Social Control Versus Social Learning: Self-Efficacy for Future Academic Success and Peer Delinquency as Mediators of the Parental Support

Abstract

The purpose of this study was to determine whether low self-efficacy for future academic success and peer delinquency mediated the relationship between weak parental support and delinquency, and if so, establish the order of mediation. Members of the Flint (Michigan) Adolescent Study (N = 850) served as participants in this study, which compared a pathway based on social control principles (weak parental support → low self-efficacy → peer delinquency → participant delinquency) with a pathway based on social learning principles (weak parental support → peer delinquency → low self-efficacy → delinquency). Path analysis was conducted and revealed that the social control pathway was significant, the social learning pathway was nonsignificant, and the difference between the two pathways was significant. These results suggest that low parental support facilitates delinquency, in part, by lowering self-efficacy for future academic success, a component of the larger construct of self-efficacy for conventional behavior, which, in turn, promotes delinquent peer associations. From a theoretical perspective, these results are more consistent with a social control interpretation of the parental support–delinquency relationship than with a social learning interpretation, although aspects of social learning theory were also supported by the results of this study.

Keywords

parental support self-efficacy for future academic success peer delinquency

In his social control theory of delinquency, Hirschi (1969) proposes that delinquency is an extension of a child’s weak investment in conventional behavior. Formation of a social bond with society and, in particular, with one’s parents can accordingly help prevent delinquency by increasing the child’s investment in a conventional lifestyle. Akers (1998), by contrast, believes that delinquency is learned in close association with those already involved in crime. Hence, while social learning theory holds that children learn to commit crime, social control theory maintains that children learn not to commit crime. Each model offers a different approach to prevention, with social learning theory seeking to place limits on peer influence and social control theory stressing improved parental support. Whereas the social control and social learning theories of crime have been adequately described by Hirschi (1969) and Akers (1998), respectively, the mechanisms by which social control and social learning factors impact on delinquent behavior are less well understood. In that parenting and peer factors appear to be instrumental in shaping future delinquent outcomes, both individually and in combination (Brown & Bakken, 2010), additional research is required to clarify the relationship between these two key sources of social influence and the mechanisms responsible for their relationship.

Parenting and Peers

There is little doubt that parental control/support and peer delinquency are central to the development of delinquent behavior. When various social–environmental correlates of antisocial behavior are compared, parental control/support and delinquent peer associations are often at the top of the list of socioenvironmental variables known to predict conduct disorder, delinquency, and crime (Lipsey & Derzon, 1998; Petrosino, Derzon, & Lavenberg, 2009; Wasserman et al., 2003). Moreover, parental control and support have been found to predict delinquency after controlling for peer influence (Johnson, Giordano, Manning, & Longmore, 2011), and peer influence has been found to predict delinquency net the effects of parental control and support (Henneberger, Tolan, Hipwell, & Keenan, 2014). Rather than uncertainty over the ability of parental control/support and peer influence to affect delinquent outcomes or questions about the existence of a relationship between these two sources of social environmental influence, what is really at issue is the direction, context, and nature of the well-documented connection between parental control/support and peer influence, particularly when it comes to using these two variables to predict future delinquent behavior.

An assumption researchers investigating the parenting–peer relationship often make is that parental control and support condition the effect of delinquent peers on subsequent delinquent behavior (Poole & Regoli, 1979). In other words, it is assumed that parenting factors precede peer factors in the development of a delinquent lifestyle and that strong parental control and support protect a child against negative peer influences, whereas weak parental control and support leave a child vulnerable to negative peer influences. It is understandable that such an assumption would be made, given that young children are nearly always more closely involved with their parents than they are with their peers. In fact, research indicates that it is not until late childhood or early adolescence that peer effects begin to catch up and eventually exceed parenting effects (Rubin, Bukowski, & Parker, 2006). Researchers who have examined the direction of the parenting–peer relationship have more often than not determined that parental control and support precede peer influence effects rather than vice versa (Forgatch et al., 2016; Simons, Wu, Conger, & Lorenz, 1994), although exceptions have been noted (see, e.g., Trucco, Colder, Wieczorek, Lengua, & Hawk, 2014).

The parenting–peer relationship may also vary as a function of context. Although any variable could potentially moderate the parenting–peer relationship, the moderating effects of gender on the parenting–peer relationship have received the greatest amount of attention thus far. This may be because there is evidence that gender can affect both the parenting–delinquency and peer–delinquency relationships. Studies show that while boys tend to be more sensitive to negative peer effects than girls (Augustyn & McGloin, 2013; Piquero, Gover, MacDonald, & Piquero, 2005), girls tend to be more sensitive to positive and negative family influences than boys (Silverman & Caldwell, 2005; Walters, 2013). Evaluating parental control/support and peer deviance as predictors of participant delinquency, Bowman, Prelow, and Weaver (2007) discovered that parental control and support predicted delinquency in girls but not boys and peer deviance predicted delinquency in boys but not girls. In a study examining gender moderation of the mediating effect of opposite-sex friendships on the parental monitoring–delinquency relationship, Poulin and Denault (2012) discovered that while parental monitoring reduced delinquency by limiting opposite-sex friendships in girls, there was no mediating effect in boys. Given the degree to which the relationships between parenting factors and delinquency, peer factors and delinquency, and all three variables combined, gender moderation was tested as part of the current investigation.

A third aspect of the parenting–peer relationship for which there is limited knowledge is the mechanism of effect between these two socioenvironmental factors. A common mechanism of effect is mediation of the relationship by one or more “third variables.” In an early study on this issue, Warr (1993) determined that attachment to parents did not have a direct effect on peer influence but rather acted indirectly by inhibiting the formation of delinquent friendships, which, in turn, reduced the influence of delinquent peers on the child’s behavior. More recently, Deutsch, Crockett, Wolff, and Russell (2012) discovered that while parental control inhibited offspring delinquency by reducing the impact of delinquent peers on youth behavior, maternal support had both a direct and indirect (via a reduction in peer influence) effect on delinquency. Several other studies have determined that the parental control/support–delinquency relationship is mediated not only by peer deviance but by social cognitive variables like antisocial attitudes, impulsive thinking, and hostile attribution biases (Janssen, Eichelsheim, Deković, & Bruinsma, 2016; Simons, Simons, Chen, Brody, & Lin, 2007). In fact, after accounting for a number of these social cognitive mediators in their structural equation model, Simons, Simons, Chen, Brody, and Lin (2007) discovered that the peer influence effect disappeared.

Social Cognitive Variables and Mediation

Social cognitive variables are internal processes by which an individual interacts with his or her social environment. According to social cognitive theory, a person learns by observing the behavior of others, developing competencies, establishing goals, and responding to feedback (Bandura, 1986). Social cognitive theory highlights several different social cognitive variables. One such variable, that is, self-efficacy, served as a mediator in the current investigation. Self-efficacy is defined by Bandura (1995) as “the belief in one’s capabilities to organize and execute the courses of action required to manage prospective situations” (p. 2). There are several key elements to this definition. First, self-efficacy, like all social cognitive variables, is a belief or internal process. Second, self-efficacy is an appraisal of one’s competency to successfully perform a specific action. Third, self-efficacy highlights one’s ability to manage future situations. Self-efficacy, as the reader may have gathered, shares a great deal in common with self-confidence, but where self-confidence is a general expectancy of success in a wide variety of activities and situations, self-efficacy is a more specific expectancy of success in managing a much narrower range of tasks and future scenarios.

Self-efficacy has received only sporadic attention from researchers in the criminology and criminal justice fields. Self-esteem, by comparison, has played a much larger role in criminological and criminal justice theory, research, and practice (Belknap & Holsinger, 2006; Oser, 2006; Wells, 1989). Although there is some overlap between self-efficacy and self-esteem, they are separate concepts. For one, self-esteem is a perception of one’s worth, whereas self-efficacy emphasizes behavioral competencies. For another, self-esteem, like self-confidence, is more general and less situationally specific than self-efficacy. Finally, while self-efficacy is considered a criminogenic need by risk assessment researchers, self-esteem is not (Andrews & Bonta, 2010). Several other concepts that figure prominently in criminological theory have also been confused with self-efficacy. Low self-control (Gottfredson & Hirschi, 1990) refers to a behavioral pattern of impulsivity and poor self-restraint; self-efficacy, on the other hand, is an attitude about one’s ability to successfully perform a particular behavioral competency. An attitudinal concept from social learning theory, definitions favorable to violations of the law (Sutherland, 1947), differs from self-efficacy in that where the former focuses on criminal attitudes, the later emphasizes noncriminal competencies that serve as protective factors against the negative influence of these criminal attitudes.

The version of self-efficacy examined in the current study was self-efficacy for future academic success. This falls somewhere between self-efficacy to perform specific academic tasks, a process known as academic self-efficacy (Bandura, 1993), and self-efficacy to achieve a conventional lifestyle, a more general form of self-efficacy that has been studied previously in research on delinquency development. In the first of two studies on self-efficacy to achieve a conventional lifestyle, self-efficacy was measured by success in getting a good job, earning a good living, having a good marriage, and graduating from college (Walters, 2018b). Analyses showed that self-efficacy to achieve a conventional lifestyle successfully mediated the past delinquency–future delinquency relationship, whereas general confidence in one’s ability to avoid legal trouble did not. In a second study, self-efficacy was measured by success in achieving one’s desired education and in achieving one’s desired occupation (Walters, 2018c). According to the results of this second study, self-efficacy to achieve a conventional lifestyle mediated the relationships between parental knowledge and delinquency and parental attachment and delinquency, but self-efficacy for deviance failed to mediate either relationship. It has also been noted that future life certainty (Brezina, Tekin, & Topali, 2009) and expectations of graduating from college (Caldwell, Wiebe, & Cleveland, 2006) predicted lower levels of future delinquency in a large group of general population adolescents, whereas a strong future orientation inhibited reoffense in a group of serious delinquents (Ozkan, 2016).

Why use the term self-efficacy to explain crime deterrence instead of social control terms like educational expectations and commitment to long-term goals (Hirschi, 1969)? There are three answers to this question. First, self-efficacy is a more flexible term that can be applied to a wider range of situations. Whereas academic self-efficacy was studied in the current study, this is a core component of self-efficacy for a conventional lifestyle (Walters, 2018b, 2018c), a principal mediator of variable relationships in lifestyle theory. Second, self-efficacy lends itself to specific situations better than educational aspirations and commitment to long-term goals. Hence, the links between self-efficacy and such antecedent and postcedent conditions as parents and peers are clearer, and the demarcations between conditions are sharper. In one study, perceived parental support predicted academic self-efficacy in a group of low-income Hispanic youth whereas perceived peer support had no effect (Ramirez, Machida, Kline, & Huang, 2014). Prior to this, perceived parental support was found to correlate with academic self-efficacy and both were discovered to correlate with decreased academic cheating (Bong, 2008). Third, self-efficacy theory identifies four etiological pathways to self-efficacy: past performance accomplishments, modeled behavior, social persuasion/feedback, and physiological states (Bandura, 1995). Past performance accomplishments, the most common pathway to self-efficacy, were constrained in the current study by controlling for prior academic performance, so that if an effect was observed, it could be more properly attributed to parental support, by way of modeling, persuasion, or a change in physiological response.

The purpose of the current investigation was to evaluate whether self-efficacy for future academic success is capable of mediating another important criminological relationship, namely, the connection between parental support and peer influence. Building on the previous Walters (2018c) study and the control and moral models of criminal lifestyle development (Walters, 2017b), the current study sets out to determine whether parental support serves a protective function against negative peer influence. Although children learn both positive and negative behaviors from their friends, the principal role of peers in the parent–peer interface according to criminal lifestyle theory is an increase in risk for future criminal behavior through impartation of criminal attitudes and beliefs (Walters, 2017b). Parents can also have a positive or negative impact on a child, although their principal role in the parent–peer interface for crime is teaching the child self-control skills and moral values that serve to buffer the criminogenic risk presented by delinquent peer associations. Self-efficacy beliefs about future academic success (Walters, 2018c) should accordingly derive from parents rather than from peers if parents are serving a protective function and peers a risk function.

Current Study

The current study contrasted a social control model (low parental support produces weak academic self-efficacy, which then leads to the peer influence effect) and a social learning model (low parental support encourages peer delinquency, which then leads to low academic self-efficacy and delinquency). Using the comparison pathways approach to mediation (Walters, 2018a), the current investigation compared a target pathway based on social control theory principles (parental support → low self-efficacy → peer delinquency → participant delinquency), with a comparison pathway based on social learning theory principles (parental support → peer delinquency → low self-efficacy → participant delinquency). If, in fact, parenting serves a protective function, then it makes more sense for low parental support to produce weak academic self-efficacy than for peer delinquency to produce weak academic self-efficacy. It was therefore hypothesized that the social control model would be significant, the social learning model would be nonsignificant, and that the two models would differ significantly from one another. Prior to conducting the main analysis, gender moderation across the three paths of the target pathway was evaluated and found to be nonsignificant. In the main analysis, three demographic variables known to correlate with delinquency (i.e., age, gender, and race) and academic performance were controlled, the latter to rule out academic competence as an alternate explanation for the effect of academic self-efficacy on delinquency (i.e., students who perform well in school will tend to have higher academic self-efficacy and lower delinquency).

Method

Participants

The sample for the current study contained all 850 members of the Flint Adolescent Study (FAS; Zimmerman, 2014), a four-wave longitudinal investigation that began in 1994 when participants were in the ninth grade and ended 4 years later in 1997. Members of the FAS came from four public high schools in Flint, MI, and were considered at risk for school dropout because of a below average grade point average (<3.0). Participants ranged in age from 14 to 16 (M = 14.55, SD = .64) at Wave 1 of the FAS and were equally divided between males and females (425 each). The ethnic breakdown for the sample was 80.1% African American, 16.8% White, and 3.1% mixed (African American and White).

Measures

The independent variable for this study was Wave 1 parental support measured with a 6-item scale. The 6 items that formed the parental support scale were as follows: (1) “My parents enjoy hearing about what I think,” (2) “I rely on my parents for emotional support,” (3) “My parents are good at helping me solve problems,” (4) “I have a deep sharing relationship with my parents,” (5) “My parents encourage me to stay in school,” and (6) “I rely on my parents for moral support.” Participants rated each item on a 5-point scale (1 = not true, 2 = little true, 3 = somewhat true, 4 = pretty true, and 5 = very true), which when summed produced a score that could range from 6 to 30. The score was then inverted (multiplied by −1) to make it compatible with the majority of other measures included in this study, where higher scores reflected poorer adjustment. Because all of the scales used in the current study were short (<10 items), the mean interitem correlation was provided along with Cronbach’s α coefficient in assessing internal consistency. According to Clark and Watson (1995), the mean interitem correlation should fall somewhere between .15 and .50 for adequate internal consistency, whereas correlations in the .40 to .50 range are considered optimal for scales measuring narrow concepts like those found in the current investigation. The 6-item parental support scale achieved an α of .87 and a mean interitem correlation of .52 at Wave 1 of the FAS.

The dependent variable for this study was Wave 4 delinquency. In this portion of the survey, participants were instructed to use a 5-point scale (1 = no times, 2 = 1 time, 3 = 2 times, 4 = 3 times, and 5 = 4 or more times) to rate how often they engaged in the following 17 delinquent acts over the past year: hit a teacher or work supervisor, got into a fight at school, took part in a fight between a group of friends and another group of individuals, hurt someone bad enough they needed bandages or a doctor, used a knife or gun or weapon to get something from a person, took something that did not belong to you worth <US$50, took something that did not belong to you worth >US$50, took something from a store without paying for it, took car that didn’t belong to your family without permission, took parts from a car without permission of the owner, went into some house or building you weren’t supposed to enter, set fire to someone’s property on purpose, damaged school property on purpose, sold an illegal drug, got in trouble with the police, carried a knife or razor, and carried a gun. These items, when summed, yielded a total delinquency score that could range from 17 to 85. The 3-year (Wave 1–Wave 4) test–retest reliability (r) of the total delinquency score was .44.

There were two mediator variables included in the current investigation: self-efficacy for future academic success and peer delinquency. Self-efficacy for future academic success was measured with 2 items: “likelihood you will graduate from high school” and “likelihood you will go to a 4-year college” (phrased “likelihood you will go to trade school or college” in the Wave 1 survey). Each item on the self-efficacy scale was rated on a 5-point scale with the following descriptive anchors (1 = not at all likely, 3 = somewhat likely, and 5 = very likely) and then inverted (multiplied by −1) for the purposes of the correlational and regression analyses. The 2 items correlated .38–.43 (Pearson Product Moment Correlation) with each other. Peer delinquency was measured with a self-report estimate of how many friends (1 = none, 2 = some, 3 = many, 4 = most, and 5 = all) engage in the following delinquent behaviors: get into fights, suspended from school, carry a knife or razor, carry a gun, shoplift from stores, busted for selling drugs, and busted for burglary or robbery. The 7-item peer delinquency scale displayed internal consistency estimates of .81–.83 (α) and .38–.42 (mean interitem r).

There were four control variables included in the present investigation: age (in years), gender (1 = male and 2 = female), race (1 = non-White, 2 = White), and average grade at Wave 1. The average grade variable was assessed with a 9-point scale on which respondents were asked to select the option that “best describes your average grade” (1 = A [93–100], 2 = A− [90–92], 3 = B+ [87–89], 4 = B [83–86], 5 = B− [80–82], 6 = C+ [77–79], 7 = C [73–76], 8 = C− [70–72], and 9 = D or lower [<70]). When the independent and mediator variables in a mediation analysis are not randomly assigned, as was the case in the current investigation, it is recommended that prior measures of a predicted variable be included in the regression equation predicting that variable in order to be more confidence in the causal direction of one’s results (Cole & Maxwell, 2003). Accordingly, prior or precursor measures of the first-stage mediators, second-stage mediators, and dependent variable were included in the relevant regression equations.

Research Design

The current study employed a fixed-sample four-wave longitudinal panel design in which the independent and control variables (parental support, age, sex, race, and average grade) along with the precursors to the first-stage mediators (self-efficacy for future academic success and peer delinquency) and dependent variable (participant delinquency) were assessed at Wave 1, the first-stage mediators (self-efficacy and peer delinquency) were assessed at Wave 2, the second-stage mediators (peer delinquency and self-efficacy) were assessed at Wave 3, and the dependent variable (delinquency) was assessed at Wave 4. All possible pathways between the independent and dependent variables were explored. Wave 1 of the FAS was administered in 1994, with yearly follow-ups in 1995, 1996, and 1997. Because there was no overlap between waves, the current design qualifies as prospective in nature.

Procedure

A five-regression multiple path analysis was performed using a maximum likelihood estimator from MPlus 5.2 (Muthén & Muthén, 1998–2007). The first equation regressed Wave 2 self-efficacy onto parental support, Wave 1 self-efficacy, and the four control variables. The second equation regressed Wave 2 peer delinquency onto parental support, Wave 1 peer delinquency, and the four control variables. The third equation regressed Wave 3 self-efficacy onto Wave 2 peer delinquency, Wave 2 self-efficacy, parental support, and the four control variables, and the fourth equation regressed Wave 3 peer delinquency onto Wave 2 self-efficacy, Wave 2 peer delinquency, parental support, and the four control variables. The fifth and final equation regressed Wave 4 delinquency onto Wave 3 peer delinquency, Wave 3 self-efficacy, Wave 2 peer delinquency, Wave 2 self-efficacy, Wave 1 delinquency, parental support, and the four control variables. In a two-stage serial mediator design, the path from the independent variable to the first-stage mediator is labeled a, the path from the first-stage mediator to the second-stage mediator is labeled d, and the path from the second-stage mediator to the dependent variable is labeled b.

The comparison pathways’ approach was used to test the significance of the target and control pathways and determine whether one was significantly stronger than the other. The significance of the target or social control (Parental Support 1 → Self-Efficacy 2 → Peer Delinquency 3 → Participant Delinquency 4) and comparison or social learning (Parental Support 1 → Peer Delinquency 2 → Self-Efficacy 3 → Participant Delinquency 4) pathways was tested with bias-corrected bootstrapped confidence intervals (CIs), and the difference between the two pathways was evaluated using Preacher and Hayes’s (2008) contrast test. The comparison pathways’ approach takes a theoretically derived sequence of variables known as the target pathway and contrasts it with an alternate theoretical or less theoretically viable sequence of variables. This provides a means by which the effects of theoretically relevant mediational chains can be compared to the effect of alternate or less theoretically relevant mediational chains. The target and comparison paths in the current study included the same variables with the position of the two mediator variables switched, that is, self-efficacy preceding peer delinquency in the target pathway and peer delinquency preceding self-efficacy in the comparison pathway.

Nonparametric bias-corrected bootstrapping (b = 5,000) was used to construct 95% CIs that were then used to test the significance of indirect effects. Individual effects with a 95% CI that did not include zero were considered statistically significant. Research denotes that using bias-corrected bootstrapping is superior to normal theory (i.e., z tests) in testing indirect effects because it does a better job of modeling the nonnormality of the indirect effect and handling nonnormality in the dependent variable (Hayes, 2013; Pituch & Stapleton, 2008; Rucker, Preacher, Tormala, & Petty, 2011).

Kenny’s (2013) “failsafe ef” procedure was used to test the sensitivity of significant indirect effects. The failsafe ef is calculated with the following formulae: $(r_{m y . x}) \times (s d_{m . x}) \times (s d_{y . x}) / (s d_{m}) \times (s d_{y})$ , where r_my.x represents the partial correlation between the mediator (m) and dependent variable (y) controlling for the independent variable (x), sd_m.x represents the standard deviation of m controlling for x, sd_y.x represents the standard deviation of y controlling for x, sd_m represents the standard deviation for m, and sd_y represents the standard deviation for y. The correlation produced by the failsafe ef indicates the degree to which an unobserved covariate confounder or omitted variable would need to correlate with the mediator and dependent variable, controlling for both the independent and mediator variables in the case of the latter, to eliminate the b or d path (running from the second-stage mediator to the dependent variable or from the first-stage mediator to the second-stage mediator) of a significant indirect effect (D. A. Kenny, personal communication, February 12, 2016). This same basic formula is used to test sensitivity in Imai’s popular R-based mediation program (Imai, Keele, & Tingley, 2010).

Missing Data

Complete data were available for 86.2% of the sample, with 4.7% of participants missing data on one variable, 5.7% of participants missing data on two or three variables, and 3.4% of participants missing data on four to seven variables. Three variables had more than 5% missing data: Self-Efficacy 3 (7.9%), Peer Delinquency 3 (8.1%), and Delinquency 4 (9.4%). Full information maximum likelihood (FIML) was used to handle the missing data in this study. FIML estimates model parameters and standard errors from available data and then uses this information to calculate parameters and standard errors for the entire sample. The parameter and standard error estimates attained with FIML are significantly less biased than the parameter and standard error estimates achieved with listwise deletion or simple imputation (Allison, 2012; Peyre, Leplége, & Coste, 2011).

Results

Descriptive statistics for the 13 variables included in the current investigation are listed in Table 1. The variables in this study correlated reasonably well with one another. In fact, over two thirds of the correlations achieved Bonferroni-corrected significance. There was no evidence of multicollinearity in any of the five regression equations (tolerance = .611–.987, variance inflation factor = 1.013–1.636), and there was no indication that sex moderated any of the variable relationships proposed in the target pathway.

Table 1.
Descriptive Statistics and Correlations for the 13 Independent, Dependent, Mediator, and Control Variables Included in the Current Study.

Variable n M SD Range 2 3 4 5 6 7 8 9 10 11 12 13

1. Age 850 14.55 0.64 14–16 −.10 −.02 .19* .02 .12* .17* .14* .27* .20* .20* .18* .08

2. Sex 850 1.50 0.50 1–2 −.04 −.09 .12* .00 −.01 −.08 −.09 −.09 −.08 −.16* −.15*

3. Race 850 1.17 0.37 1–2 .05 .07 −.15* −.13* −.09 .04 .13* .14* −.08 −.01

4. Average grade 848 5.27 1.90 1–9 .07 .17* .16* .15* .38* .34* .30* .25* .17*

5. Parental support 849 24.29 5.39 6–30 .17* .10 .13* .16* .14* .16* .19* .13*

6. Peer 1 849 13.84 5.20 7–32 .53* .49* .07 .11 .11 .51* .31*

7. Peer 2 812 14.04 5.22 7–34 .60* .16* .17* .11 .43* .36*

8. Peer 3 781 12.91 4.76 7–32 .17* .19* .16* .46* .44*

9. Self-Efficacy 1 848 8.84 1.56 2–10 .47* .43* .21* .17*

10. Self-Efficacy 2 812 8.61 1.54 2–10 .58* .17* .20* .

11. Self-Efficacy 3 783 8.54 1.55 2–10 .18* .17*

12. Delinquency 1 848 22.52 7.90 17–74 .44*

13. Delinquency 4 770 21.04 7.18 17–68

Note. Parental support and self-efficacy items were inverted (multiplied by −1) prior to correlating them with other variables; n = number of participants with nonmissing data; M = mean; SD = standard deviation; range = range of scores in the current sample; age = chronological age in years at Wave 1; sex = 1 (male) or 2 (female); race = 1 (White) or 2 (non-White); average grade = self-reported average grade at Wave 1; parental support = parental support at Wave 1; Peer 1 = peer delinquency at Wave 1; Peer 2 = peer delinquency at Wave 2; Peer 3 = peer delinquency at Wave 3; Self-Efficacy 1 = self-efficacy for future academic success at Wave 1; Self-Efficacy 2 = self-efficacy for future academic success at Wave 2; Self-Efficacy 3 = self-efficacy for future academic success at Wave 3; Delinquency 1 = participant delinquency at Wave 1; Delinquency 4 = participant delinquency at Wave 4.

p < .00064 (Bonferroni-corrected α level; .05/.78 correlations).

The results of the five-regression path analysis are summarized in Table 2. As indicated, the coefficients for all three paths of the target pathway were significant or approached significance, whereas none of the paths for the control pathway was significant (see Figure 1). Testing the indirect effects of the target and control pathways with bootstrapped CIs revealed that only the indirect effect for the target pathway was significant. According to the results of the Preacher–Hayes’s contrast test, the target (social control) pathway achieved significantly better results than the comparison (social learning) pathway (see Table 3).

Table 2.
Maximum Likelihood Path Analysis of the Parental Support–Delinquency Relationship With Self-Efficacy for Academic Success and Peer Delinquency as Cross-Lagged Mediators.

Predictor b (95% CI) β z p

Self-Efficacy 2 (outcome)

Parental support 0.019 [0.001, 0.039] .066 1.95 .051

Age 0.185 [0.022, 0.358] .076 2.17 .030

Sex −0.143 [−0.327, 0.032] −.046 −1.55 .122

Race 0.404 [0.136, 0.674] .097 2.96 .003

Average grade 0.147 [0.092, 0.197] .180 5.51 <.001

Self-Efficacy 1 0.372 [0.290, 0.445] .374 9.58 <.001

Peer Delinquency 2 (outcome)

Parental support 0.020 [−0.038, 0.076] .020 0.66 .507

Age 0.756 [0.251, 1.268] .093 2.92 .004

Sex −0.120 [−0.741, 0.498] −.012 −0.38 .700

Race −0.935 [−1.607, −0.235] −.067 −2.67 .008

Average grade 0.152 [−0.007, 0.322] .055 1.81 .071

Peer Delinquency 1 0.505 [0.430, 0.576] .501 13.58 <.001

Self-Efficacy 3 (outcome)

Peer Delinquency 2 0.000 [−0.018, 0.017] −.001 −0.05 .961

Age 0.209 [0.054, 0.367] .086 2.65 .008

Sex −0.108 [−0.286, 0.067] −.034 −1.20 .232

Race 0.241 [0.019, 0.459] .058 2.09 .036

Average grade 0.082 [0.027, 0.138] .100 2.87 .004

Parental support 0.021 [0.005, 0.039] .074 2.46 .014

Self-Efficacy 2 0.523 [0.448, 0.605] .519 12.98 <.001

Peer Delinquency 3 (outcome)

Self-Efficacy 2 0.219 [0.000, 0.451] .072 1.89 .059

Age 0.153 [−0.317, 0.610] .021 0.65 .515

Sex −0.745 [−1.303, −0.212] −.078 −2.68 .007

Race −0.435 [−1.087, 0.200] −.034 −1.33 .184

Average grade 0.084 [−0.070, 0.246] .034 1.04 .299

Parental support 0.057 [0.001, 0.118] .065 1.90 .057

Peer Delinquency 2 0.525 [0.467, 0.582] .578 17.51 <.001

Delinquency 4 (outcome)

Peer Delinquency 3 0.352 [0.197, 0.532] .238 4.14 <.001

Self-Efficacy 3 0.103 [−0.309, 0.540] .023 0.48 .633

Peer Delinquency 2 0.152 [0.006, 0.303] .113 2.04 .041

Self-Efficacy 2 0.290 [−0.139, 0.736] .064 1.30 .193

Age −0.497 [−1.300, 0.313] −.045 −1.19 .234

Sex −1.339 [−2.280, −0.450] −.095 −2.92 .004

Race 0.488 [−0.521, 1.553] .026 0.91 .361

Average grade 0.100 [−0.145, 0.340] .027 0.82 .415

Parental support 0.047 [−0.045, 0.150] .036 0.95 .343

Delinquency 1 0.226 [0.115, 0.352] .253 3.70 <.001

Age with parental support 0.084 [−0.143, 0.321] .024 0.72 .474

Sex with parental support 0.321 [0.146, 0.503] .119 3.52 <.001

Race with parental support 0.146 [0.020, 0.272] .072 2.25 .025

Grade with parental support 0.786 [0.092, 1.490] .077 2.20 .028

Self-Efficacy 2 with Peer 2 0.482 [0.027, 0.954] .084 2.04 .041

Self-Efficacy 3 with Peer 3 0.210 [−0.171, 0.615] .046 1.04 .298

Note. N* = 850. Outcome = outcome measure for that particular regression equation; with = covariance; age = chronological age in years at Wave 1; sex = 1 (male) or 2 (female); race = 1 (White) or 2 (non-White); Average grade = self-reported average grade at Wave 1; parental support = parental support at Wave 1; Peer 1 = peer delinquency at Wave 1; Peer 2 = peer delinquency at Wave 2; Peer 3 = peer delinquency at Wave 3; Self-Efficacy 1 = self-efficacy for future academic success at Wave 1; Self-Efficacy 2 = self-efficacy for future academic success at Wave 2; Self-Efficacy 3 = self-efficacy for future academic success at Wave 3; Delinquency 1 = participant delinquency at Wave 1; Delinquency 4 = participant delinquency at Wave 4; 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; z = Wald Z-test statistic; p = significance level of the Wald Z-test statistic.

Figure 1.
Maximum likelihood path analysis showing the mediating effects of Wave 2 peer delinquency and self-efficacy for future academic success and Wave 3 peer delinquency and self-efficacy for future academic success on the parental support–delinquency relationship. Standardized β coefficients are reported; covariance between peer delinquency and self-efficacy at Wave 2 = .08, p < .05; covariance between peer delinquency and self-efficacy at Wave 3 = .05, p = .30; control and delinquency and first-stage mediator precursor variables are not shown. ‡p < .10. *p < .05. **p < .001.

Table 3.
Total, Direct, and Indirect Effects for the Pathway Running From Parental Support to Participant Delinquency.

Pathways BCBCI

Estimate Lower Upper

CU traits → Wave 3 delinquency

Total effect .08433 −.01473 .19269

Direct effect .04722 −.04495 .15032

Total indirect effect .03711 .00830 .07076

Specific indirect effects

Support 1 → Peer 2 → Delinquency 4 .00299 −.00467 .01758

Support 1 → Peer 3 → Delinquency 4 .02025 .00209 .05350

Support 1 → SE 2 → Delinquency 4 .00553 −.00132 .02174

Support 1 → SE 3 → Delinquency 4 .00222 −.00632 .01478

Support 1 → Peer 2 → Peer 3 → Delinquency 4 .00363 −.00674 .01618

Support 1 → SE 2 → Peer 3 → Delinquency 4 .00147 .00004 .00530

Support 1 → Peer 2 → SE 3 → Delinquency 4 .00000 −.00022 .00011

Support 1 → SE 2 → SE 3 → Delinquency 4 .00103 −.00249 .00867

Preacher–Hayes’s contrast test .00147 .00004 .00540

Note. N = 850. Support 1 = parental support at Wave 1; Peer 2 = peer delinquency at Wave 2; Peer 3 = peer delinquency at Wave 3; SE 2 = self-efficacy for future academic success at Wave 2; SE 3 = self-efficacy for future academic success at Wave 3; Delinquency 4 = participant delinquency at Wave 4; Preacher–Hayes’s contrast test = test of the difference between the target (Support 1 → SE 2 → Peer 3 → Delinquency 4, in boldface) and control (Support 1 → Peer 2 → SE 3 → Delinquency 4) indirect effects using the test described in Preacher and Hayes (2008); BCBCI = bias-corrected bootstrapped 95% confidence interval (b = 5,000); estimate = unstandardized point estimate; lower = lower boundary of the 95% confidence interval; upper = upper boundary of the 95% confidence interval.

Sensitivity testing using Kenny’s failsafe ef procedure revealed that the indirect effect of the target pathway was moderately robust to the effects of unobserved covariate confounders. Because sensitivity testing can only be tested on one mediator at a time, two different failsafe ef sensitivity tests were performed, one on the first-stage mediator and one on the second-stage mediator. The results indicated that a “third variable” would need to correlate .18 with the first-stage mediator (SE 2) and .18 with the second-stage mediator (Peer 3), controlling for Support 1 and SE 2 in the case of Peer 3, to eliminate the significant effect of the first-stage mediator on the second stage mediator (referred to as the d path). It would then need to correlate .37 with the second-stage mediator (Peer 3) and .37 with the dependent variable (Delinquency 4), controlling for SE 2 and Peer 3 in the case of Delinquency 4, to eliminate the significant effect of the second-stage mediator on the dependent variable along the b path.

Discussion

Findings from the present investigation lend support to the hypothesis that low self-efficacy for future academic success mediated the parental support–peer influence relationship by linking parental support to peer delinquency. Although prior research indicates that self-efficacy for future academic success correlates strongly with academic aspirations and mildly with delinquency (Carroll et al., 2009), data from the current study showed that self-efficacy for future academic success acted as a bridge between parental support and delinquency through its common association with parental support and delinquent peers. These results suggest that weak parental support reduces self-efficacy for future academic success, which, in turn, leaves a child vulnerable to the negative effects of delinquent peer associations. One possibility is that low self-efficacy for future academic success can make delinquent peers, who, as a group, tend to have low academic aspirations (DiPierro, Fite, Cooley, & Poquiz, 2016), more attractive to the individual. On a more positive note, it also means that strong parental support can enhance self-efficacy for academic success, which, in turn, protects the child against negative peer influence by making delinquent peers seem less attractive. Either way, these results are more consistent with a social control interpretation of the parental support–delinquency relationship than with a social learning interpretation, given that while self-efficacy predicted peer delinquency, peer delinquency failed to predict self-efficacy.

The indirect effects observed in the current study were small, as is nearly always the case when mediation analysis is performed. It is important to understand, however, that size and meaningfulness are rarely related in research on mediation. In fact, a large ratio of indirect to direct effects, commonly referred to as the PM ratio (Wen & Fan, 2015), is as likely to reflect a statistical confound as it is to signal a meaningful indirect effect. There are several reasons why indirect effects in a mediation analysis are nearly always small: from loss of information and reduced fidelity resulting from time decay, to use of less than maximally reliable indicators, coupled with the action of suppressor variables and extensive statistical control with the introduction of precursor measures (Walters, 2018a). More so, indirect effects usually assume a nonnormal distribution, which is why indirect effects should be evaluated against bias-corrected or percentile bootstrapped CIs rather than with normal theory procedures like the Sobel test. Likewise, the traditional Baron and Kenny (1986) four-step procedure for evaluating the significance of indirect effects is problematic at all four steps and should not be used to evaluate mediation results (Hayes, 2009, 2013). The results from the current study, for instance, would not have satisfied the second (significant coefficient along the a path) and third (significant coefficient along the d path) steps of Baron and Kenny’s procedure, which would have led to premature rejection of the target pathway. Given that meaningful effects in a mediation analysis are characteristically small, it is imperative that researchers follow proper procedures when conducting such analyses.

Implications

The hypothesis tested in this study held that a pathway based on social control principles (parental support → self-efficacy → peer delinquency → participant delinquency) would prove more viable than one based on social learning principles (parental support → peer delinquency → self-efficacy → participant delinquency). Support for this hypothesis was found in the results of a five-equation path analysis, in which the target or social control pathway was significant, the comparison or social learning pathway was nonsignificant, and the target and comparison pathways differed significantly from one another. This should not be interpreted to mean that the current results are incompatible with social learning theory. Rather, they suggest that self-efficacy for future academic success does a significantly better job of mediating the parental support–peer delinquency relationship than it does of mediating the peer delinquency–participant delinquency relationship. Simply put, children are more apt to acquire high self-efficacy for future academic success from their interactions with supportive parents than they are to acquire low self-efficacy for future academic success from their interactions with delinquent peers. This does not mean that social learning principles were without support in this study. After all, the construct of self-efficacy comes from Bandura’s (1986) social cognitive theory, an offshoot of social learning theory.

Another theoretical implication of the current results is that not all variables are equally capable of mediating relationships. A widely held belief in psychology is that cognitive variables make good mediators because of their malleability and reasonably good short-term stability (Bandura, 1986; Wu & Zumbo, 2008). There is, in fact, evidence from research on delinquency that while social cognitive variables are capable of mediating the relationship between past and future behavior, behavioral variables are much less capable of mediating the relationship between past and future cognition (Walters, 2016, 2017a). In the current study, self-efficacy mediated the relationship between parental support and peer delinquency, but peer delinquency failed to mediate the relationship between parental support and self-efficacy. Whereas peer delinquency technically mediated the relationships between parental support and delinquency and between low self-efficacy and delinquency, it is questionable whether peer delinquency actually mediated either relationship. In both instances, peer delinquency was functioning as the front end of a process, in this case, the peer influence effect (peer delinquency leading to participant delinquency). When it was tested as a mediator of the parental support–self-efficacy relationship, it failed to produce an effect. This underscores the importance of distinguishing between a true mediator and a variable like peer delinquency that serves as the front end of a two-variable process.

Parent-based interventions for delinquency have traditionally focused on improving parental control through parent training. Training parents to be more consistent and inductive in parenting their children has been found to be effective in alleviating and preventing early delinquency (Piquero, Farrington, Welsh, Tremblay, & Jennings, 2009). Parental control and support, it should be noted, are equally important in promoting delinquency (Hoeve et al., 2009) and may be equally effective in promoting desistance (Sampson & Laub, 1993). One study, in fact, ascertained that both weak parental control and support were linked to delinquency by way of low self-efficacy to achieve a conventional lifestyle (Walters, 2018c). This suggests that programs like parent training that seeks to improve parental control need to be supplemented by programs designed to enhance parental support. There is very little in the literature, however, to serve as a guide for treatment providers. Items from the parental support measure used in the present study could nevertheless offer clues as to how parental support can be enhanced. The 6 items from the parental support scale mentioned parents as sources of moral and emotional support, parental interest in a child’s goals and ideas, development of a meaningful parent–child relationship, parental assistance in solving problems, and parental encouragement of a child remaining in school. Using these issues as guides, it may be possible to develop a family-based intervention that fosters parental support and complements existing programs of parental control.

Limitations

There are several limitations to the current study that could affect the interpretation of results. Chief among these is the fact that the principal mediator covered only a portion of the conceptual space previously used to measure self-efficacy to achieve a conventional lifestyle. This makes comparing results across studies difficult. Academic aspirations and performance were of maximal interest to those who created the FAS, and this is why the only items available as measures of self-efficacy in this study were those that addressed future academic success (i.e., completing high school and attending college). Graduating college (Walters, 2018b) and attaining one’s desired level of education (Walters, 2018c) were key features of prior definitions of self-efficacy to achieve a conventional lifestyle, and in the study employing the broadest definition of self-efficacy (i.e., Walters, 2018b), graduating from college was only 1 of the 2 items (out of 6 total items) capable of mediating the past delinquency–future delinquency relationship on its own. Still, measuring a latent construct with just two indicators, as was done in the current investigation, is a serious limitation that needs to be remedied in future research on parenting, self-efficacy, and the peer influence effect.

One could also take issue with the delinquency measure employed in this study, given its emphasis on more serious forms of offending. Perhaps if the delinquency measure had included such acts as drug and alcohol use, truancy, and running away from home, the connection between academic self-efficacy and delinquency would have been stronger. The age of the data could also be considered a limitation. The FAS was conducted between 1994 and 1997, making the data more than 20 years old. Relationships between academic aspirations, self-efficacy, and delinquency may have changed in the intervening years as a result of historical or generational factors, making the current results less relevant to present-day concerns than they would have been had a more contemporary sample been used. Moreover, the fact that 80% of the sample was African American and the study took place in a midsized rust belt city calls the generalizability of results into question. Lastly, all variables included in the present study were from survey data supplied by the adolescents who participated in the FAS. This raises the prospect of mono-operational bias (Shadish, Cook, & Campbell, 2002) and the possibility that some of the path coefficients were inflated by shared method variance. The counterargument to this limitation is that the target and comparison pathways were both composed of variables derived from participant self-report, yet only the target pathway achieved significance.

Future Directions

The present study can be conceptualized as an ongoing attempt to create and refine a model of delinquency development, in which parents and peers are instrumental in the formation of a delinquent lifestyle. The interplay of parents and peers is a particularly noteworthy aspect of this model and suggests that while parents and peers may be reciprocally related in some situations, the principal direction of effect for these two variables flows from parents to peers, with parents providing a protective effect and delinquent peers generating a risk effect. In closing, four recommendations are offered for future research. First, it would be helpful if future research could catalogue the different forms of protective influence that parents provide their children. According to lifestyle theory (Walters, 2017b) and the results of the current study, there are at least three—parental support, parental control, and moral instruction—but more are likely. Second, there is a need to understand the temporal relationship between the parental and peer effects. Does parental influence wane and/or peer influence wax during the child–adolescence transition and does the timing of the transition vary as a function of age, gender, or social class? Third, which of the mechanisms believed to be responsible for self-efficacy (modeling, social influence/feedback, and physiological change, given that past performance accomplishments were controlled) is responsible for self-efficacy’s ability to reduce future offending. Fourth, how can programs and policies be organized to maximize the protective effect of parenting prior to the child–adolescence transition and minimize the risk of delinquent peers after the child–adolescence transition? Answering these questions could go a long way toward explaining the nature and significance of the parent–peer interface when it comes to delinquency development.

Variable	n	M	SD	Range	2	3	4	5	6	7	8	9	10	11	12	13
1. Age	850	14.55	0.64	14–16	−.10	−.02	.19*	.02	.12*	.17*	.14*	.27*	.20*	.20*	.18*	.08
2. Sex	850	1.50	0.50	1–2	−.04	−.09	.12*	.00	−.01	−.08	−.09	−.09	−.08	−.16*	−.15*
3. Race	850	1.17	0.37	1–2	.05	.07	−.15*	−.13*	−.09	.04	.13*	.14*	−.08	−.01
4. Average grade	848	5.27	1.90	1–9	.07	.17*	.16*	.15*	.38*	.34*	.30*	.25*	.17*
5. Parental support	849	24.29	5.39	6–30	.17*	.10	.13*	.16*	.14*	.16*	.19*	.13*
6. Peer 1	849	13.84	5.20	7–32	.53*	.49*	.07	.11	.11	.51*	.31*
7. Peer 2	812	14.04	5.22	7–34	.60*	.16*	.17*	.11	.43*	.36*
8. Peer 3	781	12.91	4.76	7–32	.17*	.19*	.16*	.46*	.44*
9. Self-Efficacy 1	848	8.84	1.56	2–10	.47*	.43*	.21*	.17*
10. Self-Efficacy 2	812	8.61	1.54	2–10	.58*	.17*	.20*	.
11. Self-Efficacy 3	783	8.54	1.55	2–10	.18*	.17*
12. Delinquency 1	848	22.52	7.90	17–74	.44*
13. Delinquency 4	770	21.04	7.18	17–68

Predictor	b (95% CI)	β	z	p
Self-Efficacy 2 (outcome)
Parental support	0.019 [0.001, 0.039]	.066	1.95	.051
Age	0.185 [0.022, 0.358]	.076	2.17	.030
Sex	−0.143 [−0.327, 0.032]	−.046	−1.55	.122
Race	0.404 [0.136, 0.674]	.097	2.96	.003
Average grade	0.147 [0.092, 0.197]	.180	5.51	<.001
Self-Efficacy 1	0.372 [0.290, 0.445]	.374	9.58	<.001
Peer Delinquency 2 (outcome)
Parental support	0.020 [−0.038, 0.076]	.020	0.66	.507
Age	0.756 [0.251, 1.268]	.093	2.92	.004
Sex	−0.120 [−0.741, 0.498]	−.012	−0.38	.700
Race	−0.935 [−1.607, −0.235]	−.067	−2.67	.008
Average grade	0.152 [−0.007, 0.322]	.055	1.81	.071
Peer Delinquency 1	0.505 [0.430, 0.576]	.501	13.58	<.001
Self-Efficacy 3 (outcome)
Peer Delinquency 2	0.000 [−0.018, 0.017]	−.001	−0.05	.961
Age	0.209 [0.054, 0.367]	.086	2.65	.008
Sex	−0.108 [−0.286, 0.067]	−.034	−1.20	.232
Race	0.241 [0.019, 0.459]	.058	2.09	.036
Average grade	0.082 [0.027, 0.138]	.100	2.87	.004
Parental support	0.021 [0.005, 0.039]	.074	2.46	.014
Self-Efficacy 2	0.523 [0.448, 0.605]	.519	12.98	<.001
Peer Delinquency 3 (outcome)
Self-Efficacy 2	0.219 [0.000, 0.451]	.072	1.89	.059
Age	0.153 [−0.317, 0.610]	.021	0.65	.515
Sex	−0.745 [−1.303, −0.212]	−.078	−2.68	.007
Race	−0.435 [−1.087, 0.200]	−.034	−1.33	.184
Average grade	0.084 [−0.070, 0.246]	.034	1.04	.299
Parental support	0.057 [0.001, 0.118]	.065	1.90	.057
Peer Delinquency 2	0.525 [0.467, 0.582]	.578	17.51	<.001
Delinquency 4 (outcome)
Peer Delinquency 3	0.352 [0.197, 0.532]	.238	4.14	<.001
Self-Efficacy 3	0.103 [−0.309, 0.540]	.023	0.48	.633
Peer Delinquency 2	0.152 [0.006, 0.303]	.113	2.04	.041
Self-Efficacy 2	0.290 [−0.139, 0.736]	.064	1.30	.193
Age	−0.497 [−1.300, 0.313]	−.045	−1.19	.234
Sex	−1.339 [−2.280, −0.450]	−.095	−2.92	.004
Race	0.488 [−0.521, 1.553]	.026	0.91	.361
Average grade	0.100 [−0.145, 0.340]	.027	0.82	.415
Parental support	0.047 [−0.045, 0.150]	.036	0.95	.343
Delinquency 1	0.226 [0.115, 0.352]	.253	3.70	<.001
Age with parental support	0.084 [−0.143, 0.321]	.024	0.72	.474
Sex with parental support	0.321 [0.146, 0.503]	.119	3.52	<.001
Race with parental support	0.146 [0.020, 0.272]	.072	2.25	.025
Grade with parental support	0.786 [0.092, 1.490]	.077	2.20	.028
Self-Efficacy 2 with Peer 2	0.482 [0.027, 0.954]	.084	2.04	.041
Self-Efficacy 3 with Peer 3	0.210 [−0.171, 0.615]	.046	1.04	.298

Pathways	BCBCI
CU traits → Wave 3 delinquency
Total effect	.08433	−.01473	.19269
Direct effect	.04722	−.04495	.15032
Total indirect effect	.03711	.00830	.07076
Specific indirect effects
Support 1 → Peer 2 → Delinquency 4	.00299	−.00467	.01758
Support 1 → Peer 3 → Delinquency 4	.02025	.00209	.05350
Support 1 → SE 2 → Delinquency 4	.00553	−.00132	.02174
Support 1 → SE 3 → Delinquency 4	.00222	−.00632	.01478
Support 1 → Peer 2 → Peer 3 → Delinquency 4	.00363	−.00674	.01618
Support 1 → SE 2 → Peer 3 → Delinquency 4	.00147	.00004	.00530
Support 1 → Peer 2 → SE 3 → Delinquency 4	.00000	−.00022	.00011
Support 1 → SE 2 → SE 3 → Delinquency 4	.00103	−.00249	.00867
Preacher–Hayes’s contrast test	.00147	.00004	.00540

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

Akers

R. L.

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

Allison

P. D.

(2012). Handling missing data by maximum likelihood (SAS Global Forum 2012 (Paper 312-2012)). Cary, NC: SAS Institute.

Andrews

D. A.

Bonta

(2010). The psychology of criminal conduct (5th ed.). New Providence, NJ: Matthew Bender.

Augustyn

M. B.

McGloin

J. M.

(2013). The risk of informal socializing with peers: Considering gender differences across predatory delinquency and substance use. Justice Quarterly, 30, 117–143.

Bandura

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

Bandura

(1993). Perceived self-efficacy in cognitive development and functioning. Educational Psychologist, 28, 117–149.

Bandura

(1995). Exercise of personal and collective efficacy in changing societies. In Bandura

(Ed.), Self-efficacy in changing societies (pp. 1–45). New York, NY: Cambridge University Press.

Baron

R. M.

Kenny

D. A.

(1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182.

Belknap

Holsinger

(2006). The gendered nature of risk factors for delinquency. Feminist Criminology, 1, 48–71.

10.

Bong

(2008). Effects of parent-child relationships and classroom goal structures on motivation, help-seeking avoidance, and cheating. Journal of Experimental Education, 76, 191–217.

11.

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.

12.

Brezina

Tekin

Topalli

(2009). “Might not be a tomorrow”: A multimethods approach to anticipated early death and youth crime. Criminology, 47, 1091–1129.

13.

Brown

B. B.

Bakken

J. P.

(2010). Parenting and peer relationships: Reinvigorating research on family-peer linkages in adolescence. Journal of Research on Adolescence, 21, 153–165.

14.

Caldwell

R. M.

Wiebe

R. P.

Cleveland

H. H.

(2006). The influence of future certainty and contextual factors on delinquent behavior and school adjustment among African American adolescents. Journal of Youth and Adolescence, 35, 591–602.

15.

Carroll

Houghton

Wood

Unsworth

Hattie

Gordon

Bower

(2009). Self-efficacy and academic achievement in Australian high school students: The mediating effects of academic aspirations and delinquency. Journal of Adolescence, 32, 797–817.

16.

Clark

L. A.

Watson

(1995). Constructing validity: Basic issues in objective scale development. Psychological Assessment, 7, 309–319.

17.

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.

18.

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.

19.

DiPierro

Fite

P. J.

Cooley

J. L.

Poquiz

J. L.

(2016). Academic aspirations as a moderator of the link between negative life events and delinquency in a sample of Latino youth. Child & Youth Care Forum, 45, 505–523.

20.

Forgatch

M. S.

Snyder

J. J.

Patterson

G. R.

Pauldine

M. R.

Chaw

Elish

… Richardson

E. B.

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

21.

Gottfredson

M. R.

Hirschi

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

22.

Hayes

A. F.

(2009). Beyond Baron and Kenny: Statistical mediation analysis in the new millennium. Communication Monographs, 76, 408–420.

23.

Hayes

A. F.

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

24.

Henneberger

A. K.

Tolan

P. H.

Hipwell

A. E.

Keenan

(2014). Delinquency in adolescent girls: Using a confluence approach to understand the influences of parents and peers. Criminal Justice and Behavior, 41, 1327–1337.

25.

Hirschi

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

26.

Hoeve

Dubas

J. S.

Eichelsheim

V. I.

van der Lann

P. H.

Smeenk

Gerris

J. R. M.

(2009). The relationship between parenting and delinquency: A meta-analysis. Journal of Abnormal Child Psychology, 37, 749–775.

27.

Imai

Keele

Tingley

(2010). A general approach to causal mediation analysis. Psychological Methods, 15, 309–334.

28.

Janssen

H. J.

Eichelsheim

V. I.

Deković

Bruinsma

G. J. N.

(2016). How is parenting related to adolescent delinquency? A between- and within-person analysis of the mediating role of self-control, delinquent attitudes, peer delinquency, and time spent in criminogenic settings. European Journal of Criminology, 13, 169–194.

29.

Johnson

W. L.

Giordano

P. C.

Manning

W. D.

Longmore

M. A.

(2011). Parent–child relations and offending during young adulthood. Journal of Youth and Adolescence, 40, 786–799.

30.

Kenny

D. A.

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

31.

Lipsey

M. W.

Derzon

J. H.

(1998). Predictors of violent or serious delinquency in adolescence and early adulthood: A synthesis of longitudinal research. In Loeber

Farrington

D. P.

(Eds.), Serious and violent juvenile offenders: Risk factors and successful interventions (pp. 86–105). Thousand Oaks, CA: Sage.

32.

Muthén

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

33.

Oser

C. B.

(2006). The criminal offending–self-esteem nexus. The Prison Journal, 86, 344–363.

34.

Ozkan

(2016). Reoffending among serious juvenile offenders: A developmental perspective. Journal of Criminal Justice, 46, 18–31.

35.

Petrosino

Derzon

Lavenberg

(2009). The role of the family in crime and delinquency: Evidence from prior quantitative reviews. Southwest Journal of Criminal Justice, 6, 108–132.

36.

Peyre

Leplége

Coste

(2011). Missing data methods for dealing with missing items in quality of life questionnaires: A comparison by simulation of personal mean score, full information maximum likelihood, multiple imputation, and hot deck techniques applied to the SF-36 in the French 2003 decennial health survey. Quality of Life Research, 20, 287–300.

37.

Piquero

A. R.

Farrington

D. P.

Welsh

B. C.

Tremblay

Jennings

W. G.

(2009). Effects of early family/parent training programs on antisocial behavior and delinquency. Journal of Experimental Criminology, 5, 83–120.

38.

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.

39.

Pituch

K. A.

Stapleton

L. M.

(2008). The performance of methods to test upper-level mediation in the presence of nonnormal data. Multivariate Behavioral Research, 43, 237–267.

40.

Poole

E. D.

Regoli

R. M.

(1979). Parental support, delinquent friends, and delinquency: A test of interaction effects. Journal of Criminal Law and Criminology, 70, 188–193.

41.

Poulin

Denault

A.-S.

(2012). Other-sex friendships as a mediator between parental monitoring and substance use in girls and boys. Journal of Youth and Adolescence, 41, 1488–1501.

42.

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.

43.

Ramirez

Machida

S. K.

Kline

Huang

(2014). Low-income Hispanic and Latino high school students’ perceptions of parent and peer academic support. Contemporary School Psychology, 18, 214–221.

44.

Rubin

K. H.

Bukowski

W. M.

Parker

J. G.

(2006). Peer interactions, relationships, and groups. In Eisenberg

Damon

Lerner

R. M.

(Eds.), Handbook of child psychology, vol. 3: Social, emotional, and personality development (6th ed., pp. 571–645). Hoboken, NJ: Wiley.

45.

Rucker

D. D.

Preacher

K. J.

Tormala

Z. L.

Petty

R. E.

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

46.

Sampson

R. J.

Laub

J. H.

(1993). Crime in the making: Pathways and turning points through life. Cambridge, MA: Harvard University Press.

47.

Shadish

W. R.

Cook

T. D.

Campbell

D. T.

(2002). Experimental and quasi-experimental designs for generalized causal inference. Boston, MA: Mifflin.

48.

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.

49.

Simons

R. L.

Simons

L. G.

Chen

Y.-F.

Brody

G. H.

Lin

K.-H.

(2007). Identifying the psychological factors that mediate the association between parenting practices and delinquency. Criminology, 45, 481–517.

50.

Simons

R. L.

C.-I.

Conger

R. D.

Lorenz

F. O.

(1994). Two routes to delinquency: Differences between early and late starts in the impact of parenting and negative peers. Criminology, 32, 247–276.

51.

Sutherland

E. H.

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

52.

Trucco

E. M.

Colder

C. R.

Wieczorek

W. F.

Lengua

L. J.

Hawk

L. W.

Jr. (2014). Early adolescent alcohol use in context: How neighborhoods, parents, and peers impact youth. Development and Psychopathology, 26, 425–436.

53.

Walters

G. D.

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

54.

Walters

G. D.

(2016). Crime continuity and psychological inertia: Testing the cognitive mediation and additive postulates with male adjudicated delinquents. Journal of Quantitative Criminology, 32, 237–252.

55.

Walters

G. D.

(2017a). Beyond dustbowl empiricism: The need for theory in recidivism prediction research and its potential realization in causal mediation analysis. Criminal Justice and Behavior, 44, 40–58.

56.

Walters

G. D.

(2017b). Risk, need, and responsivity in a criminal lifestyle. In Taxman

F. S.

(Ed.), Handbook on risk and need assessment: Theory and practice (pp. 193–219). New York, NY: Routledge.

57.

Walters

G. D.

(2018a). P _M effect size estimation for mediation analysis: A cautionary note, alternate strategy, and real data illustration. International Journal of Social Research Methodology, 21, 25–33.

58.

Walters

G. D.

(2018b). Getting specific about psychological inertia: Mediating the past crime–future crime relationship with self-efficacy for a conventional lifestyle. Criminal Justice Review, 23, 186–201.

59.

Walters

G. D.

(2018c). Mediating the relationship between parental control/support and offspring delinquency: Self-efficacy for a conventional lifestyle versus self-efficacy for deviance. Crime & Delinquency, 64, 606–624.

60.

Warr

(1993). Parents, peers, and delinquency. Social Forces, 72, 247–264.

61.

Wasserman

G. A.

Keenan

Tremblay

R. E.

Cole

J. D.

Herrenkohl

T. L.

Loeber

Petechuk

(2003, 4). Risk and protective factors of child delinquency. Child Delinquency Bulletin. Washington, DC: Office of Justice Programs, U.S. Department of Justice.

62.

Wells

L. E.

(1989). Self-enhancement through delinquency: A conditional test of self-derogation theory. Journal of Research in Crime and Delinquency, 26, 226–252.

63.

Wen

Fan

(2015). Monotonicity of effect sizes: Questioning kappa-squared as mediation effect size measure. Psychological Methods, 20, 193–203.

64.

A. D.

Zumbo

B. D.

(2008). Understanding and using mediators and moderators. Social Indicators Research, 87, 367–392.

65.

Zimmerman

(2014). Flint [Michigan] adolescent study (FAS): A longitudinal study of school dropout and substance use, 1994-1997 (ICPSR34598-v1). Ann Arbor, MI: Inter-university Consortium for Political and Social Research [distributor].

Pathways	BCBCI
Pathways	Estimate	Lower	Upper
CU traits → Wave 3 delinquency
Total effect	.08433	−.01473	.19269
Direct effect	.04722	−.04495	.15032
Total indirect effect	.03711	.00830	.07076
Specific indirect effects
Support 1 → Peer 2 → Delinquency 4	.00299	−.00467	.01758
Support 1 → Peer 3 → Delinquency 4	.02025	.00209	.05350
Support 1 → SE 2 → Delinquency 4	.00553	−.00132	.02174
Support 1 → SE 3 → Delinquency 4	.00222	−.00632	.01478
Support 1 → Peer 2 → Peer 3 → Delinquency 4	.00363	−.00674	.01618
Support 1 → SE 2 → Peer 3 → Delinquency 4	.00147	.00004	.00530
Support 1 → Peer 2 → SE 3 → Delinquency 4	.00000	−.00022	.00011
Support 1 → SE 2 → SE 3 → Delinquency 4	.00103	−.00249	.00867
Preacher–Hayes’s contrast test	.00147	.00004	.00540

Social Control Versus Social Learning: Self-Efficacy for Future Academic Success and Peer Delinquency as Mediators of the Parental Support–Delinquency Relationship

Abstract

Keywords

Parenting and Peers

Social Cognitive Variables and Mediation

Current Study

Method

Participants

Measures

Research Design

Procedure

Missing Data

Results

Discussion

Implications

Limitations

Future Directions

Footnotes

Declaration of Conflicting Interests

Funding

References