Delinquency During Puberty as a Biosocial Behavior: Time-Varying Effects of Social Contexts on Girls’ Delinquency in South Korea

Abstract

Using Time-Varying Effect Model (TVEM) to analyze five waves of nationally representative South Korean panel data, from Grades 8 to 12, this study investigates how the effects of delinquent peers, parental care, and compliance with school rules on delinquency likelihood change across adolescence for girls with early versus late pubertal timing. The results reveal complex nonlinear trajectories that differ by puberty group: delinquent peer influence becomes more pronounced for late maturers after Grade 10, parental care consistently suppresses delinquency only for early maturers, and school rule compliance relates negatively to delinquency in late maturers but positively in early maturers. The findings challenge the partial application of the maturity gap hypothesis to early maturers only, highlighting how both early and late maturers experience strain from off-time development but negotiate it differently through shifting interactions with key social contexts over adolescence.

Keywords

puberty maturity gap delinquency South Korea time-varying effect model

Introduction

Male adolescents commit more frequent and serious deviance than their female counterparts; thus, scholars have traditionally focused on their delinquency. However, there have been calls to investigate the unique nature of female adolescent delinquency (for review, Kruttschnitt, 2013), with pubertal delinquency among girls being one such area of interest. The relationship between the premature development of girls and delinquency has been examined through various disciplines (for review, Dimler & Natsuaki, 2015). Extant literature has repeatedly reported that a girl’s early onset of puberty is more likely to cause internal and external problems, including violent misconduct (Harden & Mendle, 2012), psychopathologic problems (Graber, 2013), sexual risky behaviors (Downing & Bellis, 2009), smoking (van Jaarsveld et al., 2007), drinking alcohol (Bucci et al., 2021), and other substance abuse (Kaltiala-Heino et al., 2011).

However, the complex interactions between the biological and social aspects of pubertal behaviors over time have not been adequately studied. One major reason for this gap is the dependence on cross-sectional data (Tsai et al., 2015). Indeed, it is crucial to acknowledge that pubertal development is a gradual process that occurs over several years, rather than a single isolated event. As a result, the responses of others to an adolescent’s pubertal development are also likely to change over time. For example, both friends and parents may regard girls who experience early puberty as more mature than their peers, which in turn leads to changes in the interplay between them. Therefore, to investigate the biosocial aspects of pubertal behaviors, it is essential to conduct longitudinal research that reveals the influence of puberty and its interplay with sociocultural factors over an extended period.

Although some research has utilized longitudinal models (e.g., Kaltiala-Heino et al., 2011; Negriff et al., 2011), they also have proven insufficient to represent the complex trajectory of biosocial dimensions of puberty. Specifically, except for Copeland et al. (2010) and Tsai et al. (2015), previous research conducted only one or two follow-up investigations instead of regularly analyzing population-based panel samples. Another issue is that these studies relied on multilevel models that presuppose a specific shape of time effect, which was linear in most cases. These limitations have led to the conclusion that pubertal development is linear throughout adolescence, which might oversimplify the effect of social factors as well as the critical period.

Another gap in previous research on the influences of girls’ pubertal timing on adolescents’ behavioral disorders arises from a focus on European descent samples in Western countries (Tsai et al., 2015). Given that the meaning of a girl’s sexual development varies by culture, religion, history, and geographic location (Egan, 2013), investigating East Asian girls’ pubertal timing and its behavioral impacts could enrich previous discussions on pathways of juvenile risk behaviors among female adolescents. Paying attention to East Asia, especially South Korea, holds social as well as academic significance. In relation to puberty timing, South Korea is unique; it is one of the countries with the most rapid decline in the average age of girls’ onset of puberty. The average menarche age decreased from 13.0 years in 2006 to 12.6 years in 2015 (Seo et al., 2020). Compared to the United States, which experienced a 0.1-year decline during a similar period (Martinez, 2020), the rate of decline in South Korea is markedly steeper. Reflecting this trend, the number of children’s hospital visits for precocious puberty in South Korea increased dramatically from 2,795 in 2004 to 136,334 in 2020, with approximately 90% of these cases involving girls (Korean Health Insurance Review & Assessment Service, 2020). This rapid change may reflect societal concerns about girls’ puberty rather than an epidemiological trend.

To address these limitations, this study aims to investigate the time-varying effects of social contexts on delinquency by puberty timing, using five waves of the Korea Child and Youth Panel Study (KCYPS). A Time-Varying Effect Model (TVEM) is employed for statistical analysis. TVEM offers a nonparametric continuous time function, allowing the trajectory of coefficients to be directly drawn from data, unlike other longitudinal models that impose a prespecified linear shape of time. Therefore, TVEM enables researchers to investigate the nuanced changes in the effects of predictors, affording more intuitive and elaborate interpretations.

The Maturity Gap as a Source of Strain

The concept of the maturity gap was introduced by Moffitt (1993) to explain why the majority of criminal offenders are teenagers. The maturity gap refers to the discrepancy between physical maturity and social maturity, where adolescents may appear physically mature but lack the social and legal privileges of adulthood. Moffitt suggested two distinct groups of adolescents: life-course persistent offenders and adolescence-limited offenders. Life-course persistent offenders engage in antisocial and criminal behaviors that begin early in childhood and continue throughout adulthood, with causes being a combination of neuropsychological impairments present at birth and adverse contextual factors. However, this group only accounts for 5% to 10% of the overall population (Moffitt, 1993; Schwartz & Beaver, 2015).

In contrast, a substantial population that engages in delinquency as a normal part of adolescent development consists of adolescent-limited offenders. Their antisocial behavior is limited to adolescence and is closely related to pubertal development. The accelerated pubertal timing in contemporary society due to nutritional progress has widened the gap between physical maturity and reaching legal adulthood. Moreover, the delay in adulthood caused by the transition in economic structure and the prolongation of life expectancy has further widened this gap (Arnett, 2000). As a result, adolescents who experience strain from the mismatch between biological and sociocultural maturity tend to demonstrate their independence by mimicking the antisocial behaviors or adult-like behaviors of life-course persistent offenders and older peers who are similarly restricted by the maturity gap (Moffitt, 1993, p. 686).

Researchers have accumulated studies illustrating the relationship between the timing of puberty and delinquency (e.g., Bucci & Staff, 2020), emphasizing that the effect of pubertal timing on delinquency is not attributable solely to either biological or social factors. Although some studies highlight the gonadal hormonal change at puberty and its impact on brain changes that cause psychopathologic status (Patton & Viner, 2007), demonstrating which hormonal secretion triggers such brain changes remains elusive (Roberts, 2015, p. 134). Harden and Mendle (2012) include genetic information and sociocultural factors in their analyses to distinguish genetic effects from environmental effects, drawing from a nationally representative twin-sibling sample. Their research showed that, after adjusting for genetic effects, social environments still have significant impacts among girls who experience early puberty.

To explain the relationship between maturity gap and delinquency, studies have mobilized the general strain theory (Agnew, 1992). General strain theory posits that crime is a result of negative relationships with others that cause individuals to experience strain or stress. This strain then leads to negative emotions such as anger and frustration, which may prompt criminal behavior as a way to cope. When applied to the maturity gap, girls who experience early puberty may have strain due to the maturity gap between their adult-like expectations and the limitations imposed by their legal age. For example, early maturers are more likely to have a negative body image of themselves than others, which leads to more delinquency (Bucci & Staff, 2020). In this case, the stress about their body image may come from the mismatch between their physically mature appearance and their actual age, as well as societal expectations and norms associated with their physical development. On the contrary, late-maturing girls may experience significant strain as a result of their delayed development. They may feel left behind or inadequate compared to their more physically developed peers, which can lead to increased levels of anxiety, depression, and a decreased sense of self-worth (Negriff & Susman, 2011). This psychological stress can manifest in engagement in adult-like activities such as substance use, as a means of demonstrating their maturity and compensating for their perceived developmental lag (Williams & Dunlop, 1999).

Social Factors in Pubertal Delinquency

Previous studies have demonstrated that pubertal timing interplays with several social contexts (e.g., Haynie, 2003). Among social contexts, this study investigates three social contexts that are expected to be associated with strain from puberty onset timing. The first social context is peer influence. Many delinquent behaviors, such as smoking and drinking alcohol, are considered adult-like behaviors. Adolescents who engage in these behaviors are often viewed by their peers as mature and privileged. Moffitt (1993) argued that delinquency caused by the maturity gap is a form of “social mimicry” of other peers’ behaviors (p. 686). Early maturers are more likely to mimic adult-like behaviors to address the strain resulting from the maturity gap. Supporting this hypothesis, previous research has demonstrated that early maturers tend to be influenced by older or deviant peers who serve as a reference group (Mrug et al., 2014). For example, recent research by Bucci and Staff (2020) found that increases in deviant peer affiliations mediate the relationship between pubertal timing and delinquency, suggesting that early pubertal development attracts delinquent peers, which in turn leads to more delinquencies. Moreover, as romantic relationships are representative of adult-like behaviors, they can be one dimension of peer effects that increases the likelihood of delinquency among early maturers. Studies demonstrated that not only are early maturers more likely to have romantic orientations during adolescence (Cavanagh, 2011) but also that their romantic relationships lead to more delinquencies (Haynie, 2003).

The second social context of delinquency during pubertal development is parenting style. Previous research shows that some parenting styles are effective in reducing delinquency among girls who experience early puberty. For instance, strong parental supervision has been shown to reduce the likelihood of early maturers engaging in risky behaviors (Klopack et al., 2020). In line with this, low parental nurturance, communication, and knowledge of the child’s activities are likely to increase the possibility of risky behaviors in early maturers (Mrug et al., 2008). Moreover, parenting style is critical for adolescents who experience the maturity gap and have desire to seek autonomy from their parents’ guardianship. Indeed, early maturers report higher levels of autonomy and more frequent conflicts with parents, particularly when their parents have a harsh parenting style (F. R. Chen & Raine, 2018). Furthermore, the relationship between pubertal timing and delinquency is mediated by the desire for autonomy as well as conflicts with parents (Haynie, 2003). These results indicate that parenting in response to a girl’s pubertal development plays a crucial role in delinquency.

The last social context is school. From the perspective of the maturity gap, school has a dual role. According to social control theory (Hirschi, 1969), school serves as an important social bond that prevents adolescents from engaging in delinquency by instilling conventional values in them. For instance, school can hinder adolescents who mature early from forming connections with older peers (Stattin et al., 2011). However, at the same time, school can also be a source of strain for some students as it provides environments for comparing each other. In the context of pubertal timing, early maturers may regard themselves as more mature than other students, thus complying with the same school rules might cause additional strain for them. In this scenario, they may resolve this strain by engaging in delinquency outside of school. As a result, although school rules can create distinctive strain for adolescents experiencing the maturity gap, few studies have considered school-related factors in their empirical models.

Building on prior studies on pubertal timing, social contexts, and delinquency, this study articulates the possibility of changing effects of social factors in different pubertal timing groups over time. Addressing this gap is critical, as the strain from the maturity gap is influenced by adolescents’ relative perceptions of their bodily development (Dalzell & Cavanagh, 2021), which continues to evolve throughout their adolescence. The dynamic nature of these perceptions and their impact on social interactions and delinquency have not been adequately explored in previous research. For instance, early maturers may initially enjoy a high social status among their peer groups due to their advanced physical development compared to their peers. They may be perceived as more mature and attractive, which can lead to increased popularity and social influence. However, this elevated status may be temporary, as it may diminish once late maturers begin their own pubertal development. As more adolescents reach puberty, the distinctiveness of early maturation is lost, and the social advantages associated with it may fade. In contrast, late maturers may initially experience social disadvantages and strain due to their delayed physical development. They may feel left behind or inadequate compared to their more physically developed peers, leading to increased levels of stress and negative emotions. However, as they progress through their own pubertal development, late maturers may gradually catch up with their peers, narrowing the maturity gap and reducing the associated strain. As a result, the influence of social contexts on delinquency among late maturers may also change over time, with the impact of factors such as peer influence and parental care potentially varying as they navigate their own developmental trajectory.

Previous studies may have overlooked these time-sensitive dynamics due to an assumed linear time effect, which simplifies the complex interplay between pubertal development and social factors. Typically, this approach depicts the relationship over time solely through the steepness of the slope, hindering the identification of critical periods in which adolescents are most likely to be influenced by social contexts. To overcome these limitations, this study aims to use the Time-Varying Effects Model (TVEM) to examine how social contexts (i.e., delinquent peers, parental care, and compliance with school rules) influence the timing of delinquency across different pubertal onset groups in South Korea.

Method

Data

This study derives a data set from the Korea Children and Youth Panel Study (KCYPS) produced by the Korean government. KCYPS has been conducted every year for seven waves of the sample since the first survey interview in 2010. The sample was systematically selected using the 2009 Ministry of Education School Statistics as a sampling frame through the stratified cluster sampling method. The survey consists of three cohorts: 2,342 students in first grade cohort, 2,378 students in fourth grade cohort, and 2,351 students in seventh grade cohort. Because this study focuses on female puberty and its time-varying effects, I used seventh grade cohort of the three cohorts. The final sample comprises 985 female adolescents from wave 2 (Grade 8) to wave 6 (Grade 12).

Measures

Delinquency in the Past 12 Months

The dependent variable is whether the respondent committed one or more delinquencies in the past 12 months. KCYPS measured 13 types of self-reported delinquency, including smoking, drinking, unexcused absence, running away, stealing, serious and malevolent teasing, gambling, threatening other students, bullying other students, gang fighting, robbing, assault, sexual intercourse, and sexual harassment. Respondents were asked if they had engaged in each activity in the past 12 months or not. Except for drinking (8.5%), smoking (2.3%), and unexcused absences (2.4%), the prevalence of other delinquencies is below 1%. Consequently, a dichotomized scale was employed for the outcome variable, assigning a value of “1” if the respondent had committed at least one delinquency in the past year, and “0” otherwise. The prevalence of engaging in any form of delinquency is measured at 12.11% (Supplemental Table S2, available in the online version of this article). This overall delinquency prevalence is relatively lower than in the United States or other Western countries due to cultural differences. First, in South Korea, any illicit drug that is frequently used with other substances, such as cigarettes or alcohol, is strictly prohibited for all populations, while 38.6% of U.S. adolescents used illicit drugs in the past year (Miech et al., 2023). Second, due to collectivism and Confucian cultures, East Asian countries have less crime among younger age groups as familial and societal expectations enforce more stringent behavioral norms (Steffensmeier et al., 2020).

Pubertal Timing Groups

This study has four core predictors that are expected to impact the probability of female adolescents’ pubertal delinquency. First, the age of menarche is the proxy for the pubertal timing of girls. It is worth noting that the objective standard of puberty has not been established yet, and there are several possible criteria to evaluate a girl’s pubertal stage. However, the majority of data on adolescent development around the world are likely to measure the age of menarche for girls’ puberty because measuring other physical changes requires longitudinal tracking, which takes time and effort, but menarche is a discrete event. Accordingly, previous research on pubertal timing and its behavioral and psychological effects have widely used the menarche item. The KCYPS data measures the school grades of menarche, which hinders the exact calculation of the onset of menstruation because age variation within the same grade exists. Following previous research using the same data set (Park et al., 2017), I calculated an approximate indicator of the age of menarche. Since every participant completed menarche by Grade 11, I calculated the gap between the grade of menarche and Grade 11 and then subtracted this gap from the actual age. For instance, when a girl is 16.8 years old in Grade 11 and started menarche in Grade 9, her age of menarche is calculated as: 16.8 − 2 = 14.8.

Based on previous research (Haynie, 2003; Park et al., 2017), 15% of participants from both the bottom and top were categorized into early and late groups: early (<11.50), normal (11.50–13.75), and late puberty (>13.75) groups. As shown in Supplemental Table S1, the total mean age of menarche is 12.68, a finding comparable to the results of other nationally representative investigations (Seo et al., 2020). For the robustness check, analyses with 10% and 20% thresholds were also conducted, and the overall results were substantially similar.

Social Factors

The first social factor assessed was the number of delinquent peers. Respondents answered the question, “How many of your close friends have engaged in the following delinquent behaviors in the past year?” using the same 13 items previously measured for self-reported delinquency. The total was derived by summing these 13 items. However, due to significant skewness in the responses (M = 2.39; SD = 8.68; Min = 0; Max = 330; Skewness = 2.85), the data were top-coded at 20 to address the extreme values.

The second social factor under investigation is the level of parental care. This variable is measured using four items: “My parents value me more than anything else,” “They are curious about my school life and ask about it,” “They always try to keep my body, clothes, blankets, etc., clean,” and “They ensure I receive appropriate treatment when I am sick.” Respondents provided answers on a 4-point scale, with the final variable being the average of these four items, which ranged from 1 to 4. Principal factor analysis revealed that all items are highly loaded on a single factor (Eigenvalue = 2.33), with factor loadings varying from 0.62 to 0.82. The Cronbach’s alpha for this scale is .75, demonstrating adequate internal consistency.

The final social factor examined is the level of compliance with school rules. This factor is composed of five items: “I work hard on class activities,” “I treat school property with care,” “I take turns well in the restroom and lunchroom,” and “I dispose of trash properly.” Answers were again provided on a 4-point scale, and the final variable represents the mean of these five items, ranging from 1 to 4. Principal factor analysis indicates that all items are highly loaded on one factor (Eigenvalue = 3.63), with factor loadings ranging from 0.83 to 0.89. The Cronbach’s alpha for this scale is .90, indicating strong internal consistency.

Control Variables

Four control variables related to participants’ family backgrounds are included in the models. First, the educational level of parents is measured, consisting of five ordered scales: middle school graduation, high school graduation, community college, university, and graduate school. The M score is 2.86 with an SD of 0.96. Second, family structure is considered; a “0” is assigned for participants living with both biological mother and father, and “1” otherwise. Of total, 89% of participants live with both biological parents. Third, dual-income household is considered: “1” is assigned if both mother and father are employed, “0” otherwise. Of total, 62% of families are dual-income households. The final family background variable is the logarithm of the household’s income. Other control variables examined in previous research on pubertal timing and delinquency are also included (Park et al., 2017). Victimization is measured as whether respondents have experienced at least one of seven victimization items (Cronbach’s alpha = .8741). Weekly time spent with peers is gauged by the number of hours respondents generally spend after school with friends. Coeducation is defined as whether respondents attend a female-only or coeducational school. Finally, the variable indicating whether respondents experienced menarche or not in each wave is also included.

Statistical Analysis: TVEM

TVEM is a relatively emerging method for investigating how the relationship between two variables changes over time (Lanza & Linden-Carmichael, 2021, p. 2). Standard longitudinal models (e.g., mixed models, latent growth curve models, and so on) are frequently described with a time regression term, generating a single-point estimate for the time impact. However, TVEM enables measuring both temporal changes and their relationship to variables directly from observations, without making any assumptions about the shape of the trajectories (i.e., linear, quadratic, or cubic). This is because TVEM estimates nonparametric regression coefficients as continuous functions of time, which allows for the investigation of complex nonlinear time-varying effects (Lanza et al., 2016). By assuming the continuous functions of time, TVEM can elaborately depict when the direction of effect is reversed, how strong the effects exist at the specific time point, what the overall trajectory of effect is, and which factors (i.e., moderators) change this trajectory for different population groups. Furthermore, TVEM estimates the time-varying effect of both time variants (e.g., the exposure to peer delinquency) and time-invariant predictors (e.g., race or biological sex):

y_{i} = β_{o} (t_{i}) + β_{1} (t_{i}) x_{i} + ε_{i},

(1)

Equation 1 is the basic form. TVEM is an extension of multiple regression in which regression coefficients are calculated as nonparametric functions of continuous time. The random error term ( $ε_{i}$ ) in Equation 1 is the difference between an observed value and a predicted value, which is supposed to be normally and independently distributed. The intercept function ( $β_{o} (t_{i})$ ) and slope function ( $β_{1} (t_{i})$ ) are called coefficient functions. Respectively, the intercept function reflects the mean of the outcome variable (y) at time $t_{i}$ , when all values of predictors (x) are 0, while the slope function is the estimated difference in outcome associated with a one-unit change in predictors at a specific time, $t_{i}$ .

Control variables can also be added to the model. In contrast to the coefficient function, the effects of control variables are considered time-fixed. Put differently, the influence of control variables is considered constant across time, as a fixed effect in traditional multilevel models. Finally, because TVEM provides a nonparametric coefficient function of time, it divides the time variable in the data set into at least a hundred separate time points (default in the TVEM SAS Macro) to make it continuous. As a result, all values of the intercept and slope function depend on different points in time:

\begin{matrix} \ln (d e l i n q u e n c y_{i} / 1 - d e l i n q u e n c y_{i}) = β_{o} (t_{i}) + β_{1} (t_{i}) d e l i n q u e n t p e e r s_{i} \\ + β_{2} (t_{i}) p a r e n t a l c a r i n g_{i} \\ + β_{3} (t_{i}) c o m p l i a n c e w i t h s c h o o l r u l e s_{i} \\ + c o n t r o l v a r i a b l e s_{i} + ε_{i} \end{matrix}

(2)

The following analyses consist of three steps. First, to illustrate how the effect of pubertal timing changes over time, a full-sample estimation is conducted. The coefficient function is applied to the maturity gap as a continuous scale (i.e., legal age of adult [19]—pubertal timing), and other variables are regarded as control variables. Second, an intercept-only model shows the prevalence of delinquency across time (i.e., school grades) separately for three puberty groups. Finally, core independent variables (delinquent peers, parental caring, and compliance with school rules) and control variables are incorporated into the model. The final model shows how the effect of predictors on the likelihood of delinquency varies across time while estimating the time-fixed effect of control variables. All TVEM models were conducted using the Weighted TVEM SAS Macro (Version 2.6.0) (2017). All models were estimated using the B-spline method, and the Akaike information criterion (AIC) and the Bayesian information criterion (BIC) were used to choose models with the right number of knots (i.e., inflection points) for each coefficient function. The results of time-fixed control variable models and the results of three group models (i.e., early, on-time, and late puberty groups) are included in the Supplemental document.

Results

Time-Varying Effects of Maturity Gap

Figure 1 illustrates the time-varying effects of the maturity gap on delinquency. The black line indicates the coefficient trend, while the red dotted line represents 95% confidence intervals (CIs). Nonoverlapping CIs with the gray line (y = 0) denote a significant effect at the 95% confidence level for that time point. For instance, in Figure 1, when the x-value is Grade 8, the lower side of the confidence level does not overlap with the y = 0 line, indicating that a greater maturity gap is associated with a higher delinquency probability (95% CI range = [0.02, 0.55]). In other words, consistent with previous studies, girls who experienced puberty early are more likely to commit delinquency at this time point. Among other covariates without time-varying functions, delinquent peers (Coef = 0.17, p < .0001), parental caring (Coef = −0.34, p = .02), compliance with school rules (Coef = −0.37, p = .02), time spent with friends (Coef = 0.14, p < .0001), and victimization (Coef = 1.85, p < .0001) significantly influence the likelihood of delinquency at all time points. However, the maturity gap–delinquency relationship only persists until Grade 10, as CIs subsequently overlap with y = 0. This implies that the independent effect of the maturity gap attenuates with age.

Figure 1:

Time-Varying Coefficients Functions Depicting the Association Between Delinquency and Maturity Gap

Prevalence of Delinquency

Figure 2 presents the intercept functions without other covariates separately for the early puberty group (solid red line) and the late puberty group (blue dashed lines) and the corresponding 95% CIs. The coefficient was converted into the prevalence of delinquency across time. A significant difference between the two puberty groups exists when the CIs of each line do not overlap, although this is considered an overly conservative estimate (Lanza & Linden-Carmichael, 2021).

Figure 2:

Estimated Prevalence of Delinquency Across Grade (Early and Late)

Figure 2 shows the trend of the average prevalence of delinquency for early and late maturers. Both trajectories are slightly U-shaped, implying that the prevalence of delinquency first decreases and then increases. In Grade 8, the prevalence for the early puberty group is 19.55% (95% CI = [15.65, 24.43]), but it decreases to 13.59% (95% CI = [10.78, 17.11]) in Grade 9.5. However, by Grade 12, the prevalence increases to 19.80% (95% CI = [15.99, 24.52]). The late puberty group exhibits a similar pattern, with the lowest delinquency prevalence occurring in Grade 9.5 (7.78%, 95% CI = [6.16, 9.83]). In line with the maturity gap hypothesis, the line of the early puberty group is consistently above that of the late puberty group. More specifically, the early maturers’ prevalence is significantly higher than that of late maturers from Grades 8.3 to 10.8 without any overlaps between CIs.

Time-Varying Effects of Social Factors

Figures 3 to 5 illustrate the results of the models including time-varying social factors and time-fixed control variables. As in Figure 2, CIs overlapping the black line (y = 0) indicate no significant effect at that time point. Positive and negative coefficients signify promotive (+) and suppressive (−) roles of each factor on delinquency, respectively. Figure 3 shows that, regardless of pubertal timing, affiliating with delinquent peers significantly increases delinquency across all grades. However, the effect trajectory varies by puberty group. First, although the overall delinquency prevalence is higher among early maturers (Figure 2), the influence of delinquent peers is greater for late maturers. The lines begin diverging at Grade 10.2, with a significant difference lasting until Grade 11.9. In other words, for approximately 2 years, late maturers are more susceptible to peer influence than early maturers. Second, the effect increases steadily for late maturers but shows little change over time among early maturers. In Grade 8, coefficients are 0.13 (95% CI = [0.09, 0.17]) and 0.07 (95% CI = [0.03, 0.12]) for early and late groups, respectively. By Grade 12, these become 0.18 (95% CI = [0.14, 0.23]) and 0.27 (95% CI = [0.22, 0.31]). Thus, the slope of peer influence is steeper for late maturers in mid-to-late adolescence. In summary, late maturers become more peer-influenced over time, while early maturers do not.

Figure 3:

Time-Varying Coefficients Functions Depicting the Association Between Delinquency and Exposure to Peer Delinquency Across Puberty Groups (Early and Late)

Figure 4:

Time-Varying Coefficients Functions Depicting the Association Between Delinquency and Parental Caring Across Puberty Groups (Early and Late)

Figure 5:

Time-Varying Coefficients Functions Depicting the Association Between Delinquency and Compliance With School Rules Across Puberty Groups (Early and Late)

Figure 4 illustrates the time-varying effect of the second social factor, parental caring. First, for the early puberty group, the negative effects of parental caring persist across time points, suggesting that higher levels of parental caring suppress delinquency among early maturers. Second, for the late puberty group, parental caring does not significantly predict the likelihood of delinquency over the first four years. During this period, the effect shifts from positive to negative at Grade 9 and continues declining, although nonsignificantly. Upon transitioning to Grade 12, the negative effect of parental caring becomes significant, as evidenced by nonoverlapping CIs. In other words, during most of the analytic period, the suppressing effect of parental supervision on delinquency is only observed in early maturers. The significant difference in effect between the two groups exists between Grades 8 and 9.4.

Finally, Figure 5 illustrates the changing effects of compliance with school rules. First, the early puberty group line stays above the y = 0 line for most time points. Significant effects occur from Grades 9.6 to 10.6, indicating that increased compliance is associated with a higher delinquency probability among early maturers. Second, conversely, compliance negatively relates to delinquency for late maturers during approximately half of the analytic periods. From Grades 8 to 8.4 and Grades 10.4 to 12, significant suppressive effects exist for late maturers. The significant difference in effect between the two groups exists between Grades 10.3 and 11.8.

It is worth noting that the effects of control variables without time-varying functions were also estimated for each group. For the early puberty group, victimization (Coef = 2.28, p < .001) and coeducation (Coef = 0.68, p = .002) were positively associated with delinquency. In contrast, for the late puberty group, time spent with friends (Coef = 0.18, p = .02) showed positive associations with delinquency.

Discussion and Conclusion

The objective of this research is to investigate the biosocial dimension of female adolescent delinquency during pubertal development. Previous studies have found that not only does early puberty lead to a higher risk of delinquency but also that pubertal timing is mediated or moderated by various social contexts such as peers and parents. However, most empirical analyses have relied on a linear function of time, which cannot illuminate the dynamic effects of puberty and social factors over time. Using TVEM analysis on five waves of nationally representative panel data in South Korea, this study demonstrates the changing effects of three social factors on the likelihood of delinquency across different pubertal timing groups.

The results revealed several underexplored findings. Beginning with the full-sample model that incorporates a time-varying function of the maturity gap (as illustrated in Figure 1), the findings align with prior research in noting the elevated risk associated with early puberty in South Korea (Park et al., 2017). Notably, the impact of the maturity gap did not persist over the entire 5-year span. Direct effects were primarily concentrated within the first 3 years, ranging from Grades 8 to 10. This observation suggests a potential diminishing of strain as the maturity gap narrows with age. However, the attenuation of such strain does not necessarily correlate with a reduction in delinquency rates. As indicated in Figure 2, the overall prevalence of delinquency, both in early and late puberty cohorts, escalated during this timeframe. Consequently, post-Grade 10 delinquency may result from intricate interactions between the maturity gap and various social contexts. This nuanced finding underscores the limitations of relying solely on 1-year cross-sectional analyses for studying the relationship between puberty and delinquency. Indeed, the TVEMs in this study show that the trajectory of three social factors’ influence diverges among puberty groups. This finding further complicates the understanding of the interplay between biological and social factors in influencing delinquent behavior.

The main findings from TVEMs can be summarized as follows. First, numerous criminological studies have shown that delinquent peers significantly influence adolescent behavior, increasing the likelihood of delinquency. This effect is particularly pronounced in late maturers, becoming more apparent after Grade 10. In other words, late maturers exhibit greater sensitivity to the influence of delinquent peers compared to early maturers over the same period. This heightened sensitivity among late maturers has not been adequately emphasized in previous research, likely because they are not perceived as experiencing the strains associated with a maturity gap (Barnes et al., 2011; Moffitt, 1993). A possible explanation for this could be that the significance of puberty onset as a marker of privileged status changes over time. On one hand, status competition among peers of similar age is a significant driver of adolescent delinquency (Hoeben et al., 2021), and the strain from a maturity gap is closely associated with the social comparison with other peers (Moffitt, 1993). On the other hand, early puberty might be viewed as a winning sign in this competition due to adult-like physical development, attracting more delinquent or older peers (Bucci & Staff, 2020). However, data from this study indicate that the majority of late maturers begin puberty before Grade 9 (M = Grade 8.51, SD = 0.60). As late maturers reach the physical development level of early maturers, the latter may lose their perceived adult-like status. Consequently, as their peers begin their pubertal development, early maturers might no longer view delinquency among peers as a sign specifically targeting them. Conversely, late maturers, previously considered too young, might begin to view their onset of puberty as indicative of adult-like behaviors. This shift in perception could prompt late maturers to increasingly imitate the adult-like behaviors of their peers, compensating for their delayed attainment of status following puberty onset. This finding is crucial as it underscores the complexity of antisocial behaviors not only in early but also in late maturers, who may engage in such behaviors due to the stress related to on-time bodily development (Tsai et al., 2015; Williams & Dunlop, 1999), with potential mechanisms related to social interaction previously unexplored.

It is important to note that the impact of coeducation is significant in the model with time-fixed control variables (Supplemental Table S3), showing that attending a mixed-gender school is associated with increased delinquencies among early maturers. This finding highlights the potential role of male peers and boyfriends in explaining the relationship between puberty timing and delinquency (Caspi et al., 1993; Haynie, 2003; Park et al., 2017). However, caution is warranted in interpreting these results. The mere presence of coeducation does not fully elucidate the mechanisms of male peer influence on delinquency. In many South Korean schools, for instance, even though the schools are coeducational, classrooms or entire buildings may be segregated by gender. Consequently, it is challenging to assert that coeducation directly results in increased interactions with male peers or boyfriends.

Second, within the early puberty group, higher levels of parental care are consistently linked with a lower probability of delinquency, showing little variance over time. Conversely, in late maturers, parental care does not influence delinquency rates until after Grade 11, where it has a negative effect. In simpler terms, parental care plays a more critical role in reducing delinquency among early maturers than it does among late maturers. This supports findings from prior research (Haynie, 2003; Mrug et al., 2008), which emphasize the significant impact of parental involvement on early maturers. In addition, Supplemental Figure S3 in the Supplemental note highlights the distinct influence of parental care in this group, as the trends for on-time and late maturers do not diverge significantly. One possible explanation for the stronger relationship between parental care and reduced delinquency in early maturers may be the heightened societal concern over early puberty in South Korea since the 2010s. Given that early maturers experience menarche before Grade 5 (M = 4.86, SD = 0.48), parents aware of their child’s advanced development may become more proactive in their involvement. They may recognize the potential risks and challenges associated with early pubertal development, such as increased vulnerability to peer influence and engagement in adult-like behaviors. As a result, parents of early maturers may be more vigilant in monitoring their children’s activities, providing emotional support, and maintaining open lines of communication. This increased parental care and involvement may serve as a protective factor against delinquency for early maturers, helping them navigate the strain of the maturity gap and resist the temptation to engage in problem behaviors. However, it is important to note that additional analyses with diverse data sets are necessary for substantiating this cohort-specific interpretation.

Finally, the effects of compliance with school rules show contrasting patterns between early and late maturers. For late maturers, increased compliance is linked with a reduced likelihood of delinquency from Grades 10.5 to 12—a pattern that differs significantly from that observed in early maturers. For late maturers, this period typically follows the onset of puberty, suggesting that enhanced social bonding within schools might mitigate the challenges associated with late puberty. This finding highlights the potential protective role of school engagement and adherence to school rules for late maturers. By fostering a sense of belonging and promoting prosocial behaviors, schools can help late maturers navigate the challenges of their delayed pubertal development and reduce the risk of delinquency. Conversely, among early maturers, greater compliance increases the risk of delinquency between Grades 9.6 and 10.6. Although the direction of coefficients for the remainder of the period is not statistically significant, they are predominantly positive. This suggests that while adherence to school rules generally serves as a social bond that discourages delinquency (Hirschi, 1969), it may intensify the strain of the maturity gap for early maturers. Compliance might lead early maturers to perceive a devaluation of their adult-like status, potentially prompting compensatory delinquent behaviors. In other words, early maturers who feel constrained by school rules and expectations may seek to assert their independence and maturity through engaging in delinquent activities outside of the school context.

Several limitations should be stated. First, the study omits the early adolescent period before Grade 8. Early maturers experience puberty onset around Grade 5; thus, their strain from the maturity gap may be present in Grades 6 and 7. This means that the current study cannot capture the potential effects of the maturity gap on delinquency during these earlier grades. The lack of data from these earlier time points limits our understanding of how the maturity gap may influence delinquency over a more extended period of adolescence. Moreover, the absence of data from Grades 6 and 7 prevents us from examining whether the effects of the maturity gap on delinquency differ between the early and later stages of adolescence. Future research should aim to include data from earlier grades to provide a more comprehensive understanding of the relationship between the maturity gap and delinquency throughout adolescence. Second, the measure of delinquent peers in this study does not differentiate between co-offending and the delinquencies of peers, a recognized limitation in the literature on peer effects (McGloin & Thomas, 2019). This means that without network data, it is impossible to determine whether early maturers are initiating contact with delinquent peers or the other way around, even though the original maturity gap hypothesis suggests the former. By distinguishing between co-offending and the delinquencies of peers, future research can better understand the directionality of peer influence and how it may vary across different pubertal timing groups. Third, the visibility of menarche makes female puberty more readily measurable than male puberty. However, as articulated in prior research, delinquent behaviors are notably gendered (Kreager & Staff, 2009). Consequently, biosocial mechanisms may not be uniform across genders. More research is needed with accurate signs of puberty that show how systematic differences in social factors affect men and women differently.

Notwithstanding its limitations, this study offers several methodological and theoretical advancements to the current understanding of pubertal delinquency. First, this research pioneers the use of TVEM to illustrate the dynamic interactions between biological pubertal timing and social factors. The nonparametric coefficient function estimated through TVEM enables more flexible curves to be drawn for each social factor, thereby enriching the understanding of the complex dynamics among these elements across different pubertal timing groups. Second, while Moffitt’s (1993) maturity gap hypothesis has been prolifically invoked to elucidate the biosocial underpinnings of delinquencies among early maturers, the study challenges this partial application. In particular, the results of TVEMs show how important it is to study the different ways late maturers experience the maturity gap, which has not been sufficiently explored thus far (X. Chen & Adams, 2010). In essence, late maturers are not exempt from the challenges posed by the maturity gap; they simply negotiate these challenges differently through unique interactions with peers and parents at varying times.

Supplemental Material

sj-docx-1-cjb-10.1177_00938548241276087 – Supplemental material for Delinquency During Puberty as a Biosocial Behavior: Time-Varying Effects of Social Contexts on Girls’ Delinquency in South Korea

Supplemental material, sj-docx-1-cjb-10.1177_00938548241276087 for Delinquency During Puberty as a Biosocial Behavior: Time-Varying Effects of Social Contexts on Girls’ Delinquency in South Korea by Heeyoung Lee in Criminal Justice and Behavior

Footnotes

Authors’ note:

I would like to express my gratitude to the editorial team of Criminal Justice and Behavior for their support throughout the review process. I am also grateful for the valuable feedback on this paper from Professor Steven Messner, Jonathan Dirlam, and Joanne M. Kaufman in the Department of Sociology at the University at Albany, SUNY. The author declares no conflicts of interest related to this study. This research did not receive any specific grant from funding agencies in the public, commercial, or not-forprofit sectors.

ORCID iD

Heeyoung Lee

Supplemental Material

Supplemental Material is available in the online version of this article at .

Heeyoung Lee is a PhD candidate in the Sociology Department at the University at Albany, SUNY. His research interests include the integration of medical research and criminology through sociological frameworks.

References

Agnew

(1992). Foundation for a general strain theory of crime and delinquency. Criminology, 30(1), 47–88. https://doi.org/10.1111/j.1745-9125.1992.tb01093.x

Arnett

J. J.

(2000). Emerging adulthood: A theory of development from the late teens through the twenties. The American Psychologist, 55(5), 469–480.

Barnes

J. C.

Beaver

K. M.

Piquero

A. R.

(2011). A test of Moffitt’s hypotheses of delinquency abstention. Criminal Justice and Behavior, 38(7), 690–709. https://doi.org/10.1177/0093854811405282

Bucci

Staff

(2020). Pubertal timing and adolescent delinquency†. Criminology, 58(3), 537–567. https://doi.org/10.1111/1745-9125.12245

Bucci

Staff

Maggs

J. L.

Dorn

L. D.

(2021). Pubertal timing and adolescent alcohol use: The mediating role of parental and peer influences. Child Development, 92(5), e1017–e1037. https://doi.org/10.1111/cdev.13569

Caspi

Lynam

Moffitt

T. E.

Silva

P. A.

(1993). Unraveling girls’ delinquency: Biological, dispositional, and contextual contributions to adolescent misbehavior. Developmental Psychology, 29(1), 19–30. https://doi.org/10.1037/0012-1649.29.1.19

Cavanagh

S. E.

(2011). Early pubertal timing and the union formation behaviors of young women. Social Forces, 89(4), 1217–1238. https://doi.org/10.1093/sf/89.4.1217

Chen

F. R.

Raine

(2018). Effects of harsh parenting and positive parenting practices on youth aggressive behavior: The moderating role of early pubertal timing. Aggressive Behavior, 44(1), 18–28. https://doi.org/10.1002/ab.21720

Chen

Adams

(2010). Are teen delinquency abstainers social introverts? A test of Moffitt’s theory. Journal of Research in Crime and Delinquency, 47(4), 439–468. https://doi.org/10.1177/0022427810375575

10.

Copeland

Shanahan

Miller

Costello

E. J.

Angold

Maughan

(2010). Outcomes of early pubertal timing in young women: A prospective population-based study. American Journal of Psychiatry, 167(10), 1218–1225. https://doi.org/10.1176/appi.ajp.2010.09081190

11.

Dalzell

E. L.

Cavanagh

(2021). Adolescents’ perceptions of physical development relative to peers and antisocial behaviors. Journal of Developmental and Life-Course Criminology, 7(2), 176–194. https://doi.org/10.1007/s40865-021-00160-0

12.

Dimler

L. M.

Natsuaki

M. N.

(2015). The effects of pubertal timing on externalizing behaviors in adolescence and early adulthood: A meta-analytic review. Journal of Adolescence, 45, 160–170. https://doi.org/10.1016/j.adolescence.2015.07.021

13.

Downing

Bellis

M. A.

(2009). Early pubertal onset and its relationship with sexual risk taking, substance use and anti-social behaviour: A preliminary cross-sectional study. BMC Public Health, 9(1), 1–11. https://doi.org/10.1186/1471-2458-9-446

14.

Egan

R. D.

(2013). Becoming sexual: A critical appraisal of the sexualization of girls. John Wiley & Sons.

15.

Graber

J. A.

(2013). Pubertal timing and the development of psychopathology in adolescence and beyond. Hormones and Behavior, 64(2), 262–269. https://doi.org/10.1016/j.yhbeh.2013.04.003

16.

Harden

K. P.

Mendle

(2012). Gene-environment interplay in the association between pubertal timing and delinquency in adolescent girls. Journal of Abnormal Psychology, 121(1), 73–87. https://doi.org/10.1037/a0024160

17.

Haynie

D. L.

(2003). Contexts of risk? Explaining the link between girls’ pubertal development and their delinquency involvement. Social Forces, 82(1), 355–397. https://doi.org/10.1353/sof.2003.0093

18.

Hirschi

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

19.

Hoeben

E. M.

Osgood

D. W.

Siennick

S. E.

Weerman

F. M.

(2021). Hanging out with the wrong crowd? The role of unstructured socializing in adolescents’ specialization in delinquency and substance use. Journal of Quantitative Criminology, 37(1), 141–177. https://doi.org/10.1007/s10940-019-09447-4

20.

Kaltiala-Heino

Koivisto

A.-M.

Marttunen

Fröjd

(2011). Pubertal timing and substance use in middle adolescence: A 2-year follow-up study. Journal of Youth and Adolescence, 40(10), 1288–1301. https://doi.org/10.1007/s10964-011-9667-1

21.

Klopack

E. T.

Simons

R. L.

Simons

L. G.

(2020). Puberty and girls’ delinquency: A test of competing models explaining the relationship between pubertal development and delinquent behavior. Justice Quarterly, 37(1), 25–52. https://doi.org/10.1080/07418825.2018.1472291

22.

Korean Health Insurance Review & Assessment Service. (2020). Disease classification (4th Class) statistics. HIRA. https://www.hira.or.kr/main.do

23.

Kreager

D. A.

Staff

(2009). The sexual double standard and adolescent peer acceptance. Social Psychology Quarterly, 72(2), 143–164. https://doi.org/10.1177/01902725090720020

24.

Kruttschnitt

(2013). Gender and crime. Annual Review of Sociology, 39(1), 291–308. https://doi.org/10.1146/annurev-soc-071312-145605

25.

Lanza

S. T.

Linden-Carmichael

A. N.

(2021). Time-varying effect modeling for the behavioral, social, and health sciences. Springer. https://doi.org/10.1007/978-3-030-70944-0

26.

Lanza

S. T.

Vasilenko

S. A.

Russell

M. A.

(2016). Time-varying effect modeling to address new questions in behavioral research: Examples in marijuana use. Psychology of Addictive Behaviors: Journal of the Society of Psychologists in Addictive Behaviors, 30(8), 939–954. https://doi.org/10.1037/adb0000208

27.

Martinez

G. M.

(2020). Trends and patterns in menarche in the United States: 1995 through 2013-2017. National Health Statistics Reports, 146, 1–12.

28.

McGloin

J. M.

Thomas

K. J.

(2019). Peer influence and delinquency. Annual Review of Criminology, 2(1), 241–264. https://doi.org/10.1146/annurev-criminol-011518-024551

29.

Miech

R. A.

Johnston

L. D.

Patrick

M. E.

O’Malley

P. M.

Bachman

J. G.

Schulenberg

J. E.

(2023). Monitoring the future national survey results on drug use, 1975-2022: Secondary school students. Institute for Social Research. https://eric.ed.gov/?id=ED627366

30.

Moffitt

T. E.

(1993). Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100(4), 674–701. https://doi.org/10.1037/0033-295X.100.4.674

31.

Mrug

Elliott

Gilliland

M. J.

Grunbaum

J. A.

Tortolero

S. R.

Cuccaro

Schuster

(2008). Positive parenting and early puberty in girls: Protective effects against aggressive behavior. Archives of Pediatrics & Adolescent Medicine, 162(8), 781–786. https://doi.org/10.1001/archpedi.162.8.781

32.

Mrug

Elliott

M. N.

Davies

Tortolero

S. R.

Cuccaro

Schuster

M. A.

(2014). Early puberty, negative peer influence, and problem behaviors in adolescent girls. Pediatrics, 133(1), 7–14. https://doi.org/10.1542/peds.2013-0628

33.

Negriff

Trickett

P. K.

(2011). Exposure to peer delinquency as a mediator between self-report pubertal timing and delinquency: A longitudinal study of mediation. Development and Psychopathology, 23(1), 293–304. https://doi.org/10.1017/S0954579410000805

34.

Negriff

Susman

E. J.

(2011). Pubertal timing, depression, and externalizing problems: A framework, review, and examination of gender differences: Pubertal timing, depression, and externalizing problems. Journal of Research on Adolescence, 21(3), 717–746. https://doi.org/10.1111/j.1532-7795.2010.00708.x

35.

Park

H.-S.

Yun

Walsh

(2017). Early puberty, school context, and delinquency among South Korean girls. International Journal of Offender Therapy and Comparative Criminology, 61(7), 795–818. https://doi.org/10.1177/0306624X15611374

36.

Patton

G. C.

Viner

(2007). Pubertal transitions in health. The Lancet, 369(9567), 1130–1139.

37.

Roberts

(2015). Puberty in crisis: The sociology of early sexual development. Cambridge University Press.

38.

Schwartz

J. A.

Beaver

K. M.

(2015). A partial test of Moffitt’s developmental taxonomy: Examining the role of genetic risk. Justice Quarterly, 32(5), 768–791. https://doi.org/10.1080/07418825.2013.805798

39.

Seo

M. Y.

Kim

S.-H.

Juul

Park

M. J.

(2020). Trend of menarcheal age among Korean girls. Journal of Korean Medical Science, 35(49), Article e406. https://doi.org/10.3346/jkms.2020.35.e406

40.

Stattin

Kerr

Skoog

(2011). Early pubertal timing and girls’ problem behavior: Integrating two hypotheses. Journal of Youth and Adolescence, 40(10), 1271–1287. https://doi.org/10.1007/s10964-011-9696-9

41.

Steffensmeier

(2020). Age and crime in South Korea: Cross-national challenge to invariance thesis. Justice Quarterly, 37(3), 410–435. https://doi.org/10.1080/07418825.2018.1550208

42.

Tsai

M.-C.

Strong

Lin

C.-Y.

(2015). Effects of pubertal timing on deviant behaviors in Taiwan: A longitudinal analysis of 7th-to 12th-grade adolescents. Journal of Adolescence, 42, 87–97. https://doi.org/10.1016/j.adolescence.2015.03.016

43.

van Jaarsveld

C. H.

Fidler

J. A.

Simon

A. E.

Wardle

. (2007). Persistent impact of pubertal timing on trends in smoking, food choice, activity, and stress in adolescence. Psychosomatic Medicine, 69(8), 798–806. https://doi.org/10.1097/PSY.0b013e3181576106

44.

Williams

J. M.

Dunlop

L. C.

(1999). Pubertal timing and self-reported delinquency among male adolescents. Journal of Adolescence, 22(1), 157–171. https://doi.org/10.1006/jado.1998.0208

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.14 MB