Williams, E. (2018) CALLS Hub conference, University of St Andrews, UK, 23 March 2018 [SLS]
Other information: Abstract:
An extensive literature investigates the association between neighbourhood context and individual outcomes such as health, educational attainment and income. A central problem for observational studies in this neighbourhood effects literature is the issue of selection bias; whether correlations between neighbourhood characteristics and individual outcomes are explained by the impact of neighbourhood context on individuals, or by individuals with different characteristics selecting into distinct neighbourhoods. One approach to strengthening the causal claims of neighbourhood effects research is the ‘tenure-split’ natural experiment, which relies on the assertion that social housing allocations are random with respect to the neighbourhood. Three recent UK studies employ the ‘tenure-split’ approach, and utilise the results obtained to cast doubt on a wide range of neighbourhood effects studies. This study critiques the claims made relating to the random nature of social housing allocations in the UK and re-examines the results of the three UK studies. In addition, new analyses are presented that use the ‘tenure-split’ approach to investigate the association between neighbourhood deprivation and occupational mobility. The results obtained support the view that, in the UK at least, the ‘tenure-split’ approach is not a valid natural experiment and does not therefore help address the issue of selection bias.