Pubertal timing and adolescent delinquency†

• DOI: http://doi.org/10.1111/1745-9125.12245
• Author: Jeremy Staff,Rebecca Bucci
• Published date: 01 August 2020
• Date: 01 August 2020

Received: 15 March 2019 Revised: 22 March 2020 Accepted: 24 March 2020

DOI: 10.1111/1745-9125.12245

ARTICLE

Pubertal timing and adolescent delinquency*

Rebecca Bucci Jeremy Staff

Department of Sociology and Criminology,

The Pennsylvania State University

Correspondence

RebeccaBucci, Depar tment of Sociology

andCr iminology,The Pennsylvania State

University,211 Oswald Tower,University Park,

PA16802.

Email:r pb5358@psu.edu

∗Anearlier version of this article was presented

att he 2018 meetingof t he American Society

ofCr iminology.The measure of peer alco-

holuse at age 11 in the MCS was supported

bygrant AA019606 from the National Insti-

tuteon Alcohol Abuse and Alcoholism. This

researchis based on analysis of data from the

U.K.Millennium Cohort Study (MCS), which

receivescore funding from the Economic and

SocialResearch Council U.K. (ESRC) and a

consortium of U.K.government departments.

Thestudy sponsors played no role in the study

design;collection, analysis, and interpretation

ofdat a; the writing of the report; or the decision

tosubmit t he manuscript forpublication. The

authors wouldalso like to thank John Iceland,

TomLoughran, Jennifer Maggs and Wayne

Osgoodfor comments on earlier drafts of this

article, as wellas editor Jody Miller, and the

three anonymousreviewers for their comments

andinsight.

Abstract

Early pubertal timing (PT) increases the risk of adolescent

delinquency, whereas late development reduces this risk;

however, the mechanisms explaining PT effects on delin-

quency remain elusive. Theoretically, the PT–delinquency

relationship is as a result of changes in parental supervision,

peer affiliations, and body-image perceptions or is a spuri-

ous reflection of early life risk factors. Using intergenera-

tional data from the Millennium Cohort Study, a prospec-

tive sample of children followed from infancy to age 14

years in the United Kingdom (N=11,556 parent–child

pairs), we find that for both boys and girls, early PT is

associated with heightened risks of delinquency, relative to

on-time puberty, whereas late PT is associated with lower

risks, even after controlling for a large share of childhood

confounders. Mediation test results indicate that changes

in parental supervision, peer affiliations, and body-image

perceptions from ages 11 to 14 partly account for asso-

ciations between off-time PT and delinquency. Our find-

ings are most consistent with criminological theories in

which the psychosocial, familial, and peer group changes

that accompany off-time pubertal development are empha-

sized. Changes in peer substance use, in particular, were

the primary explanatory factor for the relationships between

early and late PT and delinquency, for both boys and girls.

KEYWORDS

delinquency, development, Millennium Cohort Study, pubertal timing

The findings from numerous studies have shownthat puber tal timing (PT), meaning the relative puber-

tal development compared with one’s same age and gender peers (Graber, Petersen, & Brooks-Gunn,

1996), is related to adolescent delinquency. Relativeto on-time puberty, children who experience early

Criminology. 2020;58:537–567. wileyonlinelibrary.com/journal/crim © 2020 American Society of Criminology 537

538 BUCCI AND STAFF

PT face a heightened risk of delinquency and other externalizing behaviors, whereas those who experi-

ence late PT have a lower risk (Carter, Jaccard, Silverman, & Pina, 2009; Caspi, 1995; Caspi, Lynam,

Moffitt, & Silva, 1993; Felson & Haynie, 2002; Flannery, Rowe, & Gulley, 1993; Ge, Brody, Con-

ger, Simons, & Murry, 2002; Haynie, 2003; Lynne, Graber, Nichols, Brooks-Gunn, & Botvin, 2007;

Mendle, Ryan, & McKone, 2018; Mrug et al., 2014; Obeidallah, Brennan, Brooks-Gunn, & Earls,

2004; Park, Yun, & Walsh, 2017; Sentse, Dijkstra, Lindenberg, Ormel, & Veenstra, 2010; Susman

et al., 2007). One explanation for these findings is that early PT is triggered by early life risk factors

that also predict delinquency, such as single parenthood, low parental educational or occupational sta-

tus, or high levels of parental psychological distress (Kelly, Zilanawala, Sacker, Hiatt, & Viner, 2017;

Moffitt, Caspi, Belsky, & Silva, 1992; Staff, Whichard, Siennick, & Maggs, 2015). According to psy-

chosocial acceleration theory (Belsky, Steinberg, & Draper, 1991; Ellis, 2004), high levels of familial

stressors in childhood accelerate PT, whereas PT may be delayed among youth who experience a less

stressful childhood. Thus, the positive relationship between PT and delinquency is spurious.

Alternatively,early PT may heighten the likelihood of adolescent delinquency as a result of increases

in 1) unsupervised time away from parents and other adult guardians (Cohen & Felson, 1979; Petersen

& Taylor, 1980); 2) time spent with boys (as friends or dating partners) and/or older, delinquent peers

(Caspi et al., 1993; Moffitt, 1993); or 3) frustration resulting from negative perceptions of one’s body

(Agnew, 2003; Caspi & Moffitt, 1991). Relativeto on-time puber ty, late PT may be protective against

delinquency because it is linked with higher parental supervision and time spent with nondelinquent

friends, as well as with positive perceptions of body-image (Moffitt, 1993). Theoretically,if researchers

accurately capture these changes during adolescence, PT effects on delinquency should be diminished.

The findings from recent meta-analyses and summary reviews (Deardorff, Hoyt, Carter, & Shirt-

cliff, 2019; Dimler & Natsuaki, 2015; Dorn, Hostinar, Susman, & Pervanidou, 2019; Negriff & Sus-

man, 2011; Ullsperger & Nikolas, 2017) highlight the remaining gaps in our understanding of the

PT–delinquency relationship. In particular, a lack of research exists on PT effects for boys, as well as

research in which the theoretical mediators of PT and delinquency are formally tested, and research

in which similar explanatory mechanisms for boys and girls are examined. Furthermore, a lack of

large-scale prospective studies makes it difficult to rule out childhood confounders of PT and adoles-

cent delinquency. Finally, existing research on the effects of late PT on delinquency is sparse, and the

findings are mixed regarding whether late PT is protective or harmful.

Using longitudinal data from the Millennium Cohort Study (MCS), our aim is to replicate prior

research to show the effects of PT on delinquency and then test several explanations for these links.

We build on existing PT–delinquency researchin the following ways: First, whereas most studies have

been reliant on either samples of boys (e.g., Felson & Haynie, 2002; Ge, Conger, & Elder, 2001) or

girls (e.g., Caspi et al., 1993; Haynie, 2003), the MCS sample of 11,556 adolescents and their parents

allows for us to examine the effects of PT on delinquency for both boys and girls in the same birth

cohort. Second, the large sample size also allows for us to use the Pubertal Development Scale, a valid

and reliable measure of PT (Petersen, Crockett, Richards, & Boxer, 1988), for comparison of early,on-

time, and late PT effects, as opposed to comparing early PT youth with all other adolescents. Third, the

intergenerational, longitudinal, multi-informant design of the MCS helps to minimize reporting bias,

as we rely on self-reported measures of PT and delinquency (rather than parent reported), and it allows

for us to control for early life risk factors that may confound the PT–delinquency relationship, such

as parent-reported prospective measures of childhood adversity (rather than youth reported). Fourth,

using theories from criminology and developmental psychology, we test the explanatory power of

several mediators in concert, including changes in parental supervision, peer affiliations, and negative

body image perceptions, for both boys and girls. Finally, whereas most prior research on this topic has

been based on samples of adolescents from the mid-1990s and earlier, the MCS sample is contemporary