thought of as 'desirable' or 'undesirable' may not have anything to do with poverty. Design, location, proximity to facilities and environmental factors may influence people's choices; but other factors, like population density, maintenance and the history of the area come into play, shaping the way the area is perceived and directing the pattern of future occupancy. Once areas are identified as more or less desirable, though, economic status plays a major part in determining who is going to live there. It happens, in part, that people who have the resources to move to more desirable places are able to do so (Wilson, 1987). It may also be true that there are filters, or a process of selection, which exclude poor people from more desirable areas and push them towards less desirable ones. Massey et al (1994) argue that there is high mobility among the poor, and this, in conjunction with limited opportunity, pushes them towards segregation. Even in public housing, poor people are brought together by their inability to choose. The least desirable housing tends to go to people on lower incomes, living in worse conditions; the more desirable housing goes to people on higher incomes, living in better conditions, who are better able to wait until they get what they want (Clapham and Kintrea, 1986).
When there are problems in an area, poverty can make them worse. In part, this happens because poor people are likely to be people with other problems: they may be poor because they are unemployed, mentally ill, single parents or have low educational attainment. These are not only people with low resources; they are often stigmatised and excluded. Concentrating poor people in particular places means that there is also a concentration of this kind of issue, and the place itself develops a bad reputation. It also happens that poverty and a lack of resources lead to other problems in the places where poverty occurs. Areal factors like housing and the environment may cause poor health (MacIntyre et al, 1993). (Bulmer condemns a similar argument as 'the ecological fallacy given flesh', Bulmer, 1986, p.233). Poverty is associated with poor housing conditions, not just because poor people have to live in bad housing, but because their poverty makes the conditions worse; lack of resources may mean problems of heating, dampness, and lack of maintenance and poor commercial facilities in the immediate locality. In some cases, property which is adequately designed for better-off tenants has proved to be inadequate for poorer people: high rise buildings, which have been successsful for some groups, are associated with isolation, poor maintenance and a lack of play space. The problems of living in a bad area are greater when one is poor.
Do areas cause poverty?
Areas cause poverty if the social processes associated with an area create or shape the problems of poverty. There are four main arguments for this proposition.
- Environmental determinism. Some writers have argued that space, location and territory have an effect independent of economic factors (see e.g. Newman, 1973; Coleman, 1985). Industrially built housing has been associated with structural defects, inadequate estate design, faults in the internal design of dwellings, and noise; these problems are often associated with social problems. Coleman's work has serious defects; there are problems both in the kinds of indicators she uses to identify the prevalence of social problems, and the attempt to interpret these factors in terms of design (Spicker, 1987). There may still be effects which result from the interaction of the poverty of the tenants and the design of estates (Spicker, 1987).
- Areal effects on social relationships. The literature on this subject has always been controversial, not least because much of it has been tinged with sensationalism and moral condemnation of the poor. The core of the argument is the proposition that proximity generates patterns of social contact which in turn lead to identifiably different patterns of behaviour. Some commentators have considered the impact of neighbourhood on the levels of teenage pregnancy (Crane, 1991; Anderson, 1991). Anderson attributes the pattern of behaviour to sub-cultural influences, peer group influence and 'role models'.
- Area-based resources. The implication of a process in which there are greater concentrations of poor people in some areas than others - whether or not this concentration amounts to a considerable majority of people - is that the economic resources in that area are different from those elsewhere. First, facilities in the area which depend on economic participation or disposable income - like shops, post offices or chemists - are less likely to survive. Second, living in particular areas has a direct effect on the services which one receives: middle-class areas tend to have a greater resource base and lower needs. Third, the reputation of an area may also have an effect on its resources. People who live in 'bad areas' may find it difficult to get credit, insurance premiums may be higher, and residents may not be able to get commercial deliveries or taxi services.
The implication of these arguments is plain: where one lives does affect one's command over resources. If this is right, then areas do affect poverty.
- Spatial determinants of employment and the labour market. The conditions which most obviously produce poverty - the structure of the economy, and the pattern of employment - have a strong spatial dimension. Cities have developed, historically, with a mix of residential, industrial and commercial facilities. Geographically, the implications of this development is not to create a random mix, but rather differing areal locations for different activities, often reinforced by planning restrictions. It also happens that living near higher numbers of people who are unemployed affects job prospects in itself. The process is straightforward enough - where jobs become available which are accessible, there will be more competition at similar levels of skill. Residence in a particular area is likely, then, to be associated with employment prospects. Under these circumstances, it is not surprising that certain areas of cities have tended to be associated with higher levels of employment than others; indeed, it would have been astonishing if they were not.
The impact of areas on poverty
Two questions were posed at the outset of this section. The first is whether areas exacerbate poverty. Certain resources are based in particular locations, and that must mean that location directly affects command over resources. Areas can, then, exacerbate poverty. The second is whether living in certain areas can be said to cause poverty. People who live in particular areas are more likely than people in other areas to be poor as a consequence of the place where they live. The processes by which this can come about are clear, and the answer has to be 'yes'.
Are areas poor?
The idea of 'poverty' is not simply descriptive; it is a moral term, implying an evaluation of circumstances (Piachaud, 1981). To argue that an area is 'poor' is not simply to say that there are not many resources, but that there is a problem. That has to be based in the position that the area itself can be an appropriate focus for moral concern. Like many moral positions, this does not lend itself easily to argument; ultimately one either accepts it or one does not.
The central objection to describing an area as 'poor' is individualistic: that there is no such thing as an area, only the people who make it up. It is not, then, areas which have problems or low incomes, but the people who live in them. It would still follow that the area will have a higher concentration of poor people than elsewhere; it will have a variety of social problems; and the conditions of poor people are likely to be worse than if the same people lived elsewhere. Even from an individualistic perspective, then, there is still a case for considering the impact of location on poverty.
The case for shifting the focus of attention to the area becomes markedly stronger when attention moves to people who are not themselves poor. Living in a poor area can act to their detriment. It happens, for example, through the lack of community resources, the increased competition for places in the labour market, and the effect of stigmatisation on command over resources. Further, people who live in such areas are less secure than others. The fear of crime is directly associated with perceptions of the physical deterioration of an area (Painter, 1992, p.182), but the problems are not simply a matter of perception. People who are on higher incomes in lower income areas have greater vulnerability to crime than people elsewhere, including burglary, robbery, motor vehicle theft and vandalism (Evans, 1992, pp 42-46). These people are not likely to be made poor in consequence - that would happen only if the effect of living in the area was to bring their level of resources down sufficiently to consider them as deprived - but anyone in this position has lower resources, other things being equal, than others who have desirable, well-maintained environments.
This is a crucial point. What the findings about non-poor people show is that the problems of poor areas cannot be reduced to problems of poor people within those areas. The issues which I have discussed are related to the problems of poor individuals and families, but they are not always the same problems.
This argument still starts from the premise that the poverty of an area can be understood in individual terms. If the idea of poverty is not confined to individuals - and there seems to be no reason beyond assertion that it should be - there is no difficulty in applying the concept to collective social units. Although poor areas are likely to have more poor people than other areas, they are not simply defined by the relative numbers of poor people. Poor areas are identifiable in terms of their characteristics as areas; they have poor housing, a run-down environment, a lack of security, and low status. The environment, the economic base, the social status of the area and the infrastructure of services are developed at an area level. The question to ask is not just whether individuals are poor, but whether areas are. There is a constellation of inter-related deprivations which has to be understood at the level of the area. It follows that there is such a thing as a poor area.
Implications for policy
Holtermann's critique became, in the UK, the basis for an intellectual assault on the legitimacy of area-based policy. For nearly twenty five years, urban renewal in England has been separated from anti-poverty work, and where urban policies are pursued, they could not be represented as a response to poverty. Lawless comments that the effect of this shift has been in practice to ignore the problems of urban poverty when they are encountered (Lawless, 1989, pp 11-12). Recent changes in policy remain vulnerable to the same objections.
'From the point of view of social intervention', Bulmer comments, 'policies directed at areas per se are not likely to be successful, since they rest on the ecological fallacy' (Bulmer, 1986, p.236). This embodies a fallacy all of its own: it would not follow, because a problem occurs at the level of the individual, that policies have to be targeted at individuals. For example, the World Bank has argued that 'indicator targeting', a focus on broadly defined groups with higher proportions of problems than others, is often a more effective way of distributing resources than attempting to identify individuals (World Bank, 1990). By the same token, it would not follow, because problems are experienced at an area level, that they should necessarily be tackled at an area level. Townsend argues that 'the pattern of inequality within [areas] is set nationally' (Townsend, 1979, p.560); similarly, the Community Development Projects argued that the problems of poor areas could only be dealt with at a structural level in the wider economy (Loney, 1983). Contrariwise, it could also be argued that giving poor people money individually would remove many of the problems which occur when they live in close proximity to each other. But it should at least be possible to devise a strategy which makes some sense in relation to the problems which are experienced. On that basis, a policy for poverty must address the poverty of poor areas.
REFERENCES
Anderson, E. (1991) 'Neighbourhood effects on teenage pregnancy', in C. Jencks, P. Peterson (eds.), The urban underclass , Washington D.C.: Brookings.
Baratz, M. Grigsby, W. (1971) 'Thoughts on poverty and its elimination', Journal of Social Policy 1(2): 119-134.
Bulmer, M. (1986) Social Science and Social Policy, London: Allen and Unwin.
Clapham, D. Kintrea, K. (1986) 'Rationing, choice and constraint', Journal of Social Policy 15(1): 51-68.
Coffield, F. Sarsby, J. (1980) A cycle of deprivation? London: Heinemann.
Coleman, A. (1985) Utopia on Trial: vision and reality in planned housing, Shipman.
Crane, J. (1991) 'Effects of neighborhoods on dropping out of school and teenage childbearing', in C. Jencks, P. Peterson (eds.), The urban underclass, Washington D.C.: Brookings.
Evans D. (1992) 'Left realism and the spatial study of crime', in D. Evans, N. Fyfe, D. Herbert (eds.), Crime, policing and place: essays in environmental criminology, London: Routledge, pp 42-46.
Evans, D. Fyfe, N. Herbert, D. (eds.) (1992) Crime, policing and place: essays in environmental criminology, London: Routledge.
Holtermann, S. (1975) 'Areas of deprivation in Great Britain', in Central Statistical Office, Social Trends 6: 43-48, London: HMSO.
Jencks, C. Peterson, P. (eds.) (1991) The urban underclass, Washington D.C.: Brookings.
Kolvin, I. Miller, F. Scott, D. Gatzanis, S. Fleeting, M. (1990) Continuities of deprivation? Aldershot: Avebury.
Lawless, P. (1989) Britain's inner cities, London: Paul Chapman.
Loney, M. (1983) Community against government, London: Heinemann.
Mack, J. Lansley, S. (1985) Poor Britain, London: Allen and Unwin.
Massey, D. Gross, A. Shibuya, K. (1994) 'Migration, segregation and the geographic concentration of poverty', American Sociological Review 59(3): 425-445.
Macintyre, S. Maciver, S. Sooman, A. (1993) 'Area, class and health: should we be focusing on places or people?' Journal of Social Policy 22(2): 213-234.
Narayna, D. Chambers, R. Shah, M. Petesch, P. (2000) Voices of the poor: crying out for change, World Bank/Oxford University Press.
Newman, O. (1973) Defensible space: people and design in the violent city, Architectural Press.
Painter, K. (1992) 'Different worlds', in D. Evans, N. Fyfe, D. Herbert (eds.), Crime, policing and place: essays in environmental criminology, London: Routledge,
Piachaud, D. (1981) 'Peter Townsend and the holy grail', New Society 10th September: 421.
Przeworski, A. Teune, H. (1970) The logic of comparative social inquiry, New York: Wiley.
Schwartz, S. (1994) 'The fallacy of the ecological fallacy', American Journal of Public Heath 84(5): 819-824.
Spicker, P. (1987) 'Poverty and depressed estates: a critique of Utopia on Trial', Housing Studies, 2 (4): 283-292.
Spicker, P. (1999) 'Definitions of poverty: eleven clusters of meaning', In D. Gordon, P. Spicker (eds.) The International Glossary on Poverty, London: Zed Books.
Wilson, W. J. (1987) The Truly disadvantaged, University of Chicago Press
World Bank (1990) World development report 1990, Oxford: Oxford University Press.
Paul Spicker
Department of Political Science and Social Policy
University of Dundee
Dundee DD1 4HN
Scotland
Tel: (1382) 344 929 (direct line)
E-mail: p.spicker@dundee.ac.uk
