youth recreation centre participation and criminal offending: a 20-year longitudinal study of...

13
http://jbd.sagepub.com/ Development International Journal of Behavioral http://jbd.sagepub.com/content/25/6/509 The online version of this article can be found at: DOI: 10.1080/01650250042000456 2001 25: 509 International Journal of Behavioral Development Joseph L. Mahoney, Hakan Stattin and David Magnusson Swedish boys Youth recreation centre participation and criminal offending: A 20-year longitudinal study of Published by: http://www.sagepublications.com On behalf of: International Society for the Study of Behavioral Development can be found at: International Journal of Behavioral Development Additional services and information for http://jbd.sagepub.com/cgi/alerts Email Alerts: http://jbd.sagepub.com/subscriptions Subscriptions: http://www.sagepub.com/journalsReprints.nav Reprints: http://www.sagepub.com/journalsPermissions.nav Permissions: http://jbd.sagepub.com/content/25/6/509.refs.html Citations: What is This? - Nov 1, 2001 Version of Record >> at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014 jbd.sagepub.com Downloaded from at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014 jbd.sagepub.com Downloaded from

Upload: joseph

Post on 28-Jan-2017

218 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Youth recreation centre participation and criminal offending: A 20-year longitudinal study of Swedish boys

http://jbd.sagepub.com/Development

International Journal of Behavioral

http://jbd.sagepub.com/content/25/6/509The online version of this article can be found at:

 DOI: 10.1080/01650250042000456

2001 25: 509International Journal of Behavioral DevelopmentJoseph L. Mahoney, Hakan Stattin and David Magnusson

Swedish boysYouth recreation centre participation and criminal offending: A 20-year longitudinal study of

  

Published by:

http://www.sagepublications.com

On behalf of: 

  International Society for the Study of Behavioral Development

can be found at:International Journal of Behavioral DevelopmentAdditional services and information for    

  http://jbd.sagepub.com/cgi/alertsEmail Alerts:

 

http://jbd.sagepub.com/subscriptionsSubscriptions:  

http://www.sagepub.com/journalsReprints.navReprints:  

http://www.sagepub.com/journalsPermissions.navPermissions:  

http://jbd.sagepub.com/content/25/6/509.refs.htmlCitations:  

What is This? 

- Nov 1, 2001Version of Record >>

at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014jbd.sagepub.comDownloaded from at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014jbd.sagepub.comDownloaded from

Page 2: Youth recreation centre participation and criminal offending: A 20-year longitudinal study of Swedish boys

http://www.tandf.co.uk/journals/pp/01650254.html DOI: 10.1080/01650250042000456

Youth recreation centre participation and criminaloffending: A 20-year longitudinal study of Swedish boys

Joseph L. MahoneyYale University, New Haven, USA

HaÊ kan StattinOrebro University, Sweden

David MagnussonStockholm University, Sweden

This study assessed whether participation in Swedish youth recreation centres (FritidsgaÊ rdar) isrelated to long-term criminality assessed from late childhood to mid adulthood. A prospective,longitudinal investigation of a representative cohort of 498 boys from a medium-sized Swedishcommunity was employed. A pattern-analysis identi� ed � ve con� gurations of boys who showeddifferent pro� les of social and academic competence at the age of 10. The con� gurations werecompared with respect to juvenile and adult criminality for boys who did, and who did not, make thedecision to participate in a youth recreation centre at age 13. Results showed that participation inyouth centres was nonrandom. Boys with a multiple problem pro� le of both social and academicproblems in school at age 10 showed more frequent participation in recreation centres at age 13. Thefrequency of criminal offending increased for all con� gurations of boys who became involved in arecreation centre. Frequent participation in youth centres was linked to high rates of juvenileoffending and persistent offending (i.e., registered for one or several offences both as a juvenile and asan adult). These � ndings held after controlling for individual, family, and economic factors prior toinvolvement in the youth centre. The limitations of the � ndings and their implications for socialpolicy are discussed.

The study of youth participation in leisure activities has gainedinterest in the � eld of psychology, including developmentalresearch (Masten & Coatsworth, 1998). Available evidencesuggests that leisure pursuits can in� uence individual compe-tence and social relationships and may have consequences forlong-term adjustment. Some aspects of individual functioninglinked to leisure activity participation include: short- and long-term behavioural and cognitive functioning (e.g., Brown,Kohrs, & Lazarro, 1991; Jones & Offord, 1989; Mahoney,2000), social relationships with peers and adults (e.g., Kinney,1993; Mahoney & Stattin, 2000), social status and persistencein the school system (e.g., Eccles & Barber, 1995; Eder, Evans,& Parker, 1995), and experience with the judicial and mentalhealth systems (e.g., Mahoney, 2000; McCord, 1978; Osgood,Wilson, O’Malley, Bachman, & Johnston, 1996). Leisureactivity components have also been included as part ofexperimentally designed prevention research that has impactedthe subsequent adjustment of the participants (e.g., Allen,Philliber, Herrling, & Gabriel, 1997; Jones & Offord, 1989;McCord, 1978). This suggests that the leisure activity-adjustment association may go beyond correlation. However,whether leisure activities in� uence adjustment in a positive ornegative fashion seems to depend on several features of theindividual and the activity pursued.

A variety of mechanisms have been proposed by whichleisure activities may enhance individual competence andprotect against adjustment problems. Although it is likely thatdifferent individuals bene� t from leisure pursuits for different

reasons, high levels of structure, skill-building aims, exposureto conventional values, and the presence of nondeviant peersappear to be particularly critical aspects when the reduction ofantisocial behaviour is of interest (Csikszentmihalyi, 1990;Jones & Offord, 1989; Kinney, 1993; Mahoney, 2000).

Studies showing that leisure activities have no impact, or anegative impact on adjustment often assess activities that arequalitatively different from those described above. Forinstance, transient and solitary leisure activities may beincluded in the assessment (e.g., watching television, readingcomics, ‘‘hanging out with peers’’). Additionally, the activitiesevaluated may maintain low structure, little or no adultsupervision, few opportunities for skill building, and theparticipants may be composed primarily of antisocial youth(e.g., Agnew & Peterson, 1989; Andersson, 2000; Hirschi,1969; McCord, 1978). Leisure settings wherein these condi-tions are dominant may promote, rather than decreaseadjustment problems and antisocial behaviour (Osgood et al.,1996).

Swedish youth recreation centres

Swedish youth recreation centres (FritidsgaÊ rdar) are a nation-ally sponsored organisation of leisure activities and manycommunities in Sweden are home to several centres. Thecentres were developed in the 1960s in response to a growingconcern over the lack of available leisure contexts for youth tospend their free time during the evenings. Special interest was

International Journal of Behavioral Development # 2001 The International Society for the2001, 25 (6), 509–520 Study of Behavioural Development

Correspondence should be addressed to Joseph L. Mahoney, Depart-ment of Psychology, Yale University, P.O. Box 208205, New Haven,CT 06520-8205, USA; email: [email protected].

Joseph L. Mahoney’s research was supported, in part, by NIMHGrant No. F32-MH11850-02. The research was also supported by

funds from the Swedish Council for Social Research, Axel andMargaret Ax:son Johnsons Stiftelse, and the Swedish Council forPlanning and Coordination of Research to HaÊ kan Stattin and DavidMagnusson.

at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014jbd.sagepub.comDownloaded from

Page 3: Youth recreation centre participation and criminal offending: A 20-year longitudinal study of Swedish boys

510 MAHONEY, STATTIN, MAGNUSSON / YOUTH CENTRES AND CRIMINALITY

focused on youth with so-called ‘‘expressive lifestyles’’ forwhom the centres were intended to provide an alternative towandering the city streets at night. Thus, an explicit aim of theyouth centres was to reduce youth antisocial activities bykeeping adolescents out of troubled settings during theevening. The centres are available to youth between the agesof 13 and 19. They are usually accessible every evening of theweek, opening around dinnertime and closing as late as11:30 p.m. during the weekend and summer. Today, manycentres are integrated in the schools and are open duringschool hours.

The characteristics of the youth centres vary somewhatacross communities, but the national agenda ensures thatcommonalities exist in the centres’ goals and activities. Theoverall philosophy for the youth centres has remained similaracross the last three decades, emphasising that youth should beallowed to develop their own interests. Attendance and activityparticipation at the youth centres are strictly voluntary. Typicalactivities include: pool, ping pong, video games, darts, TV,music, and drinking coffee. The centres may also sponsorvarious special events such as outdoor � eld trips, movie night,girls’ night, or weekly basketball. The available activities couldbe described as generally low in structure. Individual skill-building is not generally part of the activity goal. One or moreadults are present at the centre, but often do not direct or placedemands on the youths’ activities. The education and back-ground of the adult supervisors is variable.

Expectations for long-term adjustment

There are two primary expectations that follow from acomparison of the Swedish youth centres to the availableevidence on youth leisure activities and adjustment. First,participation in the youth recreation centres is not expected tobe a random affair. A recent cross-sectional investigation ofyouth recreation centres in central Sweden (Mahoney &Stattin, 2000) showed that adolescents who attend the centresare signi� cantly more antisocial, have more antisocial peers,and have parents who are less effective monitors of theirbehaviour, than adolescents not choosing to attend the centres.However, because of the research design, the direction ofassociation in that study cannot be clari� ed. The highantisocial behaviour of youth centre participants could bedue to antisocial persons being more likely to attend thecentres (selection), the social environment of the centresintroducing or augmenting antisocial behaviour (socialisation),or both.

Here it is anticipated that both processes—selection andsocialisation—account for the youth centre-antisocial beha-viour association. Youth with social-academic problems maybe more likely to participate in the recreation centres for severalreasons. First, the centres were designed to accommodate suchyouth and they will be welcome, if not encouraged, to attend.Second, the low structure and lack of demands at the youthcentres may be an attractive alternative for students experien-cing little success in the more structured social and academicclimate of the school. Finally, a con� uence of � nancial,knowledge, skill, and social barriers can inhibit disadvantagedyouth and those with problem behaviours from participating instructured leisure activities (e.g., Eder et al., 1995; Hultzman,1992, 1995; Kinney, 1993; Mahoney & Cairns, 1997; McNeal,1998).

The second expectation is that unstructured leisure settings

that are overrepresented by deviant youth will be associatedwith increased antisocial behaviour through peer socialisation.Children who show antisocial behaviour are often charac-terised by a pro� le of other behaviour problems (e.g., low peeracceptance, school failure, hyperactivity; Bergman & Magnus-son, 1991; Cairns, Cairns, & Neckerman, 1989; Mahoney,2000). Longitudinal studies provide convergent evidence thatpeer in� uences help to regulate individual patterns of problembehaviour over time (Cairns, Cairns, Neckerman, Gest, &Garie py, 1988; Dishion, McCord, & Poulin, 1999; Elliott,Huizinga, & Ageton, 1985; Patterson, 1993). Moreover,deviant peer relationships appear not only to maintain, butalso to promote antisocial patterns across development (Cairns& Cairns, 1994; Coie, Terry, Lenox, Lochman, & Hyman,1995; Patterson, Reid, & Dishion, 1992). Because the youthcentres become available to children at an age when peerin� uence is particularly salient (i.e., early adolescence), thisform of unstructured engagement with deviant peers may beespecially likely to introduce or augment behaviour patterns ofa negative, antisocial sort.

We address both of these expectations from a holistic,interactionistic perspective to studying development thatbegins with the assumption that persons function aswholes—as total, integrated organisms (Magnusson, 1988;Magnusson & Stattin, 1998). Aspects of the developmentalsystem function interdependently. Accordingly, key compo-nents that are relevant to the phenomenon under investigation(i.e., criminal offending) should be studied together aspackages of operating factors within persons. This does notexclude the possibility of assessing relations with moreordinary, variable-oriented methods. However, it is assumedhere that the identi� cation of individual pro� les of operatingfactors will provide complementary information on thedevelopment of criminal offending not available from vari-able-based methods (Magnusson & Bergman, 1988).

In summary, this study aims to describe the relationbetween participation in the Swedish youth recreation centresand subsequent criminal offending across adolescence andadulthood. We expect that youth with multiple problempro� les of social and academic disadvantage will be mostlikely to choose to participate in the recreation centres. Wefurther expect an associated increase in criminal behaviour foryouth who make the decision to participate in the recreationcentres compared with similar youth who do not.

Method

Participants

The participants include 498 boys and their parents who arepart of the Main Cohort in the ongoing longitudinal investiga-tion, Individual Development and Adaptation (IDA; Magnus-son, 1988; Magnusson & Bergman, 1998; Magnusson, Dune r,& Zetterblom, 1975). Sample recruitment involved all childrenin the town of Orebro, Sweden who attended the third gradeduring the spring of 1965.1 Over 90% of participants were 10years old when � rst assessed. The 498 persons involved in thisstudy represent 96% (498/519) of the total sample of males at

1 Children who attended special schools due to serious intellectua lshortcomings, and those with grave physical impairments are not included inthe present study.

at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014jbd.sagepub.comDownloaded from

Page 4: Youth recreation centre participation and criminal offending: A 20-year longitudinal study of Swedish boys

age 10 and were selected by virtue of having completeinformation on teacher-rated and peer-assessed behaviours,and achievement test data. In 1968, at sixth grade (age 13) asecond data collection was completed. At this time, 85% (423/498) of the study sample completed an extensive school surveythat included questions concerning their involvement in theyouth recreation centre. The 75 persons who did not respondto the age 13 survey had either moved out of the sampling area,or were not able to complete the survey during the testing days.A comparison of these 75 persons with the remaining 423participants revealed no signi� cant differences for aggression,hyperactivity, achievement, peer preference, or criminaloffending.

At the outset of the investigation in 1965, the city of Orebroincluded about 80,000 inhabitants and has expanded overtime. In 1985, there were approximately 120,000 personsliving in Orebro. Given the population of Sweden (currentlyabout 8.7 million), Orebro is a relatively large city by nationalstandards. The economy of Orebro was dominated by aprominent footwear industry, but has diversi� ed over time intoengineering plants, printing shops, and a food and paperindustry. Additionally, Orebro is home to a large hospital,university, and well-known psychiatric clinic.

Stattin, Magnusson, and Reichel (1986) compared rates ofcriminality in Orebro with other major cities in Sweden, andwith the nation as a whole, across the years 1965–1980.Overall, the crime rate in Orebro was higher than the Swedishnational average during this time; however, the gap has closedin the more recent years. This was the result of a rise innarcotic and fraud offences that increased in the nation as awhole, but not in Orebro. A city-based comparison showedthat Orebro had comparable rates of crime to similar sizedcities in Sweden (50,000–100,000 inhabitants), and moder-ately lower rates of crime compared to Sweden’s largest cities(i.e., Stockholm, Goteborg, Malmo).

Analytic strategy

The analytic strategy combines both variable-based methods(i.e., regression) and person-oriented methods (i.e., pattern-analysis). The use of pattern-analysis re� ects a growingappreciation for studying individual patterns of development(Cairns, Bergman, & Kagan, 1998). Grouping of the boys withrespect to their individual patterns of social and academicfunctioning was guided by two criteria: (1) that the variablescover a range of individual functioning speci� cally relevant tothe phenomenon under investigation (i.e., criminality), and(2) that the variables chosen function simultaneously in aprocess that operates as an organised whole. Previous studiesusing this dataset have independently linked aggression(Magnusson, Stattin, & Dune r, 1983), hyperactivity (Magnus-son, 1996), school achievement (Torestad & Magnusson,1996), and peer preference (Bergman & Magnusson, 1991) toantisocial behaviour.

Measures

Social behaviour. Behavioural adjustment was determinedfrom teacher-ratings and peer nominations when the partici-pants were 10 years of age. Data included teacher-ratingsconcerning Aggression and Hyperactivity. Hyperactivity wasobtained by the combination of ratings of poor concentrationand motor restlessness (alpha reliability ˆ .78). Teacher-

ratings were made on a 7-point scale (Aggression: M ˆ 3.72,Median ˆ 4.00, SD ˆ 1.57, Hyperactivity: M ˆ 3.98, Medianˆ 4.00, SD ˆ 1.56), where higher values indicate more of therated construct. Backteman and Magnusson (1981) reporthigh stability for the foregoing items using independent teacherassessments (i.e., different raters) over a three-year interval(ages 10 to 13).

Peer preference. To assess peer preferences at age 10, studentswere asked to imagine that they were being transferred toanother classroom, but that not all of their classmates could goto the new room (cf. Magnusson et al., 1975). They were thenasked to rank all of their classmates in the order in which theywould like them to move. The students were asked to rank girlsand boys separately. The rankings were then combined acrossall informants and standardised to form a continuum of peer-rated preference. Higher scores re� ect a greater number ofpreference nominations.

School achievement. Academic Achievement was assessedusing averaged results of a nationally standardised achievementtest in mathematics and Swedish when the participants were 10years of age (alpha reliability ˆ .79). The achievement itemsconsisted of knowledge in different areas: In Swedish thisinvolved reading comprehension, spelling, word knowledge,grammar, and punctuation; in mathematics this involvedmechanical arithmetic, fractions, and mental arithmetic.Achievement scores ranged from 1 to 5 for math and Swedish(M ˆ 3.04, Median ˆ 3.00, SD ˆ .90), with higher scoresre� ecting greater achievement. National measures of internalitem reliability are usually above .90 for the achievement tests.

Criminal offending. Permission was obtained from the Swed-ish legal authorities to collect information on of� ciallyregistered offences for all participants beginning at birth.Accordingly, of� cial records were available on the full range ofcriminal offences during their lifetime for the entire sample.Two age periods were considered in the present study:(1) Juvenile offences ˆ offences between the ages of 13 to17; and (2) Adult offences ˆ offences occurring between theages of 18 to 30. Persons who had registered offences at bothage periods were de� ned here as persistent criminals.

Criminal offences were limited to more serious offencecategories including: violent crimes against persons, propertydamage, crimes for personal gain, and drug/narcotic offences.2

For analytic purposes, criminality was treated either as adichotomous variable (i.e., offence or no offence as a juvenile;offence or no offence as an adult; persistent or nonpersistentoffender (offences as both a juvenile and as an adult)), or as acontinuous variable that was divided along a 5-point scale: nooffence ˆ 0, one offence ˆ 1, two or three offences ˆ 2, four to

INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2001, 25 (6), 509–520 511

2 The speci� c offences included in these categories are: violent crimes againstpersons (unlawful threat, threat to a public of� cial, assault, aggravated assault,molestation, violence to a public of� cial, breach of the peace, causing another’sdeath, violent resistance, unlawful coercion, sexual assault, indecent conduct,attempt to aid someone to escape, creating danger to another, outrageousconduct toward a public of� cial); property damage (trespassing, causing a � re,arson, false alarm, misuse of alarm, endangering of the public, in� ictingdamage); crimes for personal gain (unlawful possession of property, unlawful use,unlawful disposal, petty theft, theft, grand theft, unlawful taking of a vehicle,family theft, fraudulent conduct, fraud, grand fraud, petty receiving stolen goods,receiving stolen goods, smuggling of goods, tax offence, robbery, extortion,embezzlement); and drug/narcotic offences (petty offences against the Drug Act,offences against the Drug Act).

at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014jbd.sagepub.comDownloaded from

Page 5: Youth recreation centre participation and criminal offending: A 20-year longitudinal study of Swedish boys

512 MAHONEY, STATTIN, MAGNUSSON / YOUTH CENTRES AND CRIMINALITY

ten offences ˆ 3, eleven or more offences ˆ 4. The reader isdirected to Stattin and Magnusson (1991) and Stattin et al.(1986) for details on the measurement, coding, and frequencyof criminality in the IDA sample.

Youth recreation centre involvement. When the participantswere 13 years old they reported the frequency with which theyattended any of the youth recreation centres operating in thetown of Orebro. Centre participation was coded on a 5-pointscale ranging from no participation, to participation everyevening of the week. This scale was collapsed into a 3-pointindicator of the frequency of youth centre participation;namely, 0 ˆ no participation, 1 ˆ periodic participation(participation one to four times a month), and 2 ˆ frequentparticipation (participation two to seven times a week). The 3-point scale was used to ensure that a suf� cient number ofparticipants were represented in each frequency category forpurposes of statistical analyses.

It should be added that the number of youth recreationcentres in Orebro have varied over time. Typically, 18 to 20 areoperating in Orebro, and the location of the centres is spreadthroughout the neighbourhoods of the town.

Parental concerns for child behaviour. When the participantswere 10 years old their parents were asked to report their levelof concern for several of their child’s behaviours. A total of 486parents (99%) completed the child concerns items. Five itemswere included in this study: (1) watching television, (2) spend-ing money, (3) doing homework, (4) time to be home in theevening, and (5) making the child obey. Parent concern wascoded on a 5-point scale (1 ˆ no concern, to 5 ˆ strongconcern). The � ve items were averaged to create a parentalconcern scale (M ˆ 1.68, SD ˆ .58, Range ˆ 1–5, alphareliability ˆ .67).

Family socioeconomic status. A composite measure of familysocioeconomic status (SES) was developed using parent-reported information when the participants were 10 years ofage. Three items were included in the assessment: (1) familyincome, measured on an 8-point scale (M ˆ 4.99; Median ˆ5.00; SD ˆ 1.45); (2) parent education, measured on a 7-pointscale (M ˆ 1.91; Median ˆ 2.00; SD ˆ 1.61); and size of theliving environment, using an 8-point scale to show the numberof rooms in the home/apartment (M ˆ 4.34; Median ˆ 4.00;SD ˆ 1.36).3 Correlations between the three items rangedfrom .52 to .64, and were statistically signi� cant. The threeitems were standardised and averaged to create a measure ofSES (alpha reliability ˆ .86) which was available for 92% (460/498) of the sample.

Caregiver status. When the participants were 10 years old,parents identi� ed the person(s) primarily involved in raisingthe child. Eighty-nine per cent (444/498) of parents reportedthat both the mother and father raised the child, and 11% (52/498) reported that the mother or father raised the child alone.

These were typical patterns in Sweden in 1965. Caregiverstatus was therefore coded dichotomously (1 ˆ child raised bymother and father, 2 ˆ child raised by single parent). Twoparents reported an alternative rearing situation and were notconsidered in this investigation.

Statistical Analyses

Cluster analysis was performed using the Sleipner program forpattern-analysis (Bergman & El-Khouri, 1998). Three featuresof the program were used to identify the � nal cluster solutions:(1) a RESIDUE procedure that identi� es outliers showingextreme patterns across cluster variables; (2) a CLUSTERprocedure that identi� es initial con� gurations (Ward’s cluster-ing algorithm was employed); and (3) a RELOCATIONprocedure that reassigns persons to alternate clusters if doingso will reduce the error-sums-of-squares of the cluster solution.

The speci� c number of clusters chosen in this study wasdetermined on the basis of four criteria: (1) expectations ofcluster patterns identi� ed in previous analyses of the dataset;(2) the extent to which each additional cluster solutionprovided distinctly new and relevant patterns; (3) reductionin error sums of squares and increase in variance accounted forby each additional cluster solution; and (4) sample size (i.e., atleast 5% of the sample had to be represented in each resultantcon� guration). Variables were standardised prior to clustering.

Results

Youth recreation centre participation

Forty-one per cent (178/498) of boys participated in the youthrecreation centre at age 13. Twenty per cent (88/498) of theboys showed periodic involvement (one to four evenings amonth) and 21% (90/498) showed frequent involvement (twoto seven evenings a week).

Hierarchical regression analysis. To assess whether youthcentre participation was nonrandom, a hierarchical regressionanalysis was performed. In Step 1, the frequency of youthcentre participation was regressed on three indicators of parentbehaviour and demographic status when the child was age 10(i.e., parental concerns for child behaviour, caregiver status,SES). In Step 2, four indicators of child social-academicadjustment at age 10 (i.e., aggression, hyperactivity, achieve-ment, peer preference) were added to the model. Correlationsbetween the variables involved in this and subsequent analysesare shown in Table 1.

Results of the hierarchical regression are shown in Table 2.The measures of demographic status and parent behaviour inStep 1 were signi� cant indicators of youth centre participation.Low SES and high levels of parental concern for childbehaviour were signi� cantly related to more frequent youthcentre participation. In Step 2, the inclusion of the social-academic competence information signi� cantly improved themodel � t. However, given other variables in the model, none ofthe social-academic behaviours was signi� cantly related toyouth centre involvement in and of themselves.

Pattern-analysis. To complement the regression analysis,pattern-analysis was performed in order to determine ifindividual pro� les of social-academic adjustment at age 10

3 The mean income rating of 4.99 corresponds to an annual salary aroundSkr30,000 in 1965. The mean education rating of 1.91 corresponds tocompletion of the people’s high school in Sweden, 1965. Due to changes ineducational practice, economy, currency, in� ation, and overall standard of living,direct comparison of these statistics to the present goes beyond the scope of thisstudy. However, the � gures were fully representative of Swedish families duringthe period when the information was obtained.

at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014jbd.sagepub.comDownloaded from

Page 6: Youth recreation centre participation and criminal offending: A 20-year longitudinal study of Swedish boys

were particularly related to youth centre participation at age13. The pattern-oriented analysis identi� ed � ve clusters ofboys at the age of 10 (see Table 3). The resulting solutionsaccounted for 59.10% of the error sums of squares in thecluster variables (cf. Bergman, 1998). Con� guration 1 (HighCompetence) contains 94 boys (19% of the sample) withpositive adjustment in both social behaviour and academicachievement. Con� guration 2 (Aggressive Achievers) included119 boys (24% of the sample) with above average aggression,peer preference, and school achievement. Con� guration 3(Low Achievers and Low Behaviour Problems) included 86boys (17% of the sample) with below average academicachievement, aggression, and hyperactivity. Con� guration 4(Peer and Achievement Problems) involved 86 boys (17% ofthe sample) with poor academic achievement and very low peeracceptance. Finally, Con� guration 5 (Multiple Problems)contained 113 boys (23% of the sample) with severeadjustment problems on all four areas assessed.

The EXACON program (Bergman & El-Khouri, 1987) for

single cell contingency table analysis was used to assesswhether the observed number of youth centre participantsdiffered signi� cantly from expected values for the � ve con� g-urations. Patterns that are observed signi� cantly more oftenthan expected under a null hypothesis of stochastic indepen-dence between all patterns are referred to as ‘‘types’’, whereasthose less likely than expected under the same assumption aretermed ‘‘antitypes’’ (Bergman & El-Khouri, 1999; von Eye,1990).

Table 4 shows that persons in the high competencecon� guration were likely not to participate in the youth centres(a signi� cant type) and were unlikely to be frequent partici-pants (a signi� cant antitype). By contrast, persons in themultiple problems con� guration were underrepresentedamong persons who did not attend the youth centres (asigni� cant antitype) and were 1.6 times more likely thanexpected to become frequent participants of the youth centres(signi� cant type). Boys in the low achievement and lowbehaviour problems con� guration were 1.33 times more likely

INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2001, 25 (6), 509–520 513

Table 1Correlation matrix of parent aspects, child behaviour, youth centre attendance, and criminality

2 3 4 5 6 7 8 9

Parent aspects1. Child concerns .18*** ¡.17*** .19*** .26*** ¡.27*** ¡.09* .17*** .21***2. Marital status ¡.30*** .03 .04 ¡.03 .01 .07 .10*3. SES ¡.12* ¡.13** .27*** .06 ¡.20*** ¡.18***

Child behaviour4. Aggression .63*** ¡.29*** .26*** .31*** .19***5. Hyperactivity ¡.60*** ¡.35*** .30*** .22***6. Achievement .33*** ¡.21*** ¡.21***7. Peer preference ¡.17*** ¡.03

Other8. Youth centre attendance .25***9. Criminality

*p < .05; **p < .01; ***p < .001.

Table 2Hierarchical regression analysis of antecedents to youth recreation centre participation at age 13

Variables in the model B SE Standardised B t-value p-value

Step 1. Parent behaviour and demographic information at age 10Parent concern .25 .07 .18 5.88 .000Caregiver status .10 .15 .04 .72 .471Parent SES ¡.14 .05 ¡.15 ¡2.98 .003

Model F(3, 392) ˆ 9.64, p < .001; R2 ˆ .07

Step 2. Add social and academic adjustment at age 10Parent concern .15 .07 .11 2.09 .038Caregiver status .16 .14 .05 1.08 .281Parent SES ¡.12 .05 ¡.12 ¡2.32 .021

Aggression .05 .05 .06 1.02 .309Hyperactivity .10 .07 .12 1.48 .148Achievement .07 .06 ¡.09 ¡1.32 .188Peer preference .05 .04 .06 1.17 .241

Model F(7, 388) ˆ 6.75, p < .001; R2 ˆ .11

at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014jbd.sagepub.comDownloaded from

Page 7: Youth recreation centre participation and criminal offending: A 20-year longitudinal study of Swedish boys

514 MAHONEY, STATTIN, MAGNUSSON / YOUTH CENTRES AND CRIMINALITY

than expected to be frequent participants of the youth centreand this approached a signi� cant type (p ˆ .07). Rates of youthcentres participation did not differ signi� cantly from expectedvalues for the remaining con� gurations.

To evaluate whether the differential participation rates ofthe � ve con� gurations remained signi� cant after accountingfor demographic and parenting aspects, a univariate analysisof covariance (ANCOVA) was performed. The mean fre-quency of participation across the � ve con� gurations wascompared after the in� uence of parental concerns for childbehaviour, caregiver status, and SES were entered into themodel as covariates. Two of the covariates were signi� cant:parental concerns, F(1, 388) ˆ 6.31, p < .05, and SES,F(1, 388) ˆ 6.52, p ˆ .01. The main effect of con� gurationwas also signi� cant, F(4, 388) ˆ 5.29, p < .001; eta-squaredˆ .052. A post-hoc one-way analysis of variance (ANOVA)using Scheffe ’s method con� rmed that boys in the multiple

problem con� guration participated in the youth centressigni� cantly more often than did boys in the remaining fourcon� gurations.

Youth recreation centre participation and criminaloffending

Overall, 30.5% (152/498) of boys were registered by the policefor one or several offences by the age of 30.

Survival analysis. To assess whether the onset of criminaloffending differed as a function of frequency of youth centreparticipation, survival analysis was employed using theKaplan–Meier product-limit method (e.g., Kaplan & Meier,1958). To retain maximum precision in the timing of criminaloffences, each respondent’s age at the time of their � rstregistered offence was recorded in hundredths of years.

Table 3Person-oriented configurations at age 10

Means and (standard deviations) of cluster variables(Standardised scores)

AGG HYP ACH PEERCon� guration Description N M (SD) M (SD) M (SD) M (SD)

C1 High competence 94 ¡0.92 (0.74) ¡1.32 (0.47) 1.28 (0.57) 0.64 (0.82)C2 Aggressive achievers 119 0.43 (0.55) 0.11 (0.58) 0.43 (0.59) 0.49 (0.69)C3 Low achievers and

low behaviour problems86 ¡0.88 (0.50) ¡0.44 (0.65) ¡0.42 (0.66) 0.42 (0.67)

C4 Peer and achievement problems 86 ¡0.11 (0.63) 0.11 (0.49) ¡0.41 (0.71) ¡1.21 (0.67)C5 Multiple problems 113 1.07 (0.70) 1.23 (0.51) ¡0.88 (0.69) ¡0.45 (0.81)

AGG, Aggression; HYP, Hyperactivity; ACH, Achievement; PEER, Peer Preference.

Table 4Number of persons in each age 10 configuration who participated in the youth recreation centre at age 13

Con� guration Frequency of youth centre Ratio Hypergeometricat age 10 (N) participation at age 13 Observed Expected Observed/Expected probability

High (83) None 58 48.07 1.21 .009 Tcompetence Periodic 18 17.27 1.04 .465

Frequent 7 17.66 .40 .000 A

Aggressive (97) None 62 56.18 1.07 .106achievers Periodic 18 20.18 .89 .320

Frequent 17 20.64 .82 .188

Low achievers (74) None 40 42.86 .93 .270and low behaviour problems Periodic 13 15.39 .84 .300

Frequent 21 15.74 1.33 .071

Achievement (69) None 45 39.96 1.13 .112and peer problems Periodic 13 14.35 .91 .399

Frequent 11 14.68 .75 .153

Multiple (100) None 40 57.92 .69 .000 Aproblems Periodic 26 20.80 1.25 .094

Frequent 34 21.28 1.60 .000 T

T, signi� cant type; A, signi� cant antitype.

at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014jbd.sagepub.comDownloaded from

Page 8: Youth recreation centre participation and criminal offending: A 20-year longitudinal study of Swedish boys

Figure 1 summarises onset of offending for boys who did notparticipate in the youth centre, were periodic participants, orwere frequent participants.

The cumulative onset of criminal offending differedsigni� cantly across the three groups of boys, Log Rank (2) ˆ37.00, p < .001. Boys who participated periodically orfrequently in the youth recreation centres at age 13 had agreater cumulative offence frequency than boys who did notparticipate in the youth centres. The median age of � rst offenceacross the three groups was similar (nonparticipants ˆ 17.46;periodic participants ˆ 17.63; frequent participants ˆ 17.50).This indicated that youth centre participants were more likelyto have juvenile and adult offending onset compared tononparticipants. Accordingly, it was of interest to explorejuvenile and long-term criminal offending across the threegroups.

Juvenile and persistent offending. To assess whether thefrequency of youth centre participation was related to patternsof juvenile and adult offending, offence information wasdivided into three categories: (1) no offending (no registeredoffence as a juvenile or adult), (2) juvenile offending only, and(3) persistent offending (registered offences both as a juvenileand adult). The number of participants in the three categorieswas compared to the frequency of their participation in theyouth centres using the EXACON procedure for single-cellcontingency table analysis. Results are shown in Table 5.

Nonparticipants of the youth centres were unlikely to bejuvenile offenders or persistent offenders (signi� cant anti-

types). For periodic youth centre participants, persistentcriminal offending occurred signi� cantly more often thanexpected (a signi� cant type). Frequent participation in theyouth centres was associated with high rates of juvenileoffending and persistent offending that were over twice theexpected value (both were signi� cant types). Among frequentyouth centre participants 44% were registered for some offenceas juveniles, compared to 22% of periodic participants, and12% of nonparticipants. The corresponding percentages forpersistent criminal offending were 22% for frequent partici-pants, 17% for periodic participants, and 4% for nonpartici-pants. Accordingly, participation in the youth centres waslinked to criminal offending that initiated in adolescence and,in many cases, persisted into adulthood. The associatedincrease in juvenile offending was particularly marked forfrequent participants of the youth centres.

Hierarchical logistic regression analysis. In an evaluation ofwhether the association between frequencies of youth recrea-tion centre participation and criminal offending was linked toindividual characteristics at age 10, a hierarchical logisticregression was performed. In Step 1, a dichotomous indicatorof criminal offending (i.e., registered for some offence or not)was regressed on three indicators of parent behaviour anddemographic status when the child was age 10 (i.e., parentalconcerns about child behaviour, caregiver status, SES). InStep 2, the four indicators of child social-academic compe-tence at age 10 (i.e., aggression, hyperactivity, achievement,peer preference) were added to the model. Finally, Step 3

INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2001, 25 (6), 509–520 515

Figure 1. Onset of first arrest as a function of frequency of youth centre participation. at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014jbd.sagepub.comDownloaded from

Page 9: Youth recreation centre participation and criminal offending: A 20-year longitudinal study of Swedish boys

516 MAHONEY, STATTIN, MAGNUSSON / YOUTH CENTRES AND CRIMINALITY

introduced the trichotomous indicator of the frequency ofparticipation in the youth centres at age 13. Table 6 shows theresults of the logistic regression.

Parental concerns for child behaviour and SES weresigni� cantly associated with criminality at Step 1. At Step 2,the inclusion of the social-academic competence indicatorssigni� cantly improved the model � t. Given the other variablesin the model, high levels of hyperactivity, parental concerns,

and SES were signi� cantly linked to later offences. Finally, inStep 3, adding information on youth recreation centreparticipation at age 13 also improved the model � t signi� -cantly. High levels of hyperactivity, parental concerns for childbehaviour, and low SES continued to show signi� cant relationsto criminality; however, youth centre participation was thesingle best indicator of subsequent criminality in this model.The logistic regression was repeated separately for juvenile

Table 5Patterns of criminal offending as a function of frequency of youth centre participation at age 13

Frequency ofyouth centre Criminal Ratio Hypergeometricparticipation (N) offending Observed Expected Observed/Expected probability

None (227) None 199 180.75 1.10 .000 TJuvenile offending 17 22.51 0.76 .039 APersistent offending 11 23.73 0.46 .000 A

Periodic (72) None 56 57.33 0.98 .387Juvenile offending 4 7.14 0.56 .120Persistent offending 12 7.52 1.60 .049 T

Frequent (74) None 42 58.92 0.71 .000 AJuvenile offending 16 7.34 2.20 .000 TPersistent offending 16 7.73 2.07 .001 T

T, signi� cant type; A, signi� cant antitype.

Table 6Hierarchical logistic regression of youth recreation centre participation and criminal offending to age 30

Variables in the model B SE R Odds p-valueratio

Step 1. Parent behaviour and demographic information at age 10Parent concern .891 .204 .188 2.44 .000Caregiver status ¡.087 .397 .000 1.00 1.00Parent SES ¡.472 .558 ¡.120 0.62 .003

Model x2 (3,396) ˆ 35.80, p < .001

Step 2. Add social and academic adjustment at age 10Parent concern .621 .217 .117 1.86 .004Caregiver status ¡.087 .410 .000 0.92 1.00Parent SES ¡.470 .169 ¡.114 0.62 .005

Aggression .226 .164 .000 1.25 .169Hyperactivity .547 .208 .104 1.73 .008Achievement .064 .171 .000 1.06 .707Peer preference ¡.102 .133 .000 0.90 .442

Improvement in Model x2 (4, 396) ˆ 30.79, p < .001

Step 3. Add youth recreation centre participation at age 13Parent concern .551 .224 .099 1.73 .014Caregiver status ¡.010 .417 .000 0.99 .981Parent SES ¡.407 .173 ¡.091 0.67 .018

Aggression .194 .169 .000 1.21 .252Hyperactivity .492 .213 .089 1.64 .021Achievement .108 .176 .000 1.11 .538Peer preference ¡.142 .138 .000 0.87 .304

Youth centre participation .571 .150 .174 1.77 .000

Improvement in Model x2 (1, 396) ˆ 14.62, p < .001

at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014jbd.sagepub.comDownloaded from

Page 10: Youth recreation centre participation and criminal offending: A 20-year longitudinal study of Swedish boys

offences and persistent offences, and the � ndings weresubstantially similar to what is reported earlier. Additionally,a continuous measure of criminal offending was employedusing a logarithmic transformation of offence frequency.Standard hierarchical regression analysis using this transforma-tion again showed a pattern of results consistent to the logisticregression analysis reported earlier.

Pattern-analysis. To complement the logistic regression,offence rates were compared across the � ve con� gurationsidenti� ed at age 10 using the 5-point scale of offence frequencydescribed earlier. The mean offence frequencies ranged from.19 in the high competence con� guration to 1.12 among boysin the multiple problems con� guration (see Figure 2) and thedifference between con� gurations was signi� cant, F(1, 493) ˆ11.46, p < .001.

An ANCOVA was performed to assess whether the amountof criminal offending differed across the con� gurations as afunction of youth centre participation. Family level informa-tion and SES were � rst entered as covariates in the model, andmain effects and the interaction between con� guration andparticipation in youth centres were then tested. The results areshown in Figure 3. Family SES was a signi� cant covariate tocriminal offending, F(1, 378) ˆ 8.61, p < .01. Main effects forcon� guration, F(4, 378) ˆ 8.08, p < .001; eta-squared ˆ .08,and participation in the youth centre were signi� cant,F(2, 378) ˆ 6.23, p < .01; eta-squared ˆ .03). The interactionbetween con� guration and participation in the youth centreswas not signi� cant. Post-hoc one-way ANOVAs using Scheffe ’smethod were performed for con� guration and frequency of

youth centre participation. Boys in the multiple problemscon� guration had signi� cantly higher rates of criminal offend-ing than boys in the remaining four con� gurations. Boys whowere periodic or frequent participants of the youth centre hadsigni� cantly higher criminal offending than boys not attendingthe youth centres. Accordingly, criminal offending was high foryouth centre participants regardless of their pro� le of social-academic competence at age 10.

Discussion

In this study we argue that whether leisure activities in� uenceadjustment in a positive or negative fashion depends on severalfeatures of the individual and the activity pursued. Inparticular, when the activities maintain low structure, little orno adult supervision, few opportunities for skill building, andwhere the participants are largely composed of antisocialyouth, the potential for these activities to reduce the risk ofdeveloping antisocial behaviour is limited. The Swedish youthrecreation centres, which are available to adolescents at the ageof 13 and older, are commonly low in structure and are notexplicitly directed at individual skill building. Moreover,previous research shows that those adolescents who attendregularly are, as a group, more antisocial, have more antisocialpeers, have more con� icted parent relations, and have parentswho are less effective monitors of their behaviour, thanadolescents not choosing to attend the centres (Mahoney &Stattin, 2000).

INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2001, 25 (6), 509–520 517

Figure 2. Frequency of criminal offending (age 13 to 30) as a function of adjustment at age 10. at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014jbd.sagepub.comDownloaded from

Page 11: Youth recreation centre participation and criminal offending: A 20-year longitudinal study of Swedish boys

518 MAHONEY, STATTIN, MAGNUSSON / YOUTH CENTRES AND CRIMINALITY

Against this background, we examined the associated long-term patterns of criminality for adolescents who attendedyouth recreation centres at the age of 13 in one medium-sizedSwedish town. The two questions addressed were: who choseto attend these centres, and what was the likelihood of futurecriminality for participants compared to nonparticipants? Thestudy design involved a 20-year longitudinal investigation. Thesample was representative of Swedish youth in terms ofdelinquency and other characteristics (Stattin et al., 1986),there was a breadth of assessments with low attrition rate, andcomplete information existed on registered criminality. Thegeneral � ndings were that boys with a multiple problem pro� leof social and academic de� cits at the age of 10 were most likelyto attend the youth centres at age 13 and that youth whoparticipated in centres at the age of 13 had signi� cantly higherrates of future criminal offences than similar youth who did notparticipate at the centres.

To understand better which adolescents chose to attend theyouth centres, we compared relevant individual and familycharacteristics at the age of 10 for those who seldom or never,periodically, or frequently attended the centres at the age of 13.Our results show that the decision to participate in the youthcentres was nonrandom. The regression analyses revealed thatchildren from low SES families whose parents reported highconcerns about their social behaviour were likely to participatein the youth centres at the age of 13. The pattern-orientedanalysis showed that children with a pro� le of social andacademic adjustment problems at the age of 10 had the mostfrequent participation in the youth centres at the age of 13.Thus, adolescents with poor social and/or academic adjust-

ment and those coming from disadvantaged homes wereoverrepresented at the youth centres.

Participation in the youth centres, whether periodic orfrequent, was associated with an increased risk for criminality,including high juvenile crime and persistent offending (i.e.,registered for one or several offences both as a juvenile and as anadult). Frequent attendance was particularly linked to highrates of juvenile offending. But, given the evidence for aselection effect with respect to attending these centres, onequestion was whether the increased risk for future criminalactivity prevailed after controlling for individual and familyfactors at the age of 10. The regression analysis showed that theassociated increase in crime held after controlling for family anddemographic factors and several aspects of child social-academic competence prior to involvement in the centres.Indeed, youth centre participation was the single best indicatorof criminal offending in a model that included child aggression,hyperactivity, school achievement, peer preference, SES,marital status, and parental concerns for child behaviour. Thepattern-analysis further revealed that although youth withmultiple problems pro� les were most likely to be registeredfor some offence, the associated increase in criminality wasproportional for all � ve con� gurations of boys. Accordingly, theassociated negative outcomes linked to youth centre participa-tion were apparent regardless of the individual’s social-academic competence prior to attending the youth centre.

The issue of the mechanisms by which attending youthrecreation centres is associated with a higher risk for futuredelinquency is an important one. Multiple problem youth wereover-represented among youth who decided to participate in

Figure 3. Frequency of criminal offending (ages 13 to 30) as a function of adjustment at age 10 and youth centre participation.

at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014jbd.sagepub.comDownloaded from

Page 12: Youth recreation centre participation and criminal offending: A 20-year longitudinal study of Swedish boys

centres and the decision to attend was linked to high criminaloffending prospectively. These � ndings are consistent with theproposal that the combination of unstructured leisure andsocialisation in� uences among problem-oriented peers maypromote antisocial behaviour (e.g., Cairns et al., 1988; Dishionet al., 1999; Mahoney & Stattin, 2000). However, otherexplanations are possible. For instance, youth centre involve-ment could be indicative of a more general preference forunstructured leisure pursuits. This possibility could not beclari� ed in the present investigation. Indeed, the level ofstructure in the individual youth centres was not directlyassessed, and other aspects of the participants’ leisure timewere not considered. These are unanswered questions, and,accordingly, this study should be used as a basis for conductingmore detailed evaluations of the processes by which someleisure activities are associated with criminal behaviour.

The � ndings we present are subject to misinterpretation.First, this is primarily a descriptive study and the results arestrictly correlational. The study did not follow an experimentaldesign and causal interpretations of youth centre participationand crime cannot be made on the basis of these data. This istrue despite the years of longitudinal study involved and thefact that a number of demographic, family, and individualfactors were accounted for in the investigation. Although it canbe concluded that the decision to participate in the youthcentres was signi� cantly related to juvenile and persistentcriminality, we cannot conclude that attendance at the youthcentres was solely or directly responsible for the increasedoffending. Unmeasured in� uences could account both foryouth centre attendance and increased criminality.

Second, the historical context of the present study must beconsidered. The longitudinal investigation began in 1965 andit is an empirical question whether Swedish youth recreationcentres are associated with high long-term criminality atpresent. The policies and practices of the youth centres,however, have not changed substantially over the past 35 years.In addition, a recent cross-sectional investigation found thatyouth centre involvement continues to be associated with highantisocial behaviour (Mahoney & Stattin, 2000). Accordingly,there are reasonable grounds to expect that the long-termadjustment of youth presently attending the centres would beconsistent with the � ndings of this investigation.

Third, the descriptive nature of this study providesinformation on the overall or average associated in� uence ofyouth centre attendance and crime. Nonetheless, there weremultiple centres operating in the town (18 to 20) from whichthe sample was drawn. Some of these centres may haveoperated differently, and it is possible that the associatedcriminality may have been higher or lower depending on thespeci� c youth centre attended. The present data did not permitus to evaluate this proposal. However, North American studieshave shown that youth centres can operate in highly effectiveand potentially bene� cial ways for disadvantaged youth (Heath& McLaughlin, 1993; McLaughlin, 1994). Elsewhere (Maho-ney & Stattin, 2000), we have reviewed the general correlatesof leisure activities showing a positive association to individualadjustment. They concerned, for example, activities that arevoluntary, rule-guided, focus on increasingly complex skill-building, are led by a suf� cient number of competent adults,provide clear feedback on performance, and follow regularparticipation schedules. We suggest that these serve as guide-lines for a process-oriented assessment of the Swedish youthrecreation centres.

We believe that this study has implications for social policy.The Swedish youth centres represent a leisure context that issupported politically and � nancially by the national govern-ment, and they are thought to promote positive adjustment forthe participating youth. However, our � ndings suggest thatregional youth centres explicitly aimed to provide adolescentswith a meaningful leisure time should not, by default, beviewed as bene� cial. We suggest that an evaluation componentshould be built into the structure of the centres. Note that wedo not suggest that our � ndings provide evidence that theyouth centres should be closed. It is open to speculation whatmay have happened if youth centre participants had chosen notto attend the recreation centres, or if the youth centres werenot available.

An experimental manipulation of the social context andlevel of structure at the youth centres would be one way toevaluate the participation-crime relation more rigorously.Quasi-experimental alternatives might also be explored. Forexample, archival data could be used to assess criminality inSwedish communities prior to and following the introductionof the youth centres. Additionally, communities that do nothave such centres could be compared to those that do withrespect to criminal offending. The present dataset did notpermit either alternative to be evaluated, but we believe suchstudies are logical next steps in the research.

In summary, the results suggest that we cannot take forgranted that well-intentioned public funds to provide youthwith meaningful leisure experiences will inevitably be bene-� cial. The critical questions for future research are to specifywhat aspects of leisure activities are bene� cial, the conditionsunder which leisure activities may be problematic, and theways in which peer groups can be effectively organised duringleisure time to facilitate positive social adjustment.

Manuscript received May 1999Revised manuscript received September 1999

References

Agnew, R., & Peterson, D.M. (1989). Leisure and delinquency. Social Problems,36, 332–350.

Allen, J.P., Philliber, S., Herrling, S., & Gabriel, K.P. (1997). Preventing teenpregnancy and academic failure: Experimental evaluation of a developmen-tally based approach. Child Development, 64, 729–742.

Andersson, T. (2000). Developmental patterns and the dynamics of alcoholproblems in adolescence and young adulthood. In L.R. Bergman, R.B.Cairns, L.-G. Nilsson, & L. Nystedt (Eds.), Developmental science and theholistic approach (pp. 377–391). Mahwah, NJ: Erlbaum.

Backteman, G. & Magnusson, D. (1981). Longitudina l stability of personalitycharacteristics. Journal of Personality, 49, 148–160.

Bergman, L.R. (1998). A pattern-oriente d approach to studying individualdevelopment: Snapshots and processes. In R.B. Cairns, L.R. Bergman, &J. Kagan (Eds.), Methods and models for studying the individual (pp. 83–121).Thousand Oaks, CA: Sage.

Bergman, L.R., & El-Khouri, B.M. (1987). EXACON: A fortran 77 program forthe exact analysis of single cells in a contingency table. Educational andPsychological Measurement, 47, 155–161.

Bergman, L.R., & El-Khouri, B.M. (1998). Sleipner: A statistical package forpattern-oriented analysis. Version 2.0. Stockholm: Department of Psychology,Stockholm University, Sweden.

Bergman, L.R., & El-Khoui, B.M. (1999). Studying individual patterns ofdevelopment using I-states as objects analysis (ISOA). Biometrical Journal, 41,753–770.

Bergman, L.R., & Magnusson, D. (1991). Stability and change in patterns ofextrinsic adjustment. In D. Magnusson, L.R. Bergman, G. Rudinger, &B. Torestad (Eds.), Problems and methods in longitudinal research (pp. 323–326). Cambridge, UK: Cambridge University Press.

Brown, B.B., Kohrs, D., & Lazarro, C. (1991). The academic costs andconsequences of extracurricular participation in high school. Paper presented at

INTERNATIONAL JOURNAL OF BEHAVIORAL DEVELOPMENT, 2001, 25 (6), 509–520 519

at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014jbd.sagepub.comDownloaded from

Page 13: Youth recreation centre participation and criminal offending: A 20-year longitudinal study of Swedish boys

520 MAHONEY, STATTIN, MAGNUSSON / YOUTH CENTRES AND CRIMINALITY

the annual meeting of the American Educational Research Association,Chicago.

Cairns, R.B., Bergman, L.R., & Kagan, J. (Eds.) (1998). Methods and models fromstudying the individual: Essays in honor of Marian Radke-Yarrow. ThousandOaks, CA: Sage.

Cairns, R.B., & Cairns, B.D. (1994). Lifelines and risks: Pathways of youth in ourtime. New York: Cambridge University Press.

Cairns, R.B., Cairns, B.D., & Neckerman, H.J. (1989). Early school drop out:Determinants and con� gurations. Child Development, 60, 1437–1452.

Cairns, R.B., Cairns, B.D., Neckerman, H.J., Gest, S.D., & Garie py, J-L.(1988). Social networks and aggressive behavior: Peer support or peerrejection? Developmental Psychology, 24, 815–823.

Coie, J.D., Terry, R., Lenox, K.F., Lochman, J.E., & Hyman, C. (1995).Childhood peer rejection and aggression as predictors of stable patterns ofadolescent disorder. Development and Psychopathology, 7, 697–713.

Csikszentmihalyi, M. (1990). Flow: The psychology of optimal experience. NewYork: Harper & Row.

Dishion, T.J., McCord, J., & Poulin, F. (1999). When interventions harm: Peergroups and problem behavior. American Psychologist, 54, 755–764.

Eccles, J.L., & Barber, B.L. (1995, May). Adolescents’ activity involvement:Predictors and longitudinal consequences . In D. Mekos & G.H. Elder, Jr.(Chairs), Developmental perspectives on adolescent extracurricular involvement.Symposium conducted at the Biennial Meeting of the Society for Research inChild Development, Indianapolis, IN.

Eder, D., Evans, C., & Parker, S. (1995). School talk: Gender and adolescentculture. New Brunswick, NJ: Rutgers University Press.

Elliott, D., Huizanga, D., & Ageton, S. (1985). Explaining delinquency and druguse. Beverly Hills, CA: Sage.

Hirschi, T. (1969). Causes of delinquency. Berkeley, CA: University of CaliforniaPress.

Heath, S.B., & McLaughlin, M.W. (1993). Identity and inner-city youth: Beyondethnicity and gender. New York: Teacher’s College Press.

Hultzman, W.Z. (1992). Constraints to activity participation in early adoles-cence. Journal of Early Adolescence, 12, 280–299.

Hultzman, W.Z. (1995). The in� uence of others as a barrier to recreationparticipation among early adolescents . Journal of Leisure Research, 25, 150–164.

Jones, M.B., & Offord, D.R. (1989). Reduction of antisocial behavior in poorchildren by nonschool skill-development . Journal of Child Psychology andPsychiatry, 30, 737–750.

Kaplan, E.L., & Meier, P. (1958). Nonparametric estimation for incompleteobservations. Journal of the American Statistical Association, 53, 457–481.

Kinney, D.A. (1993). From nerds to normals: The recovery of identity amongadolescents from middle school to high school. Sociology of Education, 66, 21–40.

Magnusson, D. (1988). Individual development from an interactional perspective: Alongitudinal study. Hillsdale, NJ: Erlbaum.

Magusson, D. (1996). The patterning of antisocial behavior and autonomicreactivity. In D.M. Stoff & R.B. Cairns (Eds.), The neurobiology of clinicalaggression (pp. 291–308). Hillsdale: NJ: Erlbaum.

Magnusson, D., & Bergman, L.R. (1988). Individual and variable-basedapproaches to longitudina l research on early risk factors. In M. Rutter(Ed.), Studies of psychosocial risk: The power of longitudinal data (pp. 45–61).Cambridge, UK: Cambridge University Press.

Magnusson, D., & Bergman, L.R. (1990). A pattern approach to the study ofpathways from childhood to adulthood. In L.N. Robbins & M. Rutter (Eds.),Straight and devious pathways from childhood to adulthood (pp. 101–115).Cambridge, UK: Cambridge University Press.

Magnusson, D., & Bergman, L.R. (1998). Individual development andadaptation: The IDA program. Reports from the Department of Psychology,841, 1–17. Stockholm University, Stockholm, Sweden.

Magnusson, D., Dune r, A., & Zetterblom, G. (1975). Adjustment: A longitudinalstudy. New York: Wiley.

Magnusson, D., & Stattin, H. (1998). Person-context interaction theories. InW. Damon (Series Ed.) and R.M. Lerner (Vol. Ed.), Handbook of childpsychology: Vol. 1. Theoretical models of human development (5th ed., pp. 685–760). New York: Wiley.

Magnusson, D., Stattin, H., & Dune r, A. (1983). Aggression and criminality in alongitudina l perspective. In K. Teilman van Dusen & S.A. Mednick (Eds.),Prospective studies of crime and delinquency (pp. 277–301). Boston: Kluwer-Nijhoff.

Mahoney, J.L. (2000). Participation in school extracurricula r activities as amoderator in the development of antisocial patterns. Child Development, 71,502–516.

Mahoney, J.L., & Cairns, R.B. (1997). Do extracurricula r activities protectagainst early school dropout? Developmental Psychology, 33, 241–253.

Mahoney, J.L. & Stattin, H. (2000). Leisure time activities and adolescent anti-social behavior: The role of structure and social context. Journal of Adolescence,23, 113–127.

Masten, A.S., & Coatsworth, J.D. (1998). The development of competencein favorable and unfavorable environments. American Psychologist. 53, 205–220.

McCord, J. (1978). A 30-year follow-up of treatment effects. AmericanPsychologist, 33, 284–289.

McLaughlin, M.W. (1994). Urban sanctuaries: Neighborhood organizations in thelives and futures of inner-city youth. San Francisco, CA: Jossey-Bass.

McNeal, R.B. (1998). High school extracurricul a activities: Closed structuresand stratifying patterns of participation. The Journal of Educational Research,91, 183–191.

Osgood, D.W., Wilson, J.K., O’Malley, P.M., Bachman, J.G., & Johnston, L.D.(1996). Routine activities and individual deviant behavior. American Socio-logical Review, 61, 635–655.

Patterson, G.R. (1993). Orderly change in a stable world: The antisocial trait aschinera. Journal of Consulting and Clinical Psychology, 61, 911–919.

Patterson, G.R., Reid, J.B., & Dishion, T.J. (1992). A social learning approach:Vol. 4. Antisocial boys. Eugene, OR: Castalia.

Stattin, H., & Magnusson, D. (1991). Stability and change in criminal behaviourup to age 30. The British Journal of Criminology, 31, 327–346.

Stattin, H., Magnusson, D., & Reichel, H. (1986). Criminality from childhood toadulthood (Report No. 63). University of Stockholm, Individual andDevelopment and Adjustment Program.

Torestad, B., & Magnusson, D. (1996). Basic skills, early problematic behaviourand social maladjustment. Educational Studies, 22, 165–176.

von Eye, A. (1990). Introduction to con� gural frequency analysis: The search for typesand antitypes in cross-classi� cations. Cambridge, UK: Cambridge UniversityPress.

at MEMORIAL UNIV OF NEWFOUNDLAND on July 7, 2014jbd.sagepub.comDownloaded from