Journal of Clinical Child & Adolescent Psychology, 0(0), 2014 2015 44(3),1–14, 480–493, 1–13, Copyright © Taylor & Francis Group, LLC ISSN: 1537-4416 print/1537-4424 online DOI: 10.1080/15374416.2013.873985 10.1080/15374416.2013.873983
Early Childhood Precursors and Adolescent Sequelae Reciprocal Associations Between Interpersonal and Values of Grade School PeerClimate Rejection Victimization Dimensions of School andand Peer Victimization in Elementary School Children Karen L. Bierman, Carla B. Kalvin, and Brenda S. Heinrichs Department of Psychology, The Pennsylvania State University
Bonnie Leadbeater and Paweena Sukhawathanakul Department of Psychology, University of Victoria This study examined the early childhood precursors and adolescent outcomes associated David Smith among children oversampled for with grade school peer rejection and victimization of Education, University of Ottawa aggressive-disruptiveFaculty behaviors. A central goal was to better understand the common and unique developmental correlates associated with these two types of peer adversity. There were 754 participants (46% African American, François Bowen 50% European American, 4% other; 58% male; average age = 5.65 at kindergarten entry) followed into seventh grade. Faculty of Education, University of Montreal Six waves of data were included in structural models focused on three developmental periods. Parents and teachers rated aggressive behavior, emotion dysregulation, and internalizing problems in kindergarten and Grade 1 (Waves 1–2); peer sociometric nominations tracked “least liked”relations and victimization in Gradesand 2, 3,parents’ and 4 (Waves and We examine longitudinal among children’s reports3–5); of peer youth reported on social problems, depressed mood, school adjustment difficulties, and victimization and their perceptions of school climate dimensions reflecting school interdelinquent activities in(relationships early adolescence (Grade 7, Wave 6). Structural models personal relationships among children and their teachers and peers,revealed and of that early aggression and emotion dysregulation (but not internalizing behavior) made parents and principals) and values (fairness and equity of access to resources). Children unique school peer=rejection; only emotion made were in contributions Grades 3 andto4 grade at Time 1 (Mage 9.32, SDage = .74; 49% dysregulation boys). Bidirectional unique contributions to grade Early internalizing problems and influences of school climate andschool reportsvictimization. of peer victimization were investigated in path grade school victimization uniquely predicted adolescent social problems and depressed models across grade (Time 1 to Time 2) and within a grade (Time 2 to Time 3). Child and mood. Early aggression and grade school peer rejection uniquelystability. predicted adolescent parent reports of school climate dimensions showed considerable Hypothesized school adjustment difficulties and delinquent activities. Aggression and emotion dysregreciprocal relationships between each of the school climate dimensions and peer victimizaulation at school entry increased risk for peer rejection and victimization, and these two tion were significant. Child-reported frequency of parent involvement in school activities types of peer adversity had distinct as well as shared risk and adjustment correlates. was associated with lower levels of peer victimization within a school year. Parent percepResults suggest thatinthe emotional functioning and peer experiences of community aggressivetions of involvement school activities and the schools’ connection with the disruptive children deserve further attention in developmental and clinicalnegative research. were unrelated to the children’s reports of peer victimization. Children’s cognitions or “worldviews” coupled with peer victimization may fuel problems with school connectedness, safety, and help seeking. Findings shed light on possible pathways for reducing risk for chronic aggression and emerging Children with behavior problems atby school entry are ataspects amplify peer victimization leveraging specific of the social climate within schools.
risk for peer rejection and peer victimization, and these forms of peer adversity, in turn, are linked with significant maladajustment in adolescence, including school School climate is a complex multidimensional construct difficulties, social problems, delinquent activity, that typically refers to the overall or shared qualityand of compromised mental health (Hanish & values, Guerra,and 2004). school life—including normative beliefs, ideHowever, long-term longitudinal studies have not yet als that prevail in a school community—that is manifested examined the differential precursors of these two types in the relationships among all people involved the school of peer adversity, or explored their differential associa(Cohen, McCabe, Michelli, & Pickeral, 2009; Mitchell, tions with adolescent maladjustment. Bradshaw, & Leaf, 2010, O’Malley, Katz, Renshaw, & Developmental research that children who Furlong, 2012). Reflectingsuggests this complexity, current enter school exhibiting aggressive-disruptive behavior are at risk for peer rejection, and in turn, rejection appears to Correspondence should be addressed to Bonnie Leadbeater, Department of Psychology, University of Victoria, PO Box 3050, STN Correspondence should be addressed Karen L. Bierman, CSC, Victoria, British Columbia, V8W to 2Y2, Canada. E-mail: Department of Psychology, The Pennsylvania State University, 1100 [email protected] Moore Building, University Park, PA 16801. E-mail: [email protected]
delinquent activities (Coie & Dodge, 1998). In elementary school, peer victimization is also linked with early aggression some studies (Buhs, Ladd, &dimensions Herald, 2006; research hasinbegun to differentiate several of Crick, Murray-Close, Marks, & Mohajeri-Nelson, 2009) school climate including interpersonal relationships and with emotional dysregulation (Schwartz, Proctor, & among teachers, students, parents and principals; organiChien, 2001) and internalizing problems in others zational aspects of school functioning; quality of instruc(Hawker & Boulton, 2000).school Peer victimization does not tion; access to resources; safety; environmental appear to increase future risk for delinquent activities conditions; and school values (Bear, Gaskins, Blank, in & the same way as peer rejection Chen, 2011; Brand, Feiner, Shim,(Pouwels Seitsinger,& &Cillessen, Dumas, 2013), but instead, victimization with 2003; Cohen et al., 2009; Emmons,appears Haynes,linked & Comer, long-term elevations in social avoidance and emotional 2002; Freiberg, 1998; Van Houtte, 2005). Measures of distress (Boivin, typically Hymel, &assess Bukowski, 1995; Buhs et al., school climate individuals’ (students’, 2006; Nishina, Juvonen, & Witkow, 2005). Although teachers’, or parents’) perceptions of school climate, peer and rejection and used victimization interdependent these are often at the levelrepresent of individuals’ responses forms of peer adversity (HanishRecchia, & Guerra, 2004), their (see meta-analysis by Steffgen, & Viechtbauer, developmental precursors and adolescent outcomes may
2
LEADBEATER ET AL.
2013) but are also sometimes aggregated at the school or classroom level, in large-scale studies including many schools (e.g., Khoury-Kassabri, 2011; McCoy, Roy, & Sirkman, 2013). Considerable research has shown that positive school climate is associated with enhancements in school engagement and academic success of students (e.g., Goddard, Sweetland, & Hoy 2000; O’Malley et al., 2012) and in the engagement of their parents (Goldkind & Farmer, 2013). School climate is also associated with reductions in an array of concerns including school violence (Furlong & Morrison, 2000; Manvell, 2012; Steffgen et al., 2013); deviant student lifestyles (Zaykowski & Gunter, 2012); students’ social, emotional, and behavioral risks (Gottfredson, Gottfredson, Payne, & Gottfredson, 2005; Klein, Cornell, & Konold, 2012; Kuperminc, Leadbeater, Emmons, & Blatt, 1997; Wang & Dishion, 2012); and peer victimization and bullying (e.g., Goldstein, Young, & Boyd, 2008; Gottfredson et al., 2005; Kasen, Berenson, Cohen, & Johnson, 2004; Khoury-Kassabri, 2011; Khoury-Kassabri, Benbenishty, Zeira, & Astor, 2004; Swearer et al., 2006). Despite this extensive and growing literature, however, most research is cross-sectional (see McCoy et al., 2013, for an exception), and school climate is conceptualized as a static characteristic of the school environment. Moreover, although most assessments rely on individuals’ reports of components of school climate, little is known about the stability of or variability in individuals’ perceptions of school climate, how their perceptions change over time, or what factors influence individuals’ perceptions of school climate. In this three-wave longitudinal study, we investigate how peer victimization influences stability and change in students’ and their parents’ perceptions of school climate. Children who are victimized are unlikely to share the average students’ perceptions of many aspects of school climate, and children who continue to experience victimization over time may show worsening perceptions in salient dimensions of school climate. Both attachment and information-processing theories suggest that children develop internalized representations or “working models” of their socialization experiences that are influenced by the ways in which they are treated by peers and important adults in their social networks. These cognitive representations may be generalized to new experiences and color children’s beliefs and expectations of their social environments over time (Ladd, Ettekal, KochenderferLadd, Rudolph, & Andrews, 2014). Experiences of peer victimization, teasing, and rejection may, in particular, enhance children’s negative expectations of student–peer relationships at school (Asher, Rose, & Gabriel, 2001) but may also extend to other components of school climate. For example, chronic victimization may lead children (and possibly their parents) to believe that principals, teachers, and
SCHOOL CLIMATE AND PEER VICTIMIZATION
481
peers cannot be counted on to be caring and respectful and, given their own experiences, that not all students in the school are treated fairly. Influenced by their children’s experiences, parents of victimized children may also have negative perceptions of their child’s school climate that worsen over time. SCHOOL CLIMATE AND PEER VICTIMIZATION Growing research has found that peer victimization is related to school climate dimensions that tap student– teacher relationships and the application of consistent, clear, and fair rules. Positive relationships between students and teachers were associated with less physical and verbal/relational bullying in several studies (Flaspohler, Elfstrom, Vanderzee, Sink, & Birchmeier, 2009; Richard, Schneider, & Mallet, 2012). In a study of Israeli fourthto sixth-grade students, Khoury-Kassabri (2011) also found that although levels of victimization varied between classes and schools, the majority of variation was at the individual level. Moreover their study found that physical and emotional victimization by teachers affected levels of students’ reports of victimization by their peers. Goldstein et al. (2008) found that exposure to relational victimization was associated with adolescents’ perceptions of their schools’ safety and social atmosphere. Similarly, social support from teachers and from peers buffered the impact of bullying experiences on victimized children’s quality of life (RasKauskas, Gregory, Harvey, Rifshana, & Evans, 2010) and on adolescents’ reports of internalizing symptoms (Yeung & Leadbeater, 2010). Roland and Galloway (2004) also found that primary school students reported lower levels of victimization when they perceived that the school policies and rules were clear, consistent, and fair, and that student–teacher relationships were positive. Classroom norms regarding aggression and bullying also influence students’ tendencies to exhibit aggressive behaviour (Rodkin & Hodges, 2003). It is clear that perceptions of school interpersonal relationships and fairness characterize an important facet of school climate and have implications for reducing or exacerbating peer victimization. However, given the lack of longitudinal research, the direction of effects between school climate and peer victimization is not known. In this longitudinal study, we examined the effects of individual students’ reports of victimization and their parents’ perceptions of school interpersonal relationships and fairness across changes in grades and teachers, and across time within a grade. Stability across both grade and time would suggest that child and parent perceptions of school climate are not only a function of current classroom experiences. Children’s experiences of peer victimization are also expected to influence child and parent perceptions of school climate. Specifically,
482
LEADBEATER ET AL.
bidirectional influences between school climate dimensions and children’s reports of peer victimization were expected, as victimized children’s (and their parents’) school climate perceptions worsen over time. Data from Canadian children in Grades 3 and 4 and their parents were assessed across three waves of data collected during two school years (spring 2011, fall 2011, and spring 2012). We tested autoregressive, cross-lagged, path models (Kline, 2005) to examine how interpersonal and fairness dimensions of school climate related to peer victimization across the three waves of data. All analyses controlled for the influence of age, sex, maternal education, and the presence or absence of the WITS peer victimization prevention program. The WITS Program (see www.witsprogram.ca) is a whole school program that engages community members, school staff, students, and parents to create communities of individuals who are responsive when children report victimization or peer conflict. The program had not been fully implemented when data were collected, and investigations of its effects are beyond the scope of this article. Hence, we controlled any effect of being or not being in a program school on peer victimization and on the school climate dimensions at each time point. Although no differences were anticipated, cross-lagged models were also assessed for relational and physical victimization separately, but results showed few differences, hence types of victimization are combined in all analyses. Sex differences were not indicated in past research and are also controlled in our analyses. METHOD
SCHOOL CLIMATE AND PEER VICTIMIZATION
3
486 children who moved from Grade 2 into Grade 3 between T1 and T2) and 1,208 children at T3. Hence, data from T1 to T2 represent data collected from the same children across a change in grade and classroom, whereas data from T2 to T3 represent data from the children in the same classroom at the beginning and end of the school year. Demographic information was also collected from parents at baseline. The majority (95%) of participants were European Caucasian, 2% Aboriginal, 1% Asian, and 1% of mixed ethnicity; less than 1% were Hispanic, Indo-Canadian, or African Canadian. Parents indicated that 74% of children lived in a two-parent household; 25% of mothers and 32% of fathers had no postsecondary education after high school; 45% of mothers and 46% of fathers completed “some college or technical training” beyond high school; and 30% of mothers and 22% of fathers had some postsecondary education. Ninety-two percent of the children had attended a maximum of two schools in their lifetime, and 8% had attended three or more schools. Attrition was minimal across the three waves of data collection. Selective attrition was assessed by testing for differences at T1 on demographic and key variables for the T1 group of 800 children who remained in the longitudinal study and those who did not participate at T2 (n = 63, 7.9%; 54% boys) and T3 (n = 84, 10.5%; 54% boys). Findings did not reveal any significant demographic differences. Children who dropped out of the study reported slightly lower student-interpersonal relations at T1 (M = 2.26, SD = .50) than students who remained in the study (M = 2.39, SD = .48), t(798) = –2.43, p = .02; d = .27.
Participants Children were recruited from 137 classrooms in 27 predominately rural schools across three provinces in Canada. They were involved in the first three waves of an ongoing evaluation of the WITS Program for the prevention of peer victimization (see Leadbeater & Sukhawathanakul, 2011). At Time 1 (T1) 1,659 children participated, and an additional 105 joined the study at Time 2 (T2) due to delays in receiving consent forms back from parents (n = 1,764). Children completed the peer victimization scale at each wave. However, children completed the child-reports for school climate scales only when they were in Grade 3 and higher. Parents of children at all grade levels completed the parent-report of the school climate (including 1,653 at T1, 99% of eligible parents; 1,566 at T2, 89% of eligible parents; and 1,389 at T3, 79% of eligible parents). Children ranged in age from 6.67 to 11.33 years (M = 9.32, SD = .74) at T1; 49% were boys. Data were collected from 800 children at T1; 1,217 children at T2 (including 731 of the original 800 children plus
Procedure Teachers sent home evaluation packages to parents of children in participating schools informing them of the study and seeking consent for participation. Parents provided written consent for their child’s participation and completed a demographic and school climate questionnaire and returned it to the school in a sealed envelope for pick-up by a research assistant. Data were collected from participating children in their classrooms. A trained research assistant read the child questionnaire items aloud to the class, and children completed the questionnaires individually and privately. A research assistant also circulated to check completion of the questionnaires and to help children who needed assistance. Measures Peer victimization was measured using the Social Experience Questionnaire (SEQ; Crick & Grotpeter,
4
LEADBEATER ET AL.
1996) that was modified for administration to young children. This scale was adapted from Crick and Grotpeter’s (1996) original 5-point scale (ranging from never to all the time) to facilitate ease of understanding for younger children (Leadbeater & Hoglund, 2009; Leadbeater, Hoglund, & Woods, 2003; Leadbeater & Sukhawathanakul, 2011). In the modified version, children rated how often they experienced relational victimization (e.g., “How often does another kid tell lies about you to make others not like you anymore?”) and physical victimization (e.g., “How often do you get pushed or shoved by another kid at school?”) on a 3-point scale: 0 (never), 1 (sometimes), and 2 (almost all the time). Each subscale contained five items. Prior research has demonstrated measurement invariance across time, gender, and grade level among elementary school children using this revised scale (Desjardins, Yeung Thompson, Sukhawathanakul, Leadbeater, & MacDonald, 2013). Two practice items were also used to familiarize children with the response scale. During administration, adults read the SEQ items aloud, instructed children to follow along using a brightly colored piece of paper, and circulated around the room to ensure that children are responding to appropriate items. Relational and physical peer victimization were associated at each time point (rs = .69, .70, and .72 consecutively) so the items were summed to create an overall peer victimization construct. The reliabilities were adequate at each time point (αs = .86, .88, and .89 consecutively). School climate was measured using the School Climate Survey (Haynes, Emmons, & Comer, 1993) to assess the general social environment of the school and the quality of relationships that exist among students and adults in the school building. The elementary and middle school student version includes 30 items that assess five aspects of school climate. Student–teacher relations refers to the level of caring, respect, and trust that exists between students and teachers (e.g., “My teachers care about me”). Student interpersonal relations refers to the levels of caring, respect, and trust that exists among students in the school (e.g., “Children at my school respect one another”). Fairness refers to the equal treatment of students regardless of ethnicity and socioeconomic status (e.g., “At my school, boys and girls are treated equally well”). Equity in sharing of resources refers to equal student opportunity to participate in school activities, and with materials and equipment (e.g., “The same children do not always get to use things, like a computer, a ball or a piano, when we play”). Students responded on a 3-point scale, according to how much they agree with the statement (1 = disagree, 2 = not sure, 3 = agree). The reliabilities for the subscales were adequate at each time point (αs = .70 – .88). Using the same 3-point scale, the frequency of parent participation in school activities was assessed as a total of children’s “agree” response to five
SCHOOL CLIMATE AND PEER VICTIMIZATION
483
items (e.g., “Parents often come to my school to help with special projects”). The parent version of the School Climate Survey consists of 23 descriptive statements about prevailing school conditions. Items represented four aspects of school climate. Student–teacher relations refers to the level of caring, respect, and trust that exists between students and teachers in the school (e.g., “Teachers at my child’s school help my child with school problems”). Principal caring and sensitivity refers to the extent to which the principal shows consideration for the students, parents, and school staff and cares about their needs (e.g., “At my child’s school, the principal cares about the needs of the children”). School–community relations refer to the support and involvement of the community in the life of the school (e.g., “Community members work with staff at my child’s school to help improve the school”). Parents rated on a 5-point scale how well they think the statement describes their child’s school, from 1 (strongly disagree) to 5 (strongly agree). The reliabilities for the subscales were adequate at each time point (αs = .87–.95). Using the same 3-point scale, the frequency of parent participation in school activities was assessed as a total of parents’ “agree” and “strongly agree” responses to four items (e.g., “I often help with special school projects such as bus trips, fund-raisers etc.”). Correlations between the two overlapping child and parent school climate dimensions (i.e., student–teacher relations and frequency of parent participation) were modest. Specifically, concurrent correlations between child and parent reported student–teacher relations were .24, .26, and .25 (ps < .05), respectively. Concurrent correlations between child- and parent-reported frequency of parent participation were .30, .12, and .20 (ps < .05), respectively. Demographic covariates. We included a series of four demographic covariates in our analyses: mother’s education (used as a proxy for socioeconomic status), program status (i.e., whether schools had a peer victimization prevention program), child’s age, and sex. Mean levels of peer victimization and school climate variables were similar for girls and boys, except that girls reported significantly more positive student–teacher relations and parent involvement at baseline. Correlations for demographic covariates and study variables were small (rs = –.05 to –.21). Data Analytic Strategy We employed autoregressive, cross-lagged path analyses to assess the associations and directionality of links between children’s experiences of peer victimization and each dimension of school climate. Each model included
484 5 1–3 1–3 0–4 0–2 0–2 0–2
2.36 (.52)
2.73 (.41) 2.29 (1.16) .56 (.45) .51 (.45) .51 (.46)
.36**
.16**
.36** .24** .41**
.48**
.16**
.48** .24** .55**
.68**
.26**
—
1
.12**
.37**
.21** .03 .20**
.20**
.46**
.27** .05 .26**
.26**
—
2
.44**
.20**
.40** .25** .38**
.59**
.19**
.52** .25** .47**
—
3
.31**
.22**
.47** .25** .31**
.40**
.21**
— .28** .36**
4
.14**
.01
.19** .48** .11**
.19**
.02
— .17**
5
.41**
.21**
.46** .18** .51**
.63**
.28**
—
6
.20**
.51**
.26** .03 .27**
.27**
—
7
.57**
.23**
.50** .26** .43**
—
8
.34**
.23**
— .26** .33**
9
.19**
.03
— .14**
10
.67**
.40**
—
11
.32**
—
12
—
13
14
15
16
17
.88 .28** .22** .32** .47** .18** .31** .26** .33** .56** .21** .54** .35** .52** — NA .21** .003 .17** .24** .48** .15** .06* .18** .19** .54** .15** .05 .24** .28** — .86 –.38** –.32** –.45** –.27** –.02 –.28** –.24** –.27** –.23** −.04 −.22** −.23** −.21** −.19** −.05 — .88 –.29** –.23** –.32** –.24** –.04 –.38** –.30** –.41** –.31** −.03 −.30** −.27** −.31** −.25** −.05 .57** – .89 –.22** –.18** –.28** –.21** –.05 –.28** –.25** –.30** –.29** −.06 −.43** .33** −.44** −.38** −.06 .49** .64**
.86
.79
.84 NA .80
.84
.72
.84 NA .77
.83
.70
.78
α
Note: T = Time; α = Cronbach’s alpha; NA = not applicable. *p < .05. **p < .001.
1–3
2.21 (.64)
1–3
2.33 (.57)
1–3 0–4 1–3
1–3 0–4 1–3
2.77 (.36) 2.33 (1.14) 2.41 (.52)
2.78 (.34) 1.98 (.95) 2.42 (.55)
1–3
2.38 (.48)
1–3
1–3
2.30 (.58)
2.39 (.48)
1–3
2.40 (.54)
1. Fairness T1 2. Equal Access to Resources T1 3. Student Interpersonal Relations T1 4. Student–Teacher Relations T1 5. Parent Involvement T1 6. Fairness T2 7. Equal Access to Resources T2 8. Student Interpersonal Relations T2 9. Student–Teacher Relations T2 10. Parent Involvement T2 11. Fairness T3 12. Equal Access to Resources T3 13. Student Interpersonal Relations T3 14. Student–Teacher Relations T3 15. Parent Involvement T3 16. Peer Victimization T1 17. Peer Victimization T2 18. Peer Victimization T3
Range
M (SD)
Variable
TABLE 1 Correlations Between Child School Climate And Peer Victimization Variables
6
SCHOOL CLIMATE AND PEER VICTIMIZATION
LEADBEATER ET AL.
autoregressive pathways, within-time correlations, and cross-lagged pathways between school climate dimensions and peer victimization for the T1–T2 and T2–T3 waves. Autoregressive, cross-lagged path analysis accounts for the temporal sequencing of the variables, controlling for the level of the previous measurement occasion, and permits examination of the stability within variables as well as time-specific relations among the variables (Kline, 2005). To improve the statistical validity and model fit of the cross-lagged approach, robust standard errors that take into account the nesting of students within schools were used in the analyses (Muthén & Muthén, 2000). To control for demographic influences, mother’s education, age, sex, and program status were included as covariates with each construct at each time point. Following established guidelines (Bollen, 1989; Byrne, 2010; Hu & Bentler, 1999; Kline, 2005; Schreiber, Stage, King, Nora, & Barlow, 2006), model fit for all analyses were evaluated using the comparative fit indices (CFI), root mean square error of approximation (RMSEA), and standardized root-mean of the residual (SRMR). CFI values of >.95 represent an excellent fitting model and >.90 represent an adequate model fit; RMSEA and SRMR values
Early Childhood Precursors and Adolescent Sequelae Reciprocal Associations Between Interpersonal and Values of Grade School PeerClimate Rejection Victimization Dimensions of School andand Peer Victimization in Elementary School Children Karen L. Bierman, Carla B. Kalvin, and Brenda S. Heinrichs Department of Psychology, The Pennsylvania State University
Bonnie Leadbeater and Paweena Sukhawathanakul Department of Psychology, University of Victoria This study examined the early childhood precursors and adolescent outcomes associated David Smith among children oversampled for with grade school peer rejection and victimization of Education, University of Ottawa aggressive-disruptiveFaculty behaviors. A central goal was to better understand the common and unique developmental correlates associated with these two types of peer adversity. There were 754 participants (46% African American, François Bowen 50% European American, 4% other; 58% male; average age = 5.65 at kindergarten entry) followed into seventh grade. Faculty of Education, University of Montreal Six waves of data were included in structural models focused on three developmental periods. Parents and teachers rated aggressive behavior, emotion dysregulation, and internalizing problems in kindergarten and Grade 1 (Waves 1–2); peer sociometric nominations tracked “least liked”relations and victimization in Gradesand 2, 3,parents’ and 4 (Waves and We examine longitudinal among children’s reports3–5); of peer youth reported on social problems, depressed mood, school adjustment difficulties, and victimization and their perceptions of school climate dimensions reflecting school interdelinquent activities in(relationships early adolescence (Grade 7, Wave 6). Structural models personal relationships among children and their teachers and peers,revealed and of that early aggression and emotion dysregulation (but not internalizing behavior) made parents and principals) and values (fairness and equity of access to resources). Children unique school peer=rejection; only emotion made were in contributions Grades 3 andto4 grade at Time 1 (Mage 9.32, SDage = .74; 49% dysregulation boys). Bidirectional unique contributions to grade Early internalizing problems and influences of school climate andschool reportsvictimization. of peer victimization were investigated in path grade school victimization uniquely predicted adolescent social problems and depressed models across grade (Time 1 to Time 2) and within a grade (Time 2 to Time 3). Child and mood. Early aggression and grade school peer rejection uniquelystability. predicted adolescent parent reports of school climate dimensions showed considerable Hypothesized school adjustment difficulties and delinquent activities. Aggression and emotion dysregreciprocal relationships between each of the school climate dimensions and peer victimizaulation at school entry increased risk for peer rejection and victimization, and these two tion were significant. Child-reported frequency of parent involvement in school activities types of peer adversity had distinct as well as shared risk and adjustment correlates. was associated with lower levels of peer victimization within a school year. Parent percepResults suggest thatinthe emotional functioning and peer experiences of community aggressivetions of involvement school activities and the schools’ connection with the disruptive children deserve further attention in developmental and clinicalnegative research. were unrelated to the children’s reports of peer victimization. Children’s cognitions or “worldviews” coupled with peer victimization may fuel problems with school connectedness, safety, and help seeking. Findings shed light on possible pathways for reducing risk for chronic aggression and emerging Children with behavior problems atby school entry are ataspects amplify peer victimization leveraging specific of the social climate within schools.
risk for peer rejection and peer victimization, and these forms of peer adversity, in turn, are linked with significant maladajustment in adolescence, including school School climate is a complex multidimensional construct difficulties, social problems, delinquent activity, that typically refers to the overall or shared qualityand of compromised mental health (Hanish & values, Guerra,and 2004). school life—including normative beliefs, ideHowever, long-term longitudinal studies have not yet als that prevail in a school community—that is manifested examined the differential precursors of these two types in the relationships among all people involved the school of peer adversity, or explored their differential associa(Cohen, McCabe, Michelli, & Pickeral, 2009; Mitchell, tions with adolescent maladjustment. Bradshaw, & Leaf, 2010, O’Malley, Katz, Renshaw, & Developmental research that children who Furlong, 2012). Reflectingsuggests this complexity, current enter school exhibiting aggressive-disruptive behavior are at risk for peer rejection, and in turn, rejection appears to Correspondence should be addressed to Bonnie Leadbeater, Department of Psychology, University of Victoria, PO Box 3050, STN Correspondence should be addressed Karen L. Bierman, CSC, Victoria, British Columbia, V8W to 2Y2, Canada. E-mail: Department of Psychology, The Pennsylvania State University, 1100 [email protected] Moore Building, University Park, PA 16801. E-mail: [email protected]
delinquent activities (Coie & Dodge, 1998). In elementary school, peer victimization is also linked with early aggression some studies (Buhs, Ladd, &dimensions Herald, 2006; research hasinbegun to differentiate several of Crick, Murray-Close, Marks, & Mohajeri-Nelson, 2009) school climate including interpersonal relationships and with emotional dysregulation (Schwartz, Proctor, & among teachers, students, parents and principals; organiChien, 2001) and internalizing problems in others zational aspects of school functioning; quality of instruc(Hawker & Boulton, 2000).school Peer victimization does not tion; access to resources; safety; environmental appear to increase future risk for delinquent activities conditions; and school values (Bear, Gaskins, Blank, in & the same way as peer rejection Chen, 2011; Brand, Feiner, Shim,(Pouwels Seitsinger,& &Cillessen, Dumas, 2013), but instead, victimization with 2003; Cohen et al., 2009; Emmons,appears Haynes,linked & Comer, long-term elevations in social avoidance and emotional 2002; Freiberg, 1998; Van Houtte, 2005). Measures of distress (Boivin, typically Hymel, &assess Bukowski, 1995; Buhs et al., school climate individuals’ (students’, 2006; Nishina, Juvonen, & Witkow, 2005). Although teachers’, or parents’) perceptions of school climate, peer and rejection and used victimization interdependent these are often at the levelrepresent of individuals’ responses forms of peer adversity (HanishRecchia, & Guerra, 2004), their (see meta-analysis by Steffgen, & Viechtbauer, developmental precursors and adolescent outcomes may
2
LEADBEATER ET AL.
2013) but are also sometimes aggregated at the school or classroom level, in large-scale studies including many schools (e.g., Khoury-Kassabri, 2011; McCoy, Roy, & Sirkman, 2013). Considerable research has shown that positive school climate is associated with enhancements in school engagement and academic success of students (e.g., Goddard, Sweetland, & Hoy 2000; O’Malley et al., 2012) and in the engagement of their parents (Goldkind & Farmer, 2013). School climate is also associated with reductions in an array of concerns including school violence (Furlong & Morrison, 2000; Manvell, 2012; Steffgen et al., 2013); deviant student lifestyles (Zaykowski & Gunter, 2012); students’ social, emotional, and behavioral risks (Gottfredson, Gottfredson, Payne, & Gottfredson, 2005; Klein, Cornell, & Konold, 2012; Kuperminc, Leadbeater, Emmons, & Blatt, 1997; Wang & Dishion, 2012); and peer victimization and bullying (e.g., Goldstein, Young, & Boyd, 2008; Gottfredson et al., 2005; Kasen, Berenson, Cohen, & Johnson, 2004; Khoury-Kassabri, 2011; Khoury-Kassabri, Benbenishty, Zeira, & Astor, 2004; Swearer et al., 2006). Despite this extensive and growing literature, however, most research is cross-sectional (see McCoy et al., 2013, for an exception), and school climate is conceptualized as a static characteristic of the school environment. Moreover, although most assessments rely on individuals’ reports of components of school climate, little is known about the stability of or variability in individuals’ perceptions of school climate, how their perceptions change over time, or what factors influence individuals’ perceptions of school climate. In this three-wave longitudinal study, we investigate how peer victimization influences stability and change in students’ and their parents’ perceptions of school climate. Children who are victimized are unlikely to share the average students’ perceptions of many aspects of school climate, and children who continue to experience victimization over time may show worsening perceptions in salient dimensions of school climate. Both attachment and information-processing theories suggest that children develop internalized representations or “working models” of their socialization experiences that are influenced by the ways in which they are treated by peers and important adults in their social networks. These cognitive representations may be generalized to new experiences and color children’s beliefs and expectations of their social environments over time (Ladd, Ettekal, KochenderferLadd, Rudolph, & Andrews, 2014). Experiences of peer victimization, teasing, and rejection may, in particular, enhance children’s negative expectations of student–peer relationships at school (Asher, Rose, & Gabriel, 2001) but may also extend to other components of school climate. For example, chronic victimization may lead children (and possibly their parents) to believe that principals, teachers, and
SCHOOL CLIMATE AND PEER VICTIMIZATION
481
peers cannot be counted on to be caring and respectful and, given their own experiences, that not all students in the school are treated fairly. Influenced by their children’s experiences, parents of victimized children may also have negative perceptions of their child’s school climate that worsen over time. SCHOOL CLIMATE AND PEER VICTIMIZATION Growing research has found that peer victimization is related to school climate dimensions that tap student– teacher relationships and the application of consistent, clear, and fair rules. Positive relationships between students and teachers were associated with less physical and verbal/relational bullying in several studies (Flaspohler, Elfstrom, Vanderzee, Sink, & Birchmeier, 2009; Richard, Schneider, & Mallet, 2012). In a study of Israeli fourthto sixth-grade students, Khoury-Kassabri (2011) also found that although levels of victimization varied between classes and schools, the majority of variation was at the individual level. Moreover their study found that physical and emotional victimization by teachers affected levels of students’ reports of victimization by their peers. Goldstein et al. (2008) found that exposure to relational victimization was associated with adolescents’ perceptions of their schools’ safety and social atmosphere. Similarly, social support from teachers and from peers buffered the impact of bullying experiences on victimized children’s quality of life (RasKauskas, Gregory, Harvey, Rifshana, & Evans, 2010) and on adolescents’ reports of internalizing symptoms (Yeung & Leadbeater, 2010). Roland and Galloway (2004) also found that primary school students reported lower levels of victimization when they perceived that the school policies and rules were clear, consistent, and fair, and that student–teacher relationships were positive. Classroom norms regarding aggression and bullying also influence students’ tendencies to exhibit aggressive behaviour (Rodkin & Hodges, 2003). It is clear that perceptions of school interpersonal relationships and fairness characterize an important facet of school climate and have implications for reducing or exacerbating peer victimization. However, given the lack of longitudinal research, the direction of effects between school climate and peer victimization is not known. In this longitudinal study, we examined the effects of individual students’ reports of victimization and their parents’ perceptions of school interpersonal relationships and fairness across changes in grades and teachers, and across time within a grade. Stability across both grade and time would suggest that child and parent perceptions of school climate are not only a function of current classroom experiences. Children’s experiences of peer victimization are also expected to influence child and parent perceptions of school climate. Specifically,
482
LEADBEATER ET AL.
bidirectional influences between school climate dimensions and children’s reports of peer victimization were expected, as victimized children’s (and their parents’) school climate perceptions worsen over time. Data from Canadian children in Grades 3 and 4 and their parents were assessed across three waves of data collected during two school years (spring 2011, fall 2011, and spring 2012). We tested autoregressive, cross-lagged, path models (Kline, 2005) to examine how interpersonal and fairness dimensions of school climate related to peer victimization across the three waves of data. All analyses controlled for the influence of age, sex, maternal education, and the presence or absence of the WITS peer victimization prevention program. The WITS Program (see www.witsprogram.ca) is a whole school program that engages community members, school staff, students, and parents to create communities of individuals who are responsive when children report victimization or peer conflict. The program had not been fully implemented when data were collected, and investigations of its effects are beyond the scope of this article. Hence, we controlled any effect of being or not being in a program school on peer victimization and on the school climate dimensions at each time point. Although no differences were anticipated, cross-lagged models were also assessed for relational and physical victimization separately, but results showed few differences, hence types of victimization are combined in all analyses. Sex differences were not indicated in past research and are also controlled in our analyses. METHOD
SCHOOL CLIMATE AND PEER VICTIMIZATION
3
486 children who moved from Grade 2 into Grade 3 between T1 and T2) and 1,208 children at T3. Hence, data from T1 to T2 represent data collected from the same children across a change in grade and classroom, whereas data from T2 to T3 represent data from the children in the same classroom at the beginning and end of the school year. Demographic information was also collected from parents at baseline. The majority (95%) of participants were European Caucasian, 2% Aboriginal, 1% Asian, and 1% of mixed ethnicity; less than 1% were Hispanic, Indo-Canadian, or African Canadian. Parents indicated that 74% of children lived in a two-parent household; 25% of mothers and 32% of fathers had no postsecondary education after high school; 45% of mothers and 46% of fathers completed “some college or technical training” beyond high school; and 30% of mothers and 22% of fathers had some postsecondary education. Ninety-two percent of the children had attended a maximum of two schools in their lifetime, and 8% had attended three or more schools. Attrition was minimal across the three waves of data collection. Selective attrition was assessed by testing for differences at T1 on demographic and key variables for the T1 group of 800 children who remained in the longitudinal study and those who did not participate at T2 (n = 63, 7.9%; 54% boys) and T3 (n = 84, 10.5%; 54% boys). Findings did not reveal any significant demographic differences. Children who dropped out of the study reported slightly lower student-interpersonal relations at T1 (M = 2.26, SD = .50) than students who remained in the study (M = 2.39, SD = .48), t(798) = –2.43, p = .02; d = .27.
Participants Children were recruited from 137 classrooms in 27 predominately rural schools across three provinces in Canada. They were involved in the first three waves of an ongoing evaluation of the WITS Program for the prevention of peer victimization (see Leadbeater & Sukhawathanakul, 2011). At Time 1 (T1) 1,659 children participated, and an additional 105 joined the study at Time 2 (T2) due to delays in receiving consent forms back from parents (n = 1,764). Children completed the peer victimization scale at each wave. However, children completed the child-reports for school climate scales only when they were in Grade 3 and higher. Parents of children at all grade levels completed the parent-report of the school climate (including 1,653 at T1, 99% of eligible parents; 1,566 at T2, 89% of eligible parents; and 1,389 at T3, 79% of eligible parents). Children ranged in age from 6.67 to 11.33 years (M = 9.32, SD = .74) at T1; 49% were boys. Data were collected from 800 children at T1; 1,217 children at T2 (including 731 of the original 800 children plus
Procedure Teachers sent home evaluation packages to parents of children in participating schools informing them of the study and seeking consent for participation. Parents provided written consent for their child’s participation and completed a demographic and school climate questionnaire and returned it to the school in a sealed envelope for pick-up by a research assistant. Data were collected from participating children in their classrooms. A trained research assistant read the child questionnaire items aloud to the class, and children completed the questionnaires individually and privately. A research assistant also circulated to check completion of the questionnaires and to help children who needed assistance. Measures Peer victimization was measured using the Social Experience Questionnaire (SEQ; Crick & Grotpeter,
4
LEADBEATER ET AL.
1996) that was modified for administration to young children. This scale was adapted from Crick and Grotpeter’s (1996) original 5-point scale (ranging from never to all the time) to facilitate ease of understanding for younger children (Leadbeater & Hoglund, 2009; Leadbeater, Hoglund, & Woods, 2003; Leadbeater & Sukhawathanakul, 2011). In the modified version, children rated how often they experienced relational victimization (e.g., “How often does another kid tell lies about you to make others not like you anymore?”) and physical victimization (e.g., “How often do you get pushed or shoved by another kid at school?”) on a 3-point scale: 0 (never), 1 (sometimes), and 2 (almost all the time). Each subscale contained five items. Prior research has demonstrated measurement invariance across time, gender, and grade level among elementary school children using this revised scale (Desjardins, Yeung Thompson, Sukhawathanakul, Leadbeater, & MacDonald, 2013). Two practice items were also used to familiarize children with the response scale. During administration, adults read the SEQ items aloud, instructed children to follow along using a brightly colored piece of paper, and circulated around the room to ensure that children are responding to appropriate items. Relational and physical peer victimization were associated at each time point (rs = .69, .70, and .72 consecutively) so the items were summed to create an overall peer victimization construct. The reliabilities were adequate at each time point (αs = .86, .88, and .89 consecutively). School climate was measured using the School Climate Survey (Haynes, Emmons, & Comer, 1993) to assess the general social environment of the school and the quality of relationships that exist among students and adults in the school building. The elementary and middle school student version includes 30 items that assess five aspects of school climate. Student–teacher relations refers to the level of caring, respect, and trust that exists between students and teachers (e.g., “My teachers care about me”). Student interpersonal relations refers to the levels of caring, respect, and trust that exists among students in the school (e.g., “Children at my school respect one another”). Fairness refers to the equal treatment of students regardless of ethnicity and socioeconomic status (e.g., “At my school, boys and girls are treated equally well”). Equity in sharing of resources refers to equal student opportunity to participate in school activities, and with materials and equipment (e.g., “The same children do not always get to use things, like a computer, a ball or a piano, when we play”). Students responded on a 3-point scale, according to how much they agree with the statement (1 = disagree, 2 = not sure, 3 = agree). The reliabilities for the subscales were adequate at each time point (αs = .70 – .88). Using the same 3-point scale, the frequency of parent participation in school activities was assessed as a total of children’s “agree” response to five
SCHOOL CLIMATE AND PEER VICTIMIZATION
483
items (e.g., “Parents often come to my school to help with special projects”). The parent version of the School Climate Survey consists of 23 descriptive statements about prevailing school conditions. Items represented four aspects of school climate. Student–teacher relations refers to the level of caring, respect, and trust that exists between students and teachers in the school (e.g., “Teachers at my child’s school help my child with school problems”). Principal caring and sensitivity refers to the extent to which the principal shows consideration for the students, parents, and school staff and cares about their needs (e.g., “At my child’s school, the principal cares about the needs of the children”). School–community relations refer to the support and involvement of the community in the life of the school (e.g., “Community members work with staff at my child’s school to help improve the school”). Parents rated on a 5-point scale how well they think the statement describes their child’s school, from 1 (strongly disagree) to 5 (strongly agree). The reliabilities for the subscales were adequate at each time point (αs = .87–.95). Using the same 3-point scale, the frequency of parent participation in school activities was assessed as a total of parents’ “agree” and “strongly agree” responses to four items (e.g., “I often help with special school projects such as bus trips, fund-raisers etc.”). Correlations between the two overlapping child and parent school climate dimensions (i.e., student–teacher relations and frequency of parent participation) were modest. Specifically, concurrent correlations between child and parent reported student–teacher relations were .24, .26, and .25 (ps < .05), respectively. Concurrent correlations between child- and parent-reported frequency of parent participation were .30, .12, and .20 (ps < .05), respectively. Demographic covariates. We included a series of four demographic covariates in our analyses: mother’s education (used as a proxy for socioeconomic status), program status (i.e., whether schools had a peer victimization prevention program), child’s age, and sex. Mean levels of peer victimization and school climate variables were similar for girls and boys, except that girls reported significantly more positive student–teacher relations and parent involvement at baseline. Correlations for demographic covariates and study variables were small (rs = –.05 to –.21). Data Analytic Strategy We employed autoregressive, cross-lagged path analyses to assess the associations and directionality of links between children’s experiences of peer victimization and each dimension of school climate. Each model included
484 5 1–3 1–3 0–4 0–2 0–2 0–2
2.36 (.52)
2.73 (.41) 2.29 (1.16) .56 (.45) .51 (.45) .51 (.46)
.36**
.16**
.36** .24** .41**
.48**
.16**
.48** .24** .55**
.68**
.26**
—
1
.12**
.37**
.21** .03 .20**
.20**
.46**
.27** .05 .26**
.26**
—
2
.44**
.20**
.40** .25** .38**
.59**
.19**
.52** .25** .47**
—
3
.31**
.22**
.47** .25** .31**
.40**
.21**
— .28** .36**
4
.14**
.01
.19** .48** .11**
.19**
.02
— .17**
5
.41**
.21**
.46** .18** .51**
.63**
.28**
—
6
.20**
.51**
.26** .03 .27**
.27**
—
7
.57**
.23**
.50** .26** .43**
—
8
.34**
.23**
— .26** .33**
9
.19**
.03
— .14**
10
.67**
.40**
—
11
.32**
—
12
—
13
14
15
16
17
.88 .28** .22** .32** .47** .18** .31** .26** .33** .56** .21** .54** .35** .52** — NA .21** .003 .17** .24** .48** .15** .06* .18** .19** .54** .15** .05 .24** .28** — .86 –.38** –.32** –.45** –.27** –.02 –.28** –.24** –.27** –.23** −.04 −.22** −.23** −.21** −.19** −.05 — .88 –.29** –.23** –.32** –.24** –.04 –.38** –.30** –.41** –.31** −.03 −.30** −.27** −.31** −.25** −.05 .57** – .89 –.22** –.18** –.28** –.21** –.05 –.28** –.25** –.30** –.29** −.06 −.43** .33** −.44** −.38** −.06 .49** .64**
.86
.79
.84 NA .80
.84
.72
.84 NA .77
.83
.70
.78
α
Note: T = Time; α = Cronbach’s alpha; NA = not applicable. *p < .05. **p < .001.
1–3
2.21 (.64)
1–3
2.33 (.57)
1–3 0–4 1–3
1–3 0–4 1–3
2.77 (.36) 2.33 (1.14) 2.41 (.52)
2.78 (.34) 1.98 (.95) 2.42 (.55)
1–3
2.38 (.48)
1–3
1–3
2.30 (.58)
2.39 (.48)
1–3
2.40 (.54)
1. Fairness T1 2. Equal Access to Resources T1 3. Student Interpersonal Relations T1 4. Student–Teacher Relations T1 5. Parent Involvement T1 6. Fairness T2 7. Equal Access to Resources T2 8. Student Interpersonal Relations T2 9. Student–Teacher Relations T2 10. Parent Involvement T2 11. Fairness T3 12. Equal Access to Resources T3 13. Student Interpersonal Relations T3 14. Student–Teacher Relations T3 15. Parent Involvement T3 16. Peer Victimization T1 17. Peer Victimization T2 18. Peer Victimization T3
Range
M (SD)
Variable
TABLE 1 Correlations Between Child School Climate And Peer Victimization Variables
6
SCHOOL CLIMATE AND PEER VICTIMIZATION
LEADBEATER ET AL.
autoregressive pathways, within-time correlations, and cross-lagged pathways between school climate dimensions and peer victimization for the T1–T2 and T2–T3 waves. Autoregressive, cross-lagged path analysis accounts for the temporal sequencing of the variables, controlling for the level of the previous measurement occasion, and permits examination of the stability within variables as well as time-specific relations among the variables (Kline, 2005). To improve the statistical validity and model fit of the cross-lagged approach, robust standard errors that take into account the nesting of students within schools were used in the analyses (Muthén & Muthén, 2000). To control for demographic influences, mother’s education, age, sex, and program status were included as covariates with each construct at each time point. Following established guidelines (Bollen, 1989; Byrne, 2010; Hu & Bentler, 1999; Kline, 2005; Schreiber, Stage, King, Nora, & Barlow, 2006), model fit for all analyses were evaluated using the comparative fit indices (CFI), root mean square error of approximation (RMSEA), and standardized root-mean of the residual (SRMR). CFI values of >.95 represent an excellent fitting model and >.90 represent an adequate model fit; RMSEA and SRMR values
