Neighborhood Context,Youthful Offending,and Peer Selection

Abstract

The purpose of this study was to determine whether neighborhood context, as measured by maternal ratings of neighborhood disorder, predicted future offending behavior and association with antisocial peers. Putative relationships between neighborhood context, peers, and offending behavior were tested in all 1,725 adolescent members of the National Youth Survey. Consistent with the hypothesis that neighborhood context can either promote or inhibit the delinquent peer selection process, neighborhood disorder predicted peer selection (own offending → delinquent peers) but not peer influence (delinquent peers → own offending). Gender moderated this relationship such that an effect was found only for boys. These findings indicate that a negative psychological environment resulting from neighborhood disorder and weak informal social control can encourage offending behavior in neighborhood boys and facilitate the selection of antisocial peers. As with the parent–peer selection relationship observed in an earlier study on parental role modeling and youth offending, neighborhood climate has a bearing on future offending by virtue of its ability to support or block conditions known to suppress early delinquency and the formation of antisocial peer networks.

Keywords

neighborhood disorder peer selection peer influence mediation analysis

With the advent of the industrial revolution came a reduction in the informal social control that helped keep the crime rate low in colonial America. Historians trace the loss of collective informal social control to a shift in the geographic distribution of the American population as people began moving from the small towns and villages into the cities and urban areas to support the expanding industrial complex (Gurr, 1989). Informal social control accordingly weakened and the crime rate grew during the early part of the 19th century (Lane, 1980). What should be kept in mind, however, is that while urbanization reduced informal social control, it did not eliminate it. There were still pockets of informal social control even in some of the largest U.S. cities. These pockets of informal social control were often organized along racial/ethnic lines and were frequently marked by strong citizen involvement in the life of the community. This research, which has been reviewed and analyzed by Sampson, Raudenbush, and Earls (1997), indicates that informal social control, collective efficacy, and neighborhood disorder explain why some parts of a city experience higher or lower rates of crime than other parts of a city. The question the current investigation seeks to answer is whether or not individual appraisals of neighborhood context relate in a meaningful way to individual-level behaviors like offending and peer associations and such crime-relevant processes as peer selection and peer influence.

Neighborhood Context and Delinquency

The Chicago school’s social disadvantage approach to crime was one of the first scholarly attempts to establish a link between neighborhood context and delinquency. In outlining the tenets of the social disadvantage model, Shaw and McKay (1942) stated that areas located closest to commerce and heavy industry often had the lowest economic status, the greatest proportion of condemned buildings, and the highest crime rates. Building on the invasion–dominance–succession paradigm popular in biology at the time, they described how neighborhoods and crime formed concentric patterns of growth in Chicago and other major metropolitan areas. They further discovered that crime remained high in those areas even after the original inhabitants were replaced by a new group of immigrants. By the 1960s, the social disadvantage approach had fallen out of favor, due largely to the discouraging results recorded by neighborhood intervention programs and a general sense that neighborhood effects were weak (Kornhauser, 1978). Some 20 years later, however, revivification of the social disorganization model began to take root and neighborhood context research experienced a rebirth (Krohn, 2000). Current views on neighborhood context are summarized by Sampson, Morenoff, and Gannon-Rowley (2002) and suggest the presence of four types of neighborhood mechanisms relevant to delinquency development: social ties, collective efficacy, institutional resources, and routine activities.

The first-two mechanisms identified by Sampson et al. (2002), social ties and collective efficacy, may be particularly important in predicting and preventing delinquency. Together, they reflect informal social control, which, as I stated earlier, is reduced but not eliminated by urbanization. Informal social control and family ties, in fact, are a major reason why Asian crime rates, even in large cities, are low (Karstedt, 2001). In fact, neighborhood racial homogeneity in general may provide protection against crime (Guerrero, 2009). Neighbors who look out for one another, keep an eye on one another’s property, and are willing to watch out for and correct each other’s children provide strong informal social control even in an urban setting. In turn, these neighborhoods enjoy significantly lower crime rates than neighborhoods characterized by weak informal social control (Sampson, Raudenbush, & Earls, 1997). The activation of shared expectations for action through the development of social ties, a process known as collective efficacy, has been found to be a particularly effective deterrent to crime (Morenoff, Sampson, & Raudenbush, 2001; Sampson et al., 1997; Warner, 2014). A breakdown in neighborhood informal social control, a contextual effect commonly referred to as neighborhood disorder, by contrast, has been found to predict antisocial behavior and violence in youth (Huang & Ryan, 2014; Pardini, Loeber, Farrington, & Stouthamer-Loeber, 2012). An issue that remains unsettled and therefore unexplained is how neighborhood context overlaps with other criminogenic factors like peer associations to put certain individuals at risk for future criminality.

Peer Selection and Influence

Peer effects can be divided into two broad categories: peer selection and peer influence. The peer selection effect is marked by a pattern of offending behavior leading to delinquent peer associations (own offending → delinquent peers). The peer influence effect, on the other hand, is characterized by a pattern in which delinquent peer associations precede offending behavior (delinquent peers → own offending). Although peer selection is a major feature of Gottfredson and Hirschi’s (1990) general theory of crime (“birds of a feather”), peer influence is the guiding construct in Sutherland’s (1947) differential association model (learning from peers). Research recently conducted on the peer–parenting interface reveals that a strong parental role model may inhibit the peer selection effect but not the peer influence effect, presumably by blocking the initial step or stage of the peer selection process, namely, own offending (Walters, 2015c). Other research indicates that the direct parental control designed to instill a strong moral value system in the child (inductive discipline) can promote a peer influence effect but not a peer selection effect (Walters, 2015b). Identification with a respected adult role model, particularly a same-sex parental role model, would appear to protect against future delinquency, presumably by creating an emotional bond between the parent and child, which can then be used to provide informal social control. A similar process may be responsible for the dampening effect of neighborhood cohesion and collective efficacy on offending and the peer selection effect.

Haynie and Osgood (2005) contend that perceptual measures of peer delinquency completed by the participant rather than by the peer overestimate the relationship between own and peer delinquency because juveniles select peers as friends who are similar to them and because their perceptions of peer delinquency may be colored by their own delinquency. Weerman and colleagues, in fact, believe that youth project their own level of delinquency onto their peers, thereby giving the appearance of greater concordance in the delinquency of self and peers than is actually the case (Megens & Weerman, 2012; Weerman, 2011; Young & Weerman, 2013). Psychological research on self-serving bias comes to a very different conclusion on the relationship between peer and own externalizing behavior, however. In this research, perceptual ratings of peer externalizing behavior have been found to be more accurate than self-reports of externalizing behavior by virtue of each measure’s correlation with outside criteria (i.e., teacher ratings and behavioral observation). This suggests that juveniles are more apt to underreport their own externalizing behavior than overestimate the externalizing behavior of their peers (Henry, 2006; Peets & Kikas, 2006). In short, the relationship between peer and own delinquency is complex and may vary depending on the methodology used (perceptions of peer delinquency vs. actually reports from peers), but there is no indication at this point in time that one methodology yields more accurate results than the other.

There is some evidence that peer effects are stronger in boys than in girls (Bowman, Prelow, & Weaver, 2007; Jensen, 2003; Piquero, Gover, MacDonald, & Piquero, 2005; Walters, 2014; but see Weerman & Hoeve, 2012, for an alternate view). Sex, in fact, appeared to play a moderating role in the two previously reviewed parenting–peer studies. In Walters (2015c), for instance, boys who identified with their biological fathers experienced the lowest rates of offending and the weakest peer selection effects of any of the male subsamples in the study, whereas girls who identified with an adult female relative experienced the lowest rates of offending and the weakest peer selection effects of any of the female subsamples. In Walters (2015b) inductive parenting constrained the peer influence effect in boys but not in girls. These results support the gendered pathways theory of crime, which holds that males and females follow different paths to delinquency and crime (Chesney-Lind & Palko, 2004). Accordingly, variables instrumental in initiating and maintaining delinquency in boys may be largely ineffective in initiating and maintaining delinquency in girls, and variables instrumental in initiating and maintaining delinquency in girls may be largely ineffective in initiating and maintaining delinquency in boys. Moderation by sex was therefore evaluated in the current study by creating and testing interaction effects between the independent variable and sex and conducting separate analyses by sex if any of the interactions were significant.

The Current Study

There are at least two reasons to suspect that neighborhood context is more closely tied to offending and the peer selection process than to delinquent associations and the peer influence process. First, the peer selection effect normally precedes the peer influence effect developmentally given that one must select a peer group before being influenced by it. Shaw and McKay (1942) likewise theorized that neighborhood context had its greatest impact early in the delinquent career during the initial stages of offending (peer selection) and that developing a criminal identity and adopting the moral values of the criminal group (peer influence) do not occur until later in the delinquent career. Second, research indicates that informal social control, the primary mechanism by which neighborhoods prevent youthful offending, does a better job of inhibiting initial offending than it does of disrupting peer associations (Walters, 2015c), whereas an internalized moral value system appears to exert its greatest effect on offending by inoculating youth against peer influence rather than directly impeding initial offending (Walters, 2015b). The current study correlated neighborhood disorder with the peer selection and peer influence effects in an effort to determine whether neighborhood disorder has a significantly stronger impact on peer selection than on peer influence.

Measures of the four primary neighborhood mechanisms of informal social control identified by Sampson et al. (2002)—social ties, collective efficacy, institutional resources, and routine activities—were unavailable in the data set used in the current study. In fact, none of the mechanisms were available in this data set. Consequently, a major assumption made at the beginning of the study was that neighborhood disorder, a variable that was available in the data set, was an effective proxy for weak neighborhood informal social control. This assumption is often made in research on neighborhood disorder and crime (Skogan, 2012). Hence, brittle social ties, low collective efficacy, sparse institutional resources, and routine activities that do not protect against crime lead directly to neighborhood disorder. The hypothesis tested in this study held that neighborhood disorder would predict the peer selection effect (offending → peers) but not the peer influence effect (peers → offending) and that the difference between the two paths would be statistically significant. The significance of this research is that if neighborhood disorder is an effective proxy for weak informal neighborhood social control and if neighborhood disorder is linked to peer selection but not peer influence, then the results should have important implications for clinical intervention and theory development and integration.

Method

Participants

The sample for this study consisted of all 1,725 (918 boys and 807 girls) members of the nationally representative National Youth Survey (NYS; Elliott, 1976/1980). The average participant was 13.87 years old (SD = 1.94, range = 11–17) at the start of the study (Wave 1) and the racial breakdown for the sample was 78.9% White, 15.1% Black, 4.4% Hispanic, 1.0% Asian, 0.5% American Indian, and 0.2% Other. Data from the first-three waves of the NYS were analyzed in this study.

Measures

The independent variable for this study was neighborhood disorder as rated by the child’s parent or guardian (87.6% mother, 7.1% father, 1.9% stepmother, 0.4% stepfather, and 3.0% other relative). Eight potential types of neighborhood disorder were assessed in the NYS (vandalism, winos and junkies, traffic, abandoned houses, burglaries and thefts, run down and poor buildings, assaults and muggings, and children afraid to walk to school). The parental respondent was asked to estimate how much of a problem each of the 8 items were in their particular neighborhood using a 3-point scale (1 = no problem, 2 = somewhat of a problem, and 3 = big problem). This produced a neighborhood disorder scale that ranged from 8 to 24 and demonstrated adequate internal consistency (α = .74).

Youth ratings of own delinquency and peer delinquency were cross lagged to form the mediator and dependent variables for this study. Own delinquency (offend) was evaluated using a self-report measure of the 12 most serious crimes in the NYS, that is, damaged property, stole a vehicle, stole more than US$50, bought or sold stolen property, attacked someone in an attempt to hurt or kill them, sold marijuana, sold hard drugs, hit a teacher, hit another student, coerced someone into sexual activity, committed strong-arm robbery, and broke into a building. Items were assessed on a 9-point frequency scale (1 = not in the last year, 2 = 1–2 times a year, 3 = once every 2–3 months, 4 = once a month, 5 = once every 2–3 weeks, 6 = once a week, 7 = 2–3 times a week, 8 = once a day, and 9 = 2–3 times a day). Summing the 12 items produced a score that could potentially range from 12 to 108.

Peer delinquency (peer) was the other mediator/dependent variable included in this study. In completing this measure, juvenile respondents estimated the proportion of friends who engaged in the following seven delinquent acts during the past year (cheated on tests, destroyed property, stole less than US$5, stole more than US$50, hit someone, broke into a vehicle, and sold hard drugs). Each delinquent act was rated on a 5-point scale (1 = none of them, 2 = very few of them, 3 = some of them, 4 = most of them, and 5 = all of them). When summed, these items produced a scale that ranged in score between 7 and 35, with adequate internal consistency (α = .76 for Wave 2 and α = .79 for Wave 3). It should be noted that while the independent variable (neighborhood disorder) was based on parental reports, the mediator and dependent variables (own and peer offending) were based on child reports.

Participant age, race (White = 1 and non-White = 2), sex (male = 1 and female = 2), parental marital status (married = 1 and not married = 0), and annual income (US$6,000 or less = 1, US$6,001–10,000 = 2, US$10,001–14,000 = 3, US$14,001–18,000 = 4, US$18,001–22,000 = 5, US$22,001–26,000 = 6, US$26,001–30,000 = 7, US$30,001–34,000 = 8, US$34,001–38.000 = 9, and US$38,001 or more = 10) all served as control variables in this study. The interaction between neighborhood disorder and sex was included in the full sample path analysis to test whether sex moderated the relationship between neighborhood disorder and each of the mediating and dependent variables. Cole and Maxwell (2003) contend that prior measures of the outcome (mediator or dependent) variables should be included in the regression equations of a path analysis whenever possible. Accordingly, Offend-1 and Peer-1 were incorporated into the regression equations predicting Offend-2 and Peer-2, respectively.

Procedure

The first-three waves of the NYS were included in this study. Age, race, sex, parental marital status, annual income, the two precursor measures (Offend-1 and Peer-1), and the independent variable (neighborhood disorder) were collected at Wave 1; the mediators (Offend-2 and Peer-2) were collected at Wave 2; and the dependent variables (Peer-3 and Offend-3) were collected at Wave 3. The peer selection effect was operationally defined as the pathway running from own delinquency at Wave 2 (Offend-2) to peer delinquency at Wave 3 (Peer-3), and the peer influence effect was operationally defined as the pathway running from peer delinquency at Wave 2 (Peer-2) to own delinquency at Wave 3 (Offend-3). The cross-lagged correlations for the peer selection (Offend-2 → Peer-3) and peer influence (Peer-2 → Offend-3) effects served as the partner effects and the offending (Offend-2 → Offend-3) and peer (Peer-2 → Peer-3) autocorrelations served as the actor effects in an actor–partner interdependence research design (Kenny, Kashy, & Cook, 2006). The independent variable (neighborhood disorder) was then regressed onto the two partner effects and two actor effects.

Four regressions in all were computed for this study (one with Offend-2 as the outcome, one with Peer-2 as the outcome, one with Offend-3 as the outcome, and one with Peer-3 as the outcome). Analyses were performed with the structural equation modeling program MPlus 5.2 (Muthén & Muthén, 1998–2007) using a maximum likelihood estimator. A pathway was considered significant if the bootstrapped (b = 5,000 with replacement) bias-corrected 95% confidence interval (CI) did not include zero. Bootstrapped bias-corrected 95% CIs are currently considered the best method for evaluating the significance of indirect effects in mediation analysis (Hayes, 2013; Rucker, Preacher, Tormala, & Petty, 2011). Preacher and Hayes’ (2008) contrast method was used to determine whether neighborhood disorder predicted the peer selection effect (Offend-2 → Peer-3) significantly better than it predicted the peer influence effect (Peer-2 → Offend-3). Sensitivity was tested with Kenny’s (2013) “failsafe ef” procedure: (r_my.x) × (SD_m.x) × (SD_y.x)/(SD_m) × (SD_y).

Missing Data

Complete data were available for 70.3% of the sample. An additional 17.6% and 6.1% had missing data on one and two variables, respectively. The rest of the sample (5.9%) had missing data on 3–7 variables. Full information maximum likelihood was used to calculate model parameters and standard errors from estimated likelihood functions derived from observed relationships between the nonmissing data.

Results

Descriptive statistics and correlations for the independent variable, two mediator variables, two dependent variables, and seven control/precursor variables are summarized in Table 1. There was no evidence of multicollinearity in the set of 10 predictor variables used in this study: tolerance = .526 to .954 and variance inflation factor (VIF) = 1.048–1.901.

Table 1.
Descriptive Statistics and Correlations for the 12 Variables From the Current Study.

Variable n M SD Range 2 3 4 5 6 7 8 9 10 11 12

1. Child’s age 1,725 13.87 1.94 11–17 −.05 −.04 −.02 .05 −.07 .02 .14* .06 .12* .03 .01

2. Child’s sex 1,725 1.47 0.50 1–2 −.03 −.02 .04 .01 −.12* −.21* −.25* −.24* −.26* −.19*

3. Child’s race 1,725 1.21 0.41 1–2 −.27* −.37* .23* .05 .08 .02 .07 −.00 .02

4. Parental marital status 1,683 0.81 0.39 0–1 .41* −.20* −.11* −.09* −.05 −.10* −.08 −.09*

5. Family income 1,617 4.16 2.31 1–10 −.14* −.07 −.10* −.06 −.05 −.07 −.05

6. Neighborhood disorder 1,683 9.99 2.35 8–22 .10* .07 .07 .09 .08 .11*

7. Own delinquency W1 1,725 14.08 6.33 12–108 .48* .27* .20* .24* .15*

8. Peer delinquency W1 1,490 11.30 3.70 7–30 .34* .54* .32* .50*

9. Own delinquency W2 1,655 13.46 3.19 12–64 .56* .71* .43*

10. Peer delinquency W2 1,497 11.14 3.62 7–35 .44* .59*

11. Own delinquency W3 1,626 13.38 3.04 12–56 .50*

12. Peer delinquency W3 1,619 11.04 3.65 7–33

Note. Child’s age = child’ age in years at Wave 1; Child’s sex = 1 (male) or 2 (female); Child’s race = 1 (White) or 2 (non-White); Parental Marital Status = parents’ marital status at Wave 1; Family income = family income measured with a 10-point scale at Wave 1; Neighborhood disorder = parent’s rating of degree of disorder in the neighborhood at Wave 1; Own delinquency W1 = child’s report of own delinquency at Wave 1; Peer delinquency W1 = child’s report of peer delinquency at Wave 1; Own delinquency W2 = child’s report of own delinquency at Wave 2; Peer delinquency at Wave 2 = child’s report of peer delinquency at Wave 2; Own delinquency W3 = child’s report of own delinquency at Wave 3; Peer delinquency W3 = child’s report of peer delinquency at Wave 3; n = number of nonmissing cases; M = mean; SD = standard deviation; Range = range of scores in current sample.

p < .00075 (Bonferroni-corrected α: .05/66 correlations).

A four regression path analysis of the full sample (see Table 2) indicated that neighborhood disorder predicted the peer selection effect and not the peer influence effect, but that the two pathways were not significantly different from one another. Because sex moderated the neighborhood disorder → Offend-2 pathway (Sex × Neighborhood interaction, p < .05), separate path analyses were calculated for boys and girls.

Table 2.
Path Analysis of Neighborhood Disorder as a Predictor of Peer Selection and Influence in Total Sample.

Predictor b (95% CI) β t p

Peer-2 on

Neighborhood disorder 0.084 [−0.008, 0.172] .055 1.83 .068

Child age 0.145 [0.066, 0.227] .078 3.51 <.001

Child sex −1.200 [−1.529, −0.862] −.166 −7.06 <.001

Child race 0.139 [−0.346, 0.614] .016 0.57 .571

Parental marital status −0.576 [−1.056, −0.124] −.063 −2.39 .017

Parental income level 0.040 [−0.033, 0.111] .026 1.09 .276

Peer-1 0.322 [0.239, 0.415] .386 7.06 <.001

Sex × Neighborhood −0.007 [−0.219, 0.193] −.002 −0.06 .950

Offend-2 on

Neighborhood disorder 0.072 [−0.003, 0.150] .052 1.84 .066

Child age 0.081 [0.023, 0.151] .049 2.48 .013

Child sex −1.388 [−1.661, −1.125] −.215 −10.11 <.001

Child race −0.151 [−0.652, 0.455] −.019 −0.54 .588

Parental marital status −0.008 [−0.443, 0.439] −.001 −0.03 .973

Parental income level −0.048 [−0.106, 0.008] −.034 −1.61 .107

Offend-1 0.143 [0.089, 0.251] .280 3.08 .002

Sex × Neighborhood −0.183 [−0.365, −0.033] −.057 −2.18 .029

Peer-3 on

Offend-2 0.150 [0.069, 0.248] .133 3.31 .001

Neighborhood disorder 0.087 [0.010, 0.167] .056 2.21 .027

Child age −0.125 [−0.196, −0.058] −.067 −3.52 <.001

Child sex −0.325 [−0.628, −0.021] −.044 −2.09 .037

Child race −0.352 [−0.780, 0.072] −.039 −1.62 .104

Parental marital status −0.196 [−0.641, 0.210] −.021 −0.90 .369

Parental income level −0.013 [−0.088, 0.056] .008 −0.37 .714

Peer-2 0.524 [0.451, 0.598] .518 13.98 <.001

Sex × Neighborhood −0.089 [−0.268, 0.083] −.024 −1.00 .320

Offend-3 on

Peer-2 0.049 [0.003, 0.095] .058 2.08 .038

Neighborhood disorder 0.040 [−0.012, 0.105] .031 1.33 .185

Child age −0.032 [−0.079, 0.017] −.020 −1.30 .194

Child sex −0.450 [−0.649, −0.239] −.073 −4.28 <.001

Child race −0.391 [−0.734, −0.016] −.052 −2.14 .032

Parental marital status −0.297 [−0.676, 0.066] −.038 −1.58 .114

Parental income level −0.022 [−0.071, 0.026] −.017 −0.89 .372

Offend-2 0.626 [0.498, 0.740] .662 10.03 <.001

Sex × Neighborhood −0.073 [−0.226, 0.046] −.024 −1.08 .282

Peer-2 with offend-2 4.84 [3.469, 6.954] .514 5.57 <.001

Peer-3 with offend-3 1.74 [1.321, 2.237] .289 7.50 <.001

Note. N* = 1,725. Peer-2 on = regression equation with Wave 2 peer delinquency as the dependent variable; Offend-2 on = regression equation with Wave 2 own delinquency as the dependent variable; Peer-3 = regression equation with Wave 3 peer delinquency as the dependent variable; Offend-3 on = regression equation with Wave 3 own delinquency as the dependent variable; Child sex = child’s sex coded as 1 (male) versus 2 (female); Child race = child’s race coded as 1 (White) versus 2 (non-White); Parental marital status = parents married (1) versus not married (0); Offend-1 = own delinquency at Wave 1; Peer-1 = peer delinquency at Wave 1; Sex × Neighborhood = interaction between sex of child and neighborhood disorder; Peer-2 with Offend-2 = correlation between peer delinquency and own delinquency at Wave 2; Peer-3 with Offend-3 = correlation between peer delinquency and own delinquency at Wave 3; b (95% CI) = unstandardized coefficient and the lower and upper limits of the 95% confidence interval for the unstandardized coefficient (in parentheses); β = standardized coefficient; t = asymptotic t-test (standard z-test); p = significance level of the asymptotic t-test.

The path analysis for boys revealed that neighborhood disorder did a significantly better job of predicting peer selection than it did of predicting peer influence, and that while the former achieved significance, the latter did not (see Table 3 and Figure 1). Neither path nor the contrast test was significant when the analysis was restricted to girls (see Table 4 and Figure 2).

Table 3.
Neighborhood Disorder as a Predictor of Peer Selection and Influence in Boys.

Cross-Lagged Pairs Estimate BCBCI (Lower, Upper)

Peer selection

Neighbor-1 → Offend-2 → Peer-3 0.023 [0.005, 0.059]

Neighbor-1 → Peer-2 → Peer-3 0.047 [−0.007, 0.111]

Peer influence

Neighbor-1 → Peer-2 → Offend-3 0.005 [−0.001, 0.019]

Neighbor-1 → Offend-2 → Offend-3 0.097 [0.029, 0.190]

Preacher and Hayes contrast

Peer selection versus peer influence 0.018 [0.001, 0.052]

Note. n = 918. Neighbor-1 = neighborhood disorder at Wave 1; Peer-2 = peer delinquency at Wave 2; Peer-3 = peer delinquency at Wave 3; Offend-2 = own offending at Wave 2; Offend-3 = own offending at Wave 3; Preacher and Hayes contrast = comparison between peer selection and peer influence effects; BCBCI = bias-corrected bootstrapped 95% confidence interval (b = 5,000); Estimate = point estimate; Lower = lower boundary of the 95% confidence interval; Upper = upper boundary of the 95% confidence interval.

Figure 1.
Path analysis of neighborhood disorder as a predictor of the cross-lagged peer selection (offend → peer) and peer influence (peer → offend) effects in adolescent boys. Standardized β coefficients are reported. Neighbor = neighborhood disorder; Offend = own offending; Peer = peer delinquency. Control variables (age, race, parental marital status, family income, and offending and peer precursors) are not shown. n = 918. *p < .05. **p < .001.

Table 4.
Neighborhood Disorder as a Predictor of Peer Selection and Influence in Girls.

Cross-Lagged Pairs Estimate BCBCI (Lower, Upper)

Peer selection

Neighbor-1 → Offend-2 → Peer-3 0.000 [−0.025, 0.027]

Neighbor-1 → Peer-2 → Peer-3 0.037 [−0.031, 0.109]

Peer influence

Neighbor-1 → Peer-2 → Offend-3 0.004 [−0.003, 0.017]

Neighbor-1 → Offend-2 → Offend-3 0.000 [−0.024, 0.025]

Preacher and Hayes contrast

Peer selection versus peer influence −0.004 [−0.027, 0.021]

Note. n = 807. Neighbor-1 = neighborhood disorder at Wave 1; Peer-2 = peer delinquency at Wave 2; Peer-3 = peer delinquency at Wave 3; Offend-2 = own offending at Wave 2; Offend-3 = own offending at Wave 3; Preacher and Hayes contrast = comparison between peer selection and peer influence effects; BCBCI = bias-corrected bootstrapped 95% confidence interval (b = 5,000); Estimate = point estimate; Lower = lower boundary of the 95% confidence interval; Upper = upper boundary of the 95% confidence interval.

Figure 2.
Path analysis of neighborhood disorder as a predictor of the cross-lagged peer selection (offend → peer) and peer influence (peer → offend) effects in adolescent girls. Standardized β coefficients are reported. Neighbor = neighborhood disorder; Offend = own offending; Peer = peer delinquency. Control variables (age, race, parental marital status, family income, and offending and peer precursors) are not shown. n = 807. *p < .05. **p < .001.

Sensitivity testing using Kenny’s (2013) failsafe ef procedure revealed that an unobserved covariate confounder would need to correlate .42 with both the mediator and dependent variables to eliminate the significant mediating effect of Offend-2 on the Neighbor-1 → Peer-3 relationship.

Discussion

The results of this study are consistent with the conclusion that a neighborhood can facilitate or inhibit the peer selection process. When neighborhood context was assessed with estimates of neighborhood disorder made by the parents of the 1,725 members of the NYS, the data indicated that neighborhood disorder supported the peer selection effect but not the peer influence or socialization effect. These outcomes were obtained using a procedure (bias-corrected bootstrapping) that does a better job of modeling the nonnormality of indirect effects than the standard z-test approach (Rucker et al., 2011). Furthermore, sensitivity testing revealed that the significant mediating effect from this study was highly robust to the obfuscating effects of unobserved covariate confounders. A connection between neighborhood disorder and the peer selection effect confirms predictions made at the beginning of the study and is congruent with the theory that the neighborhood context, measured individually, can either facilitate or hinder criminal offending in neighborhood children. Theories of neighborhood context (Sampson et al., 2002) assume that neighborhood context achieves its effect by promoting a sense of informal social control, which is often viewed to be composed of two parts: (1) formation of emotional ties with the community—similar to the parent–child emotional bond believed to inhibit delinquency (Hirschi, 1969)—and (2) increased surveillance and supervision of children. These two components of informal social control work hand in hand to limit opportunities and incentives for delinquency within the neighborhood.

It should be noted that a significant interaction effect surfaced between neighborhood disorder and sex in the regression equation predicting the originating source of the peer selection effect (i.e., own offending at Wave 2). This would seem to suggest that neighborhood disorder encouraged initial offending in boys but not in girls. Sex as a moderator of the neighborhood–peer selection relationship is consistent with the bulk of prior research on sex moderation of peer deviance and delinquency in general (Bowman et al., 2007; Jensen, 2003; Piquero et al., 2005; Walters, 2014) and offers support for the gendered pathways model of crime initiation and maintenance (Chesney-Lind & Palko, 2004). Because there are points of both commonality and difference in male and female offending, sex moderation of crime-based relationships should probably be evaluated and ruled out prior to analyzing male and female data collectively. If a gender invariance or moderation analysis reveals that the relationships in question vary as a function of sex, then male and female data should probably be analyzed separately as was done in the current study. With respect to preventing female offending, it may be advisable to consider the neighborhood–parenting relationship given research suggesting that girls are more responsive to parenting effects than boys (Silverman & Caldwell, 2005; Walters, 2013).

There are both practical and theoretical implications to the current results. In developing crime prevention programs, for instance, it is vital that we find ways to bridge traditional group influences like neighborhood disorder with traditional individual factors like peer associations. The results of the present study indicate that neighborhoods may be capable of preventing crime in neighborhood children by reducing opportunities for early juvenile offending and providing for increased surveillance of neighborhood children. A strong network of neighborhood informal social control achieves its effect, not by inhibiting the peer socialization effect but by reducing the individual’s chances of selecting a deviant peer group to associate with in the first place. In this way, neighborhood informal social control can perhaps be considered a primary prevention approach. The principal implication the current study has for theory is that it demonstrates the feasibility and advisability of integrating variables and ideas from divergent models: in this case, a molar construct from the Chicago school (neighborhood informal social control) with a molecular construct from the general theory of crime (peer selection). This suggests that further integration is not only possible but necessary if we hope to achieve a comprehensive understanding of crime development and desistance.

As alluded to earlier, the results of this study are more consistent with Gottfredson and Hirschi’s (1990) contention that low self-control children select one another as associates than they are with Sutherland’s (1947) contention that children learn to become criminal by associating with those already involved in crime. At least in boys, neighborhood context correlated better with peer selection and its originating factor (own delinquency) than to peer influence and its originating factor (peer delinquency). This does not mean, however, that peer influence is any less important than peer selection in the development of a criminal lifestyle. It is just that the two effects appear to relate to different sets of variables and different developmental stages. There is evidence, for instance, that peer influence normally occurs at a later point in time than peer selection (Monahan, Steinberg, & Cauffman, 2009). Perhaps the best way to conceive of peer selection and influence is that peers must be selected before than can serve as agents of socialization or influence. Although peer selection is constrained by informal social control, social support, and active surveillance, peer influence is deterred by a strong moral code acquired through association with prosocial parents or guardians. Consequently, neighborhood context exerts both an indirect and direct effect on the behavior of adolescents, forging an emotional bond with the child while at the same time providing surveillance and monitoring of youth behavior around the neighborhood. It would appear, then, that the crime-dampening effect of neighborhood context is achieved through heightened informal social control created by surveillance and neighborhood watch programs as well as the formation of a person–community bond, both of which inhibit early offending and the peer selection effect. Informal social control should accordingly be the focus of neighborhood programs and policies.

One of the principal reasons why social disorganization theory lost adherents during the 1960s and 1970s was the perception that neighborhood interventions achieved weak results (Kornhauser, 1978). Traditionally, neighborhood disorder was assessed with structural variables (poverty level, concentrated disadvantage, and residential mobility), and intervention programs were designed to provide social support (Bursik, 1988). Since then, constructs like neighborhood cohesion and collective efficacy have been introduced to explain how structural variables affect the neighborhood–crime relationship. The informal social control that has always been at the heart of social disorganization theory is now viewed to be composed of two parts: direct surveillance and social support. Direct surveillance and situational crime control have been found effective in reducing neighborhood crime (Clarke, 1995), but social support has not been as extensively studied by proponents of social disorganization theory based on the belief that social support is relatively ineffective in promoting change (see Cullen, 1994). The relative ineffectiveness of Shaw and McKay’s (1942) original approach to neighborhood intervention, however, may have had less to do with its emphasis on social support than with the general absence of available resources for social support. Warner, Beck, and Ohmer (2010) propose that recent advances in restorative justice, reintegrative shaming, and peacemaking criminology have the potential to provide the resources required to create social support in many urban neighborhoods. Data from the current study suggest that one avenue by which neighborhood cohesion, collective efficacy, and social support might help manage crime is by reducing early offending behavior in youthful members of the community, preventing infiltration of gangs and delinquent role models into the neighborhood, and disrupting the overall peer selection process.

The individual-level approach utilized in this study whereby neighborhood context was measured with responses provided by participants’ parents rather than by the aggregate neighborhood can be considered both a strength and a weakness of this study. It is a strength in the sense that it allowed a normally collective process (i.e., neighborhood context) to be studied in conjunction with other individual-level variables, such as peer effects. This is a necessary step in creating an integrated theory of criminal behavior, in which neighborhood effects play a role. The individual-level approach adopted in this study is also a weakness to the extent that this is not how neighborhood context is normally conceptualized or measured. As such, the generalizability and relevance of the current results to the overall neighborhood context literature is uncertain. The fact that all variables were assessed at the individual level may have confounded the results to some extent and created patterns different than what would have been found had an aggregate measure of neighborhood context been used. Furthermore, a theory constructed solely on the basis of individual-level data may be untenable. The challenge confronting those conducting research in this area is finding a way to integrate the individual and collective factors described in various theories of criminology into a single model.

There are several additional limitations that should be taken into account when interpreting the results of this study. First, the study’s external validity may be limited by the fact that these data were collected over 30 years ago. Variables and relationships may have changed substantially in the past three decades, which, in turn, raises serious questions about the relevance of these data to modern day youth. Between 1976, the year the NYS began, and 2010, the year of the most recent U.S. census (U.S. Census Bureau, 2011), the proportion of White youth in the U.S. population fell from 78.9% to 63.7%, the proportion of Black youth fell from 15.1% to 12.2%, and the proportion of Hispanic youth rose from 4.4% to 16.4%. Second, the study’s internal validity is potentially limited by a major assumption made at the beginning of this study. As the reader may recall, I based selection of neighborhood disorder as the independent variable for this study on the assumption that it was an adequate proxy for weak neighborhood informal social control, as represented by social ties, collective efficacy, institutional resources, and routine activities. This assumes that events catalogued by the neighborhood disorder measure (vandalism, winos and junkies, traffic, abandoned houses, burglaries and thefts, run down and poor buildings, assaults and muggings, and children afraid to walk to school) were the direct result of weak neighborhood informal social control. If the neighborhood disorder variable is measuring something in addition to or other than weak neighborhood informal social control, then this presents a threat to the internal validity of the current results.

The development of a truly integrated theory requires that individual- and aggregate-level data be assimilated, something that is easier said than done. Lack of integration, in fact, is endemic to criminology, where the majority of theories confine themselves to a single level, if not a single variable (Ericson & Carriere, 1994). Greater integration, it has been argued, is required for the overall health of the field (Agnew, 2011; Bernard & Snipes, 1996). The current study contributes to this health by adding neighborhood context to the list of variables that have been successfully integrated with peer effects in the formation of a comprehensive theory of crime. The other variables on this list include direct and indirect parenting (Walters, 2015b; 2015c) and proactive and reactive criminal thinking (Walters, 2015a). Future research integrating all four sets of variables—neighborhood context, parenting, criminal thinking, and peer effects—is therefore required before the theory can be considered sufficiently integrated to be of use to scholars, practitioners, and policy makers. Although effective integration of individual- and group (aggregate)-level data presents a major challenge to researchers and theoreticians, it is possible given the proper measures, methods, and statistical procedures (Sabiston et al., 2009). A study that examines individuals within the context of certain naturally occurring groups (neighborhoods), gathers data at both the individual and aggregate levels, and analyses these data using a statistical procedure like multilevel modeling could provide the level of quality information capable of moving this line of research forward.

Variable	n	M	SD	Range	2	3	4	5	6	7	8	9	10	11	12
1. Child’s age	1,725	13.87	1.94	11–17	−.05	−.04	−.02	.05	−.07	.02	.14*	.06	.12*	.03	.01
2. Child’s sex	1,725	1.47	0.50	1–2		−.03	−.02	.04	.01	−.12*	−.21*	−.25*	−.24*	−.26*	−.19*
3. Child’s race	1,725	1.21	0.41	1–2			−.27*	−.37*	.23*	.05	.08	.02	.07	−.00	.02
4. Parental marital status	1,683	0.81	0.39	0–1				.41*	−.20*	−.11*	−.09*	−.05	−.10*	−.08	−.09*
5. Family income	1,617	4.16	2.31	1–10					−.14*	−.07	−.10*	−.06	−.05	−.07	−.05
6. Neighborhood disorder	1,683	9.99	2.35	8–22						.10*	.07	.07	.09	.08	.11*
7. Own delinquency W1	1,725	14.08	6.33	12–108							.48*	.27*	.20*	.24*	.15*
8. Peer delinquency W1	1,490	11.30	3.70	7–30								.34*	.54*	.32*	.50*
9. Own delinquency W2	1,655	13.46	3.19	12–64									.56*	.71*	.43*
10. Peer delinquency W2	1,497	11.14	3.62	7–35										.44*	.59*
11. Own delinquency W3	1,626	13.38	3.04	12–56											.50*
12. Peer delinquency W3	1,619	11.04	3.65	7–33

Predictor	b (95% CI)	β	t	p
Peer-2 on
Neighborhood disorder	0.084 [−0.008, 0.172]	.055	1.83	.068
Child age	0.145 [0.066, 0.227]	.078	3.51	<.001
Child sex	−1.200 [−1.529, −0.862]	−.166	−7.06	<.001
Child race	0.139 [−0.346, 0.614]	.016	0.57	.571
Parental marital status	−0.576 [−1.056, −0.124]	−.063	−2.39	.017
Parental income level	0.040 [−0.033, 0.111]	.026	1.09	.276
Peer-1	0.322 [0.239, 0.415]	.386	7.06	<.001
Sex × Neighborhood	−0.007 [−0.219, 0.193]	−.002	−0.06	.950
Offend-2 on
Neighborhood disorder	0.072 [−0.003, 0.150]	.052	1.84	.066
Child age	0.081 [0.023, 0.151]	.049	2.48	.013
Child sex	−1.388 [−1.661, −1.125]	−.215	−10.11	<.001
Child race	−0.151 [−0.652, 0.455]	−.019	−0.54	.588
Parental marital status	−0.008 [−0.443, 0.439]	−.001	−0.03	.973
Parental income level	−0.048 [−0.106, 0.008]	−.034	−1.61	.107
Offend-1	0.143 [0.089, 0.251]	.280	3.08	.002
Sex × Neighborhood	−0.183 [−0.365, −0.033]	−.057	−2.18	.029
Peer-3 on
Offend-2	0.150 [0.069, 0.248]	.133	3.31	.001
Neighborhood disorder	0.087 [0.010, 0.167]	.056	2.21	.027
Child age	−0.125 [−0.196, −0.058]	−.067	−3.52	<.001
Child sex	−0.325 [−0.628, −0.021]	−.044	−2.09	.037
Child race	−0.352 [−0.780, 0.072]	−.039	−1.62	.104
Parental marital status	−0.196 [−0.641, 0.210]	−.021	−0.90	.369
Parental income level	−0.013 [−0.088, 0.056]	.008	−0.37	.714
Peer-2	0.524 [0.451, 0.598]	.518	13.98	<.001
Sex × Neighborhood	−0.089 [−0.268, 0.083]	−.024	−1.00	.320
Offend-3 on
Peer-2	0.049 [0.003, 0.095]	.058	2.08	.038
Neighborhood disorder	0.040 [−0.012, 0.105]	.031	1.33	.185
Child age	−0.032 [−0.079, 0.017]	−.020	−1.30	.194
Child sex	−0.450 [−0.649, −0.239]	−.073	−4.28	<.001
Child race	−0.391 [−0.734, −0.016]	−.052	−2.14	.032
Parental marital status	−0.297 [−0.676, 0.066]	−.038	−1.58	.114
Parental income level	−0.022 [−0.071, 0.026]	−.017	−0.89	.372
Offend-2	0.626 [0.498, 0.740]	.662	10.03	<.001
Sex × Neighborhood	−0.073 [−0.226, 0.046]	−.024	−1.08	.282
Peer-2 with offend-2	4.84 [3.469, 6.954]	.514	5.57	<.001
Peer-3 with offend-3	1.74 [1.321, 2.237]	.289	7.50	<.001

Cross-Lagged Pairs	Estimate	BCBCI (Lower, Upper)
Peer selection
Neighbor-1 → Offend-2 → Peer-3	0.023	[0.005, 0.059]
Neighbor-1 → Peer-2 → Peer-3	0.047	[−0.007, 0.111]
Peer influence
Neighbor-1 → Peer-2 → Offend-3	0.005	[−0.001, 0.019]
Neighbor-1 → Offend-2 → Offend-3	0.097	[0.029, 0.190]
Preacher and Hayes contrast
Peer selection versus peer influence	0.018	[0.001, 0.052]

Cross-Lagged Pairs	Estimate	BCBCI (Lower, Upper)
Peer selection
Neighbor-1 → Offend-2 → Peer-3	0.000	[−0.025, 0.027]
Neighbor-1 → Peer-2 → Peer-3	0.037	[−0.031, 0.109]
Peer influence
Neighbor-1 → Peer-2 → Offend-3	0.004	[−0.003, 0.017]
Neighbor-1 → Offend-2 → Offend-3	0.000	[−0.024, 0.025]
Preacher and Hayes contrast
Peer selection versus peer influence	−0.004	[−0.027, 0.021]

Footnotes

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

References

Agnew

(2011). Toward a unified criminology: Integrating assumptions about crime, people, and society. New York, NY: New York University Press.

Bernard

T. J.

Snipes

J. B.

(1996). Theoretical integration in criminology. Crime and Justice, 20, 301–348.

Bowman

M. A.

Prelow

H. M.

Weaver

S. R.

(2007). Parenting behaviors, association with deviant peers, and delinquency in African American adolescents: A mediated moderation model. Journal of Youth and Adolescence, 36, 517–527.

Bursik

R. J.

Jr. (1988). Social disorganization and theories of crime and delinquency: Problems and prospects. Criminology, 26, 519–551.

Chesney-Lind

Palko

(2004). The female offender: Girls, women, and crime (2nd ed.). Thousand Oaks, CA: Sage.

Clarke

R. V.

(1995). Situational crime prevention. In Tonry

Farrington

(Eds.), Building a safer society: Strategic approaches to crime prevention. Crime and justice: A review of research (Vol. 19, pp. 91–150). Chicago, IL: University of Chicago Press.

Cole

D. A.

Maxwell

S. E.

(2003). Testing mediational models with longitudinal data: Questions and tips in the use of structural equation modeling. Journal of Abnormal Psychology, 112, 558–577.

Cullen

F. T.

(1994). Social support as an organizing concept for criminology: Presidential address to the academy of criminal justice sciences. Justice Quarterly, 11, 527–559.

Elliott

(1980). National youth survey (United States): Waves 1-5, 1976-1980. Ann Arbor, MI: Inter-University Consortium for Political and Social Research. (Original work published 1976)

10.

Ericson

Carriere

(1994). The fragmentation of criminology. In Nelken

(Ed.), The futures of criminology (pp. 89–109). London, England: Sage.

11.

Gottfredson

M. R.

Hirschi

(1990). A general theory of crime. Stanford, CA: Stanford University Press.

12.

Guerrero

(2009). Social disorganization theory. In Gabbidon

S. L.

Greene

H. T.

(Eds.), Encyclopedia of race and crime (pp. 762–763). Thousand Oaks, CA: Sage.

13.

Gurr

T. R.

(1989). Historical trends in violent crime: Europe and the United States. In Gurr

T. R.

(Ed.), Violence in America, vol. 1: The history of crime (pp. 21–54). Newbury Park, CA: Sage.

14.

Hayes

A. F.

(2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York, NY: Guilford.

15.

Haynie

D. L.

Osgood

D. W.

(2005). Reconsidering peers and delinquency: How do peers matter? Social Forces, 84, 1109–1130.

16.

Henry

D. B.

(2006). Associations between peer nominations, teacher ratings, self-reports, and observations of malicious and disruptive behavior. Assessment, 13, 241–252.

17.

Hirschi

(1969). The causes of delinquency. Berkeley, CA: University of California Press.

18.

Huang

Ryan

J. P.

(2014). The location of placement and juvenile delinquency: Do neighborhoods matter in child welfare? Children and Youth Services Review, 44, 33–45.

19.

Jensen

G. F.

(2003). Gender variation in delinquency: Self-image, beliefs and peers as mediating mechanisms. In Akers

R. L.

Jensen

G. F.

(Eds.), Advances in criminological theory, a guide for the new century: Social learning theory and the explanation of crime (pp. 151–177). New Brunswick, NJ: Transaction.

20.

Karstedt

(2001). Comparing cultures, comparing crime: Challenges, prospects, and problems for a global criminology. Crime, Law and Social Change, 36, 285–308.

21.

Kenny

D. A.

(2013). Mediation: Sensitivity analysis [webinar]. Retrieved from http://davidakenny.net/webinars/Mediation/Sensitivity/Sensitivity.html

22.

Kenny

D. A.

Kashy

D. A.

Cook

W. L.

(2006). Dyadic data analysis. New York, NY: Guilford.

23.

Kornhauser

R. R.

(1978). Social sources of delinquency. Chicago, IL: University of Chicago Press.

24.

Krohn

M. D.

(2000). Sources of criminality: Control and deterrence theories. In Sheley

J. F.

(Ed.), Criminology: A contemporary handbook (pp. 373–399). Belmont, CA: Wadsworth.

25.

Lane

(1980). Police and crime in nineteenth-century America. In Tonry

Morris

(Eds.), Crime and justice: An annual review (pp. 1–52). Chicago, IL: University of Chicago Press.

26.

Megens

K. C. I. M.

Weerman

F. M.

(2012). The social transmission of delinquency: Effects of peer attitudes and behavior revisited. Journal of Research in Crime and Delinquency, 49, 420–443.

27.

Monahan

K. C.

Steinberg

Cauffman

(2009). Affiliation with antisocial peers, susceptibility to peer influence, and antisocial behavior during the transition to adulthood. Developmental Psychology, 45, 1520–1530.

28.

Morenoff

J. D.

Sampson

R. J.

Raudenbush

S. W.

(2001). Neighborhood inequality, collective efficacy, and the spatial dynamics of homicide. Criminology, 39, 517–560.

29.

Muthén

(1998–2007). Mplus user’s guide (5th ed.). Los Angeles, CA: The Author. (Original work published 1998)

30.

Pardini

D. A.

Loeber

Farrington

D. P.

Stouthamer-Loeber

(2012). Identifying direct protective factors for nonviolence. American Journal of Preventive Medicine, 43, S28–S40.

31.

Peets

Kikas

(2006). Aggressive strategies and victimization during adolescence: Grade and gender differences, and cross-informant agreement. Aggressive Behavior, 32, 68–79.

32.

Piquero

N. L.

Gover

A. R.

MacDonald

J. M.

Piquero

A. R.

(2005). The influence of delinquent peers on delinquency: Does gender matter? Youth and Society, 36, 251–275.

33.

Preacher

K. J.

Hayes

A. F.

(2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40, 879–891.

34.

Rucker

D. D.

Preacher

K. J.

Tormala

Z. L.

Petty

R. E.

(2011). Mediation analysis in social psychology: Current practices and new recommendations. Social & Personality Psychology Compass, 5, 359–371.

35.

Sabiston

C. M.

Lovato

C. Y.

Ahmed

Pullman

A. W.

Hadd

Campbell

… Brown

K. S.

(2009). School smoking policy characteristics and individual perceptions of the school tobacco context: Are they linked to students’ smoking status? Journal of Youth and Adolescence, 38, 1374–1387.

36.

Sampson

R. J.

Morenoff

J. D.

Gannon-Rowley

(2002). Assessing neighborhood effects: Social processes and new directions in research. Annual Review of Sociology, 28, 443–478.

37.

Sampson

R. J.

Raudenbush

S. W.

Earls

(1997). Neighborhoods and violent crime: A multilevel study of collective efficacy. Science, 277, 918–924.

38.

Shaw

C. R.

McKay

H. D.

(1942). Juvenile delinquency and urban areas. Chicago, IL: University of Chicago Press.

39.

Silverman

J. R.

Caldwell

R. M.

(2005). The influence of parental emotional support and monitoring on self-reported delinquent impulsive behavior and noncompliance among juvenile offenders: An examination of gender differences. International Journal of Forensic Mental Health, 4, 159–174.

40.

Skogan

W. G.

(2012). Disorder and crime. In Welsh

B. C.

Farrington

D. P.

(Eds.), Oxford handbook of crime prevention (pp. 173–188). New York: Oxford University Press.

41.

Sutherland

E. H.

(1947). Principles of criminology (4th ed.). Philadelphia, PA: J.B. Lippincott.

42.

U.S. Census Bureau. (2011). Summary file—Census demographic profile. Washington, DC: The Author.

43.

Walters

G. D.

(2013). Delinquency, parental involvement, early adult criminality, and sex: Evidence of moderated mediation. Journal of Adolescence, 36, 777–785.

44.

Walters

G. D.

(2014). Sex as a moderator and perceived peer pressure as a mediator of the externalizing-delinquency relationship: A test of gendered pathways theory. Journal of Criminal Justice, 42, 299–305.

45.

Walters

G. D.

(2015a). Friends, cognition, and delinquency: Proactive and reactive criminal thinking as mediators of the peer influence and peer selection effects among male delinquents. Justice Quarterly. doi:10.1080/07418825.2015.1039048

46.

Walters

G. D.

(2015b). Parent-peer interface: Does inductive parenting reduce the influence of delinquent peers? Youth Violence and Juvenile Justice. doi:10.1177/1541204015581058

47.

Walters

G. D.

(2015c). Someone to look up to: Effect of role models on delinquent peer selection and influence. Youth Violence and Juvenile Justice. doi:10.1177/1541204015569317

48.

Warner

B. D.

(2014). Neighborhood factors related to the likelihood of successful informal social control efforts. Journal of Criminal Justice, 42, 421–430.

49.

Warner

B. D.

Beck

Ohmer

M. L.

(2010). Linking informal social control and restorative justice: Moving social disorganization theory beyond community policing. Contemporary Justice Review, 13, 355–369.

50.

Weerman

F. M.

(2011). Delinquent peers in context: A longitudinal network analysis of selection and influence effects. Criminology, 49, 253–286.

51.

Weerman

F. M.

Hoeve

(2012). Peers and delinquency among girls and boys: Are sex differences in delinquency explained by peer factors? European Journal of Criminology, 9, 228–244.

52.

Young

J. T. N.

Weerman

F. M.

(2013). Delinquency as a consequence of misperception: Overestimation of friends’ delinquent behavior and mechanisms of social influence. Social Problems, 60, 334–356.