The fact that social problems cluster in space is not new. Spatial clustering of social problems has been described for several issues such as low educational achievement, crime and drug use, among others. One key factor that has been linked to those problems is the geographical concentration of contextual disadvantage. It has been argued that this observed correlation is only to be attributed to the fact that housing and labour markets create incentives for vulnerable people to cluster in space. Some believe that this clustering generates additional effects leading to poorer outcomes that would not have been observed in the absence of spatial clustering. The literature is unclear on the question of whether there is a "neighbourhood effect" of contextual disadvantage on problems like antisocial behaviour, and how this effect might be transmitted. Neighbourhood studies have been subject to persistent methodological and conceptual shortcomings. These may be partly related to the high costs involved in producing new datasets with adequate spatial measures. The availability of datasets with contextual data is scarce, thus many of the published papers on the subject have drawn on a low number of different studies (usually from the US and Europe). Consequently, the possibility to generalize their findings seems to be limited. In addition, the availability of high quality data (e.g. longitudinal datasets) that can help to rule out known methodological problems is even more restricted. In order to contribute to improving the understanding of how contextual characteristics might influence adolescent antisocial behaviour, firstly, a systematic review of longitudinal neighbourhood effects studies was conducted. In the first part of the thesis, results from the review suggest that the evidence supporting the existence of a direct neighbourhood effect of poverty and concentrated disadvantage on antisocial behaviour is mixed. Contextual effects of concentrated disadvantage also seemed to be highly dependent on model specification, whereby most studies finding significant main effects usually failed to include potentially relevant confounders in regression models. Commonly omitted confounders were related to baseline antisocial behaviour, parenting and peer differential association. Furthermore, evidence was generally unsupportive of the idea that neighbourhood level residential instability, neighbourhood disorder and incivility, social capital and collective efficacy or exposure to violence may have a direct effect on antisocial behaviour. Regarding institutional resources, mixed results were found. Some evidence pointed to the idea that "subcultural" variables (e.g. community level tolerance to deviance) may have an effect on reduced individual level violence. At times, it seemed that more complex models regarding how neighbourhood influences may influence behavioural outcomes might be needed. In the second part of the thesis, data from a longitudinal study, representative of the school population of Santiago de Chile, was merged with independent contextual level information (Census tract, schools and police records) and analysed. By examining the case of Santiago de Chile, a series of ideas regarding how contextual characteristics of activity spaces might relate to the growth of antisocial behaviour diversity over time were explored and tested. Specific attention was paid to examine and discuss how contextual effects might operate, in particular, how contextual disadvantage may influence criminogenic processes of strain, social control and contagion (peer effects). In order to test the proposed hypotheses, a series of hierarchical linear growth models were estimated. No evidence supporting the idea that different types of activity spaces (home based or school based activity spaces) may have differential effects on antisocial behaviour was found. However, results suggest that higher levels of contextual concentrated disadvantage across activity spaces significantly predicted a steeper growth of antisocial behaviour diversity over time. In spite of this, no support was found for the existence of a direct contextual effect once other covariates (i.e. baseline antisocial behaviour, strain, family level social control, contagion effects, among others) had been controlled for. The effect of concentrated disadvantage on antisocial behaviour appears to be mainly indirect; that is, mediated by other covariates. Baseline antisocial behaviour and contagion effects (peer effects) seem to play a relevant role in explaining away the effect of contextual concentrated disadvantage on the growth of antisocial behaviour scores over time. Only partial support for the idea that strain indicators may predict growth in antisocial behaviour diversity over time was found. Additionally, mediation analysis suggests that it may seem unlikely that the effect of contextual concentrated disadvantage on antisocial behaviour would be mediated by increased levels of strain. In spite of this, the effect of family level SES on the growth of antisocial behaviour diversity does seem to be partially mediated by some of the measured strain indicators. Measurement limitations (antisocial behaviour scale could only increase or remain stable) made it difficult to interpret some unexpected findings regarding strain effects. Regarding social control variables, evidence suggested that, even though family level monitoring predicts antisocial behaviour, neither parental attachment nor monitoring seemed to mediate the effect of contextual disadvantage on antisocial behaviour. In relation to school level social control, none of the relevant measures (school value added education and school attachment) significantly predicted antisocial behaviour in the fully specified model. Moreover, none of the hypothesized mediation effects held up, after controlling for other covariates. Regarding contagion effects (measured using peer variables), macro level concentration of juveniles with arrest records failed to predict individual level growth in antisocial behaviour diversity over time. Nevertheless, micro level concentration of antisocial peers in school and/or in activity spaces did predict growth in antisocial behaviour diversity. Results on micro level concentration of antisocial peers where subject to multicollinearity problems and thus were assessed separately. The effect of both variables (concentration in schools and concentration in activity spaces) was partially mediated by best friend's antisocial behaviour. Furthermore, concentrated disadvantage and concentration of deviant schoolmates in activity space interacted to predict a stronger relationship between affiliation to deviant peers and antisocial behaviour . Results are consistent with both geographic propinquity and co-offending process. , because of a low ecometric reliability found for "concentration of antisocial peers in activity space", results regarding this variable are regarded as tentative. An explanatory hypothesis of observed effects was proposed. Results may suggest that the effect of contextual disadvantage on antisocial behaviour is mainly indirect. Contextual disadvantage might be regarded as an expression of spatial clustering (social sorting) of low SES families due to housing and other governmental policies. In average, low SES families display poorer parenting skills, which might provide at least a partial explanation as regards to why higher concentration of antisocial peers (in school or activity spaces) and increased baseline antisocial behaviour scores are observed in disadvantaged contexts. In turn, higher concentration of deviant peers may be facilitating contagion effects. Results suggest that effects of concentrated disadvantage on antisocial behaviour might be due to simultaneous occurrence of compositional and contextual effects. / Study limitations, policy implications, and recommendations for future research are discussed.
Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:711920 |
Date | January 2015 |
Creators | Hein Willius, Andreas Alexander |
Contributors | Gardner, Frances Ellen Mary |
Publisher | University of Oxford |
Source Sets | Ethos UK |
Detected Language | English |
Type | Electronic Thesis or Dissertation |
Source | http://ora.ox.ac.uk/objects/uuid:f8a96ec7-c87b-4a5e-8e0f-2dcb67df291a |
Page generated in 0.0023 seconds