Abstract
Adolescents who attend boarding schools share more time with peers than do students from day schools which, in turn, could provide more opportunities for bullying. Furthermore, some students attend boarding schools because of former social problems. In order to analyse the role of these factors, we examined the bullying behaviour of 706 German adolescents at boarding and day schools in a cross-sectional design. Adolescents at boarding schools showed higher levels of bullying by comparison to adolescents in day schools. Elevated levels of overt bullying in boarding schools could be explained by selection effects. In addition, more time spent with peers was associated with elevated levels of relational victimization. Furthermore, being overtly bullied showed stronger negative associations with life-satisfaction in students from boarding school than in students living at home.
Although the family continues to exert great influence during adolescence, peers become increasingly important (Steinberg, 2010). A special situation arises when adolescents do not live with their parents, or in a similar family environment, but rather at a boarding school. Boarding schools provide a semi-permanent institution for education, accommodation, and food for students (Chang, 2011). There are about 300 boarding schools in Germany (Boarding-School Finder, 2012).
Research on behavioural outcomes in boarding schools
While academic achievement is probably the most commonly assessed outcome of research in schools, many researchers and practitioners are also interested in externalizing and internalizing problem behaviour (Miller & Jome, 2010). Problem behaviours lead to important social and clinical problems such as anxiety and depression (Williford et al., 2012).
There are no quantitative German studies that compare adolescents from boarding schools and day schools (Little, Kohm, & Thompson, 2005). A study from Spain (Garcia & Monteoliva, 2000; Monteoliva Sanchez, & Garcia Martinez, 2001, 2007) found that adolescents from boarding schools had a lower socio-economic status, were less satisfied with their school, showed less empathy, and had more adjustment problems, such as anti-social behaviour, than did adolescents from day schools. Furthermore, while boarding schools are common in several other countries around the World, typically, studies have not compared students from boarding schools with those from day schools (e.g. Liang, Flisher, & Lombard, 2007; Zindi, 1994).
Additionally, there are studies from boarding schools for students with disabilities. In a study by Lifshitz, Hen, and Weisse (2007), teachers considered the social and emotional adjustment of students with visual impairment as less strong when they attended boarding schools rather than day schools. Furthermore, adolescents with hearing impairment who attended a boarding school reported lower social skills than their peers from day schools (Cartledge, Cochran, & Paul, 1996). However, teachers did not perceive such a difference in another study (Cartledge, Paul, Jackson, & Cochran, 1991). These authors warned against interpreting the results as evidence for a detrimental impact of boarding schools, and drew attention to selection effects, such as difficult family conditions, which may have caused the choice of a boarding school. To our knowledge such selection effects have not yet been assessed.
In addition, there are many studies about bullying in residential homes or foster care that show adolescents in these settings with elevated levels of bullying (e.g. Baker, Kurland, Curtis, Alexander, & Papa-Lentini, 2007). These living-arrangements differ from boarding schools. Adolescents often live in residential homes for limited periods, and generally do not all attend the same school (e.g. Golmaryami & Barry, 2010). Studies of bullying in boarding schools are important, because students meet the same peers as well as the same bullies at school and in their residential setting.
Bullying
Bullying is defined as an aggressive behaviour repeated over time with the intention to harm the victim (e.g. Smith, 2004). Physical bullying or aggression is characterized by observable behaviours including being hit, physical threats, and blackmail. In contrast, relational forms of bullying include more subtle and indirect forms of aggression such as spreading untrue rumours, and social exclusion. Furthermore, new possibilities of relational bullying arise in new media by using the internet or social networks (cyberbullying; see Cassidy, Faucher, & Jackson, 2013).
According to a comprehensive internally-focused review by Altomare, McCrimmon, and Beran (2013) 11% to 53% of children (aged 11- to 15-years-old) across 29 developed countries reported ‘being bullied at least once in the past couple of months’. Male adolescents tend to show higher levels of bullying than female peers (Altomare et al., 2013). Several longitudinal studies have found associations between bullying and decreased psychosocial well-being (e.g. depression, poor self-image, greater dependency on adults), somatic symptoms, school absences, and even completed suicide (Reijntjes, Kamphuis, Prinzie, & Telch, 2010).
Selection effects and environmental/socialization effects as reason for higher levels of bullying at boarding schools
About 100,000 German students change their school each year (Bellenberg & Forell, 2012). Some of these students attended boarding schools because of problems at their former schools or at home (e.g. Monteoliva Sanchez & Garcia Martinez, 2007). For example, parental separation or divorce may be reasons for attending boarding school which may reduce effective parental measures for preventing problem behaviours. These reasons could lead to higher levels of bullying by and of adolescents from boarding schools when compared to day schools. Nonetheless, not all selection effects may promote elevated levels of bullying in boarding schools. Boarding schools are more expensive and generally unavailable to the mainstream (Walford, 2009). Although many German boarding schools offer financial assistance to enable students from low-income families access, it is likely that students of boarding schools more often come from affluent families. Higher socio-economic status of students from boarding schools when compared to day schools could reduce levels of bullying (Cook, Williams, Guerra, Kim, & Sadek, 2010). Although students from families with higher SES may have better access to computers as a precondition for cyberbullying, recent studies show that almost all German students have access to information technology communication (Feierabend & Rathgeb, 2012).
Differences between students from day and boarding schools may also arise from the special environment and socialization effects. For example, living at a boarding school is associated with separation from parents, former friends, and familiar surroundings on school days and often also during weekends. This is likely to restrict the influence of parenting practices aimed at preventing or reducing problem behaviour. Furthermore, students from boarding schools share more time with schoolmates, because single rooms are the exception (White, 2004). Higher levels of contact with schoolmates provide more opportunities for bullying in boarding schools in comparison to day schools.
Research questions for the study
To our knowledge, no quantitative comparative studies on bullying in boarding schools have so far been published. In Zimbabwe, Zindi (1994) reported that 16% of students from a boarding school were occasionally bullied, and 18% were bullied weekly or more often. However, this and other studies (e.g. White, 2004) do not provide comparative data from students at day schools. Due to the lack of quantitative comparative studies, we aimed to examine the level of bullying in boarding schools and possible explanations for differences between day and boarding schools.
Students from boarding schools have been found to show more problem behaviours and lower adjustment than students from day schools (e.g. Monteoliva Sanchez & Garcia Martinez, 2001, 2007). Therefore, we assume that there are also higher levels of bullying in boarding schools as compared to day schools (Hypothesis 1). Furthermore, these differences may be, at least in part, based on the fact that some adolescents attend boarding schools because they had to leave a day school due to behaviour problems or because of the dynamics of elevated family problems that could also cause adolescent behaviour problems (Monteoliva Sanchez & Garcia Martinez, 2007). Therefore, the differences between levels of bullying of students from boarding schools and day schools were expected to decrease after excluding these adolescents from the analysis (Hypothesis 2). In addition, more time spent with peers may provide more opportunities for bullying. Therefore, we expected higher levels of bullying if students from boarding schools spend their weekends with peers at dormitories rather than in their parental homes (Hypothesis 3). Finally, bullied students in boarding schools cannot avoid perpetrators easily because these individuals are around them almost 24-hours a day. Therefore, we assumed that being bullied shows a stronger negative association with life-satisfaction in boarding schools than in day schools (Hypothesis 4).
Method
Participants
The present study included 12- to 19-year-old students from six German day schools and nine boarding schools. All participants were students from the highest school track, which qualifies students for university upon graduation. The study was approved by the Ethics Committee of the German Psychological Society. After we had received the permission of the school boards and informed consent from parents and adolescents, the students completed a questionnaire with all variables during class.
In total, 723 adolescents participated; 17 of them were excluded from the analyses because they were younger than 12 years or older than 19 years. The response rate was 82.7%. The final sample includes 300 adolescents from boarding schools and 406 adolescents from day schools. The two groups of students differed in nearly all socio-demographic variables. Adolescents from boarding schools were older (Mboarding = 15.99, Mday = 15.00; F(1,704) = 50.06; p < 0.001; η2 = 0.07), more likely male (boarding school: 65.4% day school: 44.1%; F(1,704) = 32.86; p < 0.001; η2 = 0.05), and their parents had more often completed the highest school track (Fathers: 78.7% versus 61.1%; F(1,704) = 36.06; p < 0.001; η2 = 0.05; Mothers: 77.3% versus 47.1%; F(1,704) = 37.04; p < 0.001; η2 = 0.05). Thus, we controlled for these variables in our analyses.
Measures
Bullying
The Social Experience Questionnaire—Self-Report was used to assess bullying and victimization (SEQ-S; Crick & Grotpeter, 1996). It measures adolescents’ reports of relational and overt bullying as well as victimization. Parallel versions ask about being a victim and a perpetrator
Reason for attending boarding school
We asked students if they joined their boarding school because of having problems at their former school or at home. They marked their answers on a five-point Likert-type scale from ‘1 = totally disagree’ to ‘5 = totally agree’.
Life-satisfaction
Life-satisfaction was assessed with the Satisfaction with Life Scale (Diener, Emmons, Larsen, & Griffin, 1985) which is one of the most widely used life-satisfaction measures. A sample item is ‘I am satisfied with my life’. Answers were marked on a seven-point Likert-type scale from ‘1 = strongly disagree’ to ‘7 = strongly agree’. In the present study, Cronbach’s α was 0.84.
Socio-demographic variables
These characteristics were assessed with single-item indicators: age (in years), gender (0 = male, 1 = female), habitation (living with ones’ parents = 0 versus living in boarding school = 1), father’s and mother’s educational attainment (0 = no school completed, 1 = lowest school track, 2 = middle school track, 3 = highest school track, 4 = university degree), and week-end stay (0 = at home, 1 = at boarding school).
Results
First, mean scores were created for the subscales of the Social Experience Questionnaire. Next, we computed four univariate covariance analyses with boarding school as the independent variable, age, gender, father’s and mother’s completion of highest school track as control variables, and the four subscales of the social experience questionnaire as the dependent variables.
Results indicate that adolescents from boarding schools show more relational and overt bullying in comparison to adolescents who live at home (Relational aggression: Mboarding = 1.77, Mday = 1.48; F(1,694) = 22.88; p < 0.001; η2 = 0.03; Overt aggression: Mboarding = 1.73, Mday = 1.46; F(1,694) = 7.40; p < 0.01; η2 = 0.01). Furthermore, adolescents who live in boarding schools are more often victims of relational and overt bullying than adolescents who live at home (Relational aggression: Mboarding = 2.12, Mday = 1.71; F(1,694) = 64.61; p < 0.001; η2 = 0.09; Overt aggression: Mboarding = 1.75, Mday = 1.37; F(1,694) = 30.61; p < 0.001; η2 = 0.04). Effect sizes were small to moderate. Hypothesis 1 was confirmed. Between-group differences were significantly higher in reported victimization than in reported bullying (rbully_verbal + habitation = 0.17, rvictim_verbal + habitation = 0.29, z = 2.42, p < 0.05; rbully_physical + habitation = 0.10, rvictim_physical + habitation = 0.21, z = 2.06, p < 0.05).
As mentioned earlier, higher SES may reduce bullying. Thus, statistical control for parental socioeconomic status might have increased the observed elevated levels of bullying at boarding schools. Four additional ANCOVAs were computed that did not include parental education as a covariate. However, the results did not change, and students from boarding schools still reported elevated levels of bullying.
In the next step, we tested whether the between-group difference still existed after excluding adolescents who attended boarding schools because of problems at home or at their former day schools
In the next step of analysis, we tested whether higher levels of contact with mates of similar ages might explain the observed higher levels of bullying at boarding schools. Because all of these students spend the whole day with their peers during their days at school, we had to use the information on where these students spend their weekends. We split the adolescents from boarding schools into a group who spend the weekends at boarding school (N = 92) and a group who stays at home (N = 214). Then four ANCOVAs were calculated with the new group variable as independent variable, gender, father’s and mother’s completion of the highest school track as control variables (the two groups did not differ in age), and perpetrator of relational/overt bullying as well as victim of relational/overt bullying as dependent variables. We found that adolescents who spend the weekends at boarding school did not differ from the other adolescents from boarding schools with regard to the frequencies of showing relational bullying (MWEboarding = 1.84, MWEhome = 1.71, F(1,283) = 1.54; n.s.), and overt bullying (MWEboarding = 1.72, MWEhome = 1.66, F(1,283) = 0.22; n.s.), as well as being a victim of overt bullying (MWEboarding = 1.71, MWEhome = 1.72, F(1,283) = 0.07; n.s.). However, adolescents who spent the weekends at boarding school reported higher levels of being relationally bullied in comparison to other adolescents from boarding schools who spent the weekends at the parental home (MWEboarding = 2.30, MWEhome = 2.05, F(1,283) = 3.98; p < 0.05; η2 = 0.01). Therefore, Hypothesis 3 received only weak empirical support.
Finally, we analysed whether being bullied shows a stronger negative association with life-satisfaction in adolescents from boarding schools than in their peers from day schools. We split the levels of victimization through relational and overt aggression by median, and computed two ANCOVAs with habitation and the two forms of victimization as independent variables. Age, gender, and parent’s completion of the highest school track were used as covariates, and life-satisfaction as dependent variable. The first ANCOVA showed main effects of habitation (F(7,684) = 17.12, p < 0.001, η2 = 0.02), and relational victimization (F(7,684) = 31.30, p < 0.001, η2 = 0.04) on life satisfaction, but no interaction effect between habitation and relational victimization (F(7,684) = 0.02; n.s.). The second ANCOVA showed main effects of habitation (F(7,684) = 21.53, p < 0.001, η2 = 0.03) and overt victimization (F(7,684) = 11.12, p < 0.001, η2 = 0.02), as well as an interaction effect between habitation and overt victimization (F(7,684) = 3.11; pone-tailed < 0.05; η2 = 0.01) on life-satisfaction. The main effects indicate that life-satisfaction was lower when being victimized and when living in boarding schools. Sobel-Test showed significant difference between the correlations between overt victimization and life-satisfaction within the groups (rboarding = −0.22, rday = −0.04, z = 2.39, p < 0.05). The interaction effect indicates that the association between overt aggression and life-satisfaction was more negative in students from boarding schools than in students from day schools. Therefore, Hypothesis 4 was confirmed for overtly bullying, but not for relational bullying.
Discussion
In this first quantitative, comparative study on bullying in boarding schools, we found that adolescents from German boarding schools were more often perpetrators and victims of relational and overt bullying than were adolescents living at home. The effect sizes were small to moderate. Further analysis showed that between-group differences in showing overt bullying could be fully explained by selection effects, and that spending more time with peers is associated with elevated levels of relational victimization in boarding schools. However, between-group differences in relational bullying could not be explained by our assessed variables. Finally, associations between experiencing overt bullying and life-satisfaction were more negative in boarding schools than in day schools.
The observed elevated levels of bullying in boarding schools are consistent with former results on behaviour problems in boarding schools (e.g. Monteoliva Sanchez, & Garcia Martinez, 2001, 2007). Between-group differences were higher in reported victimization than in reported bullying. This probably indicates that many bullies show this behaviour towards more than one schoolmate (Jimerson, Swearer, & Espelage, 2009). Thus, bullies at boarding schools might have a larger number of victims than at day schools.
Furthermore, only the elevated levels of overt bullying in boarding schools could be explained by selection effects. Overt aggressive and delinquent behaviour is more visible to teachers than relational bullying, and mandatory expulsion recommendations refer to consequences of overt bullying, rather than relational bullying (e.g. Morrison et al., 2001). This probably leads to a selection of overtly aggressive rather than relationally aggressive students at boarding schools when parents are overwhelmed by their children’s behaviour problems.
As elevated levels of relational bullying at boarding schools could not be explained by selection effects, factors of the school environment could be expected to explain this phenomenon. We had suggested that more time spent with peers gives more opportunities for bullying but we found such an effect only for relational victimization when comparing students from boarding schools who spend the weekends with their parents and those who stay in the boarding school and could be expected to have more contact with peers. However, this measure of peer contact is rather global because students who live in the parental home on weekends may also spend some time with peers (from the neighbourhood) and have some opportunities for bullying and being bullied during the weekends. More differentiated assessments of time spent with peers might reveal stronger effects of peer contact on bullying and victimization. Future studies should also examine other variables that might explain elevated levels of relational bullying in boarding schools. For example, students at boarding schools have less space for privacy and more unsupervised time which could lead to increased relational bullying in dormitory settings in boarding schools (White, 2004). In addition, German boarding schools have a long tradition of elite-education associated with higher competition between students (Chang, 2011) which could, again, lead to higher levels of bullying (Cook et al., 2010). Also, the reduced influence of parents could be a reason for the higher amount of relational bullying at boarding schools.
Similar to other studies (e.g. Proctor, Linley, & Maltby, 2009), we found a negative association of being bullied with life-satisfaction. As expected, this association was stronger at boarding schools than at day schools, although such a moderating effect was only observed for physical victimization and the size of the moderating effect of habitation was small. Because avoiding contact with perpetrators is more complicated at boarding schools, these students are likely to experience higher distress.
We also found lower levels of life-satisfaction in students from boarding schools compared to day schools. This difference may have been based on lower opportunities for privacy in boarding schools, lower contact with family members, or other factors.
Limitations, implications and conclusion
Our study had clear limitations. We included only subjects from the highest school track that gives access to university after completing school; nonetheless, this kind of boarding schools is most common in Germany. The analysis of correlational data does not allow for the analysis of causal relationships. Only adolescents’ self-reported data were available. Future studies on bullying should include other sources such as peer nominations and teacher reports. Nonetheless, self-report and peer reports from the social experience questionnaire were significantly correlated, thus supporting the validity of self-reports (Crick & Bigbee, 1998). In addition, teacher reports might be invalid, because teachers from boarding schools might frame their perceptions based on broader opportunities than solely classroom variables. Further, the results of our study may not be generalized to other countries with other criteria for access to boarding schools.
Despite these limitations, several conclusions can tentatively be drawn from this pilot study. Bullying is a greater problem in German boarding schools than in day schools. Longitudinal research is needed to determine how adolescents from boarding schools who are perpetrators or victims of bullying develop. For example, longitudinal studies could assess whether effects of bullying on change of life-satisfaction and other indicators of psychological well-being differ between boarding schools and day schools. As bullying is more common in the boarding schools in this study than in day schools and shows stronger associations with life-satisfaction of victims from boarding schools, interventions are needed to reduce bullying in boarding schools. Research on available bullying prevention programs (not specific to boarding schools) indicates that these interventions have meaningful, although small effects (for an overview, see Merrell, Gueldner, Ross, & Isava, 2008). Such bullying intervention programs in boarding schools are justifiable, even if effect sizes for day school interventions are small. Finally, specific interventions are needed for boarding schools. Because more time spent with peers was associated with higher levels of relational bullying, intervention may focus on reducing the amount of unsupervised time with peers. As larger numbers of students attend boarding schools because of problems at their former day schools (including externalizing behaviour problems), general classroom-based interventions may be combined with individual-focused interventions that could start immediately after these students enter boarding school.
Footnotes
Acknowledgements
The study was approved by the Ethics Committee of the German Psychological Society and was supported by a grant of the German Research Council (Pi291/9-1). We want to thank Kathrin Dolle and Silke Sievers for their support in data collection.
