Low Self-Control in “Bad” Neighborhoods: Assessing the Role of Context on the Relationship Between Self-Control and Crime

Abstract

Although a wealth of research has substantiated the relationship between self-control and offending independent of an array of theoretically relevant covariates, little is known about the contextual variability of this relationship. Our study contributes to the literature by assessing neighborhood variability in the explanatory effect of self-control on individual offending in two Eastern European cities: Lviv, Ukraine and Nizhni Novgorod, Russia. Using data elicited from interviews with 1,431 respondents across 41 neighborhoods, we examine the extent to which the relationship between self-control and offending is moderated by neighborhood socioeconomic status (SES), and investigate the role of illegal opportunities and neighborhood morality as intervening processes accounting for the cross-level interaction between self-control and neighborhood SES. Estimates from hierarchical linear models indicate that self-control effects on offending are contingent upon ecological characteristics. However, neighborhood morality, and not neighborhood SES or neighborhood opportunities for crime, is a direct moderator of these effects.

Keywords:

• morality
• self-control
• neighborhoods
• Russia
• Ukraine

Notes

1. We acknowledge an alternative theoretical argument under which self-control effects are attenuated rather than amplified in communities of lower socioeconomic standing, where criminal temptations may be sufficiently strong to “crowd out” the effects of individual-level predictors of crime (Muraven et al., 1998). The results of at least one related study suggesting that impulsivity effects are more pronounced in economically prosperous neighborhoods support this scenario (Zimmerman, 2010). We also recognize the possibility that self-control effects are hegemonic and therefore not contingent upon any social context. In support of this argument, Vazsonyi et al. (2006) failed to uncover any effect of neighborhood socioeconomic welfare on the link between impulsivity and individual behavior (see also Wikström & Loeber, 2000).

2. Each neighborhood consisted of approximately 4,500 apartments/houses or 10,000-12,000 residents, slightly larger than a US census tract, which usually includes between 1,200 and 8,000 people, with an average of 4,000 people.

3. Approximately 65% of households had to be randomly replaced; replacements had to be made for different reasons including unavailability of respondents, refusal to participate, and commercial leasing of apartments

4. Because this measure is a variety index, and not a trait or trend where items should be highly correlated, Cronbach’s alpha (α = .54) is not necessarily relevant.

5. The scale was examined with confirmatory and exploratory factor analysis using maximum likelihood estimation. The results of both analyses suggested a one-factor solution.

6. Given the level-one and level-two sample sizes, the aggregate neighborhood score is calculated from approximately 35 individuals (1,431/41). While modest, we believe that this number of cases is sufficient to generate reliable aggregate scores, given the homogeneity within micro neighborhoods.

7. Although an aggregate-level α for this index is .66, it should be noted that because this measure is a cumulative Guttman-type scale, lower inter-item correlations are expected.

8. We experimented with abbreviated versions of the neighborhood criminal opportunities scale. In one iteration, we removed the littering item. In a second iteration, we removed the littering and arson items. Because the pattern of findings was substantively unchanged, we chose to retain the original scale items.

9. We examined the robustness of our results to the operationalization of our outcome variable. In one iteration, we reestimated our best-fit models using a Poisson distribution with overdispersion. The outcome variable (mean = 3.06; standard deviation = 2.40; range 0-7) was a count of the number of offending items that the respondent answered in the affirmative (0 = no chance of offending; 1 = at least some chance of offending). In a second iteration, we re-estimated the best-fit models using a logged transformation of projected offending as the outcome variable. Because the results were substantively unchanged from those presented in the manuscript, and for ease of presentation and coefficient interpretation, we present the results from the original projected offending measure (as described in the “Research Design and Methods” section above) in the “Results” section below (the alternative results are available from the authors upon request).

10. An alternative approach to modeling the cross-level interactions would be to first test the significance of a random slope variance on self-control, and subsequently think of neighborhood-level variables that could explain the random slope. However, basing the cross-level interactions on a priori substantive arguments is preferable. The power of the statistical tests of the cross-level interaction fixed effects is considerably higher than the power of tests based on the random slopes. In addition, one can test these interactions irrespective of whether a random slope on self-control is found (see Snijders and Bosker 1999, pp. 74-75, 95-96). Nevertheless, we did examine neighborhood variability in the effect of self-control on projected offending by allowing the coefficient for “Self-Control” to vary randomly across neighborhoods. The results indicated that there is significant variation in the slopes of both behavioral self-control (τ_self-control = .01, p < .001, not shown here) and attitudinal self-control (τ_self-control = .03, p < .001; not shown here) across neighborhoods.

11. Note that in additional analyses (available from the authors upon request), we found a positive and significant effect of neighborhood SES on projected offending, which was mediated by neighborhood morality.

Additional information

Notes on contributors

Gregory M. Zimmerman

Gregory M. Zimmerman is an Assistant Professor of Criminology and Criminal Justice at Northeastern University. His research focuses on the interrelationships among individual and contextual causes of criminal offending. He has recently been published in Criminology and the American Sociological Review.

Ekaterina V. Botchkovar

Ekaterina Botchkovar is an Assistant Professor of Criminology and Criminal Justice at Northeastern University. Her interests include theory testing, comparative criminology, and theory development.

Olena Antonaccio

Olena Antonaccio is Assistant Professor of Sociology at the University of Miami. Her interests include theory testing and development and comparative criminology.

Lorine A. Hughes

Lorine Hughes is Associate Professor of Criminology and Criminal Justice at the University of Nebraska Omaha. Her interests include youth street gangs, criminological theory, social network analysis, and quantitative methods.