8
Investigating the effects of peer association and parental inuence on adolescent substance use: A study of adolescents in South Korea Eunyoung Kim a, , Dae-Hoon Kwak b , Minwoo Yun c a Department of Criminology, Law, and Society, University of Florida, Gainesville, FL 32611-5950, United States b Department of Behavioral Sciences, Texas A&M International University, Laredo, TX 78041-1900, United States c Korean Institute of Criminology, Seoul, 137-715, South Korea abstract article info The central purpose of this study was to examine whether peer inuence has a greater impact on adolescent substance use than parental inuence. This was a comparative study that examined cross-cultural applicability by applying the established ndings and theoretical suggestions, such as social learning theory and social bonding theory in the United States, to a traditionally non-Western social context (South Korea). Although the theories have rmly established their explicability on adolescent delinquency and substance use in U.S. society, there are relatively few empirical studies to establish its generalizability in societies outside the U.S. and even fewer in such traditionally non-Western societies as South Korea. Using a nationwide sample of self-reported data from 3,188 junior high school students, estimations from multivariate analyses were used to compare the relative importance of peer and parental inuence on adolescent substance (alcohol and tobacco) use. The ndings from the current study supported both social learning theory and social bonding theory, suggesting that both peer and parental inuence are signicant in predicting the risks of adolescent substance use. Although parental inuence was slightly greater than peer factors, the difference was negligible. The limitations, the unique social context of Korean society, and future research implications are then discussed. © 2010 Elsevier Ltd. All rights reserved. Introduction One quintessential quest of criminology is the search for the causes of crime or delinquency. Since Lombroso, various criminologists have proposed different answers and perspectives. Over the past one hundred years, many criminological theories have been introduced, and their empirical validity has been thoroughly tested. As a result, modern criminology has at its disposal many sound, valid explana- tions for crime or delinquency, at least in the United States. Among these explanations, social learning theory and social bonding theory enjoy wide acceptance and great respect in modern criminology. Social learning theory and social bonding theory commonly highlight the causal signicance of parental and peer inuence on later crime or delinquency. Social learning theory suggests that crime is learned behavior, acquired through contact with parents or peers. Social bonding theory contends that an individual moves toward crime when he or she experiences detachment from parents or peers. Although their theoretical explanations may differ, both parental and peer factors are indicated as important causes of later crime or delinquency. Adolescent substance use (alcohol and tobacco use) has been widely used as a measure of delinquency in much theory-testing research that supports the empirical validity of the causal importance of parental and peer factors. The soundness of utilizing substance use as a measure is supported by numerous studies revealing that adolescents' alcohol and/or tobacco use is positively associated with other illicit drug use and later delinquency or crime (Akers, 1985; G. M. Barnes, Welte, & Hoffman, 2002; Fergusson & Horwood, 1996; Hawkins et al., 2000; Huizinga, Loeber, & Thornberry, 1995; Reed & Rountree, 1997). This means that adolescent alcohol and tobacco use is an important gatewaydeviance that leads to future, more serious types of delinquency or crime. In a study examining the relationship between substance use and youth delinquency, G. M. Barnes et al. (2002) found a link between New York high school students' alcohol and other illicit substance use and later delinquency. Hawkins et al. (2000) reported similar ndings showing the importance of juvenile alcohol and drug abuse as a measure of deviant behavior. Alarmed by the linkage of adolescent substance use to more serious criminal behavior in later adulthood, many criminologists consider such behaviors to be important subsets of delinquent behaviors. Both Akers' (1985, 1998) social learning theory and Hirschi's (1969) social bonding theory have been applied when explaining the impact of parental inuence and peer relationships on adolescent alcohol and tobacco use. Consistent with the basic premise of these two theories, a Journal of Criminal Justice 38 (2010) 1724 Corresponding author. Tel.: +1 713 540 6559; fax: +1 352 392 5065. E-mail address: eunyoungkim@u.edu (E. Kim). 0047-2352/$ see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.jcrimjus.2009.11.003 Contents lists available at ScienceDirect Journal of Criminal Justice

Investigating the effects of peer association and parental influence on adolescent substance use: A study of adolescents in South Korea

Embed Size (px)

Citation preview

Page 1: Investigating the effects of peer association and parental influence on adolescent substance use: A study of adolescents in South Korea

Journal of Criminal Justice 38 (2010) 17–24

Contents lists available at ScienceDirect

Journal of Criminal Justice

Investigating the effects of peer association and parental influence on adolescentsubstance use: A study of adolescents in South Korea

Eunyoung Kim a,⁎, Dae-Hoon Kwak b, Minwoo Yun c

a Department of Criminology, Law, and Society, University of Florida, Gainesville, FL 32611-5950, United Statesb Department of Behavioral Sciences, Texas A&M International University, Laredo, TX 78041-1900, United Statesc Korean Institute of Criminology, Seoul, 137-715, South Korea

⁎ Corresponding author. Tel.: +1 713 540 6559; fax:E-mail address: [email protected] (E. Kim).

0047-2352/$ – see front matter © 2010 Elsevier Ltd. Aldoi:10.1016/j.jcrimjus.2009.11.003

a b s t r a c t

a r t i c l e i n f o

The central purpose of this study was to examine whether peer influence has a greater impact on adolescentsubstance use than parental influence. This was a comparative study that examined cross-culturalapplicability by applying the established findings and theoretical suggestions, such as social learning theoryand social bonding theory in the United States, to a traditionally non-Western social context (South Korea).Although the theories have firmly established their explicability on adolescent delinquency and substanceuse in U.S. society, there are relatively few empirical studies to establish its generalizability in societiesoutside the U.S. and even fewer in such traditionally non-Western societies as South Korea. Using anationwide sample of self-reported data from 3,188 junior high school students, estimations frommultivariate analyses were used to compare the relative importance of peer and parental influence onadolescent substance (alcohol and tobacco) use. The findings from the current study supported both sociallearning theory and social bonding theory, suggesting that both peer and parental influence are significant inpredicting the risks of adolescent substance use. Although parental influence was slightly greater than peerfactors, the difference was negligible. The limitations, the unique social context of Korean society, and futureresearch implications are then discussed.

+1 352 392 5065.

l rights reserved.

© 2010 Elsevier Ltd. All rights reserved.

Introduction

One quintessential quest of criminology is the search for the causesof crime or delinquency. Since Lombroso, various criminologists haveproposed different answers and perspectives. Over the past onehundred years, many criminological theories have been introduced,and their empirical validity has been thoroughly tested. As a result,modern criminology has at its disposal many sound, valid explana-tions for crime or delinquency, at least in the United States. Amongthese explanations, social learning theory and social bonding theoryenjoy wide acceptance and great respect in modern criminology.

Social learning theory and social bonding theory commonlyhighlight the causal significance of parental and peer influence onlater crime or delinquency. Social learning theory suggests that crimeis learned behavior, acquired through contact with parents or peers.Social bonding theory contends that an individual moves towardcrime when he or she experiences detachment from parents or peers.Although their theoretical explanations may differ, both parentaland peer factors are indicated as important causes of later crime ordelinquency.

Adolescent substance use (alcohol and tobacco use) has beenwidely used as a measure of delinquency in much theory-testingresearch that supports the empirical validity of the causal importanceof parental and peer factors. The soundness of utilizing substance useas a measure is supported by numerous studies revealing thatadolescents' alcohol and/or tobacco use is positively associated withother illicit drug use and later delinquency or crime (Akers, 1985;G. M. Barnes, Welte, & Hoffman, 2002; Fergusson & Horwood, 1996;Hawkins et al., 2000; Huizinga, Loeber, & Thornberry, 1995; Reed &Rountree, 1997). This means that adolescent alcohol and tobacco useis an important “gateway” deviance that leads to future, more serioustypes of delinquency or crime. In a study examining the relationshipbetween substance use and youth delinquency, G. M. Barnes et al.(2002) found a link between New York high school students' alcoholand other illicit substance use and later delinquency. Hawkins et al.(2000) reported similar findings showing the importance of juvenilealcohol and drug abuse as a measure of deviant behavior. Alarmed bythe linkage of adolescent substance use to more serious criminalbehavior in later adulthood, many criminologists consider suchbehaviors to be important subsets of delinquent behaviors.

Both Akers' (1985, 1998) social learning theory and Hirschi's (1969)social bonding theory have been applied when explaining the impact ofparental influence and peer relationships on adolescent alcohol andtobacco use. Consistent with the basic premise of these two theories, a

Page 2: Investigating the effects of peer association and parental influence on adolescent substance use: A study of adolescents in South Korea

18 E. Kim et al. / Journal of Criminal Justice 38 (2010) 17–24

large body of empirical research confirms the significant predictabilityof parental influence and peer relationship on juvenile substance abusebehavior (Bahr, Hawks, & Wang, 1993; Bahr, Hoffmann, & Yang, 2005;Brown, Mounts, Lamborn, & Steinberg, 1993; Dishion, Andrews, &Crosby, 1995; Dishion, Patterson, Stoolmiller, & Skinner, 1991;Fergusson, Swain-Campbell, & Horwood, 2002; Hwang & Akers, 2006;G. Lee, Akers, & Borg, 2000;Mebly, Conger, Conger, & Lorenz, 1993; Neff&Waite, 2007; Reed & Rountree, 1997;Walden,McGue, Lacono, Burt, &Elkins, 2005; Warr, 1993a, 1993b).

Despite this large body of research, there seems to be some doubtleft as to which factor is a stronger predictor of adolescent substanceabuse. In other words: is parental influence or peer association morestrongly correlated to adolescent substance use? A sufficient numberof studies have sought the answer to this question. Overall, peerinfluence is considered the more robust causal factor in accounting foradolescent substance use and deviant behaviors although parentalinfluence is still an important cause (Bahr et al., 2005; Brook, Lukoff, &Whiteman, 1997; Dishion & Loeber, 1985; Hwang & Akers, 2006).

Although aforementioned theoretical propositions and empiricalsupports have been well established as sound, their applicability maybe limited to the United States. This is because the theories wereinitially designed for American society, and most empirical studieshave been conducted on American subjects. As Akers (1998) rightlypointed out, a theory may need to meet the standard of generaliz-ability to become a more robust and better theory. In other words,both social learning theory and social bonding theory are wellestablished because parental factors and peer factors are significantpredictors of adolescent substance use and future, more seriousdelinquency. Between the two, peer factors are more influentialpredictors than parental factors, but there is some doubt as towhether this established knowledge may be limited in scope to theUnited States only. The general worth of such theories is thusgeographically limited at best. In this context, there should beempirical tests to determine whether the established knowledge inthe United States can be duplicated in other countries, and in differentsocial and cultural contexts. The greater the number of empiricalstudies duplicating the results in different settings, the better theoriesbecome due to their increased generalizability. Herein lays the truevalue of a comparative study.

There have been some empirical studies outside the United Statesthat measured the causal impacts of parental and peer factors onjuvenile substance abuse. Fergusson et al. (2002) conducted theirstudy in New Zealand. Rumpold et al. (2006) applied the aforemen-tioned theoretical perspectives to Austrian society. Both studies’findings were quite consistent with those in the United States, whichis not surprising given that Austrian and New Zealander society sharesimilar cultural features with the U.S. These nations are predom-inantly influenced by Western European culture and belong to theWestern world; thus, this commonality of cultural and social fabricmay explain the consistent findings.

What if the same theoretical perspectives of social learning andsocial bonding are applied to society beyond this so-called Westernworld? Can one still find the same consistent results? This might be aninteresting cross-cultural comparative study, whichwould strengthenthe two theories’ generalizability if the consistencies can be found.The current study began with this question. Empirical studiesconducted outside traditional Western societies are still very rare.For this reason, applying the established theoretical perspectives ofsocial learning and social bonding to South Korea—which istraditionally believed to be a non-Western society—has a certainvalue. The findings from this application may contribute to thegeneralizability of criminological theories well established in theWestern world. For this purpose, this study tested the causalsignificance of parental and peer factors on Korean adolescents'substance use. It also examined which predicting variables hadstronger effects on the adolescents' substance use.

A previous study conducted in South Korea found results con-sistent with those in the Unites States and other Western societies(Hwang & Akers, 2006). That study, however, had a limitation. It usedadolescent samples collected only from the city of Busan, creating apossible sample selection bias. Busan is the largest port in the nation,and it is both highly liberalized and highly urbanized. Due to thenature of the metropolitan city, its culture is greatly individualizedand westernized. There is also some significant level of familydisruption in that city. Samples taken only from Busan, therefore,may not represent the general South Korean adolescent population.The current study responded to the limitation of the previous study byattempting to expand the research to the entire nation's adolescentpopulation.

Literature review

Theoretical perspectives

Empirical research, examining parental and peer impact onadolescent substance use, is inseparable from criminological theories.Amajority of these studies were based on and developed by two of themost respected theories: Akers's (1985, 1998) social learning andHirschi's (1969) social bonding theory (Ellis & Walsh, 1999; Hwang &Akers, 2006). Both theories indicate parental and peer factors asstrong predictors of adolescents' substance use and later severedeviance.

Akers' social learning theory is a general theory that has beenapplied to a range of deviant and criminal behaviors, including drugand substance use (Akers, 1985; Akers & Lee, 1996; Hwang & Akers,2006; Sellers, Cochran, & Branch, 2005). It proposes that individualconformity and deviant behavior are products of a learning processsignificantly influenced by contact with parents or peers. Theselearning mechanisms play equal roles in influencing an individual'sbehavior, functioning as four key mechanisms: differential associa-tion, definitions, differential reinforcement, and imitation. Thelearning depends on: (1) differential association with conformingand deviating others who are in their primary groups (i.e., family andfriends), which has a powerful effect on one's own definition andbehaviors; (2) differential reinforcement referring to the net balanceof the past, present, and anticipated costs and rewards (includingsocial and nonsocial reinforcements), which relates to a givenbehavior, abstinence, and substance use; (3) learning values andattitudes favorable and unfavorable toward substance use andsmoking (definitions); and (4) the extent to which one observessubstance use by and obstinate behavior from people whom theindividual admires (imitation). The empirical validity of sociallearning theory has been supported repeatedly by a large body ofempirical research (Agnew, 1993; Akers, 1985; Akers & Lee, 1996;Akers & Sellers, 2004; Bahr et al., 2005; Hwang & Akers, 2006; Neff &Waite, 2007; Reed & Rountree, 1997; Sellers et al., 2005; Warr, 1993a,1993b, 1996; Warr & Stafford, 1991).

Hirschi's (1969) social bonding theory also emphasizes the signifi-cance of the two predicting variables (parental and peer factors) inforeseeing adolescents' substance use and future deviance, althoughthis theory is based on a philosophical paradigm that differs from thatof social learning theory. The theory posits that an individual'smotivation for deviance commission is normal, due to the deviance-prone nature of human beings, and thus it explains why one conformsto the norm instead of why one deviates from the norm. The theoryproposes that the more closely an adolescent bonds to his or herfamily or friends, the more likely he or she is to act in non-deviantways (Akers & Sellers, 2004; Cullen & Agnew, 2003; Hirschi, 1969;Hwang & Akers, 2003, 2006). In this context, adolescents are morelikely to refrain from substance use if (1) they are more stronglyattached to others (i.e., attachment); (2) they have greater stakes in orambition tied to conformity (i.e., commitment); (3) the majority of

Page 3: Investigating the effects of peer association and parental influence on adolescent substance use: A study of adolescents in South Korea

19E. Kim et al. / Journal of Criminal Justice 38 (2010) 17–24

their time is spent involved in conventional activities (i.e., involve-ment), and (4) they are more likely to accept conventional moral lawand conventional social belief (i.e., belief). As is true with sociallearning theory, a large number of studies have also supported theempirical validity of social bonding theory (Agnew, 1991a, 1991b;Akers & Sellers, 2004; Bahr et al., 2005; Costello, 2000; Cullen &Agnew, 2003).

Peer versus parental influences

The relative strength of parental and peer influence on adolescentsubstance use and deviance is an important issue that needs to beresolved, and both social learning and social bonding theoriesrecognize the importance of these two predicting factors. Despitethe difficulty of clear-cut theoretical distinction, social learning theoryseems to place greater emphasis on the differential association withpeers as the more important predicting variable, while social bondingtheory emphasizes parental influence as the stronger predicting factor(Akers, 1998; Akers & Sellers, 2004; Hwang & Akers, 2006). Thistheoretical distinction is, nevertheless, still up for debate.

Overall, peer factors seem to be a stronger predictor of adolescentsubstance use than parental influence, at least in the United States.Akers and Sellers (2004) argue that peer influence has strong directeffects on adolescent substance use, superior to parental attachment.Supporting this proposition, empirical research evidence verifies theimportance of peer association and delinquent peer influence over theimportance of parental influence (Akers & Cochran, 1985; Allen,Donohue, Griffin, Ryan, & Turner, 2003; Bahr et al., 1993; G. M. Barneset al., 2002; Brown et al., 1993; Dishion & Loeber, 1985; Erickson,Crosnoe, & Dornbusch, 2000; Hoffman, 1994; Hwang & Akers, 2003,2006; Neff & Waite, 2007; Triplett & Payne, 2004; Warr, 1993b). Inexplaining adolescents' initiation into marijuana use, Kendal (1985)stated that peer influence was more important than the parentalinfluence. The study by G. M. Barnes et al. (2002) revealed that ayoung student who had a more peer-oriented attitude, plus closefriends who were drinking and showing behavioral problems, wasmore likely to engage in binge drinking and problematic behaviors.G. E. Barnes, Barnes, and Patton (2005) also found that peer drug usehad the strongest effect on adolescents' drug use compared toparental drug attitudes, parental attachment, and monitoring. Usinga meta-analysis technique, Allen et al. (2003) summarized studiescomparing peer and parental influence on adolescents' substance use.The analysis confirmed that peers, siblings, and friends were moreinfluential predicting factors on substance use than parents.

Psychological and developmental empirical research, furthermore,suggests that peers play a significant role in accounting for thedynamics of adolescent substance use (Farrell, Kung, White, & Valois,2000; Fergusson & Horwood, 1999; Friedman & Glassman, 2000;Granic & Dishion, 2003;Walden et al., 2005). Walden et al. (2005), forexample, found that delinquent peer influence was the primaryenvironmental influence on early substance use. Granic and Dishion(2003) also reported that deviant talk among friends was animportantmechanism in the development of serious deviant behaviorand drug use.

Various aforementioned empirical findings, however, neverundervalued parental influence on substance use. Parents indeedplayed a key role in explaining adolescents' substance use (Broman,Reckase, & Freedman-Doan, 2006; Chassin, Curran, Hussong, & Colder,1996; Hwang & Akers, 2006; Mebly et al., 1993; Simons & Robertson,1989). Mebly et al. (1993) reported that a mother's tobacco usesignificantly increased the risk of her children smoking. Chassin et al.(1996) also found that parental alcoholism, specifically the father'salcohol use and his level of supervision, was positively related to thechildren's alcohol use. The peer factor, nevertheless, was a relativelystronger predictor than parental influence on explaining adolescents'substance use and later, more serious deviance.

Cross-cultural application of peer and parental influences outside theUnited States and South Korean context

There has been some research into parental and peer influence onadolescent substance use and delinquency outside the United States(Fergusson & Horwood, 1999; Fergusson et al., 2002; Hwang & Akers,2003, 2006; Rumpold et al., 2006). In New Zealand, Fergusson et al.(2002) assessed deviant peer influence on adolescent illegal beha-viors and substance use. Utilizing longitudinal data, the study found asignificant impact of deviant peers on both psychosocial and criminaloutcomes, such as violent crime, property crime, and alcohol andother drug abuse. In a study conducted with an Austrian youthsample, Rumpold et al. (2006) found that peer group influence wasparticularly associated with adolescent substance and drug use, ascompared to variables reflecting negative family atmosphere, schooldifficulties, and other individual factors.

Over the past decade, adolescent substance use has emerged as aserious social concern in South Korean society. Perhaps due to thenation's westernization, the level of adolescent alcohol and tobaccouse in South Korea is equivalent to that in the United States and otherWestern countries (Johnston, O'Malley, & Bachman, 2003; Park, Kim,Kim, & Sung, 2007). According to the National Youth Commission ofSouth Korea (2002), approximately 70 percent of adolescents agedfourteen to sixteen reported that they drank alcohol, and 29 percent ofthem smoked tobacco in 2001. Using a sample of 14,296 junior highschool students under age fifteen, Jung (2003) found the prevalence ofsmoking among youth: 18 percent of boys and 8 percent of girls usedtobacco.

There were few empirical studies explaining causal factors ofsuch high prevalence of adolescent substance abuse in South Korea.One of those rare studies was conducted by Hwang and Akers(2003). They tested three different criminological theories—sociallearning, social bonding, and self-control theory—in a South Koreansetting and found that peer influence outweighed parental influencein explaining South Korean youths' alcohol and tobacco use. Hwangand Akers (2006) tested the rival causal effects of peer versusparental influences on South Korean adolescents using cross-sectional data from a sample of 1,035 high school students fromthe Busan metropolitan area. In that study, they asked whether thefindings and theories in Western countries were replicable in andapplicable to South Korea, which they presumed to be a non-Western society. They hypothesized that results might differ on thegrounds that youth culture is more influential in Western society(and thus the peer factor has stronger predicting power thanparental influence), whereas the traditional family oriented tradi-tion is still strong in South Korean society (and thus parentalinfluence has stronger predicting power than the peer factor). They,surprisingly, found that peer related variables had a greaterinfluence than parental variables on South Korean adolescents'drug and other substance abuse. They concluded that their findingswere similar to those in the United States and other Westerncountries, and therefore peer factors had stronger predicting powereven in the traditionally non-Western society of South Korea, wherestrong family-oriented tradition persists.

The current study shared a similar nature with the previous study(Hwang & Akers, 2006). It was another empirical study explainingcausal factors of the high level of adolescent substance abuse in SouthKorea. In doing so, it attempted to respond to the necessity of suchempirical causal studies. This study was also built on the previouswork (Hwang & Akers, 2006). The authors noticed the possibleproblem of the sample selection in the previous study. In that study,samples were drawn solely from the city of Busan, which may notrepresent the whole of contemporary South Korean society due to itsnature as a port city and highly urban area. Thus, the authors wantedto conduct a similar study, expanded to the entire South Koreansociety by using a more geographically representative sample. In this

Page 4: Investigating the effects of peer association and parental influence on adolescent substance use: A study of adolescents in South Korea

20 E. Kim et al. / Journal of Criminal Justice 38 (2010) 17–24

context, the specific hypotheses for the current study could besummarized as follows:

Hypothesis 1. Peer and/or parental influence will significantly affectSouth Korean adolescents' substance use (alcohol and tobacco use)controlling for demographic characteristics of adolescents.

Hypothesis 2. Like the result of Hwang and Akers' (2006) study, thepeer effect will be greater than the parental effect on the South Koreanadolescents' substance use, despite the geographical expansion of thesample to the entire South Korean society.

Methods

Data—Korea Youth Panel Survey

The data used in the current study were drawn from six years ofthe Korea Youth Panel Survey (KYPS) projects (2003-2008) designedto follow two selected groups of cohorts (at the beginning year of theprojects, this was the fifth grade in elementary school and the secondgrade in junior-high school at the beginning year of the projects).The National Youth Policy Institution (NYPI), a government-fundedinstitution, supported the project to fulfill a demand for constructing alongitudinal youth data base as a central resource of research relatedto adolescent development, education, and behavior in South Korea.The NYPI aimed to contain a wide range of information in order toevaluate the various aspects of the subject's current developmentaland conceptual phase, which could affect future positive or negativesocial behaviors.

Using a stratified multistage cluster sampling method, KYPSrecruited participants through two methods. First of all, a paper andpencil method was used only for the questionnaire about problematicbehaviors. Specifically, the survey included instruments assessingvarious learning and life environments of subjects, the relationshipsbetween the subjects and the social groups, and the variety ofexperiences encountered by the subjects. Second, face-to-face inter-views were conducted by well-trained interviewers who visitedschools to complete other questionnaires for adolescents. For parents,a telephone surveywas conducted to obtain other information relatedto participants' family and parents (e.g., familymonthly income, livingstatus with parents). Finally, confidentiality was assured by NYPI (K. S.Lee & Baek, 2007; Rhee, Yun, & Khang, 2007).

The participants consisted of the fifth-grade elementary schoolstudents (similar to fifth grade in the U.S.) and the second-gradejunior-high school students (i.e., eighth grade in the U.S.) at the timethe first wave of data collection began. For the first wave, theresearchers selected samples of the second-grade junior-high cohortgroup from 104 schools across South Korea; a total of 3,697 studentsand their parents were invited to participate in the survey. Of those, atotal of 3,449 gave parental consent and took part in the Wave 1 datacollection with a response rate of 93 percent. Data collection for Wave2 was conducted in 2004. A total of 3,188 took part in theWave 2 datacollection. The follow-up rate over the two-year period of the datacollection was remarkable, with more than 90 percent of first-yearadolescent participants completing questionnaires during eachsubsequent year.1

The present study examined the first- and second-wave datacollected from the second-grade junior-high cohort group partici-pants. Of 3,449 participants who took part in Wave 1 interviews, only261 participants were excluded, which resulted in 3,188 respondentsfor this analysis.

Dependent variables

The dependent variable of this study was adolescents' substanceuse, which was the combined measure of adolescents' alcohol or

tobacco use. In the survey questionnaire, the participants were askedhowmany times they had drunk and/or smoked during the past year.This dependent variable was treated using the negative binomialregression analysis; the frequency of the dependent variable,adolescents' substance use during the past year, was included toexamine the effect of peer and parental variables on the amount ofsubstance use. The dependent variable was taken from the Wave 2data set, which allowed the authors to investigate one year laggedeffects of independent variables on dependent variable.

Independent variables

The primary independent variables included two parental vari-ables and four peer variables. All these variables were taken from theWave 1 data set and measured one year prior to the measurement ofthe dependent variables.

Two parental variables included parental attachment and parentalsupervision. Parental attachment was measured by two Likert scaleditems that measured participants' perception of a positive relationshipwith their parents. The reliability coefficient of the scale was .72 asshown in Appendix A. For the parental supervision variable, responsesto two Likert scaled items were combined to indicate participants'perception of parental supervision of their behavior (see Appendix A).The reliability coefficient for parental supervision was .80.

The four peer variables included differential peer association,intensity of differential peer association, peer attachment, and peersubstance abuse. Differential peer association was measured bycombining six items asking various questions about the existenceand number of close delinquent friends as shown in Appendix A. Thetotal numbers of delinquent peers of respondents who admittedthe existence of delinquent friends to each item were summed up inone composite scale that ranged from 0 to 62 (Cronbach's α=.83).Intensity of differential peer association was measured by a singleitem as seen in Appendix A. This item was rated on a five-point Likertscale with the response categories ranging from “strongly disagree”(1) to “strongly agree” (5). Peer attachment was indexed by fouritems measuring respondents' attachment to their close friends asshown in Appendix A. The reliability coefficient was .75 for this four-item scale. Peer substance use was summed up by two itemsseparately asking respondents to indicate the number of close friendsdrinking alcohol or smoking. Responses were recorded into twodifferent categories (i.e., 0=none and 1=one or more friends). Thereliability coefficient was .83.

Control variables

In order to control the potential intervening effect of respondents'demographic background, four demographic variables were includedin the analysis: gender (0=male, 1=female), age (0=fifteen yearsold, 1=sixteen years old), monthly family income (0=above $3,500,1=$1,500-$3,500, 2=lower than $1,500), and living status (0=livewith none or single parent, 1=live with both parents). In addition,previous substance usewas also included as a control variable, becausethis variable might have a spurious effect.

Analytic strategy

Negative binomial regression technique was used to examine theeffects of parent and peer influence on adolescents' substance use byusing the frequency of dependent variables (count variable). Ordinaryleast squares (OLS) regression usually assumes that dependentvariables be continuous and unbounded, both of which are violatedwith count data. In this case, neither OLS nor Poisson regressionanalysis are appropriate, because those analyses may overestimatethe true response of participants. This study, therefore, employed thenegative binomial regression analysis to take into account the fact that

Page 5: Investigating the effects of peer association and parental influence on adolescent substance use: A study of adolescents in South Korea

Table 2Correlations between the variables

Variables (1) (2) (3) (4) (5) (6) (7)

(1) Substance usea 1.00(2) Parental attachment -.05⁎⁎ 1.00(3) Parental supervision -.05⁎⁎ .40⁎⁎ 1.00(4) DAb-peer delinquency .03 .01 .02 1.00(5) DAb-intensity .02 .09⁎⁎ .10⁎⁎ -.01 1.00

21E. Kim et al. / Journal of Criminal Justice 38 (2010) 17–24

the expected variances exceeded the means of an appropriateregression analysis method (Land, McCall, & Nagin, 1996; Long &Freese, 2006).2 The regression model was run with all of the peer andparental influence variables, as well as control variables for theanalysis of frequency substance use variable.3 Finally, Roncek's semistandardized coefficients (βR) were calculated by multiplying the bcoefficient by its standard deviation to compare the relative strengthof the effects (Miller, Miller, & Barnes, 2007; Roncek, 1997).

(6) Peer attachment -.02 -.04 -.02 .06⁎⁎ -.03 1.00(7) Peer substance use .01 .01 .01 .30⁎⁎ -.02 .11⁎⁎ 1.00

aFrequency.bDA=differential association.⁎P≤ .05.; ⁎⁎P≤ .01.

Table 3

Findings

Descriptive statistics are presented in Table 1. The percentage ofparticipantswho had experienced substance use (i.e., alcohol drinkingand/or smoking) during the past year was approximately 31 percent.Those adolescents who had experienced substance use reportedthat they smoked or drank at least two times on average duringthe past year. The majority of adolescents were fifteen years old(about 78 percent) and lived with both parents (roughly 91 percent).Half of the participants were male (50 percent). Finally, approxi-mately 67 percent of the parents earned between $1,500 and $3,500per month during the past year.

Table 2 includes results from a correlation analysis amongindependent and dependent variables. As shown in the table, therewas no correlation between parental variables and peer variables.Regarding the correlation between independent variables anddependent variable, there was a negative relationship betweenparental variable attachment and supervision and substance use(r=-.05, r=-.05, respectively), which were statistically significant atthe .01 level. That meant that, as parental attachment and supervisionincreased, the frequency of substance use decreased. The magnitudeof the correlation coefficients was, however, relatively small. Neitherof the peer variables was statistically significant.

Table 1Descriptive statistics (N=3,158a)

Variables N Percent Mean SD Min Max

Dependent variable (Wave 2)Substance useb 2.11 6.19 0.00 240.00

Independent variables (Wave 1)Parental variablesParental attachment 6.30 1.95 2.00 10.00Parental supervision 3.40 1.05 1.00 5.00

Peer variablesDifferential association-peerdelinquency

2.10 3.81 1.00 62.00

Differential association-intensity 3.40 1.05 1.00 5.00Peer attachment 11.84 2.11 3.00 15.00Peer substance use

0=none 2,121 67.201=one or more friends 1,037 32.80

Control variables (Wave 1)Prior substance use (1=yes) 1,061 33.60Age

0=fifteen years old 2,466 78.101=sixteen years old 686 21.70

Gender (1=male) 1,577 49.90Monthly family income

0=above $3,500 701 22.201=$1,500-$3,500 2,101 66.502=lower than $1,500 356 11.30

Living statusb

0=live with none or singleparent

116 3.70

1=live with both parents 2,857 90.50

aThere were 219 missing cases excluded from the analyses.bFrequency (number of substance used).

Table 3 shows results from the negative binomial regression analyses.Two parental variables and a peer variable (differential association-intensity) were shown to be associated with adolescent substance use.The directions of predictors were shown, as expected based on thetheories. Specifically, the two variables reflecting parental influencesignificantly decreased adolescents' substance use (exp(b)=.94 and.81), while the intensity of differential association increased the expectednumber of adolescents' using substances by 21 percent (exp(b)=1.21).In addition, a control variable (family income) was statistically sig-nificantly related to substance use. Among the second-grade junior-highcohort group, youths who lived in lower income families (less than$1,500 per month) were more likely to use substances than those youthin higher-income family (above $3,500 per month) (exp(b)=1.62).

The results from negative binomial regression for the effects of peer and parentalvariables on substance use

Variables Substance use (frequency)

b Exp(b) βRa

Parental variablesParental attachment -.06 .94⁎ -.12Parental supervision -.21 .81⁎⁎ -.22

Peer variablesDAb-peer delinquency -.02 .98 -.08DAb-intensity .19 1.21⁎⁎ .18Peer attachment -.01 .99 -.02Peer substance usec

One or more friends -.14 .87 -.07

Prior substance use (1=yes) .04 1.04 .02Age (1=sixteen years old)d -.17 .84 -.07Gender (1=male) -.12 .89 -.06Monthly family incomee

$1,500-$3,500 .11 1.12 .06Lower than $1,500 .48 1.62⁎ .27

Living statusf

Live with both parents .09 1.09 .05Constant 1.13 3.10

Log likelihood -4,195.49LR chi-squared (df) 49.26 (12)⁎⁎

McFadden's R2 .06N 2,969

aRoncek's semi-standardized coefficient=b coefficient×standard deviation of thevariable.bDA=differential association.cNone was the reference category.dFifteen years old was the reference category.eAbove $3,500 was the reference category.fLive with none or single parent was the reference category.⁎Pb .05.⁎⁎Pb .01.

Page 6: Investigating the effects of peer association and parental influence on adolescent substance use: A study of adolescents in South Korea

22 E. Kim et al. / Journal of Criminal Justice 38 (2010) 17–24

To compare the relative strength of the effects, Roncek's semi-standardized coefficients were also calculated. Based on thesecoefficients, the strongest predictor for substance use in the negativebinomial regression was parental supervision (βR=-.22), followed byintensity (βR=.18) and parental attachment (βR=-.12). The absolutevalue of the semi-standardized coefficients on parental supervisionwas slightly larger than its peer variable (less than .05). The difference,however, was trivial and thus it could be said that both parental andpeer variables are significantly related to the substance use. In otherwords, findings of the negative binomial regression model supportedthat both parental variables (parental supervision and parentalattachment) and peer variable (intensity) were important predictorsof adolescent deviance (substance use).

Discussion and conclusion

The purpose of this study was to compare the predictabilitybetween peer influence and parental influence on adolescents'substance use in a traditionally non-Western society such as SouthKorea. South Korea is a unique case in the traditionally non-Westerncontext. South Korean society used to be a traditional non-Westernone with strong family cohesiveness, collectivism, and parentalauthority, but westernization and urbanization due to rapid industri-alization and democratization over the last three decades changed thesocial and cultural climate of the nation. A strong emphasis onWestern values such as individualism, equality and human rights, andthe nuclear family significantly disrupted the traditional beliefs andstructure of family cohesiveness, collectivism, and the authority ofelders. As a result, traditional values and their remnants now coexistwith Westernized values and social contexts. South Korean societytoday may be defined as a semi-Western society with a non-Westernorigin.

To compare the differential predictability of the two rival factors,this study used four peer influence variables from social learningvariables (differential association-delinquent peer affiliation andintensity, peer attachment and peer substance use), and two parentalinfluence variables from social bonding variables (attachment toparents, parental supervision). In this sense, this research was anattempt to test the cross-cultural applicability of two generallyrespected criminological theories: social learning theory and socialbonding theory, which were proposed and developed in Westernsocieties such as the United States (Hwang & Akers, 2003, 2006). Ingeneral, the findings from the current study supported not only Akers'(1985, 1998) social learning theory, but also Hirschi's (1969) socialbonding theory. Two social learning variables and social bondingvariables showed significant relationships with Korean adolescentsubstance use.

The findings on differential predictability of parental and peervariables were somewhat inconsistent with previous studies con-ducted in the United States and also with Hwang and Akers' study(2003, 2006). Both parental and peer variables significantly related toadolescent substance use. This finding was not surprising, but it wassomewhat unexpected that both parental and peer influence wereequally strong and significant factors associated with adolescentsubstance use in South Korea (parental factors were even slightlymore influential than peer factors, although the difference was small).This result was inconsistent withmuch of the previous research in theU.S. and in Hwang and Akers' studies (2003, 2006), suggesting thatthe stronger influence lay with the peers instead of the parents.

The unexpected finding of the current studymight be explained bythe cultural duality of South Korean society today, meaning thecoexistence of both traditional and Western culture. Individualismand Western values prevail in today's young Koreans, and thus peerinfluence is very influential in adolescents' beliefs and behaviors. Atthe same time, parents are still influential in the nuclear familystructure. Ordinary Korean parents pay close attention to their

children's education and future, and thus they closely supervise andinteract with their children. In addition, most Korean parents haveonly one or two children and can therefore afford to pay closeattention to them. Although individualism prevails in the generalsocial and cultural context, parental authority over children is stillvery strong in the Korean family. In this sense, Western values outsidethe nuclear family and traditional values within the family coexist,and this cultural duality may explain the findings of equal strength ofpredictability between parental and peer variables.

The disparity of the current findings with the findings of Hwangand Akers' studies (2003, 2006) might be due to the sample selection.Busan has a social and cultural context that differs from other Koreancities and rural towns. It is an international port and a metropolitancity. The majority of people work in fishing, marine trading, and lightmanufacturing industries. As a result, people frequently travel abroadfrom their home. There is also a large presence of less educated andpoor populations, because the fishing, marine trading, and lightmanufacturing industries need many of those workers. Busan was thegateway of Westernization in Korea, and thus individualism,liberalism, and other Western values prevailed much earlier therethan in other Korean cities and provinces. On the negative side ofthose social and cultural contexts, Busan has been notorious for beingthe home of organized crime, street crimes, smuggling, low-classculture, family dysfunction, and a tendency toward violence. Parentalfactors could therefore be less influential in this more Westernizedcity compared to the rest of the country. In contrast, for the currentstudy, samples were drawn from the entire nation and were thusmore representative; the current sample and findings were morelikely to reflect the nature of cultural dualism in South Korea, and thestrong influence of both parental and peer variables was found. Morestudies, nonetheless, are needed to confirm this argument.

This study found relatively fewer explanatory powers and smallercoefficients than found in the regression models in the previous studyby Hwang and Akers (2006). This would be due to the fact that theindicators in the previous studywere specifically designed tomeasurepeer influence and parental influence, while the present study usedsecondary data not specifically designed for the study of parental andpeer effects on substance use. Further, compared to the previousstudies, the current study utilized a limited number of variables. Forexample, although parents' alcohol use is a well-established riskfactor for adolescents (Chassin et al., 1996; Hawkins, Catalano, &Miller, 1992), the current study did not include these variables due tounavailability. Although these limitations are not unusual when usingsecondary data, the limitation should be recognized and a future studymay need to consider these limitations to achieve greater explanatorypowers and coefficients.

Another limitation regarding the data used in this study was thedistribution of substance use response variables. The dependentvariables were highly skewed toward the “never use” response. Sincesubstance use variables are often highly skewed, tests assumingnormality can be seriously distorted. In an attempt to overcome thatlimitation, this study operationalized negative binomial regressionanalyses. Although this multiple regressionmodel was appropriate forthis study, the coefficients of the regressionmodel were still relativelysmall compared to the findings of Hwang and Akers (2006).

Regardless of these limitations, the current research overcame animportant limitation of the previous research on Korean adolescents byemploying national level longitudinal data, which increased thegeneralizability of the findings. By using the national level data, thecurrent study found an important and interesting reality of culturaldualism in South Korea, where traditional Korean and Western valuescoexist. As a result, bothparental and peer variableswere shown tohavestrong effects on adolescents' substance use, different from the U.S. andtheWestern social context. In addition, considering the lack of empiricalstudies conducted outside ofNorthAmerica (Hwang&Akers, 2006), thecurrent study could make a considerable contribution in expanding the

Page 7: Investigating the effects of peer association and parental influence on adolescent substance use: A study of adolescents in South Korea

23E. Kim et al. / Journal of Criminal Justice 38 (2010) 17–24

extant empirical research on adolescents substance use in traditionallynon-Western countries.

Acknowledgement

The authors would like to thank Dr. Ronald, L. Akers at theUniversity of Florida for comments in the prior version of this study.

Appendix A. Scales measuring parental and peer variables

Scales

Survey questions α

Dependent variables

1. How many times did you smoke duringthe past year?

2. How many times did you use alcoholduring the past year?

Parental variables

Parental attachment 1. I am comfortable sharing my thoughts and

feelings with my parentsa

.72

2. I often talk about what happens to meoutside homea

Parental supervision

1. My parents usually know where I am whileI am away from homea

.80

2. My parents usually know with whomI am while I am away from homea

Peer variables

Differential association-peer delinquency

How many of your close friends were:

.83 1. Tough at school2. Arrested by police3. Absent without leave4. Beating5. Extorting other friends' money and property6. Stealing others' money and property

Differential association-intensity

1. I place great value on my reputation frommy close friendsa

Peer attachment

1. I want to maintain the friendship with myclose friendsa

.75

2. I enjoy hanging out with my close friendsa

3. I try to have the same feelings and thoughtsas my friendsa

4. I can share my thoughts and feelings with myclose friends in a candid waya

Peer substance use

1. Number of your close friends who use tobacco .83 2. Number of your close friends who use alcohol

a1=strongly disagree, 2=disagree, 3=neither agree nor disagree, 4=agree, and5=strongly agree.

Notes

1. So far, five annual waves of KYPS data were publicly available.2. Since there was statistically significant evidence of overdispersion (e.g., G2 for

substance use=3,300.00, pb .01), the negative binomial regression models werepreferred to the Poisson regression models (Long & Freese, 2006, pp. 376-377).

3. To check any possible multicollinearity problem among the six independentvariables and four control variables, the variance inflation factor (VIF) was calculated.The results confirmed that there was no serious multicollinearity problem among thevariables (VIF≤1.27 and Tolerance≥ .79). Thus, the current study simultaneouslyincluded all variables in the final model.

References

Agnew, R. (1991a). Social control theory and delinquency: A longitudinal test. Crimi-nology, 29, 47−71.

Agnew, R. (1991b). A longitudinal test of social control theory and delinquency. Journalof Research in Crime and Delinquency, 28, 126−156.

Agnew, R. (1993). Why do they do? An examination of the intervening mechanismsbetween social control variables and delinquency. Journal of Research in Crime andDelinquency, 30, 245−266.

Akers, R. L. (1985).Deviant behavior: A social learning approach. Belmont, CA: Wadsworth.Akers, R. L. (1998). Social learning and social structure: A general theory of crime and

deviance. Boston: Northeastern University Press.Akers, R. L., & Cochran, J. (1985). Adolescent marijuana use: A test of three theories of

deviant behavior. Deviant Behavior, 6, 323−346.

Akers, R. L., & Lee, G. (1996). A longitudinal test of social learning theory: Adolescentsmoking. Journal of Drug Issues, 26, 317−346.

Akers, R. L., & Sellers, C. S. (2004). Criminological theories: Introduction, evaluation, andapplication (4th ed.). Los Angeles: Roxbury.

Allen, M., Donohue, W. A., Griffin, A., Ryan, D., & Turner, M. M. (2003). Comparing theinfluence of parents and peers on the choice to use drugs. Criminal Justice andBehavior, 30, 163−186.

Bahr, S. J., Hawks, R. D., & Wang, G. (1993). Family and religious influences onadolescent substance abuse. Youth and Society, 24, 443−465.

Bahr, S. J., Hoffmann, J. P., & Yang, X. (2005). Parental and peer influences on the risk ofadolescent drug use. Journal of Primary Prevention, 26, 526−551.

Barnes, G. E., Barnes, M. D., & Patton, D. (2005). Prevalence and predictors of “heavy”marijuana use in a Canadian youth sample. Substance Use and Misuse, 40,1849−1863.

Barnes, G. M., Welte, J. W., & Hoffman, J. H. (2002). Relationship of alcohol use todelinquency and illicit drug use in adolescents: Gender, age, and racial/ethnicdifferences. Journal of Drug Issues, 32, 153−178.

Broman, C. L., Reckase, M. D., & Freedman-Doan, C. R. (2006). The role of parenting indrug use among Black, Latino, andWhite adolescents. Journal of Ethnicity in SubstanceAbuse, 5, 39−50.

Brook, J., Lukoff, I., & Whiteman, M. (1997). Peer, family, and personality domainsrelated to adolescents' drug behavior. Psychological Reports, 41, 1095−1102.

Brown, B. B., Mounts, N., Lamborn, S. D., & Steinberg, L. (1993). Parenting practices andpeer group affliction in adolescence. Child Development, 64, 469−482.

Chassin, L., Curran, P. J., Hussong, A. M., & Colder, C. R. (1996). The relation of parentalcoholism to adolescent substance use: A longitudinal follow-up study. Journal ofAbnormal Psychology, 105, 70−80.

Costello, B. J. (2000). Techniques of neutralization and self-esteem: A critical test ofsocial control and neutralization theory. Deviant Behavior, 21, 307−330.

Cullen, F., & Agnew, R. (2003). Criminological theory: Past to present (2nd ed.). LosAngeles: Roxbury.

Dishion, T. J., Andrews, D. W., & Crosby, L. (1995). Antisocial boys and their friends inearly adolescence: Relationship characteristics, quality and interactional process.Child Development, 66, 139−151.

Dishion, T. J., & Loeber, R. (1985). Adolescent marijuana and alcohol use: The role ofparents and peers revisited. American Journal of Drug and Alcohol Abuse, 11, 11−25.

Dishion, T. J., Patterson, G. R., Stoolmiller, M., & Skinner, M. L. (1991). Family, school, andbehavioral antecedents to early adolescent involvement with antisocial peers.Developmental Psychology, 27, 172−180.

Ellis, L., &Walsh, A. (1999). Criminologists' opinions about causes and theories of crimeand delinquency. The Criminologist, 24, 4−6.

Erickson, K.G., Crosnoe, R., &Dornbusch, S.M. (2000). A social processmodel of adolescentdeviance: Combining social control and differential association perspectives. Journalof Youth and Adolescence, 29, 395−425.

Farrell, A. D., Kung, E. M., White, K. S., & Valois, R. F. (2000). The structure of self-reported aggression, drug use, and delinquent behaviors during early adolescence.Journal of Clinical Child Psychology, 29, 282−292.

Fergusson, D. M., & Horwood, L. J. (1996). The role of adolescent peer affiliations in thecontinuity between childhood behavioral adjustment and juvenile offending.Journal of Abnormal Child Psychology, 24, 205−221.

Fergusson, D.M., & Horwood, L. J. (1999). Prospective childhood predictors of deviant peerafflictions in adolescence. Journal of Child Psychological Psychiatric, 40, 581−592.

Fergusson, D. M., Swain-Campbell, N. R., & Horwood, L. J. (2002). Deviant peeraffiliations, crime and substance use: A fixed effects regression analysis. Journal ofAbnormal Child Psychology, 30, 419−430.

Friedman, A. S., & Glassman, K. (2000). Family risk factors versus peer risk factors fordrug abuse: A longitudinal study of an African American urban community sample.Journal of Substance Abuse Treatment, 18, 267−275.

Granic, L., & Dishion, T. J. (2003). Deviant talk in adolescent friendships: A step towardmeasuring a pathogenic attractor process. Social Development, 12, 314−334.

Hawkins, J. D., Catalano, R. F., & Miller, J. Y. (1992). Risk and protective factors foralcohol and other drug problems in adolescence and early adulthood: Implicationsfor substance abuse prevention. Psychological Bulletin, 112, 64−105.

Hawkins, J. D., Herrenkohl, T. I., Farrington, D. P., Brewer, D., Catalano, R. F., Harachi, T.W.,et al. (2000). Predictors of youth violence. Juvenile Justice Bulletin, 4, 1−11.

Hirschi, T. (1969). Causes of delinquency. Berkeley: University of California Press.Hoffman, J. P. (1994). Exploring the direct and indirect family effects on adolescent drug

use. Journal of Drug Issues, 23, 535−557.Huizinga, D., Loeber, R., & Thornberry, T. P. (1995). Urban delinquency and substance

abuse. Washington, DC: Office of Juvenile Justice and Delinquency Prevention,Office of Justice Programs.

Hwang, S., & Akers, R. L. (2003). Substance use by Korean adolescents: A cross-culturaltest of social learning, social bonding, and self-control theories. In R. L. Akers & G.F.Jensen (Eds.), Social learning and the explanation of crime: A guide for the newcentury, advances in criminological theory (Vol. 11, pp. 39−63). New Brunswick, NJ:Transaction.

Hwang, S., & Akers, R. L. (2006). Parental and peer influences on adolescent drug use inKorea. Asian Journal of Criminology, 1, 51−69.

Johnston, L. D., O'Malley, P. M., & Bachman, J. G. (2003). Monitoring the future nationalsurvey results on drug use, 1975-2002. Bethesda, MD: National Institute of DrugAbuse.

Jung, U. S. (2003). Investigation of factors relevant to smoking among juvenile highschool student in South Korea. Journal of Korean Academic Family Medicine, 23,894−903.

Kendal, D. B. (1985). On processes of peer influences in adolescent drug use: Adevelopmental perspective. Advances in Alcohol and Substance Abuse, 4, 139−163.

Page 8: Investigating the effects of peer association and parental influence on adolescent substance use: A study of adolescents in South Korea

24 E. Kim et al. / Journal of Criminal Justice 38 (2010) 17–24

Land, K. C., McCall, P. L., & Nagin, D. S. (1996). A comparison of Poisson, negativebinomial, and semiparametric mixed Poisson regression models: With empiricalapplications to criminal careers data. Sociological Methods and Research, 24, 387−442.

Lee, G., Akers, R. L., & Borg, M. (2000). Social learning and structural factors inadolescent substance use. Western Criminology Review, 5, 17−34.

Lee, K. S., & Baek, H. J. (2007). Korea youth panel survey (KYPS) III: A summary report(Korean). Seoul: Korea Institute for Youth Development.

Long, S., & Freese, J. (2006). Regression models for categorical dependent variables usingStata. College Station, TX: Stata Press.

Mebly, J. N., Conger, R. D., Conger, K. J., & Lorenz, F. O. (1993). Effects of parentalbehavior on tobacco use by young male adolescents. Journal of Marriage and Family,55, 439−454.

Miller, J. M., Miller, H. V., & Barnes, J. C. (2007). The effect of demeanor on drug courtadmission. Criminal Justice Policy Review, 18, 246−260.

National Youth Commission. (2002). A study on environments negatively affectingadolescents. Seoul, Korea: Author.

Neff, L. J., & Waite, D. E. (2007). Male versus female substance abuse patterns amongincarcerated juvenile offenders: Comparing strain and social learning variables.Justice Quarterly, 24, 106−132.

Park, S., Kim, H. S., Kim, H., & Sung, K. T. (2007). Exploration of the prevalence andcorrelates of substance use among sheltered adolescents in South Korea. Adoles-cence, 42, 603−616.

Reed, M. D., & Rountree, P. W. (1997). Peer pressure and adolescent substance use.Journal of Quantitative Criminology, 13, 143−180.

Rhee, D., Yun, S., & Khang, Y. H. (2007). Co-occurrence of problem behaviors in SouthKorean adolescents: Findings from Korea youth panel survey. Journal of AdolescentHealth, 40, 195−197.

Roncek, D. (1997, November). Interpreting the relative importance of negative binomialand Poisson regression coefficients. Paper presented at the annual meeting of theAmerican Society of Criminology, San Diego, CA.

Rumpold, G., Klingseis, M., Dornauer, K., Kopp, M., Doering, S., Hofer, S., et al. (2006).Psychotropic substance abuse among adolescents: A structural equation model onrisk and protective factors. Substance Use and Misuse, 41, 1155−1169.

Sellers, C. S., Cochran, J. K., & Branch, K. A. (2005). Social learning theory and partnerviolence: A research note. Deviant Behavior, 26, 379−395.

Simons, R., & Robertson, J. (1989). The impact of parenting factors, deviant peers, andcoping style upon adolescent drug use. Family Relations, 38, 273−281.

Triplett, R., & Payne, B. (2004). Problem solving as reinforcement in adolescent druguse: Implication for theory and policy. Journal of Criminal Justice, 32, 617−630.

Walden, B., McGue, M., Lacono, W. G., Burt, A., & Elkins, I. (2005). Identifying sharedenvironmental contributions to early substance use: The perspective roles of peersand parents. Journal of Abnormal Psychology, 113, 440−450.

Warr, M. (1993a). Age, peers, and delinquency. Criminology, 31, 17−40.Warr, M. (1993b). Parents, peers, and delinquency. Social Forces, 72, 247−264.Warr, M. (1996). Organization and instigation in delinquent groups. Criminology, 34,

11−38.Warr, M., & Stafford, M. (1991). The influence of delinquent peers: “What they think or

what they do?” Criminology, 29, 851−866.