The Influence of Families on Early Adolescent School Connectedness: Evidence That This Association Varies with Adolescent Involvement in Peer Drinking Networks Adrian B. Kelly, Martin O’Flaherty, John W. Toumbourou, Ross Homel, George C. Patton, Angela White & Joanne Williams Journal of Abnormal Child Psychology An official publication of the International Society for Research in Child and Adolescent Psychopathology ISSN 0091-0627 Volume 40 Number 3 J Abnorm Child Psychol (2012) 40:437-447 DOI 10.1007/s10802-011-9577-4
1 23
Your article is protected by copyright and all rights are held exclusively by Springer Science+Business Media, LLC. This e-offprint is for personal use only and shall not be selfarchived in electronic repositories. If you wish to self-archive your work, please use the accepted author’s version for posting to your own website or your institution’s repository. You may further deposit the accepted author’s version on a funder’s repository at a funder’s request, provided it is not made publicly available until 12 months after publication.
1 23
Author's personal copy J Abnorm Child Psychol (2012) 40:437–447 DOI 10.1007/s10802-011-9577-4
The Influence of Families on Early Adolescent School Connectedness: Evidence That This Association Varies with Adolescent Involvement in Peer Drinking Networks Adrian B. Kelly & Martin O’Flaherty & John W. Toumbourou & Ross Homel & George C. Patton & Angela White & Joanne Williams
Published online: 12 October 2011 # Springer Science+Business Media, LLC 2011
Abstract School connectedness is central to the long term well-being of adolescents, and high quality parent–child relationships facilitate school connectedness. This study examined the extent to which family relationship quality is associated with the school connectedness of pre- and early teenagers, and how this association varies with adolescent involvement in peer drinking networks. The sample consisted of 7,372 10–14 year olds recruited from 231 schools in 30 Australian communities. Participants completed the
A. B. Kelly (*) : M. O’Flaherty : A. White Centre for Youth Substance Abuse Research, The University of Queensland, K-Floor, Mental Health Centre, Royal Brisbane and Womens’ Hospital, Brisbane, Queensland, Australia 4029 e-mail:
[email protected] R. Homel Key Centre for Ethics, Law, Justice and Governance, Griffith University, Brisbane, Australia J. W. Toumbourou Deakin University, Melbourne, Australia J. W. Toumbourou : G. C. Patton : J. Williams Centre for Adolescent Health, Melbourne, Australia J. W. Toumbourou : G. C. Patton : J. Williams Murdoch Children’s Research Institute, Melbourne, Australia G. C. Patton The University of Melbourne, Melbourne, Australia
Communities that Care youth survey. A multi-level model of school connectedness was used, with a random term for school-level variation. Key independent variables included family relationship quality, peer drinking networks, and school grade. Control variables included child gender, sensation seeking, depression, child alcohol use, parent education, and language spoken at home. For grade 6 students, the association of family relationship quality and school connectedness was lower when peer drinking networks were present, and this effect was nonsignificant for older (grade 8) students. Post hoc analyses indicated that the effect for family relationship quality on school connectedness was nonsignificant when adolescents in grade 6 reported that the majority of friends consumed alcohol. The results point to the importance of familyschool partnerships in early intervention and prevention. Keywords Adolescents . Early adolescents . School connectedness . Alcohol . Parents . Family relationship quality . Peers
Introduction School connectedness (student engagement with and a sense of belonging to school; McNeely and Falci 2004) predicts academic competence, educational attainment, job satisfaction, and psychological health (Veronneau et al. 2010; Pinquart et al. 2003; Shochet et al. 2008; Li et al. 2010). Low school connectedness is probably common in Western countries, given the high prevalence rates for truancy and related markers of low school connectedness (Moon and Ando 2009; Bonny et al. 2000). In the United States, national surveys indicate that 11% of 8th grade
Author's personal copy 438
students and 16% of 10th grade students report one or more truant days in the previous month (Henry 2007). In New Zealand, available data suggests that by the age of 16 years, just under 40% of students report at least one act of truancy, and 7.1% report truanting more than 30 times between 14 and 16 years of age (Fergusson et al. 1995). Understanding factors that are systematically related to school connectedness has potentially important implications for the prevention of poor educational, health, and occupational outcomes. The Social Developmental Model (Catalano et al. 1996) provides a useful theoretical framework for understanding psychosocial factors that influence school connectedness. This model emphasizes the importance of bonds with parents (Hirschi 1971; Catalano et al. 1996), the reciprocal influences of parenting on child behaviour and child behaviour on parenting (Patterson and Stouthamer-Loeber 1984), and the learning processes that occur when close personal relationships are formed with “pro-deviant peers” (Matsueda and Heimer 1987). Consistent with the Social Developmental Model, high family relationship quality has a direct and positive effect on school connectedness and academic performance (Steinberg et al. 1994; Heaven and Ciarrochi 2008; Shochet et al. 2007; Simons-Morton and Chen 2009; Steinberg et al. 1992). These findings hold for younger and older adolescents (Steinberg et al. 1992), and across nations (Vazsonyi et al. 2003). Also, a limited body of research is consistent with the reciprocal influences of parenting on behaviours known to erode school connectedness (notably externalising behaviours and substance abuse) (Huh et al. 2006; Stice and Barrera 1995). Finally, school connectedness is eroded when adolescents engage with delinquent peer networks, in particular those where alcohol and other drug use is common (Whitney et al. 2010; King et al. 2006; Lynskey and Hall 2000; Veronneau et al. 2010; Henry et al. 2009; Simons-Morton and Chen 2009). Collectively these studies demonstrate that family relationship quality is positively related to school connectedness, and that involvement in peer networks where alcohol and other drug use is common is negatively related to school connectedness. Very little research has examined how family relationship quality and peer drinking networks together account for school connectedness. Prior research has found that parental stress associated with rearing adolescents erodes family relationship quality (Seginer et al. 2002) and for adolescents with elevated sensation seeking and rebelliousness, the protective influence of family relationship quality on substance use is eroded (Cleveland et al. 2010; King and Chassin 2004). The results of these studies are consistent with a “protective but reactive” effect (Luthar et al. 2000), where high family
J Abnorm Child Psychol (2012) 40:437–447
relationship quality has a protective effect (associated with reduced substance use) that dissipates in the presence of high levels of individual risk. Also, low family relationship quality may indirectly erode school connectedness by increasing the likelihood of adolescents engaging with deviant peer groups. In the literature on alcohol and other drug use, parent/family factors contribute to adolescents’ engagement with and socialization by high-risk peer groups, both of which strongly predict adolescent alcohol and other drug use (Abar and Turrisi 2008; Martino et al. 2009; Bahr et al. 2005; Capaldi et al. 2009). While mediation and moderation effects have been demonstrated in models that predict adolescent alcohol and other drug use, they have not been applied to research on the interrelationships of family relationship quality, peer drinking networks, and school connectedness. This study is a first step in this direction—The study tests the cross-sectional association of these three variables. We focused on alcohol use by peers because alcohol is prevalent, early alcohol use longitudinally predicts a range of negative outcomes, and alcohol use predicts experimentation with other drugs (Masterman and Kelly 2003; Buchmann et al. 2009). Also, we focused on pre- and early teenagers (11–13 year olds) for two reasons. First, protective family influences tend to be stronger during this period than for later periods (Capaldi et al. 2009), so factors that are inter-related with family relationship quality at this age may have particular long term significance. Second, we anticipated that especially young teenagers who consume alcohol without their parent’s awareness or approval would be generally more prone to antisocial behavior than older teenagers, and so family relationship quality and school connectedness may be more negatively affected for younger versus older groups (Andrews and Duncan 1997). The key hypothesis of this study was that the association between family relationship quality and school connectedness would be lower when peer drinking networks are present, and that this finding would be stronger for preteenagers than for older teenagers. We controlled for a range of other variables that are known correlates of family relationship quality, peer drinking networks and school connectedness, including depression (Herrenkohl et al. 2009), gender (Formoso et al. 2000), sensation seeking (Romer and Hennessy 2007; Martin et al. 2002), adolescent self-reported alcohol use, and parent education (Kelly et al. 2011b, c). We also controlled for school level clustering, based on prior findings that there is significant school-level variation in early adolescent substance use (Luthar et al. 2000; Seginer et al. 2002; Kelly et al. 2011a). Finally, given evidence that school connectedness and related problems vary by gender across the transition from primary school to high school (Martínez et al. 2011), we also controlled for the interaction of gender and grade level (in Australia,
Author's personal copy J Abnorm Child Psychol (2012) 40:437–447
children are in primary school at grade 6 and in high school at grade 8).
Method Survey Procedure The original data collection involved a two-stage sampling strategy (community and school) in which 231 participating schools in 30 communities in Victoria, Queensland, and Western Australia were selected. The community sampling frame consisted of Statistical Local Areas [SLA: An Australian Standard Geographic Classification structure consisting of nonoverlapping spatial units (ABS 2009)] with greater than 17,000 inhabitants. These SLAs were stratified into quartiles of socioeconomic disadvantage based on Socio-Economic Indexes for Areas (SEIFA) (ABS 2009). Eligible communities were randomly selected from SEIFA quartiles to represent State distributions in advantage/disadvantage and urban and nonurban locations. Within each community, primary (N=164) and secondary schools (N=82) were randomly selected. Of schools invited to participate, 83% (N=443) responded, and of these, 52% agreed to participate (59% and 43% at grade 6 and 8 levels respectively). If a school declined to participate in the survey another school from the same SLA and education sector was randomly selected. If there were no additional schools in the area, then SEIFA scores for contiguous SLAs were examined and additional schools were randomly selected that had the closest SEIFA score. Participation was dependent on signed parent consent (67% response rate) and attendance at school on the day of survey administration (about 10% of all enrolments were absent on the day of administration, with the most common reason for nonattendance being illness). Because of privacy regulations, the reasons for individual nonattendance of class on the day of the survey were unavailable. The survey was web-based and completed during school class time (paper copies were provided when computer resources were not available). The survey was approved by the University of Melbourne Human Research Ethics Committee and use of the data was approved by the University of Queensland Research Ethics Committee. Further details of the survey methods are described elsewhere (Hemphill et al. 2010). Sample The initial sample consisted of 7,866 adolescents in grades 6 (last or penultimate year of primary school in all states, modal age 11) or grade 8 (first or second year of high school, modal age 13). Of the initial sample, 151
439
participants were excluded because of their response on the honesty item (see Measures), and 343 participants were excluded because of missing data on one or more key variables. The analysis dataset consisted of 7,372 children (52.9% female) in grades 6 (54.3%) and 8 (45.7%). Demographic information (split by grade and gender) are presented in Table 1. In the analysis sample, 88.7% of students came from families who spoke English only. The median highest level of education was grade 12 for both mothers and fathers, about 17% of mothers and 19% of fathers had left high school before grade 12, and about a quarter of parents had completed a post secondary qualification. The prevalence of alcohol use was significantly different across gender and grade. Alcohol use was most common among grade 8 boys (29.7% had consumed alcohol at least one in the past month, CI: 27.4%–32.0%), followed by grade 8 girls (23.3%, CI: 21.4%–25.3%), grade 6 boys (20.0%, CI: 18.2%–21.8%) and was least common among grade 6 girls (9.7%, CI: 8.4%–10.9%), χ2(3) = 248.5, p
