Neighbourhood effects on health: Does it matter where you draw the boundaries?
Introduction
Geographical variations in health exist at many scales from the global to the local (Boyle, Gatrell, & Duke-Williams, 2004). Analysis of data sources including census data, hospital admission records, and disease registration shows considerable fluctuation, with a general tendency for health to be worst in areas characterised by poverty and deprivation. In the United Kingdom, macro-scale analyses of census health data have been published by Charlton, Wallace, and White (1994) for 1991 data and by Dorling and Thomas (2004) for 2001 data. Among other studies of local-level variation, a theme issue of Geographical and Environmental Modelling was devoted to a set of papers analyzing the same local-scale health data set using different techniques (Fotheringham, 1999). Another body of work has focused on the question of whether there may be area differences in health which are independent of characteristics of individual residents. Kawachi and Berkman (2003) for example, have edited a book on Neighborhoods and Health containing review papers (Diez-Roux, 2003, Macintyre and Ellaway, 2003) and studies of different aspects of local-scale health variation. Pickett and Pearl (2001) have also reviewed this material.
The notion of neighbourhoods needs to be examined more critically. We set out a methodology to demonstrate the variable presence of significant ‘neighbourhood effects’ in area data, depending on how areas are delimited. Studies trying to identify neighbourhood effects have tended to use readily available geographical data units such as wards in the United Kingdom and census tracts in the United States (see table 1.1, Kawachi & Berkman, 2003 p. 4–5). These units may not coincide with the neighbourhoods that have an effect on health. The main objective of the paper was to determine how much conclusions about the existence and size of the neighbourhood effect on health are dependent on how neighbourhood boundaries are defined. The paper begins with a review of ideas about how neighbourhood characteristics might affect health and of the ways in which neighbourhoods might be defined. The sensitivity of analytical results to how areal units are defined is related to the modifiable areal unit problem, which is also discussed in the paper, followed by an account of zone design software which facilitates the analysis of the effects of zone design. An analysis of the neighbourhood effect using the standard system of areal units is presented, using Northamptonshire and Swindon (England) as case studies. Then, the same data are re-aggregated according to 50 different sets of ward-level zonal systems and the analysis of the neighbourhood effect is repeated for each set. This allows an evaluation of the robustness of the original results, and hence an answer to the question posed in the title.
Section snippets
Neighbourhood effects on health
In exploring neighbourhood effects on health, one important question has concerned the roles of context and composition in accounting for local variations (Macintyre et al., 1993, Pickett and Pearl, 2001). There is little doubt that health statistics in an area are influenced by the composition of the local population. Age, education, employment, ethnicity, housing, social class and other factors may all influence individuals' health. But are these compositional factors sufficient to account
How should neighbourhoods be defined?
Neighbourhoods can be defined in many different ways, but there are several problems in defining a set of neighbourhoods consistently across a whole country. Often the definition of neighbourhoods is determined by practical considerations, such as the assumption that a ward (or a census tract in the USA) is a sensible operational definition of a neighbourhood. This is convenient because data are available at the ward level, but there is little reason to expect neighbourhoods to follow ward
The modifiable areal unit problem
If we want to determine whether there is a neighbourhood effect, how should ‘neighbourhood’ be defined? It can be defined in multiple ways – but they will not necessarily give the same answers. This is an example of the modifiable areal unit problem (MAUP) – a classic problem in statistical analysis of geographical data.
The MAUP was first identified by Gehlke and Biehl (1934). Its essence is that analytical results for the same data in the same study area can be different, in some cases wildly
Criteria for designing zones
In the 1991 British census data used in this project, the smallest data unit is the Enumeration District, and hence EDs form the basic building block for designing alternative zoning systems. Because previous literature has assumed that wards are an appropriate size for identifying contextual effects, this paper makes a similar assumption, but tests a variety of different zonal systems at the same scale. In each case, EDs are aggregated to form ‘pseudo-wards’, and the analysis is concerned with
Zone design software
Investigation of the MAUP has been greatly aided in the recent past by the development of software which can be used in zone design (see Alvanides, Openshaw, & Rees, 2002). Relevant packages include SAGE (Wise, Haining, & Ma, 1997), ZDES (Alvanides, Openshaw, & Macgill, 2001) and A2Z (Daras & Alvanides, 2005). In this project, we used the AZTool system (formerly AZM), originally developed by Martin et al. for the construction of Output Areas for the 2001 census (Martin et al., 2001). AZTool
Estimating the neighbourhood effect in British census data
This study uses data from the 1991 British census to see whether there is an identifiable neighbourhood effect on people's health and, if so, to measure how big the effect is. The 1991 data were chosen because the alternative 2001 boundaries were expressly designed to homogenise neighbourhood effects. This is specifically the effect under investigation in this paper and thus the 2001 data would be inappropriate to adopt for the study. The data refer to self-reported health, and are based on
Modelling the impact of neighbourhood definition
In order to evaluate the impact of neighbourhood context on health, sets of pseudo-wards were created, using different combinations of criteria. This was done using AZTool. The results are evaluated first of all in terms of the properties of the zonal systems created, then by the correlations observed between the variables considered, and finally by the size of the neighbourhood effect identified.
Sets of pseudo-wards were generated for Swindon using various combinations of the user-defined
Properties of the zonal systems
Although the use of the population target of 8136 is intended to ensure that the number of pseudo-wards is comparable to the number of real wards, there is no guarantee that there will be exactly 21 pseudo-wards. In fact, there were 18 cases (out of 50) where the number was 21, there were 17 cases out of 20 pseudo-wards; the minimum being 16 and the maximum being 22. Condition C tended to have lower numbers than the other conditions.
The zonal systems can be evaluated according to the statistics
Conclusion – it does matter where you draw the boundaries!
The analysis reported above showed that for Northamptonshire there was a small but significant neighbourhood effect, but for Swindon there was no such effect. However, the experiments conducted with zone design showed that this result is dependent on the way the ward boundaries are drawn. Given that the ward boundaries are essentially arbitrary as far as the distribution of limiting long-term illness is concerned, the results suggest that the initial conclusion, that there is no neighbourhood
Acknowledgments
Robin Flowerdew acknowledges receipt of an Erskine Visiting Fellowship at the Department of Geography, University of Canterbury, Christchurch, New Zealand. David Martin (University of Southampton) gave us permission to use his zoning program AZTool. Jamie Pearce (University of Canterbury) provided useful insights and helped with data manipulation. The census data used in this paper are Crown Copyright and were bought by ESRC and JISC for use by the academic community. We also used digitized
References (41)
- et al.
Zone design for environmental and health studies using pre-aggregated data
Social Science and Medicine
(2005) - et al.
Constructing data zones for Scottish Neighbourhood Statistics
Computers, Environment and Urban Systems
(2007) - et al.
Childhood leukaemia in northern Scotland
Lancet
(1986) - et al.
Scales, levels and processes: studying spatial patterns of British census variables
Computers, Environment and Urban Systems
(2006) - et al.
Zone design as a spatial analysis tool
- et al.
Designing your own geographies
- et al.
Assessing deprivation in English inner city areas: making the case for EC funding for Leeds City
- et al.
Limiting long-term illness and locality deprivation in England and Wales: acknowledging the ‘socio-spatial context’
- et al.
Long-term Illness: results from the 1991 Census
Population Trends
(1994) Segregated neighbourhoods and mixed communities: A critical analysis
(2007)
Zone design in public health policy
The examination of neighbourhood effects on health: conceptual and methodological issues related to the presence of multiple levels of organisation
People and places: A 2001 census atlas of the UK
Guest editorial: local modelling
Geographical and Environmental Modelling
The modifiable areal unit problem in multivariate statistical analysis
Environment and Planning A
Geographies of health
Certain effects of grouping upon the size of the correlation in census tract material
Journal of the American Statistical Association
Introduction
Psychology and the environment
Cited by (242)
Geographic uncertainties in external exposome studies: A multi-scale approach to reduce exposure misclassification
2024, Science of the Total EnvironmentAdolescents' environmental perceptions mediate associations between streetscape environments and active school travel
2023, Transportation Research Part D: Transport and EnvironmentDo infrastructure deserts exist? Measuring and mapping infrastructure equity: A case study in Dallas, Texas, USA
2022, CitiesCitation Excerpt :Despite the widespread use of Census tracts or block groups, there are no definitive studies identifying the best spatial boundary to be used among all available options such as Census tracts, block groups, and zip codes (Flowerdew et al., 2008). Past studies have shown that the types of geographic boundaries used to aggregate data can affect variance, standard deviations, correlations, and regression analyses (Flowerdew et al., 2008). A better approach is to use perceived, resident-defined neighborhood boundaries, which may better represent the neighborhood and neighborhood-based measures such as access to destinations, walking routes, or the number of residences.