Introduction: Quantifying Neighbourhood Effects
Jörg Blasius, JĂŒrgen Friedrichs & George Galster
The policy context
The prevailing thrust of housing and urban planning policy in several Western European countries, notably France, the Netherlands, Sweden and the UK, aims to create more socioeconomically mixed residential environments (Andersen, 2002; Andersson, 2006; Musterd & Andersson, 2005; Kearns & Mason, this issue; Kintrea, 2007; Kleinhans, 2004; Meen et al., 2005; Mûrie & Musterd, 2004; Musterd, 2003; Tunstall & Fenton, 2006). Although not enshrined as national policy in the US, many programmatic efforts aimed at reducing the economic segregation of neighbourhoods are underway nonetheless, and more comprehensive efforts have been strongly advocated (de Souza Briggs, 2004; Dreier et al., 2004).
Regardless of the particular programmatic forms that this policy thrust has assumed internationally, all are founded on the belief that neighbourhoods have a strong and independent effect upon the well-being and life-chances of individuals. The adequacy of the evidence base to support this position has been the subject of spirited debates on both sides of the Atlantic (see Atkinson & Kintrea, 2001; Delorenzi,2006; Friedrichs, 2002; Galster, 2007b; Galster & Zobel, 1998; Joseph, 2006; Kearns, 2002; Kleinhans, 2004; Musterd, 2003; Joseph et al., 2006; Ostendorf et al., 2001). Indeed, although large in volume, much of the literature quantifying neighbourhood effects can be challenged on methodological grounds (for reviews see: Ellen & Turner, 2003; Friedrichs, 1998; Friedrichs et al., 2003; Galster, 2003; Gephart, 1997; Leventhal & Brooks-Gunn, 2000; Sampson et al., 2002; Van Kempen, 1997). The primary purpose in this Special Issue is to contribute to this housing policy-relevant discussion by assembling papers that quantify various types of neighbourhood effects through the use of cutting-edge social scientific techniques.
Neighbourhood Effects Investigation as a Means of Cross-disciplinary Integration
The secondary purpose of this book is to introduce these techniques to a wider array of housing researchers and to show how a variety of disciplines have offered insightful, synergistic perspectives on a field of enquiry that is central to the scholarship on housing. It is noteworthy that over the last 15 years research on neighbourhood effects has led to a research tradition extending far beyond the sociological urban research where it originated. The problem of quantifying neighbourhood effects and the use of associated methodologies (such as multi-level analysis or instrumental variables) has attracted scholars from criminology, sociology, social geography, economics and health science.
Although typically unnoticed by each of the involved disciplines, neighbourhood effect research and research methodology have grown to constitute a common core of interdisciplinary and cumulative research effort. The differences among disciplines appear to vary only by the dependent variables investigated, such as crime rates, disorder, education, earnings or health impacts. A recent illustration was the inauguration of a joint workshop at the 2006 European Network for Housing Research conference that brought together the Poverty Neighbourhood and Health Effects Working Groups. Even at the risk of overstating the cumulative evidence from the last 15 years, it is thought that the study of neighbourhood effects has made an extremely important contribution to the integration of social sciences.
About this book
An important precursor to the effort here was the Housing Studies Special Issue: âLife in Poverty Neighbourhoods, (vol. 18(6)â 2003) that we guest edited in collaboration with Sako Musterd. The papers pertained to different aspects of these deprived residential environments, such as social status mobility, residential mobility, social norms, and residentsâ bonds to the neighbourhood. Although suggestive, the papers often were descriptive in approach and did not utilize sophisticated, multivariate/multi-level methods to quantify the impact of poverty neighbourhoods. Indeed, the 2003 Special Issue also contained two papers that pointed to these shortcomings and offered directions for future research.
This book comprises a series of innovative responses to the challenges raised in our 2003 work. In contrast to the 2003 issue, contributions now focus on the multi-level problem: the impact of neighbourhood composition on the attitudes and behaviours of individual residents. Some of the chapters here use multi-level statistical techniques such as Hierarchical Linear Modelling (HLM), whereas others seek to explore these effects with techniques appropriate to small samples on the neighbourhood context level (correspondence analysis). One uses a comparison of siblings over time and another uses instrumental variables as a way to correct for selection bias associated with unmeasured parental effects. All studies provide a wealth of control variables applied in a multivariate statistical context, in an effort to isolate the independent contribution of neighbourhood context. As with the 2003 Special Issue, an international array of authors is presented here: England, Finland, Germany, Netherlands, Scotland, Sweden and the US are represented. Thus it is hoped that this book encourages continuities in the TransAtlantic dialogue. Below in this Introduction you will find a brief summary of the substantive results of the studies represented; the details of the statistical methods employed are supplied in the Appendix.
Andersson, Musterd, Galster & Kauppinen begin the Special Issue by addressing two fundamental questions: Which aspect of neighbourhood socio-economic environment is most strongly correlated with adult earnings outcomes? Does the answer depend on gender and metropolitan/non-metropolitan residence? Their analysis of Swedish social register data finds that the percentages of adult males with earnings in the lowest 30th and (to a lesser degree) the highest 30th percentiles hold greater explanatory power for all strata than domains of neighbourhood mix related to education, ethnicity or housing tenure, although the correlation is much stronger for men and for metropolitan residents.
Kearns & Mason also probe the question of which characteristics of the neighbourhood matter most to whom. They explore the relationship between different mixes of housing tenure in UK neighbourhoods and incidence of neighbourhood problems, resident-reported deficiencies, and the demand for improvements to local services and amenities. Their multivariate findings based on the Survey of English Housing indicate that the share of social renting in the area is a more important predictor of neighbourhood problems and service improvement desires than the degree of tenure mixing. Moreover, for both social renters and owner occupiers, the identification of problems in areas where social renting makes up around a quarter of the housing market is at least double than in areas where owner occupation comprises the vast majority of the housing market and social renting is at half this level or less, suggesting a social disadvantage threshold consistent with much US evidence (Galster, 2002).
Bramley & Karley address a closely related question: Is âneighbourhoodâ better operationalized as school environment or residential environment? To find an answer, they use UK student achievement data merged with data on both the schools the students attend and the small-scale neighbourhoods in which they reside. Bramley & Karley find that the proportions of homeowners in both the school and (especially) the neighbourhood strongly explain better primary and secondary school attainment outcomes, although the distinct influences of poverty and homeownership are difficult to disentangle.
Galster, Marcotte, Mandell, Wolman & Augustine investigate the consequences for children of the proportion of poor residents in the neighbourhood where they were growing up. They employ the US Panel Study of Income Dynamics (PSID) to examine the cumulative childhood conditions of children born 1968â74, and a variety of fertility, educational and earnings outcomes that they experienced as young adults in 1999. Although their application of instrumental variables substantially attenuates the apparent neighbourhood effects, they nevertheless find that cumulative neighbourhood poverty over childhood has an independent, non-trivial relationship with high school attainment and earnings.
Blasius & Friedrichs shift the emphasis toward outcomes traditionally examined by sociologists. They uncover the sizeable differentiation within an infamous âdeprived areaâ in Cologne (Germany) and relate it to residentsâ perceptions of disorder, deviance and perceived risk. They conclude that perceived disorder, observed deviant behaviour and perceived risk are quite unrelated, and thus have distinct correlates. Neighbourhood-level collective efficacy and intergenerational closure are strongly related to the perceived extent of deviance, but social status heterogeneity is not. The proportion of residents with specific characteristics, such as single-heads of households, accounts for much of the variation of the dependent variables.
Oberwittler investigates a different set of outcomes related to psychological strain and delinquent behaviours, employing a cross-sectional survey of adolescents in 61 neighbourhoods in two German cities. He finds support for neighbourhood effects, but they appear strongly contingent on ethnicity and the spatial orientation of adolescentsâ routine activities and peer networks. For example, native German girls are particularly prone to react violently to neighbourhood disadvantage. Like Bramley & Karley, Oberwittler finds that both school and neighbourhood environments are important predictors (in this case, of delinquency).
Gijsberts & Dagevos explore a particularly sensitive dimension of neighbourhood context: the proportion of ethnic minority groups. Using two large-scale Dutch social surveys, they probe the degree to which non-Dutch concentrations are related to interethnic contacts, language proficiency and mutual stereotypical attitudes. Their analysis shows that inter-ethnic contacts occur less frequently in neighbourhoods with nonDutch concentrations, and that this inhibits minoritiesâ attainment of Dutch language skills. Moreover, some mixing appears to have a positive influence on Dutch perceptions of the minorities, although a sudden and sizeable increase in the neighbourhood proportion of minorities apparently yields more negative Dutch attitudes toward minorities.
Finally, Vartanian, Buck & Gleason conduct the first investigation of intergenerational mobility across various types of neighbourhoods, using the US PSID. They are particularly interested in the extent to which children who grow up in disadvantaged neighbourhoods are likely to live in such neighbourhoods as adults. They find that whites who grow up in the most disadvantaged white neighbourhoods are far more likely to live in such places during adulthood than white children who grow up in even slightly better neighbourhoods. Although the relationsh...