Links between Pubertal Timing, Peer Influences, and Externalizing Behaviors among Urban Students followed through Middle School

Abstract

Purpose

Early pubertal maturation has been linked to higher rates of aggressive and delinquent behaviors among white, middle class females. Less is known about this association for males, minorities, or among adolescents from urban communities. The aim of this paper is to evaluate underlying mechanisms of the association between early pubertal timing and both aggression and delinquency among a sample of minority, males and females from an urban community.

Methods

The association between perceived early pubertal maturation and aggressive or delinquent behaviors for African American and Latino males and females (n = 1366) was examined as well as pathways between early maturation and these negative outcomes longitudinally across 6^th, 7^th, and 8^th grades.

Results

Early maturers reported higher mean levels of both aggression and delinquency at all time points regardless of gender or ethnicity. Associating with delinquent peers in 6^th grade fully mediated the association between early maturation and both aggression and delinquency at all time points. Early maturers did not differentially associate with greater numbers of delinquent peers in either 7^th or 8^th grade.

Conclusions

These results provide valuable information regarding at-risk groups and inform future intervention efforts.

Both aggression and delinquency escalate during middle school (i.e., 6^th–8^th grade, approximately ages 11–14) and peak at age 16 for both males and females [1–3]. To date, several risk factors for aggression and delinquency have been identified. Specifically, recent studies have demonstrated that early pubertal timing (i.e., going through puberty earlier than one’s peers) is associated with higher rates of engagement in aggressive and delinquent behaviors [4–9]. Likewise, association with delinquent peers has been identified as not only a pathway to aggression and delinquency more generally [2], but also as a factor that may potentially mediate the effects of pubertal timing on aggression and delinquency [5,10].

Particular individual or demographic characteristics may also influence rates of adolescent delinquency and aggression. Crowded, urban, inner city environments are associated with higher exposure to violent crimes and higher rates of violent, aggressive, and delinquent behaviors [1,11]. However, ethnic group differences are small such that African American and Latino adolescents report somewhat higher rates of aggressive and delinquent behaviors than their White same-gender counterparts [3]. At the same time, recent studies suggest that female delinquency and overt aggression are on the rise or, at least, that gender differences are narrowing [12]. Notably, risk factors associated with the development of aggression and delinquency do not necessarily hold across gender or ethnicity [1,13]. Therefore, investigation of the role of different risk factors on the development of externalizing behaviors for both genders across different ethnic/racial groups in urban areas is warranted, especially during the middle school years.

Numerous investigations have found early pubertal timing to be a risk for externalizing problems [2,4,5,7,9,14–16]. However, most research has examined Non-Hispanic White, middle-class adolescents with a few exceptions [4,5,8].¹ One study found that early pubertal timing was associated with violent behavior among ethnically diverse adolescent females from disadvantaged neighborhoods [8]. Another found that early pubertal timing was associated with higher rates of externalizing behaviors among African American youth [5]. While these studies support an association between early maturation and externalizing problems among minority youth, there is a need for further research with minority males and females from urban communities to clarify this association.

However, the more important question is why or how timing is linked to these problems. One theoretical explanation states that because early maturers look older or more adult-like than their same age peers, they are likely to be treated differently by peers and adults and may initiate behaviors perceived as typical of older adolescents or engage in such behaviors at an earlier age than their peers [17]. Hence, early maturation disrupts social networks with youth associating with peers of comparable physical development rather than age [18]. Girls show outwardly physical signs of development at younger ages than do boys. Thus, early maturation may be particularly problematic for girls.

The association between peer delinquency and externalizing behaviors is well established; however, the role of peer delinquency regarding early pubertal timing effects on externalizing behaviors remains unclear. Previous research with girls found that early maturers who associated with older peers reported more engagement in behaviors like alcohol use [10]. It is important to note that mediating effects of peer delinquency have not been evaluated.

The present investigation examines the role of delinquent peers in links between pubertal timing and aggression and delinquency in a large, ethnically diverse sample of urban youth who were followed longitudinally over 6^th, 7^th, and 8^th grades. To extend the prior literatures on pubertal timing, externalizing behaviors, and the role of delinquent peers, the present investigation examined three issues. First, we examined whether or not early pubertal maturation conferred risk for aggressive or delinquent behavior among a sample of young, urban males and females. It was hypothesized that early maturers would report significantly higher levels of aggressive and delinquent behaviors compared to on-time and late maturers for both boys and girls. The second aim was to examine whether this association was explained, or mediated, by engagement with delinquent peers. It was hypothesized that delinquent peer involvement would mediate the association between early pubertal timing, aggression and delinquency. Finally, these effects were examined longitudinally over 6^th, 7^th, and 8^th grades as this is a period when these behaviors frequently increase among urban youth.

Methods

Participants

The current study used data collected at 6^th, 7^th, and 8^th grades as part of a larger randomized clinical trial designed to expand and test the effectiveness of an already proven drug prevention strategy on violent and aggressive behavior. Only participants in the control condition in 6^th grade (N = 2931) were used to avoid contamination with potential intervention effects. At each grade, 1.5%-22.8% of students did not sufficiently complete the surveys. The maximum number of students with complete data was used in each of the following analyses. Twenty percent (n = 342) indicated an ethnic affiliation other than African American or Latino. These other subgroups were dropped because the group sizes were too small for meaningful analysis, resulting in a final sample of 1366 students (see discussion of attrition below). Analyses conducted on the full sample controlling for ethnicity did not differ from the following results.

The mean age for the sample of 1366 youth was 11.68 years (SD = .50) in the 6^th grade (range of 9.64 to 14.00). The sample included slightly more girls (54%) than boys. The majority of the students were African American (62%) with the remainder of the sample consisting of Latinos (38%). Over half of the students came from two-parent households (55%). Youth were enrolled in public (90%) and parochial (10%) schools. Although a measure of family SES was not available, archival public school records of participating schools showed that the majority (88%) of schools had greater than 65% student eligibility for free or reduced lunch.

Procedure

A passive consent procedure approved by Weill Cornell Medical College’s IRB was used to inform parents about the nature of the study and to provide them with an opportunity to disallow their child’s participation. A consent form describing the focus of the larger study and the self-report survey was distributed in the schools for students to take home to their parents, as well as mailed directly to students’ homes. Students whose parents indicated they did not want them to participate in the self-report survey did not complete any of the data collection activities.

A multi-ethnic team of three to five data collectors collected the surveys. Carbon monoxide breath samples were collected at 6^th, 7^th, and 8^th grades to enhance the validity of self-report cigarette smoking utilizing a variant of the bogus pipeline procedure [19]. Bogus pipeline procedures can also increase the validity of reporting on other problem behaviors [20].

Measures

Demographic data

Participants reported their gender, age, household structure, and ethnicity on the 6^th grade survey. A single dichotomous variable was created to capture household structure where 1 indicates living in a two-parent household. Students from single and no-parent households were coded as 0. Another single dichotomous variable was created to capture participants’ ethnic affiliation where 1 represented Latinos and 0 represented African Americans. A single dichotomous variable was created to indicate type of school the adolescent attended (public = 1, parochial = 0).

Early pubertal timing was assessed by a single subjective item “do you think your development was any earlier, later, or about the same as other [same gender] youth your age?” This item captured individual perceptions of the timing of puberty within the normal variation in the onset/intensity/velocity of pubertal maturation among adolescents in general. Response options were: 1 (Early), 2 (On-time), and 3 (Late). This item was part of a longer-term follow-up of the primary study (when youth were in 9^th grade). The validity of assessing pubertal timing at a later time point is supported by previous research which found that subjective measures of pubertal timing most accurately reflected objective measures of pubertal timing as youth progressed through puberty (i.e., in 8^th versus 6^th grade) [21]. Because this investigation focused on the effects of early maturation only, a single dichotomous variable was created where 1 represents participants who reported maturing early. All other responses were coded as 0.

Aggression (past month)

Ten items from the aggression scale of the Youth Self-Report [22] were used to assess general aggression in the 6^th grade (α = .92), 7^th grade (α = .93), and 8^th grade (α = .94). Students were asked how many times in the past month they had engaged in each of ten overtly aggressive behaviors (e.g., cursed at someone; tripped someone on purpose). These items do not specify the location of the aggressive act (e.g., home, school, etc.). Response options were: 1 (Never), 2 (Once), 3 (2–3 times), 4 (4–5 times), and 5 (More than 5 times). Items were rescored onto a scale of 0–4 and summed to create a continuous measure where higher scores indicated a greater number of aggressive behaviors in the past month.

Delinquency (past year)

Students reported how many times in the past year they had engaged in each of ten delinquent behaviors in 6^th grade (α = .86), 7^th grade (α = .87), and 8^th grade (α = .89). The items were from a commonly used delinquency scale [23] that tapped behaviors such as violence, vandalism, and theft (e.g., picked a fight; purposely damaged or destroyed property; taken something from a store). These items represent legal infractions but do not indicate involvement in the court or legal system. Response options were: 1 (Never), 2 (Once), 3 (2–3 times), 4 (4–5 times), and 5 (More than 5 times). Items were rescored onto a scale of 0–4 and summed to create a continuous measure where higher scores indicated a greater number of delinquent behaviors in the past year.

Friends’ delinquency (past year)

Students were asked to indicate how many of their friends had engaged in each of seven delinquent behaviors in the past year [23] in the 6^th grade (α = .88), 7^th grade (α = .91), and 8^th grade (α = .92). Response options were: 1 (None), 2 (Less than half), 3 (About half), 4 (More than half), and 5 (All or almost all). Items were rescored onto a scale of 0–4 and then summed to create a continuous measure where higher scores indicated associating with more friends who engage in a higher number of delinquent behaviors for each time point.

Attrition Analyses

To test for differences between students in the current study and those who were dropped from any of the following analyses, t-tests were run on all continuous variables and chi-square analyses were run on all background variables from the 6^th grade data. Students who were dropped from any of the current analyses were more likely to be male (χ²(1, N = 2917) = 24.64, p < .001; 53% vs 43%), more likely to attend public school (χ²(1, N = 2931) = 11.31, p = .001; 91% vs 87%), and less likely to be early maturers (χ²(1, N = 1708) = 3.74, p = .030; 25% vs 29%). The two groups did not differ on ethnicity, family structure, aggression, delinquency, or peer delinquency. Thus, the sample used in the current study had statistically equivalent baseline rates of all outcome variables as the cross-sectional sample of participants at 6^th grade. Therefore, the reduced sample does not introduce bias into the following statistical analyses.

Analysis Plan

The association between early pubertal timing and aggressive or delinquent behaviors over time was examined via two 2 (timing) × 2 (gender) × 2 (ethnicity) × 3 (grade) repeated measures ANCOVAs, controlling for family structure, and school type. Grade was the within subject variable (6^th, 7^th, & 8^th) while gender, ethnicity, and timing were the between subjects variables. To evaluate the role of delinquent peers, mediating effects were examined for each outcome using a series of regression analyses [24] with follow-up Sobel Tests of significance of the mediation.

Initial evaluations of the data suggested that the measures of aggression were roughly univariate normal in all three grades (i.e., all with skewness less than 1 and kurtosis less than 2). The measures of delinquency were slightly skewed (i.e., skewness ranged from 1.73–2.65 and kurtosis ranged from 3.29–8.79). However, due to the large sample and the robustness of the proposed analyses, even results utilizing the most conservative estimations of the F-statistics were identical to uncorrected estimations. Therefore, the slight skewness of the delinquency outcome had no impact on the reliability of the analyses.

RESULTS

Descriptive Analysis

Initial analyses checked for pubertal timing group differences on sample characteristics. There were a higher percentage of African Americans (31%) compared to Latinos (26%) who were early maturers. Given that previous research has shown that African American females and potentially males tend to mature earlier on average compared to their Latino peers, this discrepancy is not unusual [25]. No other differences were found.

Means and standard deviations for aggression and delinquency at each grade and by timing group are shown in Table 1. Average incidents of aggressive behavior within the past month ranged from 12 to 19 across 6^th, 7^th, and 8^th grades and average acts of delinquency in the past year was three in 6^th grade, increasing to six acts in 7^th grade and just under seven acts in 8^th grade. Correlations among aggression and delinquency at each time point were strong but decreased slightly over time, with r = .71, r = .67, and r = .66 in 6^th, 7^th, and 8^th grade.

Table 1.

Means and standard deviations by pubertal timing for each grade

		6th Grade	7th Grade	8th Grade
		M (SD)	M (SD)	M (SD)
Aggression	Early Maturers	13.03 (10.50)	20.83 (11.63)	21.47 (12.46)
	On-time/Late Maturers	11.91 (10.32)	18.77 (12.11)	19.10 (12.45)
	Total Sample	12.24 (10.38)	19.37 (12.01)	19.79 (12.49)
Delinquency	Early Maturers	3.97 (5.65)	6.57 (6.99)	7.39 (8.15)
	On-time/Late Maturers	3.58 (5.30)	5.94 (7.11)	6.42 (8.06)
	Total Sample	3.70 (5.40)	6.12 (7.08)	6.70 (8.09)

The Association of Pubertal Timing with Aggression and Delinquency Outcomes

Aggression

Significant main effects of grade F (2, 1850) = 17.85, p < .001, η = .02, pubertal timing F (1, 925) = 6.79, p = .009, η = .01 and ethnicity F (1, 925) = 10.15, p = .001, η = .01 were found after controlling for family structure and school type. Follow-up Bonferroni corrected pairwise comparisons revealed that there were significant increases in aggression between 6^th, 7^th, and 8^th grades and that early maturers reported significantly higher levels of aggression compared to on-time/late maturers at each grade (See Table 1). African American adolescents (M = 18.52) reported higher mean levels of aggression than Latino adolescents (M = 16.12) overall. There were no significant interactions among grade, pubertal timing, gender, and ethnicity and no main effect of gender.

Delinquency

Significant main effects of grade F (2, 1856) = 4.45, p = .012, η = .01, pubertal timing F (1, 928) = 5.47, p = .020, η = .01, and gender F (1, 928) = 28.22, p < .001, η = .03 were found after controlling for the demographic variables. Follow-up Bonferroni corrected pairwise comparisons revealed that there were significant increases in delinquency between 6^th, 7^th, and 8^th grades and early maturers reported significantly higher levels of delinquency compared to on-time/late maturers at each grade (See Table 1). Males (M = 6.83) reported significantly higher levels of delinquency compared to females (M = 4.51) overall. There were no significant interactions among grade, pubertal timing, gender, and ethnicity and no main effect of ethnicity.

The Role of Delinquent Peers: Mediation Effects

The associations between pubertal timing and the outcomes of interests do not differ by gender or ethnicity as indicated by the lack of interactions in the previous analyses. Therefore, gender and ethnicity were controlled for along with family structure and school type in subsequent analyses. In order to test the role of friend delinquency as a mediator, it was first necessary to test whether pubertal timing was significantly associated with friend delinquency. Timing was significantly associated with friend delinquency in the 6^th grade F (5, 1339) = 5.11, p < .001, but not in the 7^th or 8^th grades with early maturers reporting more friend delinquency than on-time/late maturers in the 6^th grade only (M = 4.70 and M = 3.93, respectively). Given this, further analyses testing the mediation model were conducted using friend delinquency in the 6^th grade as the mediator on all outcomes at all time points.

As indicated previously, a significant association between pubertal timing and both aggression and delinquency was found in the 6^th, 7^th, and 8^th grades. Therefore, the mediation model was tested on both outcomes at each point in time. Gender, ethnicity, family structure and school type were entered in step one while pubertal timing and 6^th grade friend delinquency were entered simultaneously in step two of the regression equations. In these analyses, only 6^th grade friend delinquency predicted aggression or delinquency in 6^th, 7^th, and 8^th grades (See Tables 2 and 3). All of the previously significant associations between pubertal timing and aggression and delinquency were non-significant. Sobel tests were conducted to evaluate the significance of the mediation effects [24,26,27]. The resulting z-statistics for all of the mediation models evaluated ranged from 2.49 to 2.61, all ps ≤ .05.

Table 2.

Mediating effects of 6th grade friend delinquency on association between early pubertal timing and aggression

	6th Grade				7th Grade				8th Grade
	B	β	p	ΔR2	B	β	p	ΔR2	B	β	p	ΔR2
Step 1				0.04***				0.02***				0.03***
School type	−0.35	−0.01	0.639		1.44	0.04	0.181		1.29	0.03	0.213
Family Structure	−0.38	−0.02	0.404		−0.57	−0.02	0.407		0.41	−0.01	0.616
Latino	−2.57	−0.12	< 0.001		−2.20	−0.09	0.003		−3.22	−0.05	0.016
Gender	1.27	0.06	0.005		−0.13	−0.01	0.854		0.52	0.15	< 0.001
Step 2				0.33***				0.10***				0.06***
Pubertal Timing	0.21	0.01	0.674		1.23	0.05	0.100		1.32	0	0.883
Friend Delinquency (6th)	1.14	0.58	< 0.001		0.72	0.31	< 0.001		0.59	0.57	< 0.001
Model R2		0.37				0.12				0.09
Model F		129.83***				24.75***				16.07***
df		6, 1337				6, 1110				6, 1020

Note. B, β, t, R2, F, and df reflect values from the final regression equation

Italics: Previously significant association between pubertal timing and aggression completely mediated

^*p < .05.

^**p < .01.

^***p < .001.

Table 3.

Mediating effects of 6th grade friend delinquency on association between early pubertal timing and delinquency

	6th Grade				7th Grade				8th Grade
	B	β	p	ΔR2	B	β	p	ΔR2	B	β	p	ΔR2
Step 1				0.05***				0.03***				0.03***
School type	0.48	0.03	0.213		0.79	0.04	0.197		1.92	0.08	0.010
Family Structure	−0.12	−0.01	0.616		−0.08	−0.01	0.832		−0.02	0	0.965
Latino	−0.59	−0.05	0.016		−0.47	−0.03	0.261		−1.16	−0.07	0.020
Gender	1.63	0.15	< 0.001		1.80	0.13	< 0.001		1.67	0.10	0.001
Step 2				0.32***				0.14***				0.08***
Pubertal Timing	0.04	0	0.883		0.19	0.01	0.661		0.30	0.02	0.562
Friend Delinquency (6th)	0.59	0.57	< 0.001		0.50	0.37	< 0.001		0.45	0.29	< 0.001
Model R2		0.37				0.17				0.11
Model F		129.88***				36.23***				21.40***
df		6, 1338				6, 1113				6, 1021

Note. B, β, t, R2, F, and df reflect values from the final regression equation

Italics: Previously significant association between pubertal timing and delinquency completely mediated

^*p < .05.

^**p < .01.

^***p < .001.

DISCUSSION

The findings from the present investigation provide an important extension of previous findings for middle-class, White youth regarding the association between early pubertal timing and externalizing behaviors in a sample of largely minority, urban males and females. This study moved beyond the demonstration of effects by showing that delinquent peer association early on in adolescence was a pathway between early pubertal timing and externalizing behaviors longitudinally over the middle school years.

As hypothesized, early maturers reported significantly higher average levels of both delinquency and aggression at all points in time, regardless of gender or ethnicity. While the negative effects of early maturation for White females is well documented in the literature for externalizing symptoms and behaviors [7,15,28], the current study extended these findings to urban, minority males and females. Additionally, the effects of early maturation did not differ significantly by gender. Previous research reported that early maturation was associated with adjustment and health in adulthood among males [29]. Future research should attempt to replicate this finding by evaluating adjustment outcomes into young adulthood.

The current study moved beyond the demonstration of main effects by investigating the role of delinquent friends as a pathway underlying the association between early pubertal maturation, aggression, and delinquency in adolescence. Prior studies have found that associating with older male peers is related to negative outcomes for early maturing white females compared to other girls [10]; however, the delinquent behavior of these peers was not evaluated. One study evaluated neighborhood conditions and parenting factors as moderators of externalizing outcomes controlling for the effects of peer delinquency in a cross-sectional sample of 10–12 year olds but peer delinquency was not evaluated as a moderator or mediator of pubertal timing effects in this study [5]. The current study expanded upon previous findings by being the first to examine the mediating effect of associating with delinquent friends in a large, ethnically diverse, urban sample of both males and females that were followed longitudinally across middle school.

Mediating effects of friend delinquency in the 6^th grade were found for aggression and delinquency at all points in time. Differential association with delinquent peers early on appears to place young adolescents on a negative trajectory, which results in higher average levels of problematic outcomes at later points in time compared to their on-time or late maturing counterparts, even if these peers subsequently associate with deviant peers. One previous study of 10–12 year old African Americans found that early maturers were significantly more likely to affiliate with delinquent peers [5]. This age range roughly corresponds with the 6^th grade sample studied here. No other research has examined the association between pubertal timing and friend delinquency in an older sample for these constructs.

Recent studies have linked romantic involvement with delinquent peers and early maturation with problem behaviors and sexual risk among girls in mid and late adolescence [30,31]. Our investigation demonstrated that delinquent peer involvement is salient to externalizing behaviors at younger ages. Perhaps these early links to externalizing behaviors during middle school establish trajectories for subsequent behavior and relationships.

The current study found the largest increases in rates of aggression and delinquency between 6^th and 7^th grades. Future research should attempt to elucidate the factors that contribute to this increase at this point in time. Future preventions attempting to reduce the rates of these problematic behaviors should ideally begin before the transition into 7^th grade or consider factors salient to this year in school. Note that all youth in the current study were enrolled in middle schools and did not change schools between 6^th and 7^th grades.

Strengths and Limitations

This study is limited in that adolescents were from urban communities only. Thus, findings can not be generalized to suburban or rural contexts. However, this is also a strength of the study given the dearth of research on all but suburban middle class communities. Another limitation is that the measures included in the present study were obtained via a self-report survey format. While steps were taken to increase the validity of these reports, some reporting bias may have occurred. Similarly, the current study used a perceived measure of early pubertal timing collected after puberty would have been completed or in advanced stages for many participants. Perceptions of pubertal timing are most highly correlated with objective measures when assessed after puberty [21] and it has been argued that perceptions of pubertal timing may be most relevant when comparing adolescents within diverse racial/ethnic groups [32]. Despite this, the limitations associated with this measure of pubertal timing are noteworthy and replication of these results with an objective measure of timing would be prudent. Recent research that has examined aggression in adolescence has distinguished between overt forms of aggression (e.g., fighting) and relational forms of aggression (e.g., rumor spreading) [33,34]. Moreover, other types of aggression (e.g., relational) may be more common among girls [35]. This study is limited in that we are unable to evaluate relational aggression as the larger prevention study did not include this information. Finally, the amount of variance explained in this study is low to moderate indicating that future studies should evaluate other potential mediators of the association between early pubertal timing and externalizing outcomes. This includes evaluations of specific skill deficits such as communication, assertiveness, and decision making skills as well as aspects of the early childhood environment such as family cohesion, presence of non-biological adult males in the home, and other stressful life events.

Implications

This study indicated that early maturation was a risk factor for aggressive and delinquent behaviors among both males and females from urban communities. Early maturation placed individuals on a negative trajectory early on resulting in higher levels of problematic behaviors in later years and this was attributable to affiliation with delinquent friends in the 6^th grade. By targeting prevention efforts to focus on developing non-delinquent peer friendships early on, it may be possible to attenuate some of the negative outcomes associated with early maturation in later years. The practical significance of even a slight reduction in rates of aggression and delinquency among adolescents would prove to be of great value socially. Further investigation of the specific risks, skill deficits, or social contexts that place early maturers at risk is warranted in order to better inform, parents, practitioners and prevention programming.