Parenting Practices and Peer Group Affiliationin Adolescence

B. Bradford BrownUniversity of Wisconsin—Madison

Mna MountsUniversity of Illinois at Urbana-Champaign

Susie D. LambornUniversity of West Florida

Laurence SteinbergTemple University

BROWN, B. BRADFORD; MOUNTS, NINA; LAMBORN, SUSIE D.; and STEINBERG, LAURENCE. ParentingPractices and Peer Group Affiliation in Adolescence. CHILD DEVELOPMENT, 1993, 64, 467-482.Social scientists have often assumed that parental influence is sharply curtailed at adolescencebecause of the rising counterinfluence of peer groups, over which parents have little control.The present study tested a conceptual model that challenged this view by arguing that parentsretain a notable but indirect infiuence over their teenage child's peer associates. Data from asample of 3,781 high school students (ages 15—19) indicated that specific parenting practices(monitoring, encouragement of achievement, joint decision making) were significantly associatedwith specific adolescent behaviors (academic achievement, drug use, self-reliance), which inturn were significantly related to membership in common adolescent crowds (jocks, druggies,etc.). Findings encourage investigators to assess more carefully parents' role in adolescents' peergroup affiliations.

Among the many transformations in across early adolescence, susceptibility tofamily relationships expected to occur as a peer pressure increases while reliance uponcbild enters adolescence, one of the most parents' opinions and advice seems to de-worrisome for adults is the shift in reference cline (Berndt, 1979), as well as by researchgroup orientation from parents to the peer indicating that association with deviantgroup (Gecas & Seff, 1990). Parents often ex- peers is one of the strongest predictors ofpress concern that their adolescent child adolescent deviant activity (Elliott, Hui-will "fall into the wrong crowd" and be per- singa, & Ageton, 1985).suaded by peer pressure to engage in behav-iors that are self-destructive and/or counter There is, however, evidence that par-to parental expectations. Such concern may ents do retain a substantial measure of in-seem justified by studies indicating that, fluence over the attitudes and activities of

A version of this paper was presented at the biennial meeting of the Society for Researchin Ghild Development, Seattle, 1991. The study on which this report is based was supported bygrants to B. Bradford Brown from the Spencer Foundation through the Wisconsin Genter forEducation Research, University of Wisconsin—Madison; to B. Bradford Brown and LaurenceSteinberg from the U.S. Department of Education through the National Center on EffectiveSecondary Schools, Wisconsin Center for Educational Research, University of Wisconsin—Madison; and to Sanford M. Dombusch. and P. Herbert Leiderman from the Spencer Foundationthrough the Genter for the Study of Families, Youth, and Ghildren at Stanford University. Theopinions expressed in this paper are those of the authors and do not necessarily reflect theopinions of the funding agencies. The authors gratefully acknowledge the helpful comments ofthree anonymous reviewers on earlier versions of this paper, as well as the assistance of Bih-HuiHuang in data analyses and Lori Folgert in preparing the manuscript. Address correspondenceto the first author at the Department of Educational Psychology, University of Wisconsin—Madison, 1025 West Johnson Street, Madison, WI 53706.

[Child Development, 1993, 64,467-482. © 1993 by the Society for Research in Child Development, Inc.All rights reserved. 0009-3920/93/6402-0008$01.00]

468 Child Development

their teenage offspring. We believe this evi-dence is key to demonstrating the significantrole that parents play in their child's choiceof peer group associates during adolescence.The present study tests a conceptual modelin which parenting behaviors affect adoles-cents' peer group affiliations significantlybut indirectly, through their infiuence on ad-olescent behavior patterns.

The Traditional View and ContemporaryCounter evidence

At mid-century, social scientists pro-moted a perspective that emphasized thecompetitive or combative role that parentsand peers play in the lives of teenagers. Par-sons (1942) portrayed the adolescent socialsystem as a monolithic youth culture whosevalue system is clearly at odds with adultexpectations. Riesman (1961) contended thatparents had abdicated much of their author-ity over adolescent offspring to the schooland the peer group. Thus, it was commonto portray parents and peers as competitorsfor teenagers' allegiance, with parentsrelatively powerless—in many domains, atleast—to countermand the infiuence of peergroup pressures (Bowerman & Kinch, 1959;Brittain, 1963; Coleman, 1961; Floyd &South, 1972).

More recent studies, however, have pro-vided a very different image of parent andpeer group infiuences on adolescents. In thefirst place, investigators have found that pa-rental influence on children's behavior re-mains extensive in adolescence. Throughparticular child-rearing practices, parentscan have a substantial impact on adolescentbehaviors of major concern to adults, such asschool achievement patterns, drug use anddeviance, and self-concept. For example,outside of the child's intellectual ability, pa-rental expectation for achievement is one ofthe strongest predictors of adolescents' aca-demic achievement levels (Featherman,1980; Seginer, 1983). Parental monitoringhas a particularly powerful infiuence on ado-lescent delinquency and drug use (Coombs& Landsverk, 1988; Loeber & Dishion,1983) as well as on academic achievement(Dombusch, Ritter, Leiderman, Roberts, &Fraleigh, 1987). Parental efforts to engagetheir adolescent in joint decision makingseem to foster self-confidence and self-reliance (Baumrind, 1991b; Grotevant &Cooper, 1986; Hauser et al., 1984), enhanceacademic performance (Dombusch et al.,1987; Yee & Flanagan, 1985), and discour-age excessive or abusive use of drugs(Glynn, 1984) or delinquent activities (Dom-busch & Ritter, 1991).

Second, in contrast to the mid-centuryimage of a unified peer culture, ethnogra-phers have consistently discovered a diversearray of peer groups in American highschools (Buff, 1970; Eckert, 1989; Eder,1985; Larkin, 1979; Schwendinger &Schwendinger, 1985). An elite group similarto Coleman's (1961) "leading crowd" is al-ways found, although it is often divided intotwo somewhat different crowds. "Populars"are portrayed as socially competent individ-uals with a strong commitment to academicachievement but also moderate involvementin delinquent behavior and illicit drug use."Jocks" are quite similar but less academi-cally oriented and more focused in theirdrug use on alcohol, which they sometimesuse to excess. Counterposed against thesecrowds is a more alienated group—"drug-gies," "burnouts," "greasers," etc.—that isnot only heavily involved in drug use anddeviant activities but also inattentive toschoolwork and often hostile toward schooladults; yet, group members seem to main-tain a fairly strong self-image. Balanced be-tween these groups are the "normals," aver-age, or "in-between" students who seem toavoid deviant activities but otherwise arenot clearly distinctive on any particular trait.In most studies there is also a group of highachievers, the "brains" or "eggheads" or"intellectuals," who thrive on academics,forge close relationships with school adults,and studiously avoid drugs and deviant ac-tivities. Their self-confidence is bolstered byacademic achievements but also eroded bytheir marginal standing in the peer statussystem. Most schools also feature a sociallyinept crowd—"loners" or "nerds"—whosemembers are generally low in socid statusand, consequently, self-esteem. Their aca-demic achievement levels are variable, butthey seem to shy away from deviant activ-ities.

The ethnographic studies not only em-phasize the diversity of life-styles that ado-lescents pursue with peers, they also empha-size the limits of peers' ability to directteenagers' behavior. Adolescents do not hap-hazardly fall into one crowd or another andthen fall victim to the normative pressuresof that crowd. Instead, they are "selectedinto" a particular crowd by virtue of the rep-utation they establish among peers (Brown,1990). One does not become a jock by quirkof fate or even simply by honing one's ath-letic skill and making the basketball team.One must be perceived by peers as actingprimarily in "jock-like" ways: being inter-ested in sports, getting adequate but not out-

Brown et al. 469

standing grades, exuding self-confidence butremaining compliant with adult authority,shying away from drugs except for alcohol,and so on. There is, to be sure, pressure toconform to crowd norms (Clasen & Brown,1985) and to select friends from fellowcrowd members (Eckert, 1989; Eder, 1985),but these do not direct adolescents to newbehaviors as much as they reinforce existingdispositions—dispositions that helped di-rect the adolescent to a particular crowd intlie first place.

In sum, parents retain substantial influ-ence over their child's behavior in ado-lescence, and peer influences are not pre-dominantly antisocial but quite variable,contingent on the norms of the peer crowdto which the adolescent belongs. Further,crowd affiliation is not a matter of chanceassociation but a function of the reputationone establishes among peers by virtue ofone's background and behavior.

Linking Parenting Practicesto Peer Group Affiliations

Ethnographers have provided a detailedanalysis of the characteristics of adolescentcrowds but have been remarkably inatten-tive to the family characteristics of crowdmembers. By the same token, researcherswho have detailed associations between par-enting strategies and adolescent behaviorsusually have not concerned themselves withadolescents' peer group relationships. Thisfosters the impression that the family andpeer group are quite separate worlds forteenagers, perhaps because by this age par-ents have lost their ability to affect peer asso-ciations. Recent studies of adolescent devi-ance counter this impression, but they alsoraise questions about the way in which par-ents influence peer group affiliations. Someinvestigators have found that if parentsmodel deviant behavior or fail to maintainclose relationships with their teenager, thechild is more likely to drift into deviant peergroups and, as a consequence, be more in-

volved in drug use (Kandel & Andrews,1987; Oetting & Beauvais, 1987) or delin-quency (Marcos, Bahr, & Johnson, 1986;Massey & Krohn, 1986). Others have sug-gested a different causal arrangement ofthese variables: parental detachment fromearly adolescents promotes antisocial behav-iors, and youngsters with such behaviorsseem to band together in antisocial cliques(Cairns, Cairns, Neckerman, Gest, & Ga-riepy, 1988; Dishion, 1990). To date, how-ever, such studies have been largely con-fined to deviant behavior and deviantlyoriented peer groups.

The intent of this study was to expandconsideration of parental influences on ado-lescent peer group affiliation to a broaderarray of peer groups and to consider bothantisocial and prosocial behaviors. The as-sumption that parenting behaviors have a di-rect effect on teenagers' peer relationshipsfollows neatly from similar research onyounger age groups (e.g., Parke & Bhavna-gri, 1989). Nevertheless, the evidence re-viewed above, indicating that parents infiu-ence many behaviors by which adolescentsare assigned to crowds, seems to favor a con-ceptual model in which parents play a sig-nificant but indirect role in adolescent peergroup affiliations (see Fig. 1). According tothis model, specific parenting behaviors aresignificantly associated with specific adoles-cent characteristics, which in turn predictthe peer group widi which the adolescentis associated. Peer group pressures serve toreinforce behavior patterns by which adoles-cents come to be associated with a particularcrowd. By fostering certain traits in theirchildren, parents essentially direct a childtoward a particular peer group, and thus ex-ercise some control over the type! of peergroup infiuences to which their child is ex-posed.

To test the conceptual model, we fo-cused on three parenting practices—emphasis on academic achievement, moni-

Parentingpractices

Adoiescent'sbehaviors

Crowdaffiiiation

Crowd norms&peer pressure

iFIG. 1.—Conceptual model of the connections between parenting behaviors, peer group influ-

ences, and adolescent behavior.

470 Child Development

toring, and efforts to engage the child in jointdecision making—that have been shown inprevious studies to be significant parentalinfluences on adolescent behavior. We alsoselected three adolescent behaviors—academic achievement, drug use, and self-reliance—that are not only highly salient toresearchers, educators, and parents, but alsoinstrumental in defining adolescent peergroups. We relied upon adolescents' owndescriptions of their social system to definethe crowds to be studied. Because thestudy's basic intent was to examine the man-ner in which parents influence peer groupaffiliation, the portion of the conceptualmodel specifying peer group influences onadolescent behavior patterns (the "feedbackloop" in Fig. 1 from peer group affiliation toadolescent behavior) was not included inour analyses.

Method

Sample and ProceduresSix public, 4-year high schools—three

in the Midwest and three on the West Coast—participated in the study. These includeda small, rural school that drew its 400 stu-dents (98% European-American) primarilyfrom farm or working-class families; a schoolwhose more ethnically diverse student body(only 70% of the 1,500 students were Euro-pean-Americans) hailed from working- andmiddle-class neighborhoods of a moderate-sized city; a mid-sized school (1,000 stu-dents) located in a pervasively European-American, upper-middle-class suburb; alarge, ethnically diverse school (45% of the2,500 students were African-American orHispanic) located in a predominantly upper-middle-class community but also drawingfrom less economically privileged familiesin neighboring towns; a school whose 1,300students were largely bifurcated betweenworking- and upper-middle-class families,with Asian-American and Hispanic studentscollectively outnumbering European-Amer-icans; and a large inner-city school with anethnically diverse student body—African-Americans formed the largest single racial/ethnic group (40%)—from a broad range ofsocioeconomic backgrounds. Collectively,the schools comprised a diverse sample interms of size, location, socioeconomic back-ground, and ethnic composition.

All students in participating schoolswere invited to complete a self-report ques-tionnaire that focused on school-relatedbehaviors (academic achievement, engage-ment in classes, extracurricular participa-

tion) but also included measures of familyrelationships and parenting behaviors, peerrelationships, deviant activities, and psycho-logical well-being. Because of its length, thequestionnaire was divided into two sections,which were administered on separate days.Students also received peer ratings of crowdaffiliation, based on the Social Type Rating(STR) procedure (Brown, 1989) that is ex-plained in the next section. STR ratingswere obtained for all students in the Mid-western schools but, because of time con-straints, only freshmen and sophomores inthe West Coast schools.

Analyses in the present study were con-fined to the 3,781 students who receivedSTR ratings and also successfully completedboth portions of the self-report question-naire, from which measures of parentingpractices and adolescent behaviors were de-rived. This constituted 72% of the combinedstudent bodies of the participating schools(excluding the California upperclassmen notincluded in STR ratings). Of the remainder,4% refused or were denied permission byparents to participate in the study, 6% failedto fill out a questionnaire completely orcredibly, and 18% were absent from class onone or both questionnaire administrationdates. The sample ranged in age from 14 to19 (M = 15.5) and was fairly evenly dividedby sex (52% female). Although a majority(61%) of the sample was European-Amer-ican, there were also substantial numbers ofAfrican-Americans (12%), Asian-Americans(12%), and Hispanic youth (13%).

Peer-rated Crowd AffiliationThe Social Type Rating (STR) proce-

dure (Brown, 1989) is an efficient mecha-nism for identifying adolescents' peer groupaffiliation, based on their reputation amongpeers. It is somewhat similar to sociometricratings by which younger children are clas-sified into comparison groups (popular, re-jected, neglected, etc.; see, e.g., Terry &Coie, 1991), except that the STR ratings takeadvantage of the more sophisticated socialsystem of adolescents in which there arecommonly shared labels for particular peergroups.

Derived from earlier studies of adoles-cent peer groups (Clasen & Brown, 1985;Poveda, 1975; Schweridinger & Schwen-dinger, 1985), the STR procedure was a two-step process. In the first step, school admin-istrators were asked to identify a set of boysand girls (within each ethnic group in multi-ethnic schools) in each grade who repre-

Brown et al. 471

sented a good cross-section of the school'ssludent body. These students were inter-viewed in small groups composed of stu-dents of the same gender, same grade level,and same ethnic group. T'hrough group dis-cussion, each group derived a list of theschool's major crowds; then each participantlisted two boys and two girls in his or hergrade who were the leaders or most promi-nent members of each crowd. From theselists a stratified sample was drawn in eachgrade (stratified by crowd type, gender, andeliinicity, but with preference given to themost frequently listed students) to become"STR raters" in the second step. Each rater,accompanied by a friend of her or his ownchoosing, was individually interviewed. Theraters were presented with the list of crowdsderived from the earlier group interviews,thien asked to place each student in theirgrade level into one of the crowds. Raterscould indicate that they did not know a stu-dent well enough to assign to a crowd. STRratings continued until each student hadbeen rated by at least 10 STR raters. Becauseraters could only deal with about 300 namesin the time allotted for STR interviews, classlists in the larger schools had to be parti-tioned and the number of raters increased toensure that all students received the re-quired number of ratings.

There are several ways of using the STRratings to describe students' crowd affilia-tion. One is to create categorical assign-ments, that is, to place each student in thecrowd to which a majority of STR raters as-signed him or her. While useful for manyanalyses, categorical assignments mask the

fact that few students are placed in the samecrowd by all raters; they also give the samescore to marginal crowd members (thoserated into a particular crowd by a bare major-ity of raters) and central crowd members(those associated with the crowd by virtuallyEJI raters). An alternative strategy, which wasemployed for this study, is to create a seriesof proportion scores representing the per-centage of raters (excluding those who didnot know a student well enough to assign toany crowd) who placed the student in agiven crowd. Thus, respondents received ascore from 0 to 1.00 for each of their school'smajor crowds. Defining the outcome mea-sures as continuous rather than categoricalvariables also permitted us to use standardregression procedures for assessing the di-rect and mediated effects of our dependentmeasures (parenting behaviors).

Analyses in the present study were con-fined to proportion scores for six majorcrowds: populars, jocks, brains, normals,druggies, and outcasts.^ These six crowdswere widely recognized by raters in all par-ticipating schools. In fact, they comprisedthe bulk of STR crowd assignments (70%, onaverage). They are also the crowds on whichmost studies of adolescent peer groups havefocused. Means, standard deviations, and in-tercorrelations among the crowd proportionscores are provided in Table 1.

As would be expected with the scoringprocedure we used, crowd proportion scoreswere negatively skewed and negatively in-tercorrelated.^ The comparatively low inter-correlation between proportion scores for

^ The outcast group combined two crowds that were sometimes differentiated by STR raters:loners and nerds. Although distinctive in some ways, these groups share the same basic imageamong peers, especially regarding the behaviors of interest in this study: limited social skills,low self-image, low involvement in deviant behavior, and average or above average academicachievement levels. Thus, they appear to be different factions of the same major crowd (Brown,Lohr, & Trujillo, 1990). In fact, some STR raters confessed that calling someone a "loner" wasbasically a "nice" way of acknowledging that the person was a nerd. It may seem odd to consideroutcasts (or at least loners) as a crowd because their label implies a lack of social interaction.Nevertheless, it is a widely recognized social category with a distinctive normative image (Brownet al., 1990) and pattern of peer pressures (Clasen & Brown, 1985). Thus, it clearly fits with ourconceptual orientation toward crowds as reputation-based entities.

^ Crowd proportion scores are not independent; collectively, all of them (including thosefor crowd types not included in these analyses) must add to 1.00. Thus, as the number of crowdtypes increases the average score expected for each type diminishes. Crowd proportion scoresin this study were low because of the substantial number of crowd types employed by STRraters. Further, as the proportion of assignments to a specific crowd increases, the proportion ofassignments to other crowds must diminish; this explains the preponderance of negative correla-tions among crowd proportion scores. The negative skew of crowd proportion scores promptedus to repeat the major analyses using log transformations of the scores. Because results weresubstantively equivalent with both sets of scores we decided to retain the analyses based on theoriginal scores.

472 Child Development

TABLE 1

MEANS, STANDARD DEVIATIONS, AND INTEKCORKELATIONSOF CROWD PROPORTION SCORES

VAWABLE

PopularJockBrainNormalDrusfiieOutcast . ...

MEAN

.11

.07

.09

.19

.06

.18

SD

.20

.16

.18

.23

.16

.25

Popular

.02-.09- .11- .11-.24

INTERCORRELATIONS

Jock

-.09-.10- .08-.18

Brain

- .07-.14-.02

Normal

-.14-.10

Druggie

- .11

the populars and jocks and for the brains andoutcasts suggests that these pairs of crowdsare more compatible and closely alignedthan other possible pairs; tbe relatively bighnegative correlation between proportionscores for populars and outcasts suggeststhat this pair of crowds is quite incompati-ble. Previous research on adolescent crowdsconfirms these patterns (see Brown, Mory, &Kinney, in press), thus bolstering the valid-ity of the STR procedure.

Other MeasuresApart from crowd affiliation (which was

based on peer ratings), all other measureswere derived from students' responses toself-report questionnaire items. These mea-sures fell into three basic sets: demograpbiccharacteristics, respondents' report of par-ents' child-rearing practices, and scores onthree measures of adolescent behavior. Eachof these sets is described below.

Demographic characteristics.—Stu-dents provided information on their back-ground and current family situation. Spe-cifically, they indicated their sex, ethnicaffiliation, family structure (classified as in-tact, single-parent, stepparent, or other), andfamily socioeconomic status (SES), based onparents' level of education.

Parenting practices.—Investigatorshave taken a variety of approaches to mea-suring parenting bebaviors. Some have at-tempted to identify global parenting stylesthat reflect an adult's child-rearing valuesand orientations (Baumrind, 1991a). Bron-fenbrenner (1991), however, has cautionedthat a given parenting style is not necessarilytranslated into the same behaviors by all par-ents. Thus, in view of our model, we choseto measure parenting practices directlyrather than inferring them from parenting

styles. There is also some controversy aboutthe best way to measure parenting behav-iors: observations of parent-child interac-tions (usually in contrived situations), par-ents' self-report, or reports from children.Each approach has strengths and limitations;observations lack ecological validity, andbotb parents and children seem to distort re-ality somewhat in their reports of parentingbehavior. Because of the size of our sample,we were forced to rely exclusively on ado-lescents' report of parenting behaviors. Tominimize distortions in their responses, wefocused on measures of concrete parentingbehaviors, ratber than asking respondents tomake more general (subjective) judgmentsabout their parents' child-rearing practices.

From the wider set of variables on thequestionnaire relating to parent attitudesand behaviors, we selected several that metour operational criteria (i.e., assessed con-crete behaviors) and defined constructs thatwere especially pertinent to the questionson which this study focused. The first con-struct was parental emphasis on achieve-ment. Included in this 15-item measurewere answers to the following questions:"How often does each of your parents (a)help you with homework when asked, (b)know how you're doing in school, (c) go toschool programs for parents, (d) watch youin sports or activities, and (e) help you inchoosing your courses?"; "What is the low-est semester grade you could get in each ofthese subject areas witbout your parents get-ting upset? (a) English, {b) math, (c) science,(d) social studies"; "How important is it toyour parents or guardians that you work hardon your schoolwork?" All items were an-swered on a 5-point Likert scale. The itemsformed a scale with acceptable internal con-sistency (alpha = .84); scores represented

Brown et al. 473

the mean of item responses (1 = low, 5 =3

Students also completed a five-itempjirental monitoring scale simileir to onesthat have been used in previous studies(Dishion, 1990; Dombusch et al., 1985; Pat-terson & Stouthamer-Loeber, 1984). On athree-point scale (don't know, know a little,know a lot) they indicated how much theirpsirents really knew about who their friendswere, how they spent their money, wherethey were after school, where they went atnight, and what they did with their free time.The scale had an internal consistency alphaof .80; scale scores (mean item score) rangedfrom 1.00 (low) to 3.00 (high).

The third measure indicated the degreeto which parents engaged their child in Jointdecision making rather than making unilat-eral decisions for the child or allowing thechild to decide things completely by himselfor herself. This measure was based on a 13-item scale in which respondents indicated,on a 5-point scale (from "my parents decidethis without discussing it with me" to "I de-cide this without discussing it with my par-ents"), how they and their parents arrived ata decision about a variety of issues com-monly faced by high school students: schoolclasses, curfew times, spending patterns, useof alcohol, and so on (Dornbusch et al., 1985;Steinberg, 1987). The scale's alpha was .82.Scale scores, ranging from 0 to 1.00, indi-cated the proportion of items on which deci-sions were derived jointly as opposed to uni-laterally (by parents only or the adolescentonly).*

Adolescent behaviors.—Three mea-sures assessed adolescent behaviors thatwere expected to be related to parentingpractices. Self-reported grade-point average(GPA), scored on the standard 4-point scale,was employed as a measure of academicachievement. Self-reported grades havebeen found to correlate highly (r = .75) withactual grades taken from student transcripts

(Donovan & Jessor, 1985; Dornbusch et al.,1987). Respondents also completed a five-item drug use scale (alpha = .86), indicating,on a four-point scale (never, once or twice,several times, often), how often in the past6 months they had used various controlledsubstances. Scale scores (the mean of itemresponses) ranged from 1.00 (low) to 4.00(high). Similar measures in previous studieshave been sbown to be valid indicators ofadolescent drug use (Needle, McCubbin,Lorence, & Hochhauser, 1983; O'Malley,Bachman, & Johnston, 1983; Polich, 1982).Finally, respondents completed the 10-itemself-reliance subscale from Greenberger'sPsychosocial Maturity Inventory (Form D;Greenberger, Josselson, Knerr, & Knerr,1974). This scale measures adolescents' feel-ings of internal control and the ability tomake decisions witbout extreme relianceupon others. Items are answered on a four-point Likert scale. Like the drug use mea-sure, scale scores (the mean of item re-sponses) ranged from 1.00 (low) to 4.00(high). In this sample the scale's alpha was.81.

Plan of AnalysesOur conceptual model postulated that

adolescent behaviors would mediate the as-sociation between parenting practices andcrowd affiliation. Analyses followed a stan-dard procedure for testing sucb mediatingeffects (Baron & Kenny, 1986). First, we ex-amined the direct association between par-enting practices and crowd affiliation. Then,we determined whether the significant di-rect effects of parenting practices dimin-ished or disappeared when adolescent be-baviors were introduced as mediatingvariables. Following this, we examined asso-ciations between parenting practices and ad-olescent behaviors and between adolescentbehaviors and crowd affiliation. Separateanalyses were conducted for each crowd af-filiation score, using multiple regressiontechniques.

^ Several other items were considered for this scale but dropped on the basis of initial factoranalyses and reliability analyses. The omitted items seemed to focus on parental strategies inresponse to poor academic performance (e.g., "How often does your parent make sure you doyour homework") rather than behaviors that underscore the value they attach to achievement.

* Previous studies have emphasized j'oini versus unilateral decision-making, without differ-entiating whether unilateral decisions are made by parents or the child (Yee & Flanagan, 1985).Those who have differentiated the two types of unilateral decisions have reported curvilineareffects, in which adolescent outcomes were highest in families practicing joint decision-makingand not significantly different in femilies practicing "youth alone" or "parent alone" strategies(Dornbusch et al., 1985, 1989). Thus, the most appropriate usage of this measure for our analysesseems to be to calculate the proportion of joint decisions.

474 Child Development

Results

Characteristics of Each Set of VariablesTable 2 presents the mean scores, stan-

dard deviations, and intercorrelations for themeasures of parenting practices and adoles-cent behaviors. Within each set of measurescorrelations were positive but relatively low,suggesting (as expected) that the measuresaddressed distinctive aspects of parenting oradolescent behavior.

To determine the strength of associationbetween student background cbaracteristicsand crowd affiliation, we correlated crowdaffiliation scores with SES and also con-ducted a series of ANGOVAs (covarying forSES) that examined the independent effectsof sex, grade level, ethnicity, family struc-ture, and school on each crowd affiliationscore. Results are summarized in Table 3.SES was positively correlated with member-ship in the popular (r = .13) and brain (r =.14) crowds but virtually uncorrelated (r's <.05) witb other crowd scores. There weresignificant sex differences in scores for allcrowds; girls had stronger associations thanboys with the popular, brain, and normalcrowds, but weaker associations with thejocks, druggies, and outcasts. Grade differ-ences were significant for half of the crowds(jocks, brains, and outcasts), although therewas no clear or consistent pattern to thesedifferences. Being from an intact familyseemed to enhance membersbip in the braincrowd and diminish membersbip in tbedruggie crowd. All crowd affiliation scoresdiffered significantly by scbool, suggestingthat there were substantial school- or com-munity-based differences in ttle size of eacbcrowd. Except for the absence of associationbetween SES and druggie crowd scores.

these demographic differences corre-sponded to ethnographers' observationsabout the crowds; ethnographers have por-trayed the druggies as dominated by lower-class youth. Collectively, demographic vari-ables accounted for between 7.3% and 13%of the variance in crowd affiliation scores(see Table 4). Because of tbese differences,we included demographic variables as con-trol variables in all subsequent analyses.

Direct Associations between ParentingPractices and Crowd Affiliation

The first step in testing the conceptualmodel was to examine the direct effects ofparenting practices on crowd affiliation. Thiswas accomplished by regressing each crowdaffiliation score on each parenting variable,after controlling for effects of the demo-graphic variables noted above as well as theother two parenting variables. Tbe paths inthese analyses that were significant at p <.05 or better are reported in Figure 2. Paren-tal emphasis on achievement was positivelyassociated with membership in the popular,jock, and normal crowds and negatively as-sociated with membership in the druggies.Parent monitoring was positively associatedwith membership in the brain crowd andnegatively associated with membership inthe druggies. Joint decision making was pos-itively associated with membership in thebrains and normals and negatively associ-ated with membersbip in tbe druggies.Tbere were no significant patbs betweenparenting variables and membersbip in theoutcast crowd.

Adolescent Behaviors as MediatingVariables

The next step in testing the conceptualmodel was to determine if the effects of par-

TABLE 2

MEANS, STANDARD DEVIATIONS, AND INTERCORRELATIONS OF PARENTINGAND PSYCHOSOCIAL ADJUSTMENT VARIABLES

INTERCORRELATIONS

VARIABLE MEAN SD

Parental emphasis onachievement (PEA) .. 3.12 .80

Monitoring (Mon) 2.31 .49Joint decision making

(JDM) 49 .25CPA 2.76 .85Drug use 1.54 .72Self-reliance 3.04 .53

*p< .001.

PEA Mon JDM CPA Drug Use

.33*

.34*

.26*-.12*

.13*

.36*

.17*-.27*

.10

.17*-.20*

.02- .27*

.19* -.07*

Brown et al. 475

TABLE 3

DEMOGRAPHIC DIFFERENCES IN MEAN CROWD AFFILIATION SCORES

CROWD AFFILIATION SCOREDEMOGRAPHIC

GROUP Popular Jock Brain Normal Druggie Outcast

Sex:Male 08 .09 .08 .16 .07 .22Female 16 .03 .10 .21 .05 .15

Crade level:Freshmen 12 .05 .07 .19 .06 .22Sophomores 12 .07 .10 .19 .06 .15Juniors 11 .06 .09 .21 .07 .17Seniors 14 .06 .10 .19 .04 .18

Family structure:Intact 12 .07 .10 .20 .05 .19Single parent 11 .06 .08 .18 .07 .18Step-parent 14 .04 .06 .19 .09 .16Other 10 .04 .06 .16 .10 .16

Ethnicity:African-American 09 .08 .05 .19 .00 .11Euro-American 14 .06 .09 .20 .09 .14Asian-American 10 .03 .18 .13 .02 .21Hispanic 07 .04 .05 .18 .02 .12

School:A 11 .10 .13 .24 .06 .13B 22 .10 .09 .12 .05 .22C 11 .06 .06 .25 .06 .14D 06 .06 .05 .19 .06 .19E 09 .02 .10 .16 .06 .13F 14 .05 .10 .18 .06 .23

NOTE.—Scores for each variable are adjusted for the effects of all other variables plus socioeconomic status.Schools are listed in the order in which they are described in the Method section. Students who did not belong toany of the four ethnic comparison groups were omitted from these analyses. Further details of the analyses areavailable from the first author.

enting practices on crowd affiliation weremediated by adolescent behaviors. Of thenine initially significant paths between par-enting practices and crowd affiliation (Fig.2), five were reduced to nonsignificance,three were reduced substantially in magni-tude, and only one (the association betweenparental emphasis on achievement andmembership in the jock crowd) remained

significant at an equivalent level of magni-tude when adolescent bebaviors were in-cluded in the regression analyses (see Fig.3). Thus, as expected, the effects of parent-ing practices were largely mediated by ado-lescent behaviors.

Collectively, parenting practices andadolescent behaviors accounted for between

TABLE 4

PROPORTION OF VARIANCE IN CROWD AFFILIATION SCORES ACCOUNTED FOR BYBACKGROUND VARIABLES, PARENTING PRACTICES, AND ADOLESCENT BEHAVIORS

Crowd

PopularsJocksBrainsNormalsDruggiesOutcasts

BackgroundVariables

.129

.073

.088

.070

.053

.087

ParentingPractices (PP)

.009

.013

.010

.004

.026

.001

PP + AdolescentBehaviors

.033

.016

.139

.007

.175

.023

Tota

.162

.089

.227

.077

.230

.111

476 Child Development

Parental emphasison achievement

Parent monitoring

Joint decision-making

FIG. 2.—Observed direct associations between parenting practices and crowd affiliation. Standard-ized betas are shown for all paths significant at at least the .05 level. *p < .05; **p < .01; ***p < .001.

2% and 15% of the variance in crowd affilia-tion scores. Except for the jock and normalscores, this was a notable improvement overthe initial regression analyses assessing thedirect effects of parenting practices oncrowd affiliation (see Table 4).

To further establish the mediating roleof adolescent behaviors in our model, we ex-amined the associations between parenting

practices and adolescent behaviors and be-tween adolescent behaviors and crowd affil-iation. Results of these analyses are depictedin Figure 3, which shows all paths that weresignificant at p < .05 or better.

First, to assess the direct, independenteffect of each parenting practice on each ad-olescent bebavior, tbe score for each of thethree adolescent behavior variables was re-

.122"'

Parentai emphasison achievement

.108'"

Parent monitoring

Joint decision-making

Grade average

i3njguse

Self-reiiance

•.0B3'

Popuiar

Jock

Brain

Nomial

Druggie

Outcast

FIG. 3.—Observed associations between parenting practices, adolescent behaviors, and crowdaffiliation. Standardized betas are shown for all paths significant at at least the .05 level. Dotted linesindicate curvilinear effects. *p < .05; **p < .01; ***p < .001.

Brown et al, 477

gressed, in turn, on scores on the other twovariables and the parenting measures (aswell as demographic control variables). Pa-rental emphasis on achievement was posi-tively related to adolescents' reports of theirself-reliance and grade-point average. Pa-rental monitoring was inversely associatedwith drug use and directly associated withself-^reliance. Also, higher rates of joint deci-sion making were predictive of higher stu-dent grade-point averages and lower levelsof drug use and self-reliance. These findingsw£;re consistent with results of previousstudies except for (o) the tendency of jointdecision making to depress, rather than en-hance, self-reliance, and {b) the absence ofa significant association between parentmonitoring and student acbievement.

Finally, regression analyses examinedthe relation between adolescent behaviorsand peer crowd affiliation. In most cases,previous studies have suggested that crowdaffiliation is directly enhanced or dimin-ished by the behaviors we considered. In afew instances, however, the relation hasbeen depicted as curvilinear, that is, jocksand normals have been depicted as moder-ate academic achievers, and brains and nor-mals have been portrayed as harboring mod-erate levels of self-esteem or self-reliance.For these four instances, we tested for possi-ble curvilinear as well as linear effects in theregression analyses. With few exceptions,the significant paths for these analyses,shown in Figure 3, confirmed the behavioralprofiles of crowds that had been observed inprevious studies. Membership in the popu-lar crowd was predicted by high scores onall three behaviors, whereas membership inthe outcasts was predicted by relatively lowscores on these behaviors (the tendency foroutcasts to have low GPAs was unexpected).Meirlbership in the brains was related tohigh grades and low drug use but not to self-reliance (a curvilinear effect had been pre-dicted). Membership in the druggies was as-sociated with low grades and high drug use;the expected positive association with self-reliance was not significant. Academicachievement was the only behavior that pre-dicted membership in tbe jock crowd; as ex-pected, the relation was curvilinear. And asexpected, drug use discouraged member-ship in the normals, whereas moderate GPAlevels enhanced it, but the expected curvi-linear association with self-reliance was notsignificant for this crowd. Each crowd affili-ation score had a unique pattern of associa-tion With the adolescent bebavior measures.

which served to confirm ethnographers' ob-servations that each crowd features a distinc-tive life-style.

SummoryCollectively, these analyses provided

clear evidence of the mediating role that ad-olescent behaviors play between parentingpractices and adolescent crowd affiliation.Parenting practices had a significant impacton adolescent behaviors, which in turn weresignificantly associated with crowd affilia-tion, sucb tbat each crowd score had aunique profile on these three behavioralmeasures. Even with the inclusion of adoles-cent behaviors as mediating variables, how-ever, some direct associations between par-enting behaviors and crowd affiliation scoresremained significant.

Crowd affiliation scores also were mod-estly associated with adolescents' back-ground characteristics, some of which (SES,family structure) reflected features of theirfamily background. The ability of thesethree sets of variables—parenting practices,adolescent behaviors, and background char-acteristics—to account for variance in crowdaffiliation scores differed considerablyamong the crowds. In most cases, the directand indirect effects of parenting practicessuperseded the effects of background char-acteristics, but in all cases these variablesexplained a relatively modest proportion ofvariance in crowd affiliation scores.

Discussion

Researchers have clearly documentedthe powerful role parents play prior to ado-lescence in shaping children's social skillsand directing their relationships with peers(Ladd, Muth, & Hart, in press; Parke &Bhavnagri, 1989; Putallaz, 1987); studies ofhow parental infiuences in this area con-tinue through adolescence are much less ex-tensive. By linking two sets of findings thathave been pursued ratber independently—examining the infiuence of parenting prac-tices on adolescent behavior patterns, andshowing how these bebaviors differentiatemembers of various adolescent crowds—thestudy joins a number of investigations dem-onstrating that although parental influenceover peer associations may diminish in ado-lescence, it is far from inconsequential.

Investigators have acknowledged thatparents retain some control over their ado-lescent child's choice of peer associatesthrough their selection of the neighborhoodin which the family lives or the schools.

478 Child Development

churches, and community organizations inwhich family members participate (Rubin &Sloman, 1984). Our study confirmed this inthe substantial differences among schools inthe size of various crowds. In this respect,our findings relate to the "community stud-ies" of several decades ago, in which the au-thors contended that the high school peergroup system reflected characteristics of'thecommunity in which it was located (Cole-man, 1961; Hollingshead, 1949; Lynd &Lynd, 1957). Often, however, these studiesasserted that peer group membership wasprimarily a function of socioeconomic divi-sions within the community, whereas in ourstudy socioeconomic status was a relativelypoor predictor of crowd affiliation—andcommunity (school) differences in crowd af-filiation scores remained highly significantafter controlling for students' socioeconomicbackground. These school-based differencesremind us of the limitations of generalizingcharacteristics of the adolescent peer groupsystem across contexts. They also under-score the need to consider contextual fea-tures when examining such questions ashow parents influence teenagers' peer groupaffiliations. Parents' efforts to direct a childtoward an athletically oriented crowd, for ex-ample, may be fruitless if such a crowd isvirtually nonexistent—as appeared to be thecase in school E (see Table 3).

Altbough ethnographers have empha-sized family socioeconomic status as an im-portant factor in adolescents' peer group af-filiations (Eckert, 1989; Hollingshead, 1949;Larkin, 1979), we found other family back-ground characteristics—family structure andethnicity—to be more salient. Students fromintact families appeared to be overrepre-sented among the brains and underrepre-sented among druggies, which was preciselythe opposite pattern of students from step-parent families, who were also underrepre-sented in the jock crowd. It may be that thehostility that many teenagers harbor towarda stepparent (Fine, 1986) "spills over" totheir relationship with school adults, thusconstraining their ability to thrive in peergroups (such as jocks and brains) tbat featureclose relationships with school staff.

The predominance of European-Amer-icans in the druggie crowd was consistentwith research evidence indicating that Euro-pean-American teenagers are more involvedin drug use than adolescents from otherethnic groups (McCord, 1990). In otherrespects, ethnic differences in crowd affil-iation scores paralleled stereotypic orienta-

tions of ethnic groups: European-Americanswere overrepresented among the high-status, popular crowd; Asian-Americans wereoverrepresented among brains; and African-American students were overrepresented inthe jock crowd. Whether or not these pat-terns reflected ethnic differences in parentalexpectations or family reward structures isan important area for further research.

Family structural characteristics such associoeconomic status, ethnic background,and marital arrangements reflect one im-portant way in which parents retain someinfluence over teenagers' peer group associ-ations, but our model was concerned primar-ily with process characteristics, namely, par-enting practices. Our confidence in thecausal relationships specified in the modelare tempered by the fact that the findingsare not based on longitudinal data. Hays andRevetto (1990) demonstrated that severalcausal arrangements can be supported incorrelational data sucb as ours. There is,however, a growing body of research evi-dence, including some longitudinal work,that supports the causal arrangement of vari-ables specified in our model. Feldman andWentzel (1990) found that harsh disciplinestrategies affected early adolescent boys' so-ciometric status indirectly, tbrough the boys'capacity to inbibit aggressive behavior.Whitbeck, Simons, Conger, and Lorenz(1989) found tbat parental values influencedthe values of their teenage offspring, which,in turn, were significantly related to theteenagers' associations with deviant peers.Working with younger age groups (fourththrough seventh graders), Dishion (1990;Dishion, Patterson, Stoolmiller, & Skinner,1991) has reported both correlational andlongitudinal evidence that parenting strate-gies (discipline and monitoring) predictchild behaviors (academic performance anddeviant activity), which in turn affect pat-terns of peer associations (sociometric statusand involvement with deviant peers). Ourcausal ordering is also consistent with otherlongitudinal evidence that deviant behaviorpredicts association with deviant peersrather than the reverse (Magnusson, 1988).

This model contrasts sharply with onesproposing that parenting strategies affectpeer group affiliations directly (Elliott et al.,1985; Kandel & Andrews, 1987). The conse-quences of these different perspectives arewell illustrated by our findings witb refer-ence to the popular crowd. Had we applieda "direct effects" perspective to our data wemight have argued that parents may

Brown et al, 479

"nudge" their child toward the popularcrowd by encouraging academic achieve-ment, a bebavior furtber enbanced by popu-lar crowd norms (considering the relativelyhigh achievement levels of this crowd). Yet,in view of the absence of direct paths be-tween otber parenting behaviors and popu-lar crowd affiliation, we probably wouldhave concluded that peer group norms,rather than parenting bebaviors, must be re-sponsible for populars' propensity to engagein drug use. By contrast, our "mediatedmodel" suggests that parents have a handin shaping both the prosocial and antisocialbehavior patterns that characterize the popu-lar crowd. Parental emphasis on achieve-ment inspires higher academic performance,which fosters association with the populars.Yet, when the emphasis on achievement isaccompanied by efforts to monitor thechild's behavior, the child is oriented moretovi'ard the brain crowd, whereas when par-ents fail to monitor the child and do not en-courage joint decision making, the child ismore likely to become associated with thepopular crowd, whose desirable behaviors(academic achievement) are accompaniedby undesirable ones (drug use).

In fact, our model proposes tbat peergroups are unlikely to countermand parentalinfluences—for example, to lead an adoles-cent into deviant activities despite parents'efforts to orient the child toward prosocialvalues and behaviors. Instead, peer groupnorms serve primarily to reinforce behaviorsand predispositions to which parents(through parenting strategies and/or familybackground characteristics) have alreadycontributed. Although this contrasts withsome traditional perspectives on deviance(Sutherland & Cressey, 1978), it is consistentwith some more recent work in this area(e.g., Dishion et al., 1991; Snyder, Dishion,& Patterson, 1986). Of course, support forthese suppositions awaits longitudinal datathat examine the full conceptual model dis-played in Figure 1, including the "feed-back" loop of crowd norms and peer pres-sures. This study was simply a modest firststep in this process, but the findings shouldencourage researchers to pursue the fullmodel in earnest.

The findings make explicit what re-mains implicit in much of the research ondeviant bebavior: tbat whereas inadequateor undesirable parenting practices may con-tribute to the child's involvement in deviantpeer groups and deviant activities, positiveparenting fosters desirable behavior patterns

and association with prosocial peer groups.Yet, in view of the amount of variance incrowd affiliation scores accounted for by ourparenting variables, it may appear as if par-ents exert some modest influence over theirchild's associations with brains or druggies,but that otherwise parental influences oncrowd affiliation are trivial. It may well bethat parents have an impact on adolescents'affiliations with certain crowds more so thanothers, but, for several reasons, conclusionsabout the magnitude of parental influencesseem premature. First, a skewed distribu-tion of scores on outcome measures (as wastbe case here) typically depresses theamount of variance accounted for, so thatparenting effects may be underestimated inour results. Second, adolescent crowds aredifferentiated on a much wider range of be-

; haviors, which in turn are likely to be af-fected by a mucb broader array of parentingpractices, than were examined in this study.Indeed, the variables we examined seemhighly salient for the brains (academic per-formance, parental empbasis on achieve-ment) and druggies (drug use, parental mon-itoring), which probably accounts for the factthat we explained more of the variance inscores for these crowds than other crowds.

Most important, our study was confinedto measures of parenting practices and chil-dren's behavior in adolescence, whereas it islikely that parental influences on adolescentcrowd affiliation operate on a more extendeddevelopmental timetable. Cairns et al.(1988) bave sbown tbat parenting character-istics contribute to the division of youngstersby middle childhood into clusters of deviantand nondeviant peers, which may be theforerunners of prosocial (e.g., brains, jocks)and antisocial (e.g., druggies) crowds in ado-lescence. Thus, a teenager's crowd affilia-tion may be influenced by parenting behav-iors many years prior to adolescence. Snyderet al. (1986) found evidence that associationswith deviant peers in adolescence seemedto be tbe result of a series of events acrosschildhood and early adolescence. What ismore, the parenting behaviors that contrib-uted to this chain of events changed withage: in childhood, parents' effectiveness inteaching social skills and disciplining antiso-cial bebavior were key factors; by early ado-lescence, parental monitoring of tbe child'sbehavior was more salient. In other words,our data did not consider the cumulative orlagged effects of parenting behaviors priorto adolescence, wbicb may bave as much ifnot more impact (as parenting practices in

480 Child Development

adolescence) on nurturing behaviors thathelp direct teenagers to particular peergroups. Researchers must now turn attentionto the task of tracing the linkages betweenparents' efforts to monitor peer associatesand foster social skills in childhood, the for-mation and metamorphosis of preadolescentcliques, and the emergence of adolescentpeer groups and processes whereby teen-agers become associated with a particularcrowd.

There may be other facets of teenagepeer relations in which parental influencesoperate rather directly, but as for adoles-cents' crowd affiliations, tbe results of tbisstudy suggest tbat parental influences arelargely indirect, through their impact on be-haviors by which teenagers become associ-ated witb a particular crowd. What is more,parents have the capacity to direct theirchild toward an essentially prosocial crowd(such as the brains) or an antisocial crowd(sucb as tbe druggies) or a peer group withmixed characteristics (such as populars oroutcasts). To be sure, parenting practicesand family background cbaracteristics can-not determine a teenager's crowd affiliation,but their influence should not be dis-counted. In view of this, researchers are en-couraged to pay more attention to the Jointinfluences of parents and peers on adoles-cent bebavior.

Referenees

Baron, R. M., & Kenny, D. A. (1986). The modera-tor-mediator variable distinction in social psy-chological research: Conceptual, strategic,and statistical considerations. Journal of Per-sonality and Social Psychology, 51, 1173-1182.

Baumrind, D. (1991a). Parenting styles and ado-lescent development. In H. M. Lerner, A. C.Petersen, & J. Brooks-Cunn (Eds.), Encyclo-pedia of adolescence (Vol. 2, pp. 746-758).New York: Carland.

Baumrind, D. (1991b). The influence of parentingstyle on adolescent competence and sub-stance use. Journal of Early Adolescence, 11,56-95.

Berndt. T. J. (1979). Developmental changes inconformity to peers and parents. Develop-mental Psychology, 15, 606—616.

Bowerman, C. E., & Kinch, J. W. (1959). Changesin family and peer orientation between the4th and 10th grades. Social Forces, 37, 206-211.

Brittain, C. V. (1963). Adolescent choices and par-ent-peer cross-pressures. American Sociologi-cal Review, 28, 385-391.

Bronfenbrenner, U. (1991, April). Discussant's

commentary. In S. Lamborn (Chair), Authori-tative parenting and adolescent develop-ment. Symposium presented at the biennialmeetings of the Society for Research in ChildDevelopment, Seattle.

Brown, B. B. (1989). Social Type Rating Manual.Madison: National Center on Effective Sec-ondary Schools, University of Wisconsin—Madison.

Brown, B. B. (1990). Peer groups and peer cul-tures. In S. S. Feldman & C. R. Elliott (Eds.),At the threshold: The developing adolescent(pp. 171-196). Cambridge, MA: Harvard Uni-versity Press.

Brown, B. B., Lohr, M. J., & Trujillo, C. (1990).Multiple crowds and multiple life-styles: Ad-olescents' perceptions of peer group stereo-types. In R. E. Muuss (Ed.), Adolescent be-havior and society, 4th ed. (pp. 30-36). NewYork: McGraw-Hill.

Brown, B. B., Mory, M., & Kinney, D. (in press).Casting adolescent crowds in a relational per-spective: Caricature, channel, and context. InR. Montemayor, C. R. Adams, & T. P. Cullotta(Eds.), Advances in adolescent development:Vol. 6. Personal relationships during adoles-cence. Newbury Park, CA: Sage.

Buff, S. A. (1970). Creasers, dupers, and hippies:Three responses to the adult world. In L.Howe (Ed.), The white majority (pp. 60-77).New York: Random House.

Cairns, R. B., Cairns, B. D., Neckerman, H. J.,Gest, S. D., & Gariepy, J.-L. (1988). Socialnetworks and aggressive behavior: Peer sup-port or peer rejection? Developmental Psy-chology, 24, 815-823.

Clasen, D. R., & Brown, B. B. (1985). The multidi-mensionality of peer pressure in adolescence.Journal of Youth and Adolescence, 14, 451-468.

Coleman, J. S. (1961). The adolescent society.New York: Free Press.

Coombs, R. H., & Landsverk, J. (1988). Parentingstyles and substance use during childhoodand adolescence. JournoZ of Marriage and theFamily, 50, 473-482.

Dishion, T. J. (1990). The peer context of trouble-some child and adolescent behavior. In P. E.Leone (Ed.), Understanding troubled andtroubling youth (pp. 128-153). NewburyPark, CA: Sage.

Dishion, T. J., Patterson, G. R., Stoolmiller, M.,& Skinner, M. L. (1991). Family, school, andbehavioral antecedents of early adolescentinvolvement with antisocial peers. Develop-mental Psychology, 27, 172-180.

Donovan, J., & Jessor, R. (1985). Structure of prob-lem behavior in adolescence and young adult-hood. Journal of Consulting and Clinical Psy-chology, 53, 890-904.

Dombusch, S. M., Carlsmith, J. M., Bushwall, P.

Brown et al. 481

L., Ritter, P. L., Leiderman, P. H., Hastorf,A. H., & Gross, R. T. (1985). Single parents,extended households, and the control of ado-lescents. Child Development, 56, 326-341.

Dornbusch, S. M., & Ritter, P. L. (1991, April).Fam,ily decision-making and authoritativeparenting. Paper presented at the biennialmeetings of the Society for Research in ChildDevelopment, Seattle.

Dornbusch, S. M., Ritter, P. L., Leiderman, P. H.,Roberts, D. F., & Fraleigh, M. J. (1987). Therelation of parenting style to adolescentschool performance. Child Development, 58,1244-1257.

Eckert, P. (1989). Jocks and burnouts: Social cate-gories and identity in the high school. NewYork: Teachers College Press.

Eder, D. (1985). The cycle of popularity: Interper-sonal relations among female adolescence.Sociology of Education, 58, 154-165.

Elliott, D. S., Huisinga, D., & Ageton, S. S. (1985).Explaining delinquency and drug use. Bev-erly Hills, CA: Sage.

Featherman, D. L. (1980). Schooling and occupa-tional careers: Constancy and change inworldly success. In O. Brim & J. Kagan (Eds.),Constancy and change in human develop-ment (pp. 675-738). Cambridge, MA: Har-vard University Press.

Feldman, S. S., & Weritzel, K. R. (1990). The rela-tionship between parenting styles, son's self-restraint, and peer relations in early adoles-cence. Journal of Early Adolescence, 10,439-454.

Fine, M. (1986). Perceptions of stepparents: Varia-tion in stereotypes as a function of currentfamily structure. JournaZ of Marriage and theFamily, 48, 537-543.

Floyd, H. H., & South, D. R. (1972). Dilemma ofyouth: The choice of parents or peers as aframe of reference for behavior. Journal ofMarriage and the Family, 34, 627-634.

Cecas, V., & Seff, M. A. (1990). Families and ado-lescents: A review of the 1980s. Journal ofMarriage and the Family, 52, 941-958.

Clynn, T. J. (1984). Adolescent drug use and thefamily environment: A review. Journal ofDrug Issues, 14, 271-295.

Gieenberger, E., Josselson, R., Knerr, C , & Knerr,B. (1974). The measurement and structure ofpsychosocial maturity. Journal of Youth andAdolescence, 4, 127-143.

Grotevant, H., & Cooper, C. R. (1986). Individua-tion in family relationships: A perspective onindividual differences in the development ofidentity and role-taking skill in adolescence.Human Development, 29, 82-100.

Hauser, S. T., Powers, S., Noam, G., Jacobsen, A.,Weiss, B., & Follansbee, D. (1984). Familialcontexts of adolescent ego development.Child Development, 55, 195-213.

Hays, R. D., & Revetto, J. P. (1990). Peer clustertheory and adolescent drug use: A reanalysis.Journal of Drug Education, 20, 191-198.

Hollingshead, A. B. (1949). Elmtown's youth: Theimpact of social class on adolescents. NewYork: Wiley.

Kandel, D. B., & Andrews, K. (1987). Processes ofadolescent socialization by parents and peers.International Journal of the Addictions, 22,319-342.

Ladd, G. W., Muth, S., & Hart, G. H. (in press).Parents' management of children's peer rela-tions: Facilitating and supervising children'sactivities in the peer culture. In R. D. Parke& G. W. Ladd (Eds.), Family-peer relations:Modes of linkage. Hillsdale, NJ: Erlbaum.

Larkin, R. W. (1979). Suburban youth in culturalcrisis. New York: Oxford.

Loeber, R., & Dishion, T. J. (1983). Early pre-dictors of male adolescent delinquency: A re-view. Psychological Bulletin, 94, 68-99.

Lynd, R. S., & Lynd, H. M. (1957). Middletown:A study in American culture. New York: Har-court. Brace, and World.

Magnusson, D. (1988). Individual developmentfrom, an interactional perspective. Hillsdale,NJ: Erlbaum.

Marcos, A. G., Bahr, S. J., & Johnson, R. E. (1986).Test of a bonding/association theory of ado-lescent drug use. Social Forces, 65, 135-161.

Massey, J. L., & Krohn, M. D. (1986). A longitudi-nal examination of an integrated social pro-cess model of deviant behavior. SocialForces, 65, 106-134.

McCord, J. (1990). Problem behaviors. In S. S.Feldman & G. R. Elliott (Eds.), At the thresh-old: The developing adolescent (pp. 414—430). Cambridge, MA: Harvard UniversityPress.

Needle, R., McCubbin, H., Lorence, J., & Hoch-hauser, M. (1983). Reliability and validity ofadolescent self-reported drug use in a family-based study: A methodological report. Inter-national Journal of the Addictions, 18,901-912.

Oetting, E. R., & Beauvais, F. (1987). Peer clustertheory, socialization characteristics, and ado-lescent drug use: A path analysis. Journal ofCounseling Psychology, 34, 205-213.

O'Malley, P. M., Bachman, J. G., & Johnston, L.D. (1983). Reliability and consistency in self-reports of drug use. International Journal ofthe Addictions, 18, 805-824.

Parke, R., & Bhavnagri, N. P. (1989). Parents asmanagers of children's peer relationships. InD. Belle (Ed.), Children's social networksand social support (pp. 241—259). New York:Wiley.

Parsons, T. (1942). Age and sex in the social struc-ture of the United States. American Sociologi-cal Review, 7, 604-616.

482 Child Development

Patterson, G. R., & Stouthamer-Loeber, M. (1984).The correlation of family management prac-tices and delinquency. Child Development,55, 1299-1307.

Polich, J. M. (1982). The validity of self-reports inalcoholism research. Addictive Behaviors, 7,123-132.

Poveda, T. G. (1975). Reputation and the adoles-cent girl: An analysis. Adolescence, 37, 127-136.

Putallaz, M. (1987). Maternal behavior and chil-dren's sociometric status. Child Develop-ment, 58, 324-340.

Riesman, D. (1961). The lonely crowd. New Ha-ven, CT: Yale University Press.

Rubin, Z., & Sloman, J. (1984). How parents in-fluence tbeir children's friendships. In M.Lewis (Ed.), Beyond the dyad (pp. 115-140).New York: Plenum.

Schwendinger, H., & Schwendinger, J. S. (1985).Adolescent subcultures and delinquency.New York: Praeger.

Seginer, R. (1983). Parents' educational expecta-tions and children's academic achievements:A literature review. Merrill-Palmer Quar-terly, 29, 1-23.

Snyder, J., Dishion, T. J., & Patterson, G. R.(1986). Determinants and consequences of as-sociating with deviant peers during preado-lescence and adolescence. Journal of EarlyAdolescence, 6, 20-34.

Steinberg, L. (1987). Single parents, step-parents,and the susceptibility of adolescents to anti-social peer pressure. Child Development, 57,269-275.

Sutherland, E. W., & Cressey, D. R. (1978). Crimi-nology (10th ed.). Philadelphia: Lippincott.

Terry, R., & Coie, J. D. (1991). A comparison ofmethods for defining sociometric statusamong cbildren. Developmental Psychology,27, 867-880.

Whitbeck, L. B., Simons, R. L., Conger, R. D., &Lorenz, F. O. (1989). Value socialization andpeer group affiliation among early adoles-cents. Journal of Early Adolescence, 9, 436—453.

Yee, D. K., & Flanagan, C. (1985). Family environ-ments and self-consciousness in early adoles-cence. Journal of Early Adolescence, 5, 59-68.