Parents and Best Friends Perceived Reaction to Participant Deviance and Self-Reported Delinquency: Moderation by Sex and Mediation by Youth Attitude Toward Deviance in Mid- to Late Adolescence

Abstract

The current study set out to determine whether youth perceptions of parental and best friends’ likely reactions to their involvement in deviant behavior influenced their own attitude toward deviance (AD) and propensity for future delinquency. It also asked whether the effect was conditional on participant sex. Relationships between time-ordered measures of youth perceptions of mother, father, and best friends’ likely reactions to participant deviance, the youth’s own AD, and the youth’s level of participation in subsequent delinquency were tested in a sample of 3,880 adolescents (mean age = 15.33 years at the start of the study, 50.6% male). A moderated mediation path analysis revealed that perceived acceptance of participant deviance by the individual’s best friends predicted a rise in participant positive attitudes toward deviance, which, in turn, led to a rise in future delinquency. There was no effect, however, for mother or father perceived attitudes toward participant deviance. These results further revealed that the effect was moderated by sex and only significant in boys. From these results, it can be surmised that best friends’ perceived reactions to deviance may have an indirect effect on future delinquency of male youth by shaping and influencing the youth’s own AD.

Keywords

perceived reaction to participant deviance attitudes toward deviance delinquency moderated mediation

Social influence is a key factor in child development and associated childhood and adolescent problems and difficulties, and delinquency is a prime example of a problem of childhood and adolescence that owes its existence, in part, to social influence from parents and peers (Deutsch et al., 2012). Several major theories of crime and delinquency, in fact, highlight these sources of social influence. Social control theories, as represented by Reiss’s (1951) early work on social controls, Hirschi’s (1969) perspective on social bonding, and Gottfredson and Hirschi’s (1990) more recent self-control model, designate parents as the primary source of social control, support, and protective influence. Social learning theories, on the other hand, view peer deviance as the main source of social influence in the development of childhood delinquency (Akers, 1998; Sutherland, 1947). Although they conceptualize social influence differently, social control and social learning theories and the focus of their theorizing (i.e., parents and peers, respectively) may contribute equally to delinquency. The purpose of the current investigation was three-fold: (1) ascertain whether peer attitudes are as strong a source of social influence as parental attitudes during mid- to late adolescence, (2) determine whether a mechanism exists that explains the transmission of relevant attitudes from parents and peers to participants, and (3) test whether a contextual factor like sex is capable of altering relationships between parent/peer and participant attitudes and behaviors.

Sources of Social Influence: Peers and Parents

Parents and peers contribute significantly to delinquency development, although the contribution may vary from one context to the next. After reviewing the literature on parenting–peer linkages, Brown and Bakken (2011) concluded that parenting and peer effects can either be concurrent or consecutive, interactive or additive. An interactive effect, for instance, was observed by Lansford et al. (2003). In this study, low-quality friendships with highly antisocial peers measured during the summer months between Grades 6 and 7 amplified the criminogenic effect of unilateral parental decision making measured in the winter of Grade 7 on adolescent externalizing behavior assessed during the spring of Grade 7. An additive effect surfaced in another study, the results of which showed that parental support led to reduced levels of subsequent deviant peer associations, followed by decreased delinquent involvement in a group of early adolescent schoolchildren (Walters, 2020). In this same study, however, deviant peer associations had no effect on subsequent parental support. Testing the effectiveness of parent management training as an intervention for aggressive children, Forgatch et al. (2016) discovered that parenting influences preceded peer influences in the development of problem behavior. The results of all three studies suggest that parenting effects may precede peer effects and congruent with the Brown and Bakken (2011) review, that parenting may be the predominate influence during childhood, but peers become increasingly more important as children make their way into adolescence.

Mechanism: Attitudes Toward Deviance

A mechanism is required to explain the relationship between parenting and peer forms of social influence and the actions of youth subjected to these influences. For instance, it has been proposed that peer deviance negatively impacts participant decision making, which then leads to participant delinquency (Hoeben & Thomas, 2019). Another possibility is that attitudes are transmitted from parents and peers to participants. According to this perspective, parental and peer attitudes on various subjects help shape participant attitudes on these same subjects and that these attitudes, in turn, give rise to participant delinquency. This may be how moral values and attitudes are acquired, particularly when it comes to learning about the wrongfulness of deviant and delinquent behavior. Prior research has shown that both parents and peers contribute to the child’s development of moral values (Svensson et al., 2017; Walker et al., 2000), and from a social learning or social cognitive perspective (Bandura, 2001), it is speculated that the learning process is based on internalized models of parental and peer attitudes acquired through observing and interacting with these two potent sources of social influence. Consequently, it may not be the parenting or peer behavior that is most influential but the individual’s perception of the attitudes that are conveyed in parenting and peer behavior.

In a series of studies, Walters (2015, 2016a) determined that proactive (planned, calculated, and amoral) criminal thinking mediated the relationship between deviant peer associations and participant delinquent behavior. Through a process of social cognitive learning, socializing with a delinquent or deviant peer group can provide an individual with the opportunity to acquire delinquent thought patterns which, in turn, increase the person’s odds of engaging in delinquent behavior. Truccob et al. (2011) likewise discovered that perceived peer approval of alcohol consumption mediated the relationship between peer deviance and participant alcohol use in a group of early adolescents. In the previously mentioned Walters’s (2015, 2016a) investigations, participant proactive criminal thinking served as a mechanism that connected peer deviance to participant delinquency. Trucco et al. (2011), on the other hand, identified perceived peer approval of alcohol consumption as the mechanism linking peer deviance to drinking behavior. In the current study, perceived parental and peer approval of deviance and participant attitudes toward deviance were incorporated into the same design, although it is also important to know whether these relationships are moderated by such variables as sex.

Context: Sex Differences in Parental and Peer Influence

Relationships need to be understood within a wider context. One of the most important contexts for parental and peer influence is sex. It is possible that boys and girls react differently to parenting or that the effect of parenting differs between boys and girls as a result of gender role differences or differences in reward sensitivity. In testing the role of self-control in the parenting–delinquency relationship, Janssen and colleagues (2017) discovered that the direct effect of parenting on delinquency was similar for boys and girls, whereas the indirect effect, using self-control as a mediator, was stronger in girls than in boys. Fagan et al. (2011), by contrast, uncovered more similarities than differences in the effect of parenting on male and female offspring. In a review of 26 studies on gender differences in susceptibility to deviant peer influence, McCoy et al. (2019) determined that adolescent males experienced greater vulnerability to deviant peer pressure for risk-taking behavior than female adolescents. The authors of this review interpret their findings as support for gender role socialization theory and speculate that males are more likely to be influenced by peer pressure to engage in risk-taking behavior as a way of aligning themselves with a masculine ideal or stereotype. Another possibility is that biologically based reward sensitivity to peer influence, as discussed in the dual systems model of adolescent risk-taking (Shulman et al., 2016; Steinberg, 2010), is less acute in adolescent girls than it is in adolescent boys. Either way, sex moderation of the parenting–delinquency relationship has produced ambiguous results, whereas research on the peer–delinquency relationship is consistent with the conclusion that a stronger effect exists for boys.

Present Study

The gap in the literature the current study sought to fill and the research question this study was designed to answer center on youth perceptions of mother, father, and best friends attitudes toward deviance and how this may influence the youth’s own attitudes toward deviance and self-reported involvement in delinquency. Using a sample of mid- to late adolescent respondents, mother, father, and best friends perceived response to participant deviance (social influence) served as independent variables and the participant’s view of their own deviance (mechanism) served as the mediator. Whether the effect differed between boys and girls was evaluated by testing the interaction (context) between sex and each of the three independent variables (mother, father, and best friends perceived response to participant deviance). Although no predictions were made with respect to the moderating effect of sex on parental attitudes toward deviance, it was predicted that boys would be significantly more influenced by their best friends’ attitudes toward deviance than girls. In an effort to test the specificity of the attitudinal effect (perceived parent and friend attitudes toward deviance leading to participant’s attitude toward deviance [AD]), peer delinquency, measured concurrent with mother, father, and best friends perceived reaction to deviance in the participant, was controlled.

Three hypotheses were tested in this study:
Hypothesis 1: Mother, father, and best friends perceived reaction to participant deviance (PRPD) should correlate significantly with subsequent participant acceptance of deviance as the first stage of a mediated pathway running from mother/father/best friends PRPD to participant acceptance of deviance to participant delinquency.

Hypothesis 2: The participant’s own AD will serve as the second stage (mediating) variable in a mediated pathway designed to predict participant delinquency.

Hypothesis 3: Sex will moderate the best friends PRPD → participant AD → participant delinquency mediated pathway, such that the indirect effect will be significant in boys but not girls. Because prior research on sex moderation of parenting effects has been mixed, no predictions were made in this regard.

Method

Participants

The sample for this study came from the Youths and Deterrence longitudinal study (Paternoster, 2001) using data collected between 1979 and 1981 in nine high schools in Columbia, South Carolina. Participants were the 3,880 members of the Youths and Deterrence study (51.5% male) who had complete data on at least one of the 12 variables included in the present study. This represents 99.9% of all youth enrolled in the Youths and Delinquency study (N = 3,882). The average age of participants at Wave 1 of this study was 15.33 years (SD = 0.70, range = 13–21; with 16% estimated from age at Wave 2 or 3) and the ethnic/racial breakdown was 82.5% White, 15.9% Black, and 1.6% other.

Although the Youths and Deterrence data set is 40 years old, these data were selected for investigation because of their relevance to the current research question. The Youths and Deterrence data set was the only place where the author could find peer, parental, and participant measures of attitude toward participant deviance as perceived and reported by the participant. Unlike the National Youth Survey (Elliott, 1976–1980), where parents provide ratings of their own attitudes toward deviance in their child and youth provide a single rating of their attitudes toward deviance in themselves and their peers, the Youth and Deterrence study had youth rate their own attitudes toward deviance and what they perceived to be the attitudes of parents and best friends toward deviance in them (i.e., the participant). We might call the parental and best friend ratings reflected attitudes, in contrast to the reflected appraisals described in labeling theory. In each case, the same four offenses (shoplifting, property damage, underage drinking, and marijuana use) were rated. This uniformity facilitated comparisons between parents and peers and made the connections between the perceived social influence sources of attitude formation and the participant’s own attitudes clearer.

Measures

Father reaction to deviance

Father PRPD was assessed with four items (father’s reaction if you [participant] were to engage in…” shoplifting,” “property damage,” “underage drinking,” and “marijuana use”), each rated on a 5-point scale (1 = strongly disapprove, 2 = disapprove, 3 = would not care, 4 = approve, and 5 = strongly approve). Scores on each of the four items were then averaged to produce a mean score that could range from 1 to 5. The internal consistency of this scale in the Youths and Delinquency sample was good (α = .82).

Mother reaction to deviance

Mother PRPD was assessed with the same four items as the father PRPD (mother’s reaction if you [participant] were to engage in…”shoplifting,” “property damage,” “underage drinking,” and “marijuana use”), with each item rated on the same 5-point scale (1 = strongly disapprove, 2 = disapprove, 3 = would not care, 4 = approve, and 5 = strongly approve). Like the father PRPD, the four items that comprise the mother PRPD were averaged to produce a mean score that could range from 1 to 5. The internal consistency of the mother PRPD scale in the current sample of participants was good (α = .85).

Best friends reaction to deviance

Best friends PRPD was assessed for “shoplifting,” “property damage,” “underage drinking,” and “marijuana use,” using the same 5-point scale as was used to assess father and mother PRPD (1 = strongly disapprove, 2 = disapprove, 3 = would not care, 4 = approve, and 5 = strongly approve). Scores on each item were then averaged to yield a score that could range from 1 to 5. The best friends PRPD achieved good internal consistency in the present sample of participants (α = .84).

Participant AD

A participant’s AD was assessed by asking them to rate the following four items, “how wrong is stealing,” “how wrong is damaging property,” “how wrong is underage drinking,” and how wrong is smoking marijuana” on a 5-point scale (1 = always wrong, 2 = usually wrong, 3 = sometimes wrong, 4 = seldom wrong, and 5 = never wrong). A mean score was then derived by averaging the individual item scores. The Participant AD scale achieved adequate internal consistency during Waves 1 and 2 of the Youths and Deterrence study (α = .75–.76).

Delinquency

Participant delinquency was assessed with a variety score based on involvement in the following 12 delinquent acts over the past year: stole something worth less than US$10, stole something worth between US$10 and US$50, damaged property, broke into a home or business to steal something, took a car without permission, carried a hidden weapon, threatened to beat someone up, damaged school property, threw objects at cars or people, stole from students, sold drugs, and beat up someone badly. A delinquency variety score was calculated as the number of different offenses participated in over the past year divided by the total number of offense categories (i.e., 12). Variety scores have been found to possess strong psychometric properties for criminological research (Sweeten, 2012) and the delinquency variety score used in the current study achieved adequate internal consistency (α = .77–.78).

Control variables

Five control variables were included in the present study: age (in years), sex (1 = male, 2 = female), race (1 = White, 2 = non-White), parent education (1 = some high school or less, 2 = high school graduate, 3 = high school plus trade or business school, 4 = some college, without graduating, 5 = four-year college graduate, and 6 = college plus graduate school; averaged across both parents), and peer delinquency. The Peer Delinquency scale was composed of four questions: “How many of your friends shoplift?” “How many of your friends damage things?” “How many of your friends drink underage?” and “How many of your friends use marijuana?” Each item was rated on a 4-point scale (1 = none, 2 = some, 3 = most, and 4 = all) and the results summed to produce a scale that could range from 4 to 16 (α = .77).

Two precursor measures were included in the current investigation. This was done in order to establish the temporal direction of the variables (Maxwell & Cole, 2003). Precursor measures are prior estimates or antecedents of a predicted variable. Including precursor measures in a regression equation creates lagged outcome variables and makes it easier to argue that the independent variable preceded the mediator and the mediator preceded the dependent variable in a mediation analysis. Hence, Wave 1 participant AD was included as a covariate in the equation predicting Wave 2 participant AD, and Wave 1 delinquency was included as a covariate in the equation predicting Wave 3 delinquency.

Research Design

A fixed sample longitudinal panel design was employed in this study. There were three nonoverlapping waves of data, each separated by 12 months, with a mean age of 15.27 years (range = 13-21) at Wave 1, 16.27 (range = 14–20) at Wave 2, and 17.24 (range = 15–21) at Wave 3. There was no overlap between adjacent waves, making this a prospective study. The three independent variables (father PRPD, mother PRPD, and best friends PRPD), five control variables (age, sex, race, parent education, and peer delinquency), two precursor measures (Participant AD-1 and Delinquency-1), and three interactions (Mother PRPD × Sex, Father PRPD × Sex, and Best Friends PRPD × Sex) were measured at Wave 1, the mediator variable (Participant AD-2) was measured at Wave 2, and the dependent variable (Delinquency-3) was measured at Wave 3. The presence of both mediating and moderating effects means that the present design was used to conduct a moderated mediation analysis.

Data Analytic Plan

A path analysis was performed to determine whether participant AD mediated the relationships between father, mother, or best friends PRPD and participant delinquency. These indirect effects, along with each direct effect from an independent variable to the dependent variable, were evaluated with bias-corrected bootstrapped confidence intervals ([BCBCIs]; 5,000 repetitions) computed with MPlus 8.3 (Muthén & Muthén, 1998-2017). Research has consistently shown that nonparametric bootstrapping does a better job of modeling the nonnormality of indirect effects than normal theory procedures like the Sobel (1982) test (Preacher, 2015). BCBCIs were likewise used to evaluate the significance of interactions between two centered variables: mother PRPD and sex, father PRPD and sex, and best friends PRPD and sex. Differences between the three pathways (Father PRPD-1 → Participant AD-2 → Delinquency-3, Mother PRPD-1 → Participant AD-2 → Delinquency-3, and Best Friends PRPD-1 → Participant AD-2 → Delinquency-3) were evaluated using the Preacher and Hayes (2008) contrast test which is also performed using bootstrapped confidence intervals. A confidence interval, it should be noted, is considered significant when it does not include zero.

A two-step procedure was employed in an effort to determine which interaction terms should be included in the final model. The first step of the procedure was to assess the Father PRPD-1, Mother PRPD-1, and Best Friends PRPD-1 main effects and interactions with sex as predictors of Participant AD-2. All main effects and interactions not achieving significance during the first step of the procedure were evaluated in combination and alone, minus the main effects achieving a significant interactive effect with sex during the first step. Only interactions achieving significance during Step 1 or Step 2 of the two-step procedure were included in the final model.

Sensitivity analyses were applied to the mediational results in an effort to assess and rule out omitted variable and endogenous selection biases. Kenny’s (2013) “failsafe ef” procedure—(r_my.x ) × (sd_m _.x) × (sd_y.x )/(sd_m ) × (sd_y )—was used to evaluate omitted variable bias. The “failsafe ef” generates a coefficient that indicates how strongly an unobserved covariate confounder would need to correlate with the mediating and dependent variables, controlling for the mediator and independent variables in the case of the dependent variable, to fully and completely eliminate the significant coefficient along the b path (M → Y) of the indirect effect. A second sensitivity analysis was used to test for endogenous selection bias or a collider effect. Because conditioning on a precursor measure can potentially inflate path coefficients (Elwert & Winship, 2014), endogenous selection bias was evaluated by redoing the analyses without precursor measures. If a collider effect is operational, the path coefficient will no longer be significant once the precursor measure is removed.

Missing Data

Slightly more than a quarter of the sample had complete data on all 12 variables (27.5%). Another 19.2% of participants were missing data on one variable, 17.7% were missing data on two variables, 4.8% were missing data on three to five variables, 29.7% were missing data on six to eight variables, and 1.2% were missing data on nine to 11 variables. Individual variables were missing between 0.2% (sex) and 43.1% (Participant AD-2) of their data. Missing data were handled with full information maximum likelihood (FIML), a procedure that estimates standard errors and population parameters from all available data, although a secondary analysis was also performed in which missing data were handled with listwise deletion. Research indicates that FIML generates significantly more accurate and significantly less biased results than traditional missing value procedures like simple imputation and listwise deletion (Allison, 2002).

Results

Descriptive statistics for the 12 independent, dependent, mediating, and control variables from this study can be found in Table 1, along with an 11 × 12 intervariable correlational matrix. Over two thirds of the correlations in this matrix achieved statistical significance using a Bonferroni-corrected α level. It should be noted that participants perceived their best friends as significantly more likely to approve of deviance than their parents (p < .001). When collinearity diagnostics were applied to the two regression equations, there was no evidence of multicollinearity between the predictor variables: tolerance = .365–.964 and variance inflation factors = 1.037–2.740.

Table 1.
Descriptive Statistics and Correlations for the Control, Independent, Moderator, and Dependent Variables Included in This Study.

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

1. Age 3,868 15.33 0.70 13–21 −.14* .14* −.11* .12* .14* .14* .09* .10* .12* .10* .11*

2. Sex 3,872 1.48 — 1–2 .01 −.03 −.16* −.06 −.01 −.20* −.07* −.16* −.32* −.37*

3. Race 3,866 1.17 — 1–2 −.12* −.11* .16* .11* −.04 −.03 −.02 −.01 −.01

4. Parent education 2,531 3.03 1.49 1–6 −.01 −.09* −.08* −.05 −.03 −.00 −.01 −.00

5. Peer delinquency 2,607 8.35 2.96 4–16 .20* .15* .63* .54* .36* .49* .36*

6. Father PRPD 2,594 1.35 0.57 1–5 .72* .33* .31* .16* .19* .10*

7. Mother PRPD 2,668 1.29 0.55 1–5 .31* .24* .14* .14* .03

8. Best friends PRPD 2,665 2.34 0.79 1–5 .52* .36* .45* .30*

9. Participant AD-1 2,638 1.71 0.75 1–5 .56* .46* .29*

10. Participant AD-2 2,209 1.71 0.76 1–5 .33* .38*

11. Delinquency-1 2,593 0.15 0.19 0–0.92 .53*

12. Delinquency-3 2,372 0.12 0.17 0–1

Note. Variable = study variables; age = participant age in years; sex = male (1) vs. female (2); race = White (1) vs. non-White (2); parent education = parent education rated on a 6-point scale and averaged between mother and father; peer delinquency = perceived peer delinquency at Wave 1; father PRPD = father perceived reaction to participant deviance at Wave 1; mother PRPD = mother perceived reaction to participant deviance at Wave 1; best friends PRPD = best friends perceived reaction to participant deviance at Wave 1; Participant AD-1 = participant’s attitude toward deviance at Wave 1; Participant AD-2 = participant attitude toward deviance at Wave 2; Delinquency-1 = participant’s self-reported delinquency variety score at Wave 1; Delinquency-3 = participant’s self-reported delinquency variety score at Wave 3; n = number of cases with complete data on that variable; M = mean, SD = standard deviation; range = range of scores achieved by participants in the current sample.

p < .00076 (Bonferroni-corrected α level; .05/66 correlations).

A two-step procedure was used to determine which interaction term(s) should be included in the final path model. During Step 1, all three independent variables (mother PRPD, father PRPD, and best friends PRPD) and their interactions with sex were entered as predictors in the regression equation predicting participant AD-2. The Step 1 results indicated that only the best friends PRPD main effect (p < .05) and its interaction (p < .01) were significant. Three additional analyses were performed at Step 2: one restricted to mother and father PRPD and their interactions with sex, one restricted to mother PRPD and its interaction with sex, and one restricted to father PRPD and its interaction with sex. None of the main or interaction effects achieved significance at Step 2 (p > .10), and so only the best friends PRPD × Sex interaction was included in the final path model.

The results of a moderated mediation path analysis with the Best Friends PRPD × Sex interaction as the only interaction term are summarized in Table 2 and Figure 1. Not only did the Best Friends PRPD × Sex interaction achieve significance, but so did the a (Best Friends PRPD → Participant AD-2) and b (Participant AD-2 → Delinquency-3) paths of the indirect effect for the Best Friends PRPD-initiated pathway. Of the three indirect effects examined in this study, only the Best Friends PRPD → Participant AD-2 → Delinquency-3 pathway achieved significance, although there were no significant differences between the three pathways (see Preacher–Hayes’s contrast test results at the bottom of Table 3). When analyzed separately by sex, the Best Friends PRPD → Participant AD-2 → Delinquency-3 pathway was significant only in the male subsample (95% BCBCI [.043, 0.193]).

Table 2.
Results of a Path Analysis With Father, Mother, and Best Friends PRPD as Independent Variables, Participant AD as the Mediator Variable, and Participant Delinquency as the Dependent Variable.

Variable b [95% BCBCI] β Z p

Outcome = Participant AD-2

Age .036 [−.011, .086] .033 1.46 .145

Sex −.136 [−.190, −.080] −.087 −4.77 <.001

Race −.020 [−.099, .066] −.010 −0.47 .637

Parent education .003 [−.018, .025] .005 0.25 .805

Peer delinquency .012 [−.002, 0027] .047 1.70 .088

Father PRPD .020 [−.086, .129] .015 0.36 .717

Mother PRPD .020 [−.093, .122] .014 0.37 .711

Best friends PRPD .065 [.009, .120] .067 2.32 .020

Best Friends PRPD × Sex −.062 [−.098, −.028] −.078 −3.51 <.001

Participant AD-1 .561 [.491, .628] .545 16.00 <.001

Outcome = Delinquency-3

Age .006 [−.006, .019] .024 0.92 .357

Sex −.067 [−.080, −.054] −.190 −10.24 <.001

Race −.003 [−.021, .016] −.007 −0.33 .738

Parent education −.005 [−.010, .000] −.040 −1.90 .058

Peer delinquency .004 [.001, .008] .071 2.31 .021

Father PRPD .008 [−.014, .029] .025 0.70 .482

Mother PRPD −.012 [−.033, .010] −.038 −1.09 .274

Best friends PRPD −.008 [−.021, .004] −.037 −1.30 .192

Participant AD-2 .053 [.039, .068] .235 7.41 <.001

Delinquency-1 .380 [.310, .449] .410 10.77 <.001

Note. N* = 3,880. Outcome = outcome measure for that particular regression equation; variable = variable name; age = participant age in years; sex = male (1) vs. female (2); race = White (1) vs. non-White (2); parent education = parent education rated on a 6-point scale and averaged between mother and father; peer delinquency = perceived peer delinquency at Wave 1; father PRPD = father perceived reaction to participant deviance at Wave 1; mother PRPD = mother perceived reaction to participant deviance at Wave 1; best friends PRPD = best friends perceived reaction to participant deviance at Wave 1; Participant AD-1 = participant’s attitude toward deviance at Wave 1; Participant AD-2 = participant’s attitude toward deviance at Wave 2; Delinquency-1 = participant’s self-reported delinquency variety score at Wave 1; Best Friends PRPD × Sex = interaction between best friends perceived reaction to participant deviance at Wave 1 and sex; Delinquency-3 = participant’s self-reported delinquency variety score at Wave 3; b [95% BCBCI] = unstandardized coefficient and 95% bias-corrected bootstrapped confidence interval (in square brackets), Z = Wald’s Z test; p = significance level of the Wald Z test.

Figure 1.
Path analysis results for Wave 1 father, mother, and best friends PRPD, sex, and the BF-PRPD × Sex interaction to Wave 2 participant AD to Wave 3 delinquency. Note. N = 3,851. Standardized β coefficients are reported; F-PRPD = father perceived reaction to participant deviance at Wave 1; M-PRPD = mother perceived reaction to participant deviance at Wave 1; BF-PRPD = best friends perceived reaction to participant deviance at Wave 1; BF-PRPD × Sex = best friends perceived reaction to participant deviance × sex interaction; P-AD = participant’s attitude toward deviance at Wave 2; Delinquency = participant self-reported delinquency variety score at Wave 3. *p < .05. **p < .001.

Table 3.
Total, Direct, and Indirect Effects for the Path Analysis.

BCBCI

Estimate Lower Upper

Effect from Father PRPD to Delinquency-3

Total effect 0.0087 −0.0132 0.0304

Direct effect −0.0077 −0.0139 0.0292

Total indirect effect 0.0011 −0.0046 0.0071

Specific indirect effect

F-PRPD-1 → P-AD-2 → Delinquency-3 0.0011 −0.0046 0.0071

Effect from Mother PRPD to Delinquency-3

Total effect −0.0108 −0.0318 0.0110

Direct effect −0.0119 −0.0331 0.0096

Total indirect effect 0.0011 −0.0048 0.0066

Specific indirect effect

M-PRPD-1 → P-AD-2 → Delinquency-3 0.0011 −0.0048 0.0066

Effect from Best Friends PRPD to Delinquency-3

Total effect −0.0047 −0.0170 0.0073

Direct effect −0.0082 −0.0208 0.0039

Total indirect effect 0.0035 0.0006 0.0068

Specific indirect effect

BF-PRPD-1 → P-AD-2 → Delinquency-3 0.0035 0.0006 0.0068

Preacher-Hayes Contrast Test

Best Friends pathway vs. Father pathway 0.0024 −0.0041 0.0094

Best Friends pathway vs. Mother pathway 0.0024 −0.0037 0.0099

Note. N = 3,880. F-PRPD-1 = father perceived reaction to participant deviance at Wave 1; M-PRPD-1 = mother perceived reaction to participant deviance at Wave 1; BF-PRPD-1 = best friends perceived reaction to participant deviance at Wave 1; P-AD-2 = participant’s attitude toward deviance at Wave 2; Delinquency-3 = participant’s self-reported delinquency variety score at Wave 3; Preacher-Hayes Contrast Test = comparisons between best friends and parental pathways using the contrast test proposed by Preacher and Hayes (2008); 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 depicts the moderating effect of sex on the prospective relationship between best friends perceived attitude toward participant deviance and participants’ own AD. Thus, while reflected attitudes about best friends’ reactions to participant deviance had little bearing on adolescent females’ future attitudes toward deviance, they had a significant bearing on adolescent males’ future attitudes toward deviance, in that boys were more likely to display future positive attitudes toward deviance if they had previously perceived that their best friends would approve of participant deviance. In addition, the best friends PRPD–participant AD relationship was significant only in boys, whereas the participant AD–participant delinquency relationship was significant in both boys and girls.

Figure 2.
Interaction between sex and best friends perceived response to deviance (approves vs. disapproves) at Wave 1 in relationship to participant’s positive attitudes toward deviance at Wave 2.

Omitted variable bias was appraised with the “failsafe ef” procedure, and endogenous selection bias was assessed by removing the precursor measures. The “failsafe ef” procedure produced a coefficient of .26, which means that a covariate confounder would need to correlate .26 with Participant AD-2 and .26 with Wave 3 delinquency, controlling for Best Friends PRPD-1 and Participant AD-2, to eliminate the coefficient along the b path of the significant Best Friends PRPD to Participant AD-2 to Delinquency-3 pathway. Endogenous selection bias, also known as a collider effect, was tested by removing the precursor measures from the regression equations and rerunning the analyses. Results revealed that the key coefficients increased rather than decreased, a finding inconsistent with the presence of a collider effect.

Even though FIML is the preferred method for handling missing data, over 70% of the participants in this study were missing data on at least one variable. Secondary analyses in which missing data were handled with listwise deletion were consequently performed. Despite a 70% reduction in power, the results paralleled those attained with FIML. Whereas the Best Friends PRPD × Sex interaction (p < .05) and Participant AD-2 → Delinquency-3 path (p < .001) were significant in the listwise deletion mixed sex sample, the Best Friends PRPD → Participant AD-2 path fell short of significance (p = .11). When the data were analyzed separately by sex, the Best Friends PRPD → Participant AD-2 → Delinquency-3 pathway was significant in males (95% BCBCI [.041, .272]) but not in females (95% BCBCI [−.026, .005]), just as it has been in the analyses where missing data had been handled with FIML.

Discussion

Three hypotheses were tested in this study and all three received at least partial support. The first hypothesis predicted that best friends PRPD would achieve significance on par with mother or father PRPD in predicting participant AD in a group of mid- to late adolescents. This hypothesis received support in the sense that best friends PRPD-initiated indirect effect achieved significance and it did so while controlling for mother and father PRPD. There was no evidence of a significant indirect effect for mother or father PRPD in this study. This was an unanticipated finding that ran contrary to the first hypothesis. The second hypothesis garnered support, again only with best friends, based on a significant mediating effect for participant AD as a link between best friends PRPD and participant delinquency in a three-wave mediated model. This implies that the perception that one’s best friends would disapprove of one’s involvement in deviant activities lessened the participant’s enthusiasm for deviant behavior. The third hypothesis was supported by the fact that sex moderated the best friends PRPD-initiated indirect effect and that when analyses were performed separately for boys and girls, the mediating effect of participant AD was significant only for boys, as predicted. It should also be noted that participants perceived their best friends as significantly more approving of deviance than their parents.

Strengths and Weaknesses

The current investigation benefited from several strengths, not the least of which was its utilization of three nonoverlapping waves of longitudinal data. Including longitudinal or prospective data in a research design can be of assistance in establishing the temporal order of variables in a study, while enhancing a researcher’s ability to make causal inferences from nonexperimental data (Miller, 1999). Another strength of this study is that prior levels of the two predicted variables (Participant AD-2 and Delinquency-3) were controlled at Wave 1 of the design (Participant AD-1 and Delinquency-1, respectively). Crafting lagged outcome variables makes it possible to estimate within-sample change over time. Controlling for the precursors to an outcome by creating lagged outcome variables also helps establish the temporal direction of one’s variables (Cole & Maxwell, 2003). A third strength of this study is that it accounted for general perceptions of peer deviance or delinquency. By controlling for general perceptions of peer delinquency, it was possible to rule out a nonspecific peer effect in favor of a more specific best friends effect marked by attitude transmission, in which a person’s perception of how their best friends might react to their involvement in deviant behavior appeared to shape their attitudes toward deviance and their future likelihood of engaging in various delinquent activities.

As counterpoints to the strengths and benefits mentioned in the previous paragraph, there were also several weaknesses and limitations that need to be acknowledged. First, the data are 40 years old, having been collected between 1979 and 1981. It could accordingly be argued that these results are less relevant to modern-day adolescents than contemporarily collected data given the changes that have taken place in information and communication technology over the past 40 years. Human behavior, however, changes more slowly than technology and so it is uncertain how much the relationships observed in this study were a function of the period in which the data were collected. Moreover, this was the only data set the author was able to get access to that allowed for construction of reflected attitudes measures of parents and peers (best friends). Second, all of the measures employed in this study were based on participant self-report, including the peer delinquency, best friends PRPD, and parental PRPD measures. Employing a single source of data like participant self-report can inflate coefficient estimates through shared method variance and mono-operational bias (Shadish et al., 2002). With respect to potential bias in the results, the current findings were moderately robust to the effects of omitted variable bias (Imai et al., 2010), and there was no evidence of endogenous selection bias when the precursor measures were removed from the regression equations (Elwert & Winship, 2014).

Theoretical Implications

As discussed previously, each one of the measures included in the present investigation was constructed from self-report data provided by the participant. Even the measures of parental and best friends PRPD were based on participant self-report. As a measure of perception, best friends and parental PRPD share certain commonalities with another cognitive process, namely, reflected appraisals. The results of cross-sectional (Brownfield & Thompson, 2005), retrospective (Walters, 2016b), and longitudinal (Lee et al., 2017; Matsueda, 1992) studies, in fact, indicate that a reflected appraisal as a lawbreaker correlates strongly with self-reported delinquency. Reflected appraisals, however, are a person’s perception of how they are viewed by others (Mead, 1934) and, as such, inform a person’s self-view or identity. The PRPD measures employed in the current study, along with participant AD, are more representative of a person’s AD than of his or her identity. In fact, these results are more consistent with a social learning theory as represented by modeling and observational learning (Akers, 1998; Bandura, 1986). People learn by developing internalized models of their environment. Acquiring criminal attitudes and the criminal behavior to which they lead is less a matter of imitating others’ actions and beliefs as it is internalizing cognitive representations of those actions and beliefs in the form of reflected attitudes.

In a 4-year longitudinal study of U.S. public schoolchildren, Zhang et al. (1997) observed a moderately sized correlation between positive or accepting attitudes toward deviance and subsequent delinquent behavior that was significantly stronger than the correlation running from delinquent behavior to subsequent positive attitudes toward deviance. Several other investigators also report that delinquents display more accepting attitudes toward deviance than nondelinquents (Landsheer & Hart, 2000; Webster & Vermeulen, 2011). These results, like those from the current study, imply that an accepting AD is shaped, in part, by one’s perception of how others are likely to react to one’s involvement in delinquent behavior. Participants in the current study were in mid- to late adolescence and the results showed that in this age-group, best friends’ PRPD effectively increased the participant’s own attitudes toward the acceptability of deviance, whereas parental perceived reactions to deviance did not. Perhaps if a younger sample had been used, parental PRPD would have had as much, if not more, impact, on participants’ attitudes toward deviance than best friends PRPD. It should also be noted that just as unsupervised activities with friends have been found to predict participant delinquency when level of peer delinquency is controlled (Haynie & Osgood, 2005), best friends PRPD also appears to facilitate participant AD after level of peer delinquency is controlled (current study).

Practical Implications

In the interests of both prevention and intervention, results from the current study suggest that it may be beneficial to assess and monitor male adolescents’ reflected attitudes of how close friends would likely view the participant’s involvement in delinquent and deviant activities. Studies conducted on youth, particularly male youth, indicate that they frequently overestimate the number of friends and peers who smoke cigarettes, drink alcohol, and use illegal drugs (Elsey et al., 2015; Haug et al., 2011; Pape, 2012; Roditis et al., 2016). Such distorted social norms for drug use and other forms of deviant behavior could also apply to attitudes and beliefs, whereby male youth overestimate friends’ and peers’ acceptance of deviant behavior. Assessing the child’s belief systems and challenging exaggerated or false perceptions of friend and peer acceptance of deviance could go a long way toward reducing the youth’s vulnerability to future deviant behavior. By encouraging male youth to check out their perceptions with friends and peers on a one-to-one basis, rather than in a group setting, where the motivation to express nonconformist attitudes for social acceptance could be stronger, it may be possible to reduce deviant behavior in youth by lowering their estimates of friend acceptance of deviance and the effect this has on their own attitudes toward deviance.

Explaining Female Attitudes Toward Deviance

The current results illustrate one mechanism by which male adolescents are socially influenced by friends to experiment with deviant behavior and enter into delinquent activity. The same cannot be said for female adolescents. Unlike males, whose perceptions of peer delinquency and best friends’ acceptance of deviance predicted their own acceptance of deviance, neither of these variables nor any other variable in the set of variables examined as part of this study, including mother and father PRPD, demonstrated an ability to predict participant AD in females. Participant AD, however, was just as predictive of participant delinquency in girls as it was in boys. The question that could not be answered in the current study, but which requires further investigation, is one that asks, what factors facilitate the development of female adolescent attitudes that support deviance? We know from the literature on gendered pathways to crime that factors such as early physical abuse, relationship difficulties, mental illness, and substance misuse are particularly important for female offending (Chesney-Lind & Pasko, 2004). Perhaps one or more of these variables is capable of shaping female adolescent attitudes toward deviance just as best friends PRPD impacted on male adolescent attitudes toward deviance in the current study. Of course, it is also possible that biologically based differences in reward sensitivity are also at work here.

Conclusion

A moderated mediation analysis performed on a group of mid- to late adolescents revealed that best friends’ perceived reactions to participant deviance (social influence) stimulated participant attitudes toward deviance in a manner that affected future delinquency (mechanism), in a pattern that was confined to males (context). What this means from a theoretical standpoint is that friends’ perceived reactions to deviance help shape male youths’ attitudes toward deviance, which, in turn, influence their propensity to engage in future delinquent behavior. What this means from a practical standpoint is that effort needs to be directed at altering perceptions of friends’ reactions to deviance, at least in boys. Because these results were restricted to male adolescents, additional research is required to determine what might shape female adolescent attitudes toward deviance in a manner similar to how perceptions of best friends’ reactions to deviance helped shape male adolescent attitudes toward deviance in the current study.

Variable	n	M	SD	Range	2	3	4	5	6	7	8	9	10	11	12
1. Age	3,868	15.33	0.70	13–21	−.14*	.14*	−.11*	.12*	.14*	.14*	.09*	.10*	.12*	.10*	.11*
2. Sex	3,872	1.48	—	1–2		.01	−.03	−.16*	−.06	−.01	−.20*	−.07*	−.16*	−.32*	−.37*
3. Race	3,866	1.17	—	1–2			−.12*	−.11*	.16*	.11*	−.04	−.03	−.02	−.01	−.01
4. Parent education	2,531	3.03	1.49	1–6				−.01	−.09*	−.08*	−.05	−.03	−.00	−.01	−.00
5. Peer delinquency	2,607	8.35	2.96	4–16					.20*	.15*	.63*	.54*	.36*	.49*	.36*
6. Father PRPD	2,594	1.35	0.57	1–5						.72*	.33*	.31*	.16*	.19*	.10*
7. Mother PRPD	2,668	1.29	0.55	1–5							.31*	.24*	.14*	.14*	.03
8. Best friends PRPD	2,665	2.34	0.79	1–5								.52*	.36*	.45*	.30*
9. Participant AD-1	2,638	1.71	0.75	1–5									.56*	.46*	.29*
10. Participant AD-2	2,209	1.71	0.76	1–5										.33*	.38*
11. Delinquency-1	2,593	0.15	0.19	0–0.92											.53*
12. Delinquency-3	2,372	0.12	0.17	0–1

Variable	b [95% BCBCI]	β	Z	p
Outcome = Participant AD-2
Age	.036 [−.011, .086]	.033	1.46	.145
Sex	−.136 [−.190, −.080]	−.087	−4.77	<.001
Race	−.020 [−.099, .066]	−.010	−0.47	.637
Parent education	.003 [−.018, .025]	.005	0.25	.805
Peer delinquency	.012 [−.002, 0027]	.047	1.70	.088
Father PRPD	.020 [−.086, .129]	.015	0.36	.717
Mother PRPD	.020 [−.093, .122]	.014	0.37	.711
Best friends PRPD	.065 [.009, .120]	.067	2.32	.020
Best Friends PRPD × Sex	−.062 [−.098, −.028]	−.078	−3.51	<.001
Participant AD-1	.561 [.491, .628]	.545	16.00	<.001
Outcome = Delinquency-3
Age	.006 [−.006, .019]	.024	0.92	.357
Sex	−.067 [−.080, −.054]	−.190	−10.24	<.001
Race	−.003 [−.021, .016]	−.007	−0.33	.738
Parent education	−.005 [−.010, .000]	−.040	−1.90	.058
Peer delinquency	.004 [.001, .008]	.071	2.31	.021
Father PRPD	.008 [−.014, .029]	.025	0.70	.482
Mother PRPD	−.012 [−.033, .010]	−.038	−1.09	.274
Best friends PRPD	−.008 [−.021, .004]	−.037	−1.30	.192
Participant AD-2	.053 [.039, .068]	.235	7.41	<.001
Delinquency-1	.380 [.310, .449]	.410	10.77	<.001

		BCBCI
Effect from Father PRPD to Delinquency-3
Total effect	0.0087	−0.0132	0.0304
Direct effect	−0.0077	−0.0139	0.0292
Total indirect effect	0.0011	−0.0046	0.0071
Specific indirect effect
F-PRPD-1 → P-AD-2 → Delinquency-3	0.0011	−0.0046	0.0071
Effect from Mother PRPD to Delinquency-3
Total effect	−0.0108	−0.0318	0.0110
Direct effect	−0.0119	−0.0331	0.0096
Total indirect effect	0.0011	−0.0048	0.0066
Specific indirect effect
M-PRPD-1 → P-AD-2 → Delinquency-3	0.0011	−0.0048	0.0066
Effect from Best Friends PRPD to Delinquency-3
Total effect	−0.0047	−0.0170	0.0073
Direct effect	−0.0082	−0.0208	0.0039
Total indirect effect	0.0035	0.0006	0.0068
Specific indirect effect
BF-PRPD-1 → P-AD-2 → Delinquency-3	0.0035	0.0006	0.0068
Preacher-Hayes Contrast Test
Best Friends pathway vs. Father pathway	0.0024	−0.0041	0.0094
Best Friends pathway vs. Mother pathway	0.0024	−0.0037	0.0099

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.

ORCID iD

Glenn D. Walters

References

Akers

R. L.

(1998). Social learning and social structure: A general theory of crime and deviance. Northeastern University Press.

Allison

P. D.

(2002). Missing data. Sage.

Bandura

(1986). Social foundations of thought and action: A social cognitive theory. Prentice Hall.

Bandura

(2001). Social cognitive theory: An agentic perspective. Annual Review of Psychology, 52, 1–26.

Brown

B. B.

Bakken

J. P.

(2011). Parenting and peer relationships: Reinvigorating research on family–peer linkages in adolescence. Journal of Research on Adolescence, 21, 153–165.

Brownfield

Thompson

(2005). Self-concept and delinquency: The effects of reflected appraisals by parent and peers. Western Criminology Review, 6, 22–29.

Chesney-Lind

Pasko

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

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.

Deutsch

A. R.

Crockett

L. J.

Wolff

J. M.

Russell

S. T.

(2012). Parent and peer pathways to adolescent delinquency: Variations by ethnicity and neighborhood context. Journal of Youth and Adolescence, 41, 1078–1094.

10.

Elliott

D. S.

(1976–1980). National youth survey (United States): Waves 1–5, 1976–1980. Inter-University Consortium for Political and Social Research.

11.

Elsey

Owiredu

Thomson

Mann

Mehta

Siddiqi

(2015). Do children overestimate the extent of smoking among their peers? A feasibility study of the social norms approach to prevent smoking. Addictive Behaviors, 41, 7–11.

12.

Elwert

Winship

(2014). Endogenous selection bias: The problem of conditioning on a collider variable. Annual Review of Sociology, 40, 31–53.

13.

Fagan

A. A.

Van Horn

M. L.

Antaramian

Hawkins

J. D.

(2011). How do families matter? Age and gender differences in family influences on delinquency and drug use. Youth Violence and Juvenile Justice, 9, 150–170.

14.

Forgatch

Snyder

J. J.

Patterson

G. R.

Pauldine

M. R.

Chaw

Elish

Harris

J. B.

Richardson

E. B.

(2016). Resurrecting the chimera: Progressions in parenting and peer processes. Development and Psychopathology, 28, 689–706.

15.

Gottfredson

M. R.

Hirschi

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

16.

Haug

Ulbricht

Hanke

Meyer

John

(2011). Overestimation of drinking norms and its association with alcohol consumption in apprentices. Alcohol and Alcoholism, 46, 204–209.

17.

Haynie

D. L.

Osgood

D. W.

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

18.

Hirschi

(1969). Causes of delinquency. University of California Press.

19.

Hoeben

E. M.

Thomas

K. J.

(2019). Peers and offender decision-making. Criminology and Public Policy, 18, 759–784.

20.

Imai

Keele

Yamamoto

(2010). Identification, inference, and sensitivity analysis for causal mediation effects. Statistical Science, 25, 51–71.

21.

Janssen

H. J.

Eichelsheim

V. I.

Deković

Bruinsma

G. J. N.

(2017). Sex differences in longitudinal pathways from parenting to delinquency. European Journal of Criminal Policy and Research, 23, 503–521.

22.

Kenny

D. A.

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

23.

Landsheer

J. A.

Hart

(2000). Punishments adolescents find justified: An examination of attitudes toward delinquency. Adolescence, 35, 683–693.

24.

Lansford

J. E.

Criss

M. M.

Pettit

G. S.

Dodge

K. A.

Bates

J. E.

(2003). Friendship quality, peer group affiliation, and peer antisocial behavior as moderators of the link between negative parenting and adolescent externalizing behavior. Journal of Research on Adolescence, 13, 161–184.

25.

Lee

J. S.

Tajima

E. A.

Herrenkohl

T. I.

Hong

(2017). Effects of formal and informal deviant labels in adolescence and crime in adulthood. Social Work Research, 41, 97–109.

26.

Matsueda

R. L.

(1992). Reflected appraisals, parental labeling, and delinquency: Specifying a symbolic interactionist theory. American Journal of Sociology, 97, 1577–1611.

27.

McCoy

S. S.

Dimler

L. M.

Samuels

D. V.

Natsuaki

M. N.

(2019). Adolescent susceptibility to deviant peer pressure: Does gender matter? Adolescent Research Review, 4, 59–71.

28.

Mead

G. H.

(1934). Mind, self, and society. University of Chicago Press.

29.

Miller

W. E.

(1999). Temporal order and causal inference. Political Analysis, 8, 119–146.

30.

Muthén

(1998-2017). Mplus user’s guide (8th ed.). Muthén and Muthén.

31.

Pape

(2012). Young people’s overestimation of peer substance use: An exaggerated phenomenon? Addiction, 107, 878–884.

32.

Paternoster

(2001). Youths and deterrence: Columbia, South Carolina, 1979–1981 [Computer file]. 2nd ICPSR version. University of Maryland [producer], 1985. Inter-university Consortium for Political and Social Research [distributor].

33.

Preacher

K. J.

(2015). Advances in mediation analysis: A survey and synthesis of new developments. Annual Review of Psychology, 66, 825–852.

34.

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.

35.

Reiss

A. J.

(1951). Delinquency as the failure of personal and social controls. American Sociological Review, 16, 196–207.

36.

Roditis

M. L.

Delucchi

Chang

Halpern-Felsher

(2016). Perceptions of social norms and exposure to pro-marijuana messages are associated with adolescent marijuana use. Preventive Medicine, 93, 171–176.

37.

Shadish

W. R.

Cook

T. D.

Campbell

D. T.

(2002). Experimental and quasi-experimental designs for generalized causal inference. Mifflin.

38.

Shulman

E. P.

Smith

A. R

Silva

Icenogle

Duell

Chein

Steinberg

(2016). The dual systems model: Review, reappraisal, and reaffirmation. Developmental Cognitive Neuroscience, 17, 103–112.

39.

Sobel

(1982). Asymptotic confidence intervals for indirect effects in structural equation models. Sociological Methods, 13, 290–312.

40.

Steinberg

(2010). A dual systems model of adolescent risk-taking. Developmental Psychobiology, 52, 216–224.

41.

Sutherland

E. H.

(1947). Principles of criminology (4th ed.). J.B. Lippincott.

42.

Svensson

Pauwels

L. J. R.

Weerman

F. M.

Bruinsma

J. N.

(2017). Explaining individual changes in moral values and moral emotions among adolescent boys and girls: A fixed effects analysis. European Journal of Criminology, 14, 290–308.

43.

Sweeten

(2012). Scaling criminal offending. Journal of Quantitative Criminology, 28, 533–557.

44.

Trucco

E. M.

Colder

C. R.

Wieczorek

W. F.

(2011). Vulnerability to peer influence: A moderated mediation study of early adolescent alcohol use initiation. Addictive Behaviors, 36, 729–736.

45.

Walker

L. J.

Henning

K. H.

Krettenauer

(2000). Parent and peer contexts for children’s moral reasoning development. Child Development, 71, 1033–1048.

46.

Walters

G. D.

(2015). Proactive criminal thinking and the transmission of differential association: A cross-lagged multi-wave path analysis. Criminal Justice and Behavior, 42, 1128–1144.

47.

Walters

G. D.

(2016a). Friends, cognition, and delinquency: Proactive and reactive criminal thinking as mediators of the peer influence and peer selection effects among male delinquents. Justice Quarterly, 33, 1055–1079.

48.

Walters

G. D.

(2016b). Reflected appraisals and self-view in delinquency development: An analysis of retrospective accounts from members of the Racine birth cohorts. Journal of Criminal Justice, 47, 100–107.

49.

Walters

G. D.

(2020). Positive parents and negative peers: Assessing the nature and order of caregiver and friend effects in predicting early delinquency. Youth Violence and Juvenile Justice, 18, 96–114.

50.

Webster

R. G.

Vermeulen

S. C.

(2011). Tolerance for law violations and social projection among offenders and nonoffenders. Criminal Justice and Behavior, 38, 1028–1041.

51.

Zhang

Loeber

Stouthamer-Loeber

(1997). Developmental trends of delinquent attitudes and behaviors: Replications and synthesis across domains, times, and samples. Journal of Quantitative Criminology, 13, 181–215.

		BCBCI
	Estimate	Lower	Upper
Effect from Father PRPD to Delinquency-3
Total effect	0.0087	−0.0132	0.0304
Direct effect	−0.0077	−0.0139	0.0292
Total indirect effect	0.0011	−0.0046	0.0071
Specific indirect effect
F-PRPD-1 → P-AD-2 → Delinquency-3	0.0011	−0.0046	0.0071
Effect from Mother PRPD to Delinquency-3
Total effect	−0.0108	−0.0318	0.0110
Direct effect	−0.0119	−0.0331	0.0096
Total indirect effect	0.0011	−0.0048	0.0066
Specific indirect effect
M-PRPD-1 → P-AD-2 → Delinquency-3	0.0011	−0.0048	0.0066
Effect from Best Friends PRPD to Delinquency-3
Total effect	−0.0047	−0.0170	0.0073
Direct effect	−0.0082	−0.0208	0.0039
Total indirect effect	0.0035	0.0006	0.0068
Specific indirect effect
BF-PRPD-1 → P-AD-2 → Delinquency-3	0.0035	0.0006	0.0068
Preacher-Hayes Contrast Test
Best Friends pathway vs. Father pathway	0.0024	−0.0041	0.0094
Best Friends pathway vs. Mother pathway	0.0024	−0.0037	0.0099