The Effects of Neighbourhood Poverty on Adolescent Problem Behaviours: A Multi-level Analysis Differentiated by Gender and Ethnicity

Abstract

This paper investigates how adolescents react to living in urban areas of concentrated poverty, and whether contextual effects on psychological strain and delinquent behaviour exist, using a cross-sectional youth survey in 61 neighbourhoods in two German cities and a rural area (n = ca. 5300). Multi-level analysis is applied to estimate neighbourhood effects controlling for individual socio-demographic composition. Results suggest that neighbourhood effects on delinquency exist which are, however, dependent on the spatial orientation of adolescents' routine activities and peer networks. Adolescents with an immigrant background do not seem to be influenced by neighbourhood conditions in the same way that native adolescents are. In particular, native girls react to neighbourhood disadvantage by resorting to violence. A tentative cross-classified multi-level model suggests that schools and neighbourhood contexts simultaneously affect adolescents' delinquency.

Keywords:

• Neighbourhood
• adolescence
• delinquency
• crime
• multi-level analysis
• poverty

Acknowledgements

This research has been supported by the Deutsche Forschungsgemeinschaft (Ob 134/3-1,2). The author was a Marie-Curie Intra-European Fellow at the Institute of Criminology, University of Cambridge (UK) during 2004 to 2006. The author would also like to thank the organizers and participants of the conference Inside Poverty Areas, the editors and anonymous reviewers of this special issue and Per-Olof Wikström for their useful comments.

Notes

¹ Survey factor score: percentage immigrant background, percentage unemployed or welfare recipient > 5 months; mean parents' highest occupational prestige, percentage university degree of parents, mean household goods; official factor score: percentage non-German < 14 years, percentage welfare recipients < 18 years, mean dwelling floor space per person.

² In ca. one-third of cases, no sufficient information was available for assigning an ISCO code; however, a rough categorization to a four-point ordinal scale yielded valid information for 95 per cent of the sample, using the highest status of either father or mother (Oberwittler & Blank, 2003 for details).

³ This survey question suffers from a relatively high item non-response (3.4 per cent), which implies that some respondents decided not to report their parents' unemployment or welfare dependency due to social desirability concerns. It seems possible that these respondents also under-reported on related scales such as relative deprivation, thus causing a systematic bias to wrongly accept the null hypothesis.

⁴ Single-context models of potential school effects on delinquency yield significant ICCs for both native (8.9 per cent) and immigrant respondents (5.6 per cent) (conditional models controlling for socio-demographic composition).