Lifespan Development

The effects of pubertal timing on externalizing behaviors in

adolescence and early adulthood: A meta-analytic review

Laura M. Dimler *, Misaki N. Natsuaki

University of California, Riverside, CA, United States

article info

Article history:

Available online 3 October 2015

Keywords:

Meta-analysis

Pubertal timing

Externalizing

Moderator

Adolescence

abstract

Using a meta-analytic approach, this investigation examines the association between early

pubertal timing and externalizing behaviors in adolescence and early adulthood. The

ﬁndings showed that the effect size of early pubertal maturation on externalizing be-

haviors wasr¼0.180. This small, yet signiﬁcant effect size is consistent with the models of

early pubertal maturation in that early maturation is associated with higher levels of

externalizing behaviors. Using contrast analyses, we examined three potential moderators

of this association: sex, the concurrent versus long-term effect of early puberty, and types

of puberty assessments. Neither sex nor type of pubertal timing assessment moderated the

effect signiﬁcantly. However, results indicated that the effect was stronger for studies that

measured pubertal timing and externalizing behaviors concurrently rather than longitu-

dinally (i.e., examining prospective effect of pubertal timing on later externalizing be-

haviors). Theﬁndings are discussed in terms of implications for future research.

Early puberty is a risk factor for the development of adolescent psychopathologies, including externalizing behaviors such

as classroom disruptions, aggression, delinquency, and social deviancy. Although there have been many literature reviews on

this topic (e.g.,Ge&Natsuaki, 2009; Graber, 2013; Mendle&Ferrero, 2012; Mendle, Turkheimer,&Emery, 2007; Negriff&

Susman, 2011; Rudolph, 2014), no meta-analysis has been conducted to date. Meta-analysis on pubertal timing and exter-

nalizing behaviors would help further advance theﬁeld for the following reasons. First, while there are many studies that do

report a positive association between early puberty and externalizing problems, there are also many studies thatdo notﬁnd

early puberty to predict adolescent externalizing behaviors (e.g.,Carter, Caldwell, Matusko, Antonucci,&Jackson, 2011;

Obeidallah, Brennan, Brooks-Gunn,&Earls, 2004; Stattin, Kerr,&Skoog, 2011). Therefore, quantiﬁcation of the pubertal

timing effect using a systematic meta-analysis would further the literature by providing a general overview of the statistical

associations between early pubertal maturation and externalizing behaviors. Second, quantiﬁcation of the effect size can

inform applied science. It can provide clues to the question of whether targeting early maturing youths is an effective and

efﬁcient strategy for interventions and prevention of externalizing behaviors. If the effect of early pubertal maturation is

found to be robust in the meta-analysis, it is especially alarming because in the context of secular trend of puberty, the age of

pubertal onset has been steadily decreasing in both sexes in the last 25 years (Anderson&Must, 2005; Golub et al., 2008;

Sørenson et al., 2012). For these reasons, the overarching aim of this study is toﬁll this critical gap in the literature by

conducting a meta-analytic review to further our understanding of the strength of the associations between timing of

*Corresponding author.

E-mail address:[email protected](L.M. Dimler).

Contents lists available atScienceDirect

Journal of Adolescence

journal homepage:www.elsevier.com/locate/jado

http://dx.doi.org/10.1016/j.adolescence.2015.07.021

Journal of Adolescence 45 (2015) 160e17 0 pubertal maturation and externalizing behavior. We do this by analyzing effect sizes and multiple moderators on more than

36,000 adolescents.

Although previous studies have shown the association between early pubertal maturation and externalizing behaviors,

the strength of this link varies greatly among studies (Ge&Natsuaki, 2009; Ge, Natsuaki, Jin,&Biehl, 2011). In this report, we

sought to examine whether the following potential moderators contribute to the mixed results in the literature: sex of the

adolescent (Graber, Lewinsohn, Seeley,&Brooks-Gunn, 1997; Najman et al., 2009); whether the researchers assessed a

concurrent or long-term effect of pubertal timing (Graber et al., 1997); and types of pubertal timing measurement (Brooks-

Gunn, Warren, Rosso,&Gargiulo, 1987; Dorn&Biro, 2011; Shirtcliff, Dahl,&Pollak, 2009). Thus, an additional aim of this

study is to examine whether these potential moderators alter the effect size of pubertal timing on externalizing behaviors.

Early puberty and externalizing behaviors

It is well documented that adolescents who experience pubertal maturation earlier than their same-age, same-sex peers

are more likely to have negative developmental outcomes, including externalizing behaviors and afﬁliation with deviant

peers, especially in adolescence (e.g.,Caspi, Lynam, Mofﬁtt,&Silva, 1993; Felson&Haynie, 2002; Ge, Brody, Conger, Simons,&

Murry, 2002; Lynne, Graber, Nichols, Brooks-Gunn,&Botvin, 2007; Mensah et al., 2013; Stattin&Magnusson, 1990). For

instance, in a longitudinal study with more than 500 young adults, early maturers tended to have signiﬁcantly higher rates of

disruptive behavior disorders (i.e., conduct disorder and oppositional deﬁant disorder) and antisocial personality traits in

early adulthood compared with on-time maturers (Graber, Seeley, Brooks-Gunn,&Lewinsohn, 2004). More recently,Mrug

et al. (2014)echo similar results, but add that early pubertal timing, compared to on-time or late pubertal timing, is

related to a wide range of externalizing behaviors including delinquency, physical aggression, relational aggression, and

nonphysical aggression.

Several hypotheses have been put forth to explain this association between early pubertal maturation and externalizing

behaviors (seeGe&Natsuaki, 2009for a review). One possible explanation is that the hormonal changes associated with

puberty increases the risk for developing externalizing problems by heightening an adolescent's novelty-seeking behaviors

and other biological systems such as stress reactivity and brain development (Rudolph, 2014). Another hypothesis suggests

that the gap between physical and psychological (i.e., cognitive and emotional) maturities places early physical maturers at

risk for developing externalizing behaviors. This maturation disparity model asserts that early maturers' externalizing be-

haviors are reﬂections of misalignment of their slow-developing neural and cognitive development and their body's fast-

paced development, lending to a host of social and emotional demands for which the adolescent is not yet cognitively or

emotionally equipped (Ge&Natsuaki, 2009; Mofﬁtt, 1993). Another hypothesis proposes that early puberty places an

adolescent into a demanding transition that is novel, uncertain, and ambiguous, which may exacerbate pre-existing vul-

nerabilities including pre-pubertal behavioral problems (Caspi&Mofﬁtt, 1991; Giletta et al., 2015; Graber, 2013; Hamilton

et al., 2014; Rudolph, 2014). Lastly,Ge and Natsuaki (2009)outline the contextual ampliﬁcation hypothesis, which states

that an adverse situation (e.g., family conﬂict, peer challenges) ampliﬁes the effects of the rapid hormonal and biological

changes occurring in an adolescent's body, exacerbating developing psychopathologies.

Despite the theoretical elaboration of the aforementioned mechanisms, the empirical literature has not demonstrated the

consistent effects of early pubertal maturation. Some studies report signiﬁcant effect sizes that are rather small (e.g.,r¼0.09,

Burt, McGue, DeMarte, Krueger,&Iacono, 2006), while other studies have reported larger effects (e.g.,r¼0.55,Storvoll&

Wichstrøm, 2002). Therefore, because of the theoretical implications and the statistical inconsistencies in the literature, it

is important to compute the aggregated effect size across studies, and the variation in effect sizes, if any, that needs to be

explained.

Moderators of the pubertal timing effect

To explain the potential heterogeneity in effect sizes, we focused on three possible moderators: sex of the adolescent,

short- vs. long-term effects of pubertal timing, and assessments of pubertal maturation.

Sex

Sex differences in the effect of pubertal timing is discussed extensively in the domain of internalizing psychopathology;

early maturing girls tend to have higher rates of depression, anxiety, and other internalizing symptoms during adolescence

and adulthood while the evidence on adolescent boys' internalizing symptoms and pubertal maturation are mixed

(Fernandez-Castelao&Kroner-Herwig, 2014; Kaltiala-Heino, Marttunen, Rantanen,&Rimpel€

a, 2003; Mendle&Ferrero,

2012; Natsuaki, Biehl,&Ge, 2009). Stronger effects of early maturation in female internalizing symptoms are theorized that

features of female pubertal development (e.g., curvy body, breast development, menarche) can elicit social and emotional

challenges, and the stress associated with these challenges deplete yet-developed early maturing girls' coping and resources

to deal with the pressures, which in turn can contribute to depression and anxiety (Natsuaki, Samuels,&Leve, 2014). For

males, on the other hand, earlier work suggests that features associated with male puberty such as voice change, facial hair,

and growth spurt bring social advantages, such that early maturing boys can enjoy potential dominance, power, and lead-

ership in peer relationships (McCabe&Ricciardelli, 2004; Simmons&Blyth, 1987). However, recent empirical research

L.M. Dimler, M.N. Natsuaki / Journal of Adolescence 45 (2015) 160e17 0161 advises that this thinking ought to be revised because this early pubertal transition is associated with elevated levels of

internalizing psychopathology for young males (Ge, Brody, Conger,&Simons, 2006; Graber, 2013; Mendle&Ferrero, 2012).

While existing reviews note that the effect of early maturation on externalizing psychopathology also depends on sex of

the child, the literature is not as clear-cut as it is for internalizing symptoms. Although a fairly consistentﬁnding is that early

maturing boys have higher levels of externalizing behaviors than late maturing counterparts (Felson&Haynie, 2002; Mendle

&Ferrero, 2012; Susman et al., 2007), the results are rather mixed for girls. For example, multiple recent studies on females

suggest early menarche does not have an impact on externalizing behaviors in females (e.g.,Boden, Fergusson,&Horwood,

2011; Carter et al., 2011), while there seem to be just as many recent studies suggesting that entering into puberty earlier than

most peers is a risk factor for future delinquency for girls (e.g.,Mendle et al., 2007; Mrug et al., 2014). In attempt to resolve this

confusion, the present meta-analysis includes sex as a moderator of the association between early pubertal maturation and

externalizing psychopathology.

Concurrent vs. longitudinal effect of pubertal timing

Another potential moderator of the association between early pubertal timing and externalizing behaviors is the short-

versus long-term effect of early puberty. Pubertal timing is associated with externalizing behaviors concurrently, as evidenced

by a number of cross-sectional studies (e.g.,Flannery, Rowe,&Gulley, 1993; Graber et al., 1997; Sontag, Graber,&Clemans,

2011). However, we have limited knowledge regarding the long-term effect of early puberty. Although there are a compar-

atively small number of studies that have investigated the long-term effects of early pubertal timing, available evidence

suggests that the adverse effect of early maturation is rather short-lived and is contained within the bounds of adolescence

(Boden et al., 2011; Natsuaki et al., 2009; Obeidallah et al., 2004). Among studies that have examined this longitudinal effect,

there is evidence that early pubertal timing in boys, who were measured once a year for four years, consistently resulted in

more externalized hostile feeling in late adolescence than on-time or late-maturing boys (Ge, Conger,&Elder, 2001), but did

not result in major psychopathology in adulthood (Graber et al., 2004). Similarly, adult females who were formerly early

maturers in adolescence were more likely than formerly on-time maturers to have antisocial behavior and conduct behavior

disorders in early adulthood, but not after the age of 21 (Boden et al., 2011; Graber et al., 2004). As such, this quantitative

review aims to combine the results of cross-sectional as well as longitudinal studies to examine whether the concurrent vs.

longitudinal effect of puberty moderates the strength of the association between early puberty and externalizing behaviors.

Pubertal timing assessment

The third potential moderator that may explain mixedﬁndings in this area of research concerns the type of pubertal

maturation assessments used in each study. Although there are many different and nuanced ways to measure the complex

phenomenon of puberty than are outlined in this meta-analysis (e.g., hormone concentrations, bone age, and gonadal ul-

trasound), we discuss theﬁve most common forms of assessing pubertal timing that are found within the literature of

externalizing behaviors (seeDorn&Biro, 2011; Dorn, Dahl, Woodward,&Biro, 2006for comprehensive reviews on puberty

assessment). Conventional measures of puberty in psychological research include the Pubertal Development Scale (PDS;

Petersen, Crockett, Richards,&Boxer, 1988), the Tanner Stages (i.e., self-report and/or clinician-report;Marshall&Tanner,

1969, 1970), perceived pubertal timing (i.e., subjective assessment of whether one feels mature earlier than their peers;

seeMendle, 2014a), and self-reported age of menarche in females.

The PDS is a subjective, noninvasiveﬁve-item questionnaire that assesses the status of puberty-related physical devel-

opment (breast development and menarche for girls; voice change and facial hair for boys; skin change, body hair, and growth

spurt for both sexes) at the time of questionnaire administration. Adolescents, parents, and clinicians respond to a four-point

scale, indicating how far along the adolescent's physical maturation has advanced. The Tanner Stages assessment uses an

illustrated scale of primary and secondary sex characteristics in which the adolescent, parent, and/or clinician select one of

theﬁve pictures that describes the adolescent's current physical development. Research demonstrates that both the PDS and

the Tanner Stages are highly correlated with puberty-related hormones (i.e., dehydroepiandrosterone, testosterone, and

estradiol;Shirtcliff et al., 2009), indicating that the subjective measures are valid and practical when hormonal proﬁling and

other biological or objective measures of pubertal timing are unfeasible to attain. The clinician-reported Tanner Stages are

known to be more accurate when purposefully isolating an off-time puberty sample (Brooks-Gunn et al., 1987; Dorn et al.,

2006; Shirtcliff et al., 2009). To compute pubertal timing scores from the pubertal status scores based on the PDS and/or

Tanner Stages, researchers typically either 1) standardize the scores within age and sex to create a continuous measure of

pubertal timing; and/or 2) classify the sample into three groups of early, on-time, and late-maturers using the score distri-

bution generated from the given sample (peer-normative comparison).

Perceived pubertal timing is another method that assesses an adolescent's perception of pubertal timing synchrony with

peers, usually by asking adolescents if they feel their physical changes are occurring before, at the same time, or after the rest

of their peers (Mendle, 2014a). Recently, studies have indicated that adolescents' subjective assessment of their timing status

among their peers may be an important factor that uniquely contributes to externalizing behaviors (Lynne et al., 2007;

Storvoll&Wichstrøm, 2002).Mendle (2014a)calls for greater awareness of the importance of the subjective synchrony of

pubertal timing within peers as these self-perceptions of feeling‘different’from peers may result from thoughts and actions

that correspond with the adolescents' impression of their physical appearance. For example, in the context of externalizing

L.M. Dimler, M.N. Natsuaki / Journal of Adolescence 45 (2015) 160e17 0 16 2 behaviors, feeling physically‘different’from peers may accentuate the desire to be accepted by peers, resulting in adolescents

associating with older peers and engaging in“adult-like”behaviors that are not age-appropriate for them (i.e., driving a car at

age 14).

Lastly, self-reported recall of age at menarche is the most frequently used measure of female pubertal timing because of its

ease of accessibility and the fact that menarche is a clearly timed event in a female's life (Dorn&Biro, 2011). However, this

type of measurement has its limitations: menarche occurs late in the pubertal process and is not the most accurate measure of

pubertal timing in terms of reliability of self-reported recall, as the reported age of menarche has been shown to vary within

longitudinal studies (Dorn et al., 2006).

Given the heterogeneity in the assessment of pubertal timing across studies, it may be the case that the strength of

observed associations between early pubertal maturation and externalizing behaviors in previous work may depend on the

types of pubertal timing measures and the reporter of pubertal timing, but this possibility has never been systematically

examined. Previous studies have shown that the potential effect of pubertal timing measurement on an adolescent's psy-

chopathology differ by continuous and categorical forms of measurement (Negriff, Fung,&Trickett, 2008) and stage-

normative and peer-normative forms of measurement (i.e.,Cance, Ennett, Morgan-Lopez,&Foshee, 2012), thus, this meta-

analysis further expands these studies by examiningﬁve types of pubertal timing measures in the context of externalizing

psychopathology.

Present study

The goal of the present study was two-fold: (1) to conduct a meta-analysis on pubertal timing and externalizing behaviors

to quantify the magnitude of the pubertal timing effect; and (2) to explore factors that potentially moderate the link between

pubertal timing and externalizing psychopathology. Based on the previous literature outlined above, we hypothesized that:

(1) early pubertal timing would lead to more externalizing behaviors; (2) the link between pubertal timing and externalizing

behaviors would be stronger for boys than for girls; and (3) this association would be stronger when pubertal timing and

externalizing behaviors are measured concurrently, as opposed to longitudinally. We did not form speciﬁc hypotheses

regarding the pubertal timing assessments.

This meta-analytic review advances previous literature reviews in two ways. First, meta-analyses compliment theoretical

reviews by focusing on effect sizes, providing a statistical tool for describing average effect sizes, and allowing for more

powerful methods of evaluating variability ofﬁndings across studies and its sources (Rosenthal, 1991). Second, this meta-

analysis allows for testing for moderation via contrast analysis.

Methods

Selection of studies

In this study, the operationalization of externalizing behaviors was based on the Diagnostic and Statistical Manual, Fifth

Edition (DSM-5) (American Psychiatric Association, 2013). According to the DSM-5, disruptive, impulse control, and conduct

disorders include symptoms that involve problems in the self-control of emotions and behaviors. While other disorders in

DSM-5 may also involve problems in emotional and/or behavioral regulation (i.e., substance use and risky sexual behaviors),

the disorders in this cluster are unique in that these problems are manifested in behaviors that violate the rights of others (e.g.

aggression, destruction of property) and/or bring the individual into signiﬁcant conﬂict with societal norms or authority

ﬁgures (American Psychiatric Association, 2013).

Peer-reviewed studies were obtained through (a) searching computerized literature in the PsycINFO and Academic Search

Complete databases using all combinations of the keywords:adolescent,externalizing,antisocial,aggression,delinquency,

conduct disorder,conduct problems,behavior problems,behavior disorder,puberty,pubertal timing,adolescence,menarche,

spermarche, andoigarche; (b) going through the authors' personal collection of pertinent studies on puberty; (c) using the

ancestry method by examining references cited in prior reviews and empirical articles; and (d) communicating with those in

theﬁeld who have conducted research in this speciﬁc area.

Studies were included in this quantitative review if they met the following criteria: (a) they used a correlational model to

assess an association between pubertal timing and externalizing behaviors; (b) they used externalizing behaviors as an

outcome variable, predicted from pubertal timing; (c) the studies were available in peer-reviewed English language journals;

and (d) the studies reported effect sizeror enough statistical information to reconstruct this effect size (i.e., means and

standard deviations).

The following studies were not included in the meta-analysis: (a) Studies that focus exclusively on risky sexual behaviors

(n¼50); (b) studies that assessed only substance/alcohol use (n¼23); and (c) studies that reported odds ratios or relative risk

ratios (n¼3), as there is no process of which to reconstitute a reliable effect sizer(Rosenthal&Rosnow, 2008; Rosenthal,

Rosnow,&Rubin, 2000). Studies with exclusive focus on sexual behaviors and substance use were excluded because these

behaviors tend to be measured as their own constructs and have their own validity, as opposed to being nested under

externalizing behaviors (e.g.,American Psychiatric Association, 2013; Timmermans, van Lier,&Koot, 2008; Winters,

Stinchﬁeld, Latimer,&Stone, 2008). Readers who are speciﬁcally interested in the effect of pubertal timing and risky sex-

ual behaviors should refer to a recent meta-analysis byBaams, Dubas, Overbeek, and van Aken (2015).

L.M. Dimler, M.N. Natsuaki / Journal of Adolescence 45 (2015) 160e17 016 3 Because we assessed sex as a potential moderator, we extracted two effect sizes from a study if it reported separate

statistical analyses for males and females. Under these criteria, 34 publications consisting of over 36,000 adolescents

(N¼36,641) and 40 effect sizes were included in the current investigation.

Extraction of information from studies

The following information was extracted from each study: sample size, correlation (or regression coefﬁcient), sex of the

sample, the type of pubertal timing measurement, the ages of the participants when the pubertal timing and externalizing

measures were completed, and the duration of the study. It is important to note that when multiple time points were used to

assess for externalizing symptoms (e.g., pubertal timing's effects on externalizing behavior in the seventh grade, tenth grade,

and twelfth grade), the furthest time point from the pubertal timing assessment was used (e.g., in a four-year longitudinal

study, the ninth grade puberty assessment and the twelfth grade externalizing assessment would be used). This was done so

that the longitudinal effect would be as long-term as possible, within each article's conﬁnes.

Coding of moderators

This study focused on three moderators: participants' sex, concurrent vs. long-term effect of pubertal timing, and the type

of pubertal timing assessment.Sexwas dummy coded with males as the reference group. Forconcurrent vs. long-term effectof

pubertal timing, we dummy coded with concurrent¼1 and long-term¼0. Long-term was deﬁned by a longitudinal design

with at least one year between the times at which puberty and externalizing psychopathology were assessed. We identiﬁed

16 effect sizes that corresponded with longitudinal studies that met the aforementioned criterion, with an average time

interval of 4.6 years (SD¼4.5) between puberty and externalizing behavior assessments. Lastly,pubertal timing measurement

was divided intoﬁve subgroups: (a) PDS; (b) medical professional-reported Tanner Stages; (c) parent- or self-reported Tanner

Stages; (d) perceived pubertal timing; and (e) age of menarche (for female-only samples). Each type of pubertal timing

assessment was dummy coded (1¼the measure of interest, 0¼all others).

Statistical analyses

Effect size calculations

The effect sizerwas the statistical basis for this meta-analysis becauserillustrates both the strength and direction of the

association between variables (Durlak, 2009; Ozer, 1985; Rosenthal&DiMatteo, 2001; Rosenthal&Rubin, 1982). As noted

earlier, studies reporting only odds ratios or relative risk ratios (and did not report descriptive statistics on the dataset) were

omitted from the meta-analysis due to difﬁculty in the reconstituted effect size (Rosenthal&Rosnow, 2008; Rosenthal et al.,

2000), unless the corresponding author was available and willing to release the data needed to computer.

For studies reporting effect sizes other thanr, the information in the published article was computed into the desired effect

size through other information within the article such as group means and standard deviations, Hedge'sg, Cohen'sd, an exact

p-value, chi-square, or anF-test with one degree of freedom in the numerator (Rosenthal, 1991; Rosenthal&Rosnow, 2008;

Rosenthal et al., 2000). If a study did not report an exactp-value for a signiﬁcant result, the one-tailedp-value was

conservatively assumed to be 0.025. Effect sizes were transformed via Fisher'sZrtransformation, and back-transformed tor's

for intuitive interpretability. Four effect sizes from the same sample in separate papers (Negriff et al., 2008; Negriff, Ji,&

Trickett, 2011; Negriff, Susman,&Trickett, 2011; Negriff&Trickett, 2010) were ensemble adjusted into one overall effect

size (thus, using aﬁxed effects approach since all effect sizes were drawn from the same population;Borenstein, Hedges,

Higgins,&Rothstein, 2009; Rosenthal&Rubin, 1983). In such cases, we transformed eachrinto Fisher'sZ

r, averaged the

effect sizes, transformed back to anr, and gave the one averaged effect size (Achenbach, McConaughy,&Howell, 1987;

Borenstein et al., 2009; Rosenthal, 1991; Rosenthal&DiMatteo, 2001; Rosenthal&Rubin, 1983; Smith&Glass, 1977; see

Appendix A). By doing so, we avoided violations of independence assumptions made when testing signiﬁcance (Card, 2010;

Rosenthal, 1991). It is important to note that when a publication separately reported male and female results, we reported two

separate effect sizes from the study.

Combination statistics

For each analysis, the unweighted mean and median effect size were computed, along with conﬁdence intervals. Using the

unweighted means and medians allows us to have a fuller picture of the data and the psychological phenomenon at hand. The

random effects model was used to compute and combine effect size statistics, using the unweighted meanrbased onk(the

total number of effect sizes included). This model allows the results to be generalized beyond the samples included in the

meta-analysis (Borenstein et al., 2009). In order to address theﬁle-drawer problem and the potential number of new, un-

published, or otherwise unretrieved studies that would need to show no effect at thep<0.05 level in order to negate our

results, we computed the tolerance level (N¼210) and fail-safeN(N¼3053.47) values (Rosenthal&Rosnow, 2008). All

statistical analyses were conducted using SPSS 22.0 and Excel Professional Plus 2013.

Moderator analyses

A chi square test for heterogeneity wasﬁrst computed to assess for heterogeneous results within the sample (Hunter&

Schmidt, 2000; Rosenthal, 1991). The chi-square statistic was used because the Type I error rate is lowest when ther-

L.M. Dimler, M.N. Natsuaki / Journal of Adolescence 45 (2015) 160e17 0 16 4 transformed-to-Fisher's-Zstatistic is used in conjunction with the chi-square test for heterogeneity (S

anchez-Meca&Marín-

Martínez, 1997). The moderators were tested in random effects models in order to assess signiﬁcant differences between

effect sizes as a function of potential moderator variables at the most generalizable level. The random effects (unweighted)

approach to analyzing moderators allows for generalization to studies that are not identical to the study sample, but are

others of the same ilk (Hedges&Vevea, 1998;Hunter&Schmidt, 2000;Rosenthal&DiMatteo, 2001). Following the guideline

by other meta-analytic studies (e.g.,Card, 2010; Connell&Goodman, 2002) that moderator analyses with less thanﬁve effect

sizes in each category are considered to be less reliable, we only ran moderator analyses with groups that containedﬁve or

more effect sizes. For moderator analyses, we computed independent samplet-tests and reportedrfor each group for

respective comparisons.

Results

In total, 40 effect sizes (34 publications) were examined in this investigation. The sample sizes for these effects ranged

from 52 to 5700, with a median sample size of 345, mean sample size of 916, and a total sample size of 36,641 adolescents.

Thirteen effect sizes did not support the association between early pubertal timing and externalizing behaviors. Fourteen

effect sizes were for males, 24 for females, and two examined both sexes without reporting separate results for each sex in the

original publication. Sixteen effect sizes measured the effect of early pubertal timing longitudinally. Eleven of the 40 effect

sizes used the PDS as the pubertal timing measure,ﬁve used clinician-report Tanner Stages, six used parent- or self-report

Tanner Stages, seven used a perceived pubertal timing assessment, and 12 effect sizes used age of menarche as the puber-

tal timing measure. For a more detailed and comprehensive summary of the included studies (with effect sizer, the signif-

icance, conclusion, and moderators for each study), please seeAppendix A.

Across the 40 effect sizes, the median correlation wasr¼0.123, with an average correlation ofr¼0.180 (95% CI: 0.109,

0.250;t(39)¼140.81,p<0.0001). These results suggest that early maturing adolescents tend to engage in more externalizing

behaviors than their on-time or late-time maturing peers. Theseﬁndings would be disputed only if there were over 3053 (i.e.,

fail-safeN) studies showing that there is in fact no association between early pubertal timing and higher levels of exter-

nalizing behaviors; this well exceeds the tolerance level ofN¼210.

The chi square test for heterogeneity was not signiﬁcant at the conventionalp<0.05 levelðc

ð39Þ ¼47:023Þ. Although the

chi-square test is not signiﬁcant by the conventional alpha level of 0.05, it has been shown that it is not necessary to have

signiﬁcantly heterogeneous results (Borenstein et al., 2009; Hall&Rosenthal, 1991) and is still important to examine for

potential moderators (Borenstein et al., 2009; Rosenthal&DiMatteo, 2001).

Results from a series oft-tests for moderation analysis are shown inTable 1. Results from these random effects analyses

showed that there was only one signiﬁcant moderator (concurrent vs. long-term effect of pubertal timing). The concurrent

effect (r¼0.226) of pubertal maturation on externalizing behaviors was found to be stronger than its longitudinal effect

(r¼0.100) (t

(38) ¼2.366,p¼0.023). Sex did not moderate the effect of early maturation on externalizing behaviors

(36) ¼0.991,p¼0.328), indicating that there was no gender difference in the effect of early pubertal maturation on

externalizing behaviors. Early puberty effect on externalizing behaviors did not signiﬁcantly differ across theﬁve types of

assessment tools.

Table 1

Moderator analyses.

Moderating variable Categories No. of ES in each grouprtsigniﬁcance test (t

(df),p,r)

Sex a Male 14 0.250t (36) ¼0.991,p¼0.328,r¼0.156

Female 24 0.148

Timing effect Concurrent 24 0.226t

(38) ¼2.366,p¼0.023,r¼0.359*

Long-term 16 0.100

PDS PDS 11 0.241t

(38) ¼ 1.394,p¼0.171,r¼0.221

Other 29 0.154

Clinician-report tanner stages Clinician tanner 5 0.322t

(38) ¼0.615,p¼0.542,r¼0.101

Other 35 0.166

Self-/parent-report tanner stages Self/parent tanner 6 0.105t

(38) ¼0.579,p¼0.566,r¼0.094

Other 34 0.192

Perceived timing Perceived timing 7 0.208t

(38) ¼ 0.607,p¼0.548,r¼0.098

Other 33 0.172

Age of menarche

b Age of menarche 12 0.102t (22) ¼0.988,p¼0.334,r¼0.206

Other 12 0.212

Note: PDS¼pubertal development scale.

*p<0.05.

aTwo effect sizes did not examine sex differences.b24 studies examined females.

L.M. Dimler, M.N. Natsuaki / Journal of Adolescence 45 (2015) 160e17 016 5 Discussion

The aims of the present study were (1) to quantify the magnitude of early pubertal timing effect on externalizing be-

haviors, and (2) to explore potential variables that may moderate this association. Results conﬁrm our hypothesis that early

pubertal timing is a signiﬁcant predictor of adolescent externalizing behaviors at the effect size of 0.180, which explains

3.24% of the variance in externalizing behaviors. Although this effect size is considered small in accordance withCohen's

(1988)guidelines, it would be hasty to conclude that small to moderate effect sizes are meaningless (Rosenthal&

Rosnow, 2008). In addition, any single predictor of highly complex psychological phenomena, such as externalizing

behaviors, is necessarily small, given that multiple predictors work in a synergetic manner (Ahadi&Deiner, 1989; Durlak,

2009).

Our moderator analysis indicated that sex did not moderate the effect of early maturation on externalizing behaviors,

suggesting that the adverse effect of early pubertal maturation on externalizing behaviors does not differ by sex of the child.

This null result may be due to the fact that there is different developmental timing of puberty for boys versus girls and/or

the different methods used to measure puberty for each sex. We may also see this result because externalizing behaviors

rise for both sexes throughout adolescence, which results in a narrower, and statistically insigniﬁcant, sex gap in exter-

nalizing behaviors than at any other point in development (Ge et al., 2006; Mendle, 2014b; Najman et al., 2009). However, it

is important to note that, as with any nullﬁndings, this interpretation must be treated with caution because a statistical null

ﬁnding is not conclusive evidence for the absence of an effect (Natsuaki, 2015; Nickerson, 2000). Replication of thisﬁnding

is needed in future research for further corroboration. Future longitudinal studies ought to also test whether adolescent

externalizing psychopathology differs by sex in relation to the physical and social developmental differences between the

sexes.

Finally, in support of our second hypothesis on concurrent vs. longitudinal effect, the effect of early pubertal timing was

more strongly related to externalizing behaviors while adolescents go through puberty, as opposed to after they complete

pubertal transition. Thisﬁnding coincides with the current literature, indicating that early pubertal timing has an effect on

adolescent externalizing behaviors only during adolescence (Obeidallah et al., 2004). This lends credence to the idea that

there is relatively little continuity in externalizing behaviors from adolescence into early adulthood and once early adulthood

is reached and puberty is complete for all normally-developing individuals, any differences in behaviors due to timing of

pubertal entrance dissipate over time (Natsuaki et al., 2009; Obeidallah et al., 2004).

The present study is one of theﬁrst to directly contrastﬁve types of puberty assessments in a meta-analytic framework

(see alsoCance et al., 2012). We did notﬁnd evidence of moderation by any speciﬁc type of pubertal timing measure.Cance

et al. (2012)reported that different types of pubertal timing assessment may tap onto different facets of pubertal develop-

ment. In their empirical study, they found that adolescents who perceived themselves maturing earlier than peers when using

the one-item perceived pubertal timing scale were not necessarily identiﬁed as early maturers when assessed by the PDS. The

current meta-analysis expands their results by suggesting that although different measures may describe different processes

of pubertal maturation, these differences do not seem to qualify the link between pubertal timing and externalizing be-

haviors. However, readers are reminded that the observed null effect may be due to the small number of studies in each

category of puberty assessments, generating too low of power to properly detect an effect. Future studies should revisit this

moderation when enough studies have accumulated.

Limitations and future directions

Theﬁndings reported here need to be interpreted within the context of the study limitations. First, although multiple

literature search strategies were employed in this study, as with any meta-analysis, it is quite possible that some empirical

articles were not identiﬁed and excluded unintentionally. For instance, theﬁle drawer problem is well known, where studies

with statistically signiﬁcantﬁndings may have had greater likelihood of publication than those with nullﬁndings, and

therefore, the effect size can be overestimated. Despite this potential drawback, the large fail-safeN(the number of

nonsigniﬁcant studies that would be necessary to reduce the effect size to a nonsigniﬁcant value) in this report made it highly

unlikely that any missed or unpublished studies would dramatically change the current results.

Second, it is possible that the deﬁnition and operationalization of externalizing behaviors may have been diverse among

the included studies. In this current study, we attempted to conduct a comprehensive search by including studies using such

terms such astheft,conduct problems,delinquency,behavior problems, andantisocial tendencies. However, each construct may

have tapped into unique components of broadband externalizing behaviors. Although this study was not able to test the types

of externalizing behaviors as a moderator due to small sample size for each category, it is imperative that future researchers

conduct a larger meta-analysis and separately analyze by types of externalizing behaviors.

Third, we also wish to remind readers that a small number of studies included in some of the pubertal timing mea-

surement groups (e.g., Clinician-reported Tanner exam,n¼5) may have resulted in statistical challenges inﬁnding effects.

Additionally, as noted earlier, measurement and computation/classiﬁcation of pubertal timing used in theﬁeld is quite

diverse and there are only few studies that use the exact same measure and classiﬁcation system. To test the types of pubertal

timing measures as moderators more fully, more studies in each measurement category are needed. Also, future investigators

should carefully describe how they measure and operationalize pubertal timing in order to aid future meta-analytic work on

this topic.

L.M. Dimler, M.N. Natsuaki / Journal of Adolescence 45 (2015) 160e17 0 16 6 Fourth, we were unable to include moderation by race because there were so few studies (n¼3) that reported necessary

statistics (e.g.,ror means and standard deviations) by racial/ethnic groups. Given that the pubertal timing effect may operate

differently depending on ethnic/cultural background (Anderson, Dallal,&Must, 2003; White, Deardoff, Liu,&Gonzales,

2013), future investigators should speciﬁcally examine this potential interaction effect or report their analyses separately

by race. Fifth, readers are reminded that there are more than theﬁve types of measurements tested here that are used to

measure and analyze pubertal timing (e.g., hormonal proﬁling, bone density;Dorn&Biro, 2011; Dorn et al., 2006).

Furthermore, even within each puberty measure, pubertal timing scores can be analyzed as a continuous or categorical

variable. It can also be computed by using a cut-off based on national norms (e.g., national average of menarche), by relying on

the distribution of the puberty variable within the study sample (e.g., identifying early maturers as the top 30% of the study

sample in the PDS score), by standardizing by age and sex within the sample, or by using a residualized puberty status score

after controlling for age. Unfortunately, this study was not equipped to address these issues due to low power, but this

nuanced approach to pubertal timing measures is a topic that deserves future attention.

Aﬁnal limitation is inherent in meta-analysis computations: we were not able to include articles that reported odds-ratios

or risk-ratios as effect size indicators. As suggested byRosenthal et al. (2000), future research should include effect size

indicators such as those inranddeven when also reporting odds-ratios or risk-ratios, so that future meta-analyses can

replicate theseﬁndings and are able to include studies that may have otherwise been excluded for computational reasons.

Despite these limitations, theﬁndings from this meta-analysis should be considered in terms of the overall effect that early

pubertal timing has on an adolescent. Although puberty is a developmental event that is universal across healthy adolescents,

the timing of its occurrence appears to hold different consequence and meaning for individuals (Natsuaki, 2013). Identifying

the risk factors and behaviors that may be inﬂuenced by adolescents' pubertal timing is of both theoretical and clinical in-

terest. By attempting to understand this complex biological and social process, researchers and clinicians can further

delineate the etiology of externalizing behaviors in adolescents.

Acknowledgments

We would like to extend our gratefulness to Dr. Robert Rosenthal for his continual statistical guidance, advice, and un-

matched expertise in meta-analysis. We would also like to thank Tricia Miller and Danielle Samuels for their feedback and

valuable help on this meta-analysis.

Appendix A. Summary of studies that tested for an association between early pubertal timing and externalizing

behaviors.

StudyNr pConclusion aSex bPuberty

measurementAge at puberty

measurementAge at ext.

measurement cTiming

effect

Boden et al. (2011)75 0 0.5 EjNE 1 Age of menarche 13.5 30 Long-term

Burt et al. (2006)688 0.090 0.009 E>NE 1 Age of menarche 14.8 14.8 Long-term

Carter, Jaccard, Silverman,

and Pina (2009)102 0.174 0.04 EjNE 1 Age of menarche 14.7 14.7 Concurrent

Carter et al. (2011)607 0.040 0.163 EjNE 1 Age of menarche 15.2 15.2 Concurrent

Caspi et al. (1993)297 0.168 0.003 EjNE 1 Age of menarche 13 15 Long-term

Copeland et al. (2010)140 0.080 0.17 EjNE 1 Self/parent tanner 16 21 Long-term

Cota-Robles, Neiss, and Rowe (2002)5550 0.120 0.001 E>NE 0 PDS 14 14 Concurrent

Deardorff et al. (2013)263 0.020 0.3732 EjNE 1 PDS 10.5 16 Long-term

Dorn, Susman, and Ponirakis (2003)56 0.279 0.0217 E>NE 0 Clinician tanner 12 13.7 Long-term

Dorn et al. (2003)52 0.226 0.0426 EjNE 1 Clinician tanner 12.7 13 Long-term

Felson and Haynie (2002)5700 0.861 0.00005 E>NE 0 PDS 14.1 14.1 Concurrent

Flannery et al. (1993)(1) 376 0.117 0.011 E>NE 0 Self/parent tanner 13.5 13.5 Concurrent

Flannery et al. (1993)(2) 397 0.209 0.00005 E>NE 1 Self/parent tanner 13.5 13.5 Concurrent

Ge et al. (2002)(1) 379 0.193 0.0001 E>NE 0 PDS 10.5 10.5 Concurrent

Ge et al. (2002)(2) 441 0.163 0.0003 E>NE 1 PDS 10.5 10.5 Concurrent

Ge et al. (2001)170 0.206 0.0031 E>NE 0 PDS 12.7 15.7 Long-term

Graber et al. (1997)(1) 667 0.006 0.4345 EjNE 0 Perceived timing 16.6 16.6 Concurrent

Graber et al. (1997)(2) 801 0.120 0.0004 E>NE 1 Perceived timing 16.6 16.6 Concurrent

Haynie (2003)5477 0.063 0.00005 E>NE 1 PDS 15.1 15.1 Concurrent

Lynne et al. (2007)1366 0.081 0.0068 E>NE 2 Perceived timing 11.7 11.7 Concurrent

Mensah et al. (2013)(1) 1773 0.270 0.001 E>NE 0 PDS 8.5 10.5 Long-term

Mensah et al. (2013)(2) 1718 0.050 0.02 E>NE 1 PDS 8.5 10.5 Long-term

Mrug et al. (2008)330 0.126 0.01 E>NE 1 Age of menarche 11.3 11.3 Concurrent

Mrug et al. (2014)2607 0.053 0.0035 E>NE 1 Age of menarche 11.1 16.1 Long-term

Najman et al. (2009)

(1) 1359 0.005 0.4313 EjNE 0 Self/parent tanner 14 20.6 Long-term

Najman et al. (2009)(2) 1425 0.060 0.0113 E>NE 1 Age of menarche 14 20.6 Long-term

(continued on next page)

L.M. Dimler, M.N. Natsuaki / Journal of Adolescence 45 (2015) 160e17 016 7 References 1

Achenbach, T. M., McConaughy, S. H., & Howell, C. T. (1987). Child/adolescent behavioral and emotional problems: implications of cross-informant cor-

relations for situational speciﬁcity.Psychological Bulletin, 101(2), 213e232.

Ahadi, S., & Deiner, E. (1989). Multiple determinants of effect size.Personality and Social Psychology, 56(3), 398e406.

American Psychiatric Association. (2013).Diagnostic and statistical manual of mental disorders(5th ed.). Washington, DC: American Psychiatric Association.

Anderson, S. E., Dallal, G. E., & Must, A. (2003). Relative weight and race inﬂuence average age at menarche: results from two nationally representative

surveys of US girls studied 25 years apart.Journal of Pediatrics, 111(4), 844e850.

Anderson, S. E., & Must, A. (2005). Interpreting the continued decline in the average age at menarche: results from two nationally representative surveys of

US girls studied 10 years apart.Journal of Pediatrics, 147(6), 753e760.http://dx.doi.org/10.1016/j.jpeds.2005.07.016.

Baams, L., Dubas, J. S., Overbeek, G., & van Aken, M. A. G. (2015). Transitions in body and behavior: a meta-analytic study on the relationship between

pubertal development and adolescent sexual behavior.Journal of Adolescent Health.http://dx.doi.org/10.1016/j.jadohealth.2014.11.019.

*Boden, J. M., Fergusson, D. M., & Horwood, L. J. (2011). Age of menarche and psychosocial outcomes in a New Zealand birth cohort.Journal of the American

Academy of Child and Adolescent Psychiatry, 50(2), 132e140.http://dx.doi.org/10.1016/j.jaac.2010.11.007. e135.

Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. (2009).Introduction to meta-analysis. West Sussex, UK: John Wiley&Sons, Ltd.

Brooks-Gunn, J., Warren, M. P., Rosso, J., & Gargiulo, J. (1987). Validity of self-report measures of girls' pubertal status.Child Development, 58(3), 829e841.

*Burt, S. A., McGue, M., DeMarte, J. A., Krueger, R. F., & Iacono, W. G. (2006). Timing of menarche and the origins of conduct disorder.Archives of General

Psychiatry, 63(8), 890e896.

Cance, J. D., Ennett, S. T., Morgan-Lopez, A. A., & Foshee, V. A. (2012). The stability of perceived pubertal timing across adolescence.Journal of Youth and

Adolescence, 41(6), 764e775.http://dx.doi.org/10.1007/s10964-011-9720-0.

Card, N. (2010). Antipathetic relationships in child and adolescent development: a meta-analytic review and recommendations for an emerging area of

study.Developmental Psychology, 46(2), 516e529.

*Carter, R., Caldwell, C. H., Matusko, N., Antonucci, T., & Jackson, J. S. (2011). Ethnicity, perceived pubertal timing, externalizing behaviors, and depressive

symptoms among black adolescent girls.Journal of Youth and Adolescence, 40(10), 1394e14 0 6 .http://dx.doi.org/10.1007/s10964-010-9611-9.

*Carter, R., Jaccard, J., Silverman, W. K., & Pina, A. A. (2009). Pubertal timing and its link to behavioral and emotional problems among‘at-risk’African

American adolescent girls.

Journal of Adolescence, 32(3), 467e481.http://dx.doi.org/10.1016/j.adolescence.2008.07.005.

*Caspi, A., Lynam, D., Mofﬁtt, T. E., & Silva, P. A. (1993). Unraveling girls' delinquency: biological, dispositional, and contextual contributions to adolescent

misbehavior.Developmental Psychology, 29(1), 19e30.

Caspi, A., & Mofﬁtt, T. E. (1991). Individual differences are accentuated during periods of social change: the sample case of girls at puberty.Journal of

Personality and Social Psychology, 61(1), 157e16 8.

Cohen, J. (1988).Statistical power for the social sciences. Hillsdale, NJ: Laurence Erlbaum and Associates.

Connell, A. M., & Goodman, S. H. (2002). The association between psychopathology in fathers versus mothers and children's internalizing and externalizing

behavior problems: a meta-analysis.Psychological Bulletin, 128(5), 746e773.http://dx.doi.org/10.1037//0033-2909.128.5.746.

*Copeland, W., Shanahan, L., Miller, S., Costello, E. J., Angold, A., & Maughan, B. (2010). Outcomes of early pubertal timing in young women: a prospective

population-based study.American Journal of Psychiatry, 167(10), 1218e1225.

*Cota-Robles, S., Neiss, M., & Rowe, D. C. (2002). The role of puberty in violent and nonviolent delinquency among Anglo American, Mexican American, and

African American boys.Journal of Adolescent Research, 17(4), 364e376.

*Deardorff, J., Cham, H., Gonzales, N. A., White, R. M., Tein, J. Y., Wong, J. J., et al. (2013). Pubertal timing and Mexican-origin girls' internalizing and

externalizing symptoms: the inﬂuence of harsh parenting.Developmental Psychology, 49(9), 1790e1804.http://dx.doi.org/10.1037/a0031016.

Dorn, L. D., & Biro, F. M. (2011). Puberty and its measurement: a decade in review.Journal of Research on Adolescence, 21(1), 180e195.http://dx.doi.org/10.

1111/j.1532-7795.2010.00722.x.

Dorn, L. D., Dahl, R. E., Woodward, H. R., & Biro, F. M. (2006). Deﬁning the boundaries of early adolescence: a user's guide to assessing pubertal status and

pubertal timing in research with adolescents.Applied Developmental Science, 10(1), 30e56.

*Dorn, L. D., Susman, E. J., & Ponirakis, A. (2003). Pubertal timing and adolescent adjustment and behavior: conclusions vary by rater.Journal of Youth and

Adolescence, 32(3), 157e16 7.

(continued)

StudyNr pConclusion

aSex bPuberty

measurementAge at puberty

measurementAge at ext.

measurement cTiming

effect

Negriff et al., ensemble adjusted

d 361 0.090 0.087 EjNE 2 Self/parent tanner 10.9 12.9 Long-term

Obeidallah et al. (2004)121 0.174 0.02675 EjNE 1 Age of menarche 13.5 15.6 Long-term

Skoog, Stattin, Ruiselova,

and Ozdemir (2013)1597 0.135 0.001 E>NE 1 Age of menarche 14.5 14.5 Concurrent

Sontag-Padilla et al. (2012)76 0.287 0.0048 E>NE 1 Clinician tanner 7.5 7.5 Concurrent

Sontag et al. (2011)(1) 98 0.030 0.385 EjNE 0 Perceived timing 12.9 12.9 Concurrent

Sontag et al. (2011)(2) 166 0.220 0.0018 E>NE 1 Perceived timing 12.9 12.9 Concurrent

Stattin et al. (2011)284 0.070 0.12 EjNE 1 Age of menarche 14.5 14.5 Concurrent

Stattin and Magnusson (1990)466 0.135 0.002 E>NE 1 Age of menarche 14.5 14.5 Concurrent

Storvoll and Wichstrøm (2002)(1) 67 0.426 0.0003 E>NE 0 Perceived timing 16 16 Concurrent

Storvoll and Wichstrøm (2002)(2) 67 0.551 0.00005 E>NE 1 Perceived timing 16 16 Concurrent

Susman et al. (2007)(1) 56 0.340 0.0073 E>NE 0 Clinician tanner 11.4 11.4 Concurrent

Susman et al. (2007)(2) 55 0.340 0.00785 E>NE 1 Clinician tanner 10.5 10.5 Concurrent

White et al. (2013)312 0.120 0.01705 E>NE 0 PDS 12.8 15.9 Long-term

Williams and Dunlop (1999)99 0.185 0.0319 E>NE 0 PDS 14.2 14.2 Concurrent

aE¼early pubertal timing; NE¼not early pubertal timing.b0¼Male; 1¼Female; 2¼Both sexes (original author did not separate analyses for the sexes).cAge of participant when responding to an externalizing behavior measure.dThe ensemble adjusted effect size was created from the following studies:Negriff et al. (2008), Negriff, Ji, and Trickett (2011), Negriff, Susman, and

Trickett (2011), Negriff and Trickett (2010).

1References marked with an asterisk indicate studies included in the meta-analysis.

L.M. Dimler, M.N. Natsuaki / Journal of Adolescence 45 (2015) 160e17 0 16 8 Durlak, J. A. (2009). How to select, calculate, and interpret effect sizes.Journal of Pediatric Psychology, 34(9), 917e928.http://dx.doi.org/10.1093/jpepsy/

jsp004.

*Felson, R. B., & Haynie, D. L. (2002). Pubertal development, social factors, and delinquency among adolescent boys.Criminology, 40(4), 967e988.

Fernandez-Castelao, C., & Kroner-Herwig, B. (2014). Developmental trajectories and predictors of externalizing behavior: a comparison of girls and boys.

Journal of Youth and Adolescence, 43(5), 775e789.http://dx.doi.org/10.1007/s10964-013-0011-9.

*Flannery, D. J., Rowe, D. C., & Gulley, B. L. (1993). Impact of pubertal status, timing, and age on adolescent sexual experience and delinquency.Journal of

Adolescent Research, 8(1), 21e40.

Ge, X., Brody, G. H., Conger, R. D., & Simons, R. L. (2006). Pubertal maturation and African American children's internalizing and externalizing symptoms.

Journal of Youth and Adolescence, 35(4), 528e537.http://dx.doi.org/10.1007/s10964-006-9046-5.

*Ge, X., Brody, G. H., Conger, R. D., Simons, R. L., & Murry, V. M. (2002). Contextual ampliﬁcation of pubertal transition effects on deviant peer afﬁliation and

externalizing behavior among African American children.Developmental Psychology, 38(1), 42e54.http://dx.doi.org/10.1037//0012-1649.38.1.42.

*Ge, X., Conger, R. D., & Elder, G. H., Jr. (2001). The relation between puberty and psychological distress in adolescent boys.Journal of Research on Adolescence,

11(1), 49e70.

Ge, X., & Natsuaki, M. N. (2009). In search of explanations for early pubertal timing effects on developmental psychopathology.Current Directions in

Psychological Science, 18(6), 327e331.

Ge, X., Natsuaki, M. N., Jin, R., & Biehl, M. C. (2011). A contextual ampliﬁcation hypothesis: pubertal timing and girls' emotional and behavioral problems. In

M. Kerr, H. Stattin, R. C. Engels, G. Overbeek, & A.-K. Andershed (Eds.),Understanding girls' problem behavior: how girls' delinquency develops in the context

of maturity and health, co-occurring problems, and relationships(Vol. 7). John Wiley&Sons.

Giletta, M., Calhoun, C. D., Hastings, P. D., Rudolph, K. D., Nock, M. K., & Prinstein, M. J. (2015). Multi-level risk factors for suicidal ideation among at-risk

adolescent females: the role of hypothalamic-pituitary-adrenal axis responses to stress.Journal of Abnormal Child Psychology, 43(5), 807e820.http://dx.

doi.org/10.1007/s10802-014-9897-2.

Golub, M. S., Collman, G. W., Foster, P. M. D., Kimmel, C. A., Rajpert-De Meyts, E., Reiter, E. O., et al. (2008). Public health implications of altered puberty

timing.Pediatrics, 121,S218eS230.

Graber, J. A. (2013). Pubertal timing and the development of psychopathology in adolescence and beyond.Hormones and Behavior, 64(2), 262e269.http://

dx.doi.org/10.1016/j.yhbeh.2013.04.003.

*Graber, J. A., Lewinsohn, P. M., Seeley, J. R., & Brooks-Gunn, J. (1997). Is psychopathology associated with the timing of pubertal development?Journal of the

American Academy of Child and Adolescent Psychiatry, 36(12), 1768e17 76.

Graber, J. A., Seeley, J. R., Brooks-Gunn, J., & Lewinsohn, P. M. (2004). Is pubertal timing associated with psychopathology in young adulthood?Journal of the

American Academy of Child and Adolescent Psychiatry, 43(6), 718e726.http://dx.doi.org/10.1097/01.chi.0000120022.14101.11.

Hall, J. A., & Rosenthal, R. (1991). Testing for moderator variables in meta-analysis: Issues and methods.Communication Monographs, 58(4), 437e448.

Hamilton, J. L., Stange, J. P., Kleiman, E. M., Hamlat, E. J., Abramson, L. Y., & Alloy, L. B. (2014). Cognitive vulnerabilities amplify the effect of early pubertal

timing on interpersonal stress generation during adolescence.Journal of Youth and Adolescence, 43(5), 824e833.http://dx.doi.org/10.1007/s10964-013-

0015-5.

*Haynie, D. L. (2003). Contexts of risk? Explaining the link between girls' pubertal development and their delinquency involvement.Social Forces, 82(1),

355e397.

Hedges, L. V., & Vevea, J. L. (1998). Fixed- and random-effects models in meta-analysis.Psychological Methods, 3(4), 486e504.

Hunter, J. E., & Schmidt, F. L. (2000). Fixed effects vs. random effects meta-analysis models: Implications for cumulative research knowledge.International

Journal of Selection and Assessment, 8(4), 275e292.

Kaltiala-Heino, R., Marttunen, M., Rantanen, P., & Rimpel€

a, M. (2003). Early puberty is associated with mental health problems in middle adolescence.Social

Science&Medicine, 57(6), 1055e10 6 4 .http://dx.doi.org/10.1016/s0277-9536(02)00480-x.

*Lynne, S. D., Graber, J. A., Nichols, T. R., Brooks-Gunn, J., & Botvin, G. J. (2007). Links between pubertal timing, peer inﬂuences, and externalizing behaviors

among urban students followed through middle school.Journal of Adolescent Health, 40(2), 181.e7e181.e13.http://dx.doi.org/10.1016/j.jadohealth.2006.

09.008.

Marshall, W. A., & Tanner, J. M. (1969). Variations in the pattern of pubertal changes in girls.Archives of Disease in Childhood, 44(235), 291e303.

Marshall, W. A., & Tanner, J. M. (1970). Variations in the pattern of pubertal changes in boys.Archives of Disease in Childhood, 45(13), 13e23.

McCabe, M. P., & Ricciardelli, L. A. (2004). A longitudinal study of pubertal timing and extreme body change behaviors among adolescent boys and girls.

Adolescence, 39(153), 144e16 6.

Mendle, J. (2014a). Beyond pubertal timing: new directions for studying individual differences in development.Current Directions in Psychological Science,

23(3), 215e219.http://dx.doi.org/10.1177/0963721414530144.

Mendle, J. (2014b). Why puberty matters for psychopathology.Child Development Perspectives, 8

(4), 218e222.http://dx.doi.org/10.1111/cdep.12092.

Mendle, J., & Ferrero, J. (2012). Detrimental psychological outcomes associated with pubertal timing in adolescent boys.Developmental Review, 32(1),

49e66.http://dx.doi.org/10.1016/j.dr.2011.11.001.

Mendle, J., Turkheimer, E., & Emery, R. E. (2007). Detrimental psychological outcomes associated with early pubertal timing in adolescent girls.Develop-

mental Review, 27(2), 151e17 1.http://dx.doi.org/10.1016/j.dr.2006.11.001.

*Mensah, F. K., Bayer, J. K., Wake, M., Carlin, J. B., Allen, N. B., & Patton, G. C. (2013). Early puberty and childhood social and behavioral adjustment.Journal of

Adolescent Health, 53,118e124.http://dx.doi.org/10.1016/j.jadohealth2012.12.018.

Mofﬁtt, T. E. (1993). Adolescent-limited and life-course-persistent antisocial behavior: a developmental taxonomy.Psychological Review, 100(4), 674e701.

*Mrug, S., Elliot, M., Gilliland, J., Grunbaum, J. A., Tortolero, S. R., Cuccaro, P., et al. (2008). Positive parenting and early puberty in girls: protective effects

against aggressive behavior.Archives of Pediatric&Adolescent Medicine, 162(8), 781e786.

*Mrug, S., Elliott, M. N., Davies, S., Tortolero, S. R., Cuccaro, P., & Schuster, M. A. (2014). Early puberty, negative peer inﬂuence, and problem behaviors in

adolescent girls.Pediatrics, 133(1), 7e14 .http://dx.doi.org/10.1542/peds.2013-0628.

*Najman, J. M., Hayatbakhsh, M. R., McGee, T. R., Bor, W., O'Callaghan, M. J., & Williams, G. M. (2009). The impact of puberty on aggression/delinquency:

adolescence to young adulthood.The Australian and New Zealand Journal of Criminology, 42(3), 369e386.

Natsuaki, M. N. (2013). Puberty in context: toward a more nuanced understanding of early maturation.Journal of Adolescent Health, 53,677e678.http://dx.

doi.org/10.1016/j.jadohealth.2013.09.019.

Natsuaki, M. N. (2015). Intervention effect fades over time: when, how, and for whom?The Lancet Psychiatry.http://dx.doi.org/10.1016/S2215-0366(15)

00158-3.

Natsuaki, M. N., Biehl, M. C., & Ge, X. (2009). Trajectories of depressed mood from early adolescence to young adulthood: the effects of pubertal timing and

adolescent dating.Journal of Research on Adolescence, 19,47e74.

Natsuaki, M. N., Samuels, D., & Leve, L. D. (2014). Puberty, identity, and context: a biopsychosocial perspective on internalizing psychopathologyin early

adolescent girls. In K. C. McLean, & M. Syed (Eds.),The Oxford handbook of identity development(pp. 389e405). New York, NY: Oxford University Press.

*Negriff, S., Fung, M. T., & Trickett, P. K. (2008). Self-rated pubertal development, depressive symptoms and delinquency: measurement issues and

moderation by gender and maltreatment.Journal of Youth and Adolescence, 37(6), 736e

746.http://dx.doi.org/10.1007/s10964-008-9274-y.

*Negriff, S., Ji, J., & 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, 293e304.

Negriff, S., & Susman, E. J. (2011). Pubertal timing, depression, and externalizing problems: a framework, review, and examination of gender differences.

Journal of Research on Adolescence, 21(3), 717e746.http://dx.doi.org/10.1111/j.1532-7795.2010.00708.x.

*Negriff, S., Susman, E. J., & Trickett, P. K. (2011). The developmental pathway from pubertal timing to delinquency and sexual activity from early tolate

adolescence.Journal of Youth and Adolescence, 40, 1343e1356.

L.M. Dimler, M.N. Natsuaki / Journal of Adolescence 45 (2015) 160e17 016 9 *Negriff, S., & Trickett, P. K. (2010). The relationship between pubertal timing and delinquent behavior in maltreated male and female adolescents.Journal of

Early Adolescence, 30(4), 518e542.http://dx.doi.org/10.1177/0272431609338180.

Nickerson, R. S. (2000). Null hypothesis signiﬁcance testing: a review of an old and continuing controversy.Psychological Methods, 5(2), 241e301.

*Obeidallah, D., Brennan, R. T., Brooks-Gunn, J., & Earls, F. (2004). Links between pubertal timing and neighborhood contexts: implications for girls' violent

behavior.Journal of the American Academy of Child and Adolescent Psychiatry, 43(12), 1460e1468.http://dx.doi.org/10.1097/01.chi.0000142667.52062.1e.

Ozer, D. J. (1985). Correlation and the coefﬁcient of determination.Psychological Bulletin, 97(2), 307e315.http://dx.doi.org/10.1037/0033-2909.97.2.307.

Petersen, A. C., Crockett, L. J., Richards, M., & Boxer, A. (1988). A self-report measure of pubertal status: reliability, validity, and initial norms.Journal of Youth

and Adolescence, 17(2), 117e13 3.

Rosenthal, R. (1991).Meta-analytic procedures for social research(Vol. 6). Newbury Park, CA: Sage.

Rosenthal, R., & DiMatteo, M. R. (2001). Meta-analysis: recent developments in quantitative methods for literature reviews.Annual Review of Psychology, 52,

59e82.

Rosenthal, R., & Rosnow, R. L. (2008).Essentials of behavioral research: methods and data analysis(Vol. 3). New York, NY: McGraw-Hill.

Rosenthal, R., Rosnow, R. L., & Rubin, D. B. (2000).Contrasts and effect sizes in behavioral research: a correlational approach. New York, NY: Cambridge

University Press.

Rosenthal, R., & Rubin, D. B. (1982). A simple, general purpose display of magnitude of experimental effect.Journal of Educational Psychology, 74(2), 166e169.

http://dx.doi.org/10.1037/0022-0663.74.2.166.

Rosenthal, R., & Rubin, D. B. (1983). Ensemble-adjustedpvalues.Psychological Bulletin, 94, 540e541.

Rudolph, K. D. (2014). Puberty as a developmental context of risk for psychopathology. In M. Lewis, & K. D. Rudolph (Eds.),Handbook of developmental

psychopathology(pp. 331e354). New York, NY: Springer.

anchez-Meca, J., & Marín-Martínez, F. (1997). Homogeneity tests in meta-analysis: a Monte Carlo comparison of statistical power and type I error.Quality&

Quantity, 31, 385e399.

Shirtcliff, E. A., Dahl, R. E., & Pollak, S. D. (2009). Pubertal development: correspondence between hormonal and physical development.Child Development,

80(2), 327e337.

Simmons, R. G., & Blyth, D. A. (1987).Moving into adolescence: the impact of pubertal change and school context. Aldine Transaction.

*Skoog, T., Stattin, H., Ruiselova, Z., & Ozdemir, M. (2013). Female pubertal timing and problem behaviour: the role of culture.International Journal of

Behavioral Development, 37(4), 357e365.http://dx.doi.org/10.1177/0165025413486761.

Smith, L. M., & Glass, G. V. (1977). Meta analysis of psychotherapy outcomes.American Psychologist, 32, 752e760.

*Sontag-Padilla, L. M., Dorn, L. D., Tissot, A., Susman, E. J., Beers, S. R., & Rose, S. R. (2012). Executive functioning, cortisol reactivity, and symptoms of

psychopathology in girls with premature adrenarche.Developmental Psychopathology, 24(1), 211e223.http://dx.doi.org/10.1017/S0954579411000782.

*Sontag, L. M., Graber, J. A., & Clemans, K. H. (2011). The role of peer stress and pubertal timing on symptoms of psychopathology during early adolescence.

Journal of Youth and Adolescence, 40(10), 1371e1382.http://dx.doi.org/10.1007/s10964-010-9620-8.

Sørenson, K., Mouritsen, A., Aksglaede, L., Hagen, C. P., Mogensen, S. S., & Juul, A. (2012). Recent secular trends in pubertal timing: implications for evaluation

and diagnosis of precocious puberty.Hormone Research in Paediatrics, 77,137e145.http://dx.doi.org/10.1159/000336325.

*Stattin, H., Kerr, M., & Skoog, T. (2011). Early pubertal timing and girls' problem behavior: integrating two hypotheses.Journal of Youth and Adolescence, 40,

1271e1287.

*Stattin, H., & Magnusson, D. (1990).Paths through life: pubertal maturation in female development(Vol. 2). Hillsdale, NJ: Lawrence Erlbaum Associates,

Publishers.

*Storvoll, E. E., & Wichstrøm, L. (2002). Do the risk factors associated with conduct problems in adolescents vary according to gender?Journal of Adolescence,

25(2), 183e202.http://dx.doi.org/10.1006/jado.2002.0460.

*Susman, E. J., Dockray, S., Schiefelbein, V. L., Herwehe, S., Heaton, J. A., & Dorn, L. D. (2007). Morningness/eveningness, morning-to-afternoon cortisol ratio,

and antisocial behavior problems during puberty.Developmental Psychology, 43(4), 811e822.http://dx.doi.org/10.1037/0012-1649.43.4.811.

Timmermans, M., van Lier, P. A. C., & Koot, H. M. (2008). Which forms of child/adolescent externalizing behaviors account for late adolescent risky sexual

behavior and substance abuse? In M. Timmermans (Ed.),Antisocial behaviors: courses and consequences from toddlerhood to late adolescence(pp. 65e79).

Duivendrecht, Netherlands: PI Research.

*White, R., Deardoff, J., Liu, Y., & Gonzales, N. A. (2013). Contextual ampliﬁcation of attenuation of the impact of pubertal timing on Mexican-origin boys'

mental health symptoms.Journal of Adolescent Health, 53, 692e698.

*Williams, J. M., & Dunlop, L. C. (1999). Pubertal timing and self-reported delinquency among male adolescents.Journal of Adolescence, 22,157e17 1.

Winters, K. C., Stinchﬁeld, R. D., Latimer, W. W., & Stone, A. (2008). Internalizing and externalizing behaviors and their association with the treatment of

adolescents with substance use disorder.

Journal of Substance Abuse Treatment, 35(3), 269e278.http://dx.doi.org/10.1016/j.jsat.2007.11.002.

L.M. Dimler, M.N. Natsuaki / Journal of Adolescence 45 (2015) 160e17 0 17 0