Examining Deviant Peer Association as a Predictor of Dual Systems Model Development:  Testing for Moderation Effects of Age

doi:10.21203/rs.3.rs-4909074/v1

Download PDF

Research Article

Examining Deviant Peer Association as a Predictor of Dual Systems Model Development: Testing for Moderation Effects of Age

https://doi.org/10.21203/rs.3.rs-4909074/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

The dual systems model is a prominent developmental framework that focuses on the role of cognitive development for understanding risk for antisocial behavior during adolescence. Prior research has implicated deviant peer association as a potential risk factor for atypical cognitive development. The present study builds on the prior literature by examining deviant peer association as a predictor of the development of sensation-seeking and impulse control and examines age of exposure as a moderator of these relationships. Data from the Adolescent Brain Cognitive Development study were analyzed. Ordinary least squares regression was used to examine relationships of interest. Greater deviant peer association was found to predict diminished impulse control. Deviant peer association did not significantly predict sensation-seeking. Neither of these relationships were significantly moderated by age. Message framing may have implications for the treatment of impulse control issues.

The dual systems model is a cognitive developmental framework that has risen to prominence in recent decades because of its capacity for predicting involvement in risky behaviors, particularly during adolescence (Ellingson et al., 2019; Rhyner et al., 2018; Wasserman et al., 2017). This theory focuses on the differential development of brain regions governing impulse control and sensation-seeking that lead to imbalance in these constructs during adolescence that precipitates the peak risk for involvement in risky behaviors during this period of the life-course (Steinberg et al., 2008; Steinberg, 2010). While the framework initially focused on typical development of these constructs driving the overall heightened propensity for adolescents to engage in risky behavior, there is also a great need to understand drivers of differential development of these constructs in order to better understand how atypical development may lead to even greater risk for involvement in a variety of antisocial behaviors. Deviant peer association has recently been identified as a risk factor that may be particularly important for understanding the atypical development of these constructs (Wojciechowski, 2023). However, there remain important gaps in our understanding of deviant peer association as a predictor of cognitive development. One area where additional research is necessary is understanding how early exposure to deviant peers during childhood and the entry into adolescence may be important for understanding cognitive development pertaining to impulse control and sensation-seeking. There is also a dearth of research that has examined the potential that the salience of deviant peer association as a predictor of this cognitive development may differ based on when the exposure to these peers occurs. The present study sought to address these gaps in the extant literature by examining deviant peer association as a predictor of impulse control and sensation-seeking and testing to determine whether deviant peer association significantly interacts with age to predict cognitive development.

The Dual Systems Model

As mentioned above, the dual systems model has been identified as a highly relevant framework for predicting involvement in risky behaviors and this is due to the imbalance in the development of impulse control and sensation-seeking (Steinberg et al., 2008; Steinberg, 2010). Sensation-seeking refers to the drive to seek out novel, exciting, and thrilling experiences. This cognitive construct is governed by the socioemotional network in the brain which is comprised in part by the dopaminergic system. The dopaminergic system is relevant because of its function of modulating the release of dopamine, a key neurotransmitter that is important for reward, reinforcement, and learning processes (Chen & Bonci, 2017; Corominas-Roso et al., 2007). The dopaminergic system tends to develop rapidly following puberty, leading to the establishment of novel neural pathways facilitating the release of dopamine in response to stimuli (Steinberg et al., 2008; Steinberg, 2010). The corresponding novel releases of dopamine in response to exciting stimuli may function to reward and reinforce engagement in particularly stimulating experiences, with risky behaviors like substance use presenting such an example. This rapid acceleration in development then tends to peak around the timing of entrance into adulthood and sensation-seeking may also plateau and decline around this time (Steinberg et al., 2008; Steinberg, 2010).

Impulse control refers to the capacity to stop and consider potential risks and/or consequences of an action before engagement. The cognitive control network is responsible for facilitating communication across various regions of the brain and this communication can aid in the consideration of potential consequences in a manner consistent with the ability to exert impulse control. The prefrontal cortex is an important component of this network and, unlike the dopaminergic system, tends to be one of the final regions of the brain to reach full maturity (Spencer-Smith & Anderson, 2009). The slower and steadier development of these systems leads to limited impulse control during adolescence. So, adolescents typically develop a strong drive to seek out novel and exciting experiences that antisocial behavior may present, but do not fully develop the capacity to stop and consider the risks involved with such behaviors until they reach adulthood. It is this developmental imbalance that is predicted to drive the increased risk for antisocial behaviors that is observed during adolescence (Steinberg et al., 2008; Steinberg, 2010). While the dual systems model has historically focused on identifying the typical patterns of development observed for these constructs, there is also a need to identify drivers of atypical development in order to best understand why some individuals present particularly high risk for engagement in antisocial behavior and why some individuals demonstrate early onset and chronic risk in this regard.

Deviant Peer Association as a Predictor of Dual Systems Model Development

Deviant peer association may be one risk factor that may drive differential development of impulse control and sensation-seeking. Deviant peer association refers to the degree to which an individual affiliates with and spends time with peers who are involved in antisocial behavior and/or endorse antisocial values and attitudes. Social learning theory has established itself in criminology as a prominent theory for understanding how differential association with antisocial peers may result in a socialization process that increases one’s own risk for involvement in antisocial behavior (Akers, 1973). The robust research base here mainly focuses on how this results to changes to deviant attitudes and values (Akers & Jensen, 2017; Pratt et al., 2010), but there is the potential that this peer socialization may lead to cognitive changes as well. Recent research has provided some evidence that association with deviant peers may lead to atypical dual systems model development. Wojciechowski (2023) observed that deviant peer association predicted diminished impulse control and increased sensation-seeking at follow-up even when controlling for prior levels of these constructs. Resistance to peer influence was found to have buffer against this effect for sensation-seeking, but not impulse control. This, however, was the only study identified which specifically examined the impact that deviant peers have on cognitive development in this manner. Other studies have yielded findings that offer some limited evidence that peers can influence cognitive processes in the short-term (Ciranka & Van den bos, 2019; O’Brien et al., 2011; Smith, 2018). For example, Chein et al (2011) examined adolescent and adult participants completing a simulated driving task with participants either not observed or observed by peers with their brains monitored by functional magnetic resonance imaging technology. Findings indicated that the participants were more likely to take risks when observed by peers than when alone. While this highlights the fact that youth may situationally engage in greater risk-taking when in the presence of peers, this and other research are generally limited by this situational nature and the fact that they cannot make broader claims about long-term changes in cognitive development across the life-course. The Wojciechowski (2023) study has been the only one to do this thus far, but it similarly suffers from some limitations. First, this study examined a sample of justice-involved youth, so generalizability was greatly limited. Second, this study examined these relationships during adolescence and emerging adulthood only. These limitations highlight the need for additional research on these relationships that address these issues.

The present study has the capacity to address the noted limitations of previous work in this area. The limited generalizability of the Wojciechowski (2023) study necessitates the use of a general population sample of youth to determine whether or not deviant peers have a similar impact on cognitive development as their justice-involved peers. Examining these processes with a more generalizable sample can help to determine the robustness of the previous findings and help to further guide design and implementation of treatment programming to attenuate the impact that deviant peers may have on cognitive development. The present study also sought to examine these relationships outside of adolescence and emerging adulthood. While this is generally the most relevant time both for the influence of peers and for dual systems model development to occur (Steinberg et al., 2008; Steinberg, 2010), there is also a great deal of utility in examining these processes during earlier periods of the life-course. Examining the relevance of deviant peers for influencing cognitive development during childhood may be pertinent as well, as this may set the stage for later development during more salient periods of the life-course.

Differential Salience of Deviant Peer Association by Age of Exposure

Another relevant consideration for the examination of the relationship between deviant peer association and dual systems model development is the role that age of exposure plays. While there is a dearth of research focused on understanding how deviant peers may influence cognitive development in this regard, there is even less research that examines how the salience of deviant peer association for predicting cognitive development may differ depending on when in the life-course individuals are exposed to the influence of these peers. The Chein et al. (2011) study provides some indication that life-course timing of exposure to peer influence may matter (at least in the short-term), as it was only adolescent participants demonstrated increased reward center activation when in the presence of peers than when alone, but that reward center activation did not vary based on peers vs. alone for adult participants. This, however, says little about how this may matter for longer-term patterns of cognitive development. Further, these findings may hold little relevance for understanding differences in salience for older children and those just entering adolescence; the groups focused on in the present study. Prior research has indicated that the salience of peer influence increases and peaks during adolescence (Warr, 1993; Warr, 2002). This would seem to indicate that the salience of peer influence on cognitive development should increase with age among older children and younger adolescents, as youth gaining into adolescence should demonstrate an elevated reactivity to the influence of their peers relative to their younger counterparts. However, this remains untested. Alternatively, it may be that precocious involvement with antisocial peers during childhood may trigger longer-term patterns of cognitive development that may lead to earlier development of sensation-seeking and/or an even slower development of impulse control. This necessitates a rigorous test of these propositions by examining age as a moderator of the relationship between deviant peer association and dual systems model constructs.

Current Study

The dual systems model is a prominent psychological framework focused on how the imbalance in the development of sensation-seeking and impulse control during adolescence may explain the observation of high prevalence of involvement in antisocial behavior during this period of the life-course (Le Blanc, 2020; Loeber, 2012; Steinberg et al., 2008). Deviant peer association has been observed by prior research to be an important predictor of development of these outcomes (Wojciechowski, 2023). That said, the existing research on these relationships is limited by a lack of generalizability beyond a high-risk population and a limited focus on adolescence and emerging adulthood as the periods of the life-course examined. While this is understandable given that the salience of peer influence peaks during adolescence (Warr, 1993; Warr, 2002), this limited focus is problematic because it says very little about the time preceding this important period of the life-course. Peer influence may not become a salient influence on cognitive development until individuals reach adolescence when both sets of processes become most salient. Alternatively, early peer relationships may set the stage for later behavioral issues if cognitive development is interrupted during childhood. The present study sought to address these gaps in the literature by testing the following hypotheses:

Hypothesis 1

Greater deviant peer association will predict increased sensation-seeking and diminished impulse control at follow-up.

Hypothesis 2

Age will significantly moderate the relationships between deviant peer association and dual systems model constructs.

The present study utilized data from the first five waves of the Adolescent Brain Cognitive Development (ABCD) study. These data were comprised of 11,880 children aged 9–10 years old at baseline and followed across the subsequent years with annual assessments, with plans to continue these assessments until they reach early adulthood. Assessments included surveys measuring behavior and psychosocial factors and brain scans with intention of understanding the biological and social underpinnings of development, mental health, substance use, and other behaviors. Participants were recruited and assessed across 21 research sites in the United States, with the formation of the ABCD consortium to conduct the study in 2015.

Measures

Dual Systems Model Constructs

The two main dependent variables examined in this study were sensation-seeking and impulse control. These constructs were measured at Wave 5 using the UPPS Impulsivity scale (Whiteside et al., 2005). This instrument assessed the various dimensions of impulsivity, of which sensation-seeking and lack of premeditation were two of them. Lack of premeditation was used as a proxy for impulse control because of the constructs’ analogous definitions as the capacity (or lack thereof) to stop and consider potential consequences of an action before engagement (Sangha & Foti, 2018). Both of these constructs were measured using a series of ordinal items that asked participants to indicate the degree to which various statements pertaining to these constructs corresponded to their own attitudes or actions (Impulse control: e.g., I like to stop and think things over before I do them; Sensation-seeking e.g., I quite enjoy taking risks). Index scores were then computed from these ordinal items for each of the constructs so that each participant had one sensation-seeking score and one impulse control score at Wave 5.

Deviant Peer Association

One of the main independent variables examined in this study was deviant peer association measured at Wave 4. This construct was measured using the Peer Behavior Profile instrument (Bingham et al., 1995). This instrument was comprised of three items assessing the degree of participants’ affiliations with deviant peers using an ordinal scale (e.g., How many friends have skipped school?). A mean score was computed from these ordinal items so that all participants had a single deviant peer association score at Wave 4.

Age

The other main independent variable examined in this study was participants’ ages at each wave. Age was measured in single-year intervals at each wave. Wave 4 and Wave 5 measures of age were included in analyses. Wave 4 age was examined as the moderator of the relationships between the independent and dependent variables because of its concurrent measurement with deviant peer association. Wave 5 age was included as a control variable because of its concurrent measurement with the dependent variables.1

Control Variables

Several control variables were also included in analyses in order to mitigate bias in estimation of effects of interest. Gender was the first of these variables, as prior research has indicated gender differences in cognitive development pertaining to dual systems model constructs (Ardila et al., 2011). Gender was measured at baseline using a binary variable which delineated male and female participants (0 = Male; 1 = Female).2

It was also important to control for race/ethnicity in these analyses, as prior research has indicated that there exists variation in cognitive development along racial/ethnic lines (Keels, 2009; Rushton & Jensen, 2005). Race was measured at baseline using a nominal variable that delineated participants into three distinct categories: White, Black, and Other Race. Three dummy variables were computed that corresponded to each of the race categories, with each variable delineating participants in one of the race categories from all other participants (e.g., 1 = Black; 0 = All other participants). The dummy variable that corresponded to White participants was then omitted from analyses as the reference group for interpreting coefficients in relation to. The other ethnicity variable included in analyses delineated participants who reported being of Latinx ethnicity from those who did not using a binary variable at baseline (1 = Latinx; 0 = Not Latinx).

Prior research has also indicated that the health of cognitive development may be stratified by social class (González et al., 2020). For this reason, participants’ parents’ annual income at baseline was controlled for in these analyses. This measure was operationalized using an ordinal variable, with higher scores corresponding to higher annual household income.

Traumatic stress exposure was at each wave was also controlled for in these analyses, as past research has indicated that experiencing traumatic stress may result in atypical cognitive development (Wilson et al., 2011). The post-traumatic stress disorder module from the Kiddie Schedule for Affective Disorders and Schizophrenia was used to assess this construct at baseline (Puig-Antich & Ryan, 1986). Part of this module entailed asking participants about the various types of traumatic stressors that they may have experienced during the prior observation period (e.g., Witnessed or caught in a natural disaster that caused significant property damage or personal injury). Participants were asked to endorse any of the traumatic stressors that they experienced during this time. A binary variable was then used in analyses which delineated participants who reported experiencing any of the traumatic stressors that they were asked about prior to baseline compared to who did not experience any of these traumatic stressors (0 = No; 1 = Yes).

The final set of control variables included in these analyses were measures of the dependent variables that were included in the corresponding models in which they matched the dependent variable of interest measured at Wave 3. This was done in order to ensure that any effects of deviant peer association from the previous wave were not just evidence of continuity of earlier levels of the dependent variable. This facilitated a greater capacity for causal inference and ensured that any differences in the dependent variable were due to deviant peer association if the coefficient for this independent variable was significant. These constructs were measured analogously to the measurement of the dependent variables described above using the UPPS Impulsivity scale.

Analytic Strategy

The present study utilized a series of ordinary least squares (OLS) regression modeling to examine the direct and moderated effects of interest. OLS regression was used here because of the normal distribution of the dependent variables and their residual errors. Two models were estimated for each of the dual systems model outcomes. Model 1 examined the direct effects of age and deviant peer association on the dual systems model outcome of interest, net of all control covariates. Model 2 then included an additional variable that modeled the interaction between age and deviant peer association to determine whether the salience of deviant peer association for predicting each of the dual systems model outcomes significantly varied depending on the age at which participants were exposed. Coefficients were then interpreted as the predicted change in the dependent variable based on a one-unit difference in a given independent variable during the prior observation period. Multiple imputation was used to manage missing data in analyses, with variables with missing data regressed on gender and race variables with complete data in order to ensure that estimates were not biased by data missing not at random.

Results were robust whether both or only one age variable were included in analyses.
There were three participants who identified with nonbinary gender identities, but they were dropped from analyses because of the limited utility of such a small sample size.

Table 1 provides descriptive statistics for all variables included in these analyses. Table 2 provides mixed effects modeling results for the models with impulse control as the dependent variable. Table 3 then provides these same mixed effects modeling results for the sensation-seeking outcome. Model 1 results pertaining to the impulse control outcome indicated that greater deviant peer association was associated with greater lack of premeditation (lower impulse control) at follow-up (Coefficient = .123, p < .001). Being older was associated with increased impulse control in this model also (Coefficient=-.196, p < .007). Lower impulse control during the previous observation period were also both associated with diminished impulse control at follow-up in this model also. Black participants also reported greater impulse control than White participants in this model. Model 2 results pertaining to the impulse control outcome indicated that greater deviant peer association remained a significant predictor of diminished impulse control at follow-up (Coefficient = .122; p < .001). Age remained a significant predictor of impulse control in this model also in the same direction (Coefficient=-.195, p < .007). The interaction between age and deviant peer association in this model was also not significant. Once again, having lower impulse control scores during the prior observation period were also associated with diminished impulse control. Black participants again reported greater impulse control than their White counterparts in this model also.

Table 1

**Descriptive Statistics**
	Mean/Proportion	Standard Deviation	Minimum	Maximum
Impulse Control (Wave 5)	7.941	2.153	4	16
Impulse Control (Wave 3)	7.784	2.244	4	16
Sensation-Seeking (Wave 5)	9.975	2.650	4	16
Sensation-Seeking (Wave 3)	9.486	2.670	3	16
Deviant Peer Association (Wave 4)	3.608	1.358	2	15
Age (Wave 5)	13.644	.728	11	17
Age (Wave 4)	12.428	.683	10	14
Annual Family Income at Baseline	7.224	2.422	1	10
Traumatic Stress Exposure (Baseline) No = 65.27% (N = 7,751) Yes = 34.73% (N = 4,125)
Gender Female = 47.83% (N = 5,680) Male = 52.17% (N = 6,196)
Race* White = 74.13% (N = 8,804) Black = 21.20% (N = 2,518) Other Race = 0.72% (N = 86) Latinx Ethnicity No = 79.43% (N = 9,312) Yes = 20.57% (N = 2,411)
* Percentages do not add up to 100% because of some overlap between categories due to multiracial participants

Table 2

**Ordinary Least Squares Regression Covariate Effects on Impulse Control at Wave 5 (N = 11,877)**
	Model 1			Model 2
	Coefficient	p-Value	95% Confidence Interval	Coefficient	p-Value	95% Confidence Interval
Deviant Peer Association at Wave 4	.123	< .001	.081 .165	.122	< .001	.080 .163
Age at Wave 4	− .196	.007	− .335 − .057	− .195	.007	− .334 − .055
Deviant Peer Association X Age at Wave 4	---	---	---	.033	.327	− .034 .099
Gender (0 = Male; 1 = Female)	.090	.057	− .003 .183	.090	.056	− .002 .182
Race (reference = White) Black Other Race	− .241 .058	.006 .889	− .407 − .075 − .780 .896	− .238 .058	.006 .890	− .405 − .071 − .782 .898
Latinx (0 = No; 1 = Yes)	− .062	.287	− .176 .053	− .062	.283	− .177 .052
Parental Annual Income at Baseline	− .005	.682	− .030 .020	− .005	.684	− .030 .020
Traumatic Stress Exposure at Baseline (0 = No; 1 = Yes)	.026	.589	− .067 .119	.025	.590	− .068 .119
Age at Wave 5	.007	.917	− .121 .135	.007	.912	− .122 .136
Impulse Control at Wave 3	.454	< .001	.431 .477	.453	< .001	.430 .476
Constant	6.358	< .001	5.183 7.534	6.348	< .001	5.171 7.524

Table 3

**Ordinary Least Squares Regression Covariate Effects on Sensation-Seeking at Wave 5 (N = 11,877)**
	Model 1			Model 2
	Coefficient	p-Value	95% Confidence Interval	Coefficient	p-Value	95% Confidence Interval
Deviant Peer Association at Wave 4	.030	.187	− .015 .075	.030	.179	− .014 .074
Age at Wave 4	.020	.785	− .169 .129	− .020	.786	− .170 .130
Deviant Peer Association X Age at Wave 4	---	---	---	< .001	.998	− .083 .083
Gender (0 = Male; 1 = Female)	− .115	.128	− .264 .034	− .115	.127	− .264 .034
Race (reference = White) Black Other Race	− .590 − .479	< .001 .353	− .821 − .358 -1.515 .557	− .590 − .479	< .001 .354	− .817 − .362 -1.517 .559
Latinx (0 = No; 1 = Yes)	.004	.953	− .122 .130	.004	.949	− .122 .130
Parental Annual Income at Baseline	.027	.137	− .009 .063	.027	.137	− .009 .063
Traumatic Stress Exposure at Baseline (0 = No; 1 = Yes)	− .017	.779	− .140 .105	− .017	.779	− .140 .105
Age at Wave 5	.152	.036	.010 .294	.152	.036	.010 .294
Sensation-Seeking at Wave 3	.497	< .001	.471 .523	.497	< .001	.471 .523
Constant	3.277	< .001	1.657 4.897	3.277	< .001	.1.663 4.891

Model 1 results for the sensation-seeking outcome indicated that deviant peer association was not a significant predictor of greater sensation-seeking at follow-up. Age at Wave 4 similarly was not a significant predictor of sensation-seeking in this model. Being older at Wave 5 and greater sensation-seeking during the previous observation period were also all associated with increased sensation-seeking at follow-up. Black participants reported lower sensation-seeking than White participants in this model also. Model 2 results indicated that the nonsignificant effects of age at Wave 4 and deviant peer association remained nonsignificant predictors of sensation-seeking. The interaction between age and deviant peer association was also nonsignificant. Being older at Wave 5 greater sensation-seeking during the prior observation period were also all significant predictors of greater sensation-seeking at follow-up in this model. Black participants reported lower sensation-seeking than White participants in this model also.

This study provided a unique examination of the influence that association with antisocial peers may have on cognitive development during childhood and early adolescence. Deviant peer association was found to predict diminished impulse control at follow-up, but did not significantly predict sensation-seeking. The impact of deviant peers on impulse control was robust to control for pre-existing levels of impulse control also. Contrary to Hypothesis 2, these relationships were not significantly moderated by age, indicating that the salience of deviant peer association was consistent during childhood and early adolescence. There are a number of important implications of these findings for intervention with youth.

Deviant peer association exerted significant main effects on the development of impulse control. These findings indicate the importance of interrupting relationships with antisocial peers among youth in order to foster healthy impulse control development and potentially reduce the risk of progression to serious antisocial behavior early in the life-course. While much of the research on the role of deviant peers for predicting involvement in antisocial behavior takes a social learning theory approach (Akers, 1973; Akers & Jensen, 2017; Pratt et al., 2010), the results of this study add to the burgeoning literature indicating that antisocial peers may also increase risk for such behaviors via atypical cognitive development. This dovetails with a biosocial approach, as the social risk factor of peers provides another example of how the social may “get under the skin” and cause changes to the biological maturation that these cognitive constructs represent. While an examination of biological processes was beyond the scope of this study, the potential that this biosocial approach presents highlights the need for additional research on this topic. Identifying specific biological mechanisms that are influenced by deviant peer association may open the door for novel treatments that can address any changes to impulse control development caused by social risk factor exposure and potentially prevent progression to serious antisocial behavior early in life. This additional research could potentially facilitate the development of therapeutic medications that could aid in limiting the dysfunction in impulse control functioning caused by affiliation with deviant peers. These medications could focus on the specific regions of the brain that govern impulse control and are impacted by association with deviant peers. Pairing this with psychosocial programming could result in a dual pronged approach to reduce the risk that deviant peer association or other social risk factors impact cognitive development and increase risk for involvement in antisocial behaviors.

With impulse control impacted by deviant peer association, this highlights the need for psychosocial interventions that are able to aid in the healthy development of this construct. Message framing has presented some utility for reducing the risk for involvement in risky health behaviors (Apanovitch et al., 2003; Keyworth et al., 2018; Toll et al., 2008). This programming has been used in medical contexts by educating patients about the risks involved with engagement in risky health behaviors. Medical professionals are used in their capacities as authorities on the topics as a means of convincing individuals that the risks involved with these behaviors are not worth the perceived rewards. It can be argued that such programming could have utility for improving both impulse control, as identifying the potential risks involved with such behaviors may facilitate greater capacity to stop and consider consequences prior to engagement. For individuals involved with deviant peers who may attempt to influence individuals to engage in antisocial behaviors themselves, this could help to facilitate greater capacity to exert impulse control when presented with opportunities to engage in risky behaviors. While this would potentially be useful for selected populations who demonstrate involvement with deviant peers already, such programming may also potentially be useful for universal populations by focusing on reducing association with deviant peers altogether. However, there remain concerns about the utility of such programming in this regard. As mentioned above, message framing has historically been used in medical contexts and provided by medical professionals. It is unclear how this programming may translate outside of these settings and provided by non-medical program staff. This indicates the need for pilot testing of such interventions in non-clinical settings by non-medical staff to determine efficacy before scaling up to broader contexts.

Another finding of this study was that the impact of deviant peer association did not vary by age for either of the dual systems model constructs. While this was contrary to hypotheses, this provides indication that programming meant to attenuate these and related relationships may be provided to both children and adolescents without a significant difference in salience. That said, there should remain some concerns about this. The sample was somewhat limited in terms of variance in age. While ages in the study did range from 8–17, there were very few participants who had reached the upper limits of this range by the final wave included in the study period. The vast majority of participants only reached aged 14, so a great deal of adolescence remains to be examined with enough statistical power to be more confident in these null interaction effects. So, while these nonsignificant interactions provide preliminary indication of a lack of differences in salience by age, this also highlights the need to re-examine these relationships as new waves of data from the ABCD study are released as participants get older. This will provide the increased statistical power to be necessary that there are truly no age differences in the salience of deviant peer influence for altering cognitive development.

While the present study provided unique insight into the relevance of deviant peers for understanding dual systems model development, there remain several important limitations. The first limitation relates to the somewhat limited nature of the deviant peer association measure used in analyses. This measure focuses only on a limited number of antisocial peer behaviors/markers (school suspension, shoplifting, skipped school). This limited focus was likely because of the more limited engagement in antisocial behaviors exhibited by youth during childhood, but this does leave out a range of other antisocial behaviors that may have been more prevalent in adolescence, like substance use and violent behavior. This indicates the need for additional research on these relationships which utilizes more comprehensive measures of peer behaviors in order to ensure that direct and interactive effects are robust to those observed in the main analyses here. Another limitation of this study pertains to limited nature of the measurement of the dual systems model constructs. The UPPS scale used to measure these constructs was only administered every two years in the ABCD study at the time this study was conducted. While this is not a fatal issue for the study, this does mean that year-to-year variation on these measures was missed and this resulted in a much coarser measure of cognitive development. This indicates the need for continued research on these processes that can provide a finer-grained examination on an annual basis in order to best understand year-by-year variation in dual systems model development and the role that deviant peers play for predicting this development. Another limitation of the study pertains to the OLS regression modeling strategy used here. The staggered measurement of the dependent variables made use of more advanced methodologies that could account for the nested nature of the data and within-person change an issue. Using a method like mixed effects modeling with more complete data would have facilitated greater capacity to truly understand the within-person change that is necessary for causal inference here. This indicates the need for continued research on this topic using richer data that addresses these issues.

Author Contribution

TW is the sole author and completed all tasks related to the manuscript.

Akers RL (1973) Deviant behavior: A social learning approach. Wadsworth Publishing Company
Akers RL, Jensen GF (2017) The empirical status of social learning theory of crime and deviance: The past, present, and future. Tak stock, 37–76
Apanovitch AM, McCarthy D, Salovey P (2003) Using message framing to motivate HIV testing among low-income, ethnic minority women. Health Psychol 22(1):60
Ardila A, Rosselli M, Matute E, Inozemtseva O (2011) Gender differences in cognitive development. Dev Psychol 47(4):984
Bingham CR, Fitzgerald HE, Zucker RA (1995) Peer behavior profile/peer activities questionnaire. Unpublished questionnaire. East Lansing: Department of Psychology, Michigan State University
Chein J, Albert D, O’Brien L, Uckert K, Steinberg L (2011) Peers increase adolescent risk taking by enhancing activity in the brain’s reward circuitry
Chen BT, Bonci A (2017) The mesocortical dopaminergic system. Hum frontal lobes: Funct disorders, 13
Ciranka S, Van den Bos W (2019) Social influence in adolescent decision-making: A formal framework. Frontiers in psychology, 1915
Corominas-Roso M, Roncero C, Bruguera E, Casas M (2007) The dopaminergic system and addictions. Revista de neurologia 44(1):23–31
Ellingson JM, Corley R, Hewitt JK, Friedman NP (2019) A prospective study of alcohol involvement and the dual-systems model of adolescent risk‐taking during late adolescence and emerging adulthood. Addiction 114(4):653–661
González L, Cortés-Sancho R, Murcia M, Ballester F, Rebagliato M, Rodríguez-Bernal CL (2020) The role of parental social class, education and unemployment on child cognitive development. Gac Sanit 34:51–60
Keels M (2009) Ethnic group differences in early head start parents’ parenting beliefs and practices and links to children's early cognitive development. Early Child Res Q 24(4):381–397
Keyworth C, Nelson PA, Bundy C, Pye SR, Griffiths CEM, Cordingley L (2018) Does message framing affect changes in behavioural intentions in people with psoriasis? A randomized exploratory study examining health risk communication. Psychol health Med 23(7):763–778
Le Blanc M (2020) On the future of the individual longitudinal age-crime curve. Criminal Behav mental health 30(4):183–195
Loeber R (2012) Does the study of the age-crime curve have a future. future Criminol, 11–19
O'Brien L, Albert D, Chein J, Steinberg L (2011) Adolescents prefer more immediate rewards when in the presence of their peers. J Res adolescence 21(4):747–753
Pratt TC, Cullen FT, Sellers CS, Winfree T Jr, Madensen L, Daigle TD, Gau LE, J. M (2010) The empirical status of social learning theory: A meta-analysis. Justice Q 27(6):765–802
Puig-Antich J, Ryan N (1986) Kiddie schedule for affective disorders and schizophrenia. Western Psychiatric Institute, Pittsburgh, PA
Rhyner KJ, Uhl CA, Terrance CA (2018) Are teens being unfairly punished? Applying the dual systems model of adolescent risk-taking to sexting among adolescents. Youth justice 18(1):52–66
Rushton JP, Jensen AR (2005) Thirty years of research on race differences in cognitive ability. Psychol public policy law 11(2):235
Sangha S, Foti D (2018) Neurobiology of Abnormal Emotion and Motivated Behaviors: Integrating Animal and Human Research. Academic, Cambridge, MA
Smith AR, Rosenbaum GM, Botdorf MA, Steinberg L, Chein JM (2018) Peers influence adolescent reward processing, but not response inhibition. Cogn Affect Behav Neurosci 18(2):284–295
Spencer-Smith M, Anderson V (2009) Healthy and abnormal development of the prefrontal cortex. Dev Neurorehabilitation 12(5):279–297
Steinberg L, Albert D, Cauffman E, Banich M, Graham S, Woolard J (2008) Age differences in sensation seeking and impulsivity as indexed by behavior and self-report: Evidence for a dual systems model. Dev Psychol 44(6):1764–1778
Steinberg L (2010) A dual systems model of adolescent risk-taking. Dev Psychobiol 52(3):216–224
Toll BA, Salovey P, O'Malley SS, Mazure CM, Latimer A, McKee SA (2008) Message framing for smoking cessation: The interaction of risk perceptions and gender. Nicotine Tob Res 10(1):195–200
Warr M (1993) Age, peers, and delinquency. Criminology 31(1):17–40
Warr M (2002) Companions in crime: the social aspects of criminal conduct. Cambridge University Press, Cambridge, England
Wasserman AM, Crockett LJ, Hoffman L (2017) Reward seeking and cognitive control: using the dual systems model to predict adolescent sexual behavior. J Res Adolescence 27(4):907–913
Wilson KR, Hansen DJ, Li M (2011) The traumatic stress response in child maltreatment and resultant neuropsychological effects. Aggress Violent Beh 16(2):87–97
Wojciechowski TW (2023) The influence of deviant peer association on dual systems model development: The protective role of resistance to peer influence. Appl Dev Sci 27(4):373–388

No competing interests reported.

Download PDF

Reviewers agreed at journal
08 Oct, 2024
Reviewers invited by journal
22 Sep, 2024
Editor assigned by journal
01 Sep, 2024
Submission checks completed at journal
14 Aug, 2024
First submitted to journal
13 Aug, 2024

You are reading this latest preprint version

Examining Deviant Peer Association as a Predictor of Dual Systems Model Development: Testing for Moderation Effects of Age

Status:

Version 1

Abstract

Introduction

The Dual Systems Model

Deviant Peer Association as a Predictor of Dual Systems Model Development

Differential Salience of Deviant Peer Association by Age of Exposure

Current Study

Methods

Measures

Dual Systems Model Constructs

Deviant Peer Association

Age

Control Variables

Analytic Strategy

Results

Discussion

Declarations

Author Contribution

References

Additional Declarations

Status:

Version 1