Running head: TRACKING VICTIMIZATION AND COGNITION Tracking the Trajectories of Peer Victimization and Negative Self-Cognitions in Children: A Longitudinal Approach to Approximating Causality Maysa Kaskas Dr. David Cole lab 1 TRACKING VICTIMIZATION AND COGNITION 2 Abstract This study investigates the effect of targeted peer victimization (TPV) on negative selfcognitions as a function of victimization type through a three-wave longitudinal study. Measures of TPV and cognitions were collected from 956 students equally distributed through grades three through eight. Two types of TPV were analyzed: physical peer victimization, which occurs when a person is controlled or injured by physical means, and relational peer victimization, which includes acts or statements intended to harm peer relationships and social acceptance. Obtained results are as follows: (a) both negative self-cognitions and TPV have high rates of stability over time; (b) while both genders experience higher levels of relational victimization than physical victimization, females experience more relational victimization than males do, and males experience more physical victimization than females do; and (c) relational victimization was more related to changes in depressive self-cognitions than was physical victimization. However, neither type of PV was significant after controlling for prior levels of the cognitive variable. Implications and avenues for future research are discussed. TRACKING VICTIMIZATION AND COGNITION 3 Introduction Targeted peer victimization, or bullying, is defined as “the experience among children of being a target of the aggressive behavior of other children” (Hawker & Boulton, 2000, p. 441). This experience is relatively common, with around 30-60% of schoolage children falling victim each semester, and it can have adverse effects on a child’s thoughts and emotions (Rigby, 2000). Research has shown that there are academic, social, emotional, and psychological consequences of peer victimization from social anxiety to an increased risk of suicide has been linked to bullying (Malti, Perren, & Buchmann, 2010). However, current research on peer victimization has failed to show effects over time across middle childhood and early adolescence, the period when rates of victimization peak (Crick, Casas, & Ku, 1999). This study will investigate the persistence and stability of physical and relational peer victimization and its effects on selfesteem over time. There are two main types of peer victimization. Overt or physical peer victimization occurs when a person is controlled or injured by physical means. This includes acts of violence such as punching, kicking, and hitting, but it also includes threats of future or potential physical harm (Cole et al., 2010). While the second type of peer victimization may not leave visible bruises, covert or relational peer victimization may have a stronger and more enduring effect on a child’s sense of self and mental health. This type of peer victimization includes acts or statements intended to harm peer relationships and social acceptance, typically involving exclusion from social groups or activities and rumor spreading. Relational peer victimization behaviors often take a more subtle form than physical types of bullying, making its detection harder and its long-term impact more insidious and less studied than physical peer victimization (Sinclair et al., 2012). This type of bullying also may have a stronger effect on internalizing TRACKING VICTIMIZATION AND COGNITION 4 problems like depression and shape the way children think about themselves (i.e., selfcognitions), as it involves victims receiving negative self-relevant information from peers on social status, competencies, and personal flaws (Hunter & Boyle, 2002). The time period of middle childhood to early adolescence is critical as it has the highest rates of peer victimization and is also the period where victimization has the strongest potential to harm a child’s self-cognitions (Crick et al., 1999). One of the major developmental tasks during this time is the construction of a self-concept, a task which can be seriously affected by negative feedback from peers (Cole et al., 2010). The looking glass model suggests that perceptions of the self are derived heavily from our perceptions of others’ views of us, and support for this model is strongest in adolescents and children (Cooley, 1902; Cole, 1991). Some children navigate the crucial task of developing a self-concept well and develop positive selfcognitions, which can serve as a protective factor against depression, while others develop negative self-cognitions, which can predispose them to depression (Cole, Martin, Powers, & Truglio, 1996; Burt, Obradovic, Long, & Martin, 2008). Not surprisingly, middle childhood to early adolescence is also a key period for the emergence of depression, sealing its significance as a period in need of more in-depth research (Cole et al., 2010). The overarching purpose of this study is to examine the reciprocal relations of physical and relational peer victimization with negative self-cognitions over time by analyzing a dataset that spans three years and follows the victimization and cognition patterns of almost one thousand third through eighth grade students. How time plays a moderating role in the frequency of peer victimization and in the development and maintenance of negative self-cognitions is unclear. Tran, Cole, and Weiss (2012) found that symptoms of depression predicted changes in physical and relational peer victimization but neither type of peer victimization predicted TRACKING VICTIMIZATION AND COGNITION 5 changes in depression symptoms; however, negative self-cognitions are inherently internal experiences, unlike symptoms of depression, which are expressed and observable to others. A child showing outward symptoms of depression may be viewed as weak and perceived as an easy target for bullies; this theory is supported by Tran et al. (2012) results, as they found that boys were particularly likely to be victimized physically if they showed symptoms of depression, perhaps due to gender role expectations. The finding that peer victimization does not predict changes in symptoms of depression contradicts previous studies, suggesting a need for more research with careful attention to moderators. Most of the previous research on the effects of peer victimization is cross-sectional (Cole et al., 2010; Hawker & Boulton, 2000). Without conducting a longitudinal study, prior levels of victimization and self-cognitions cannot be statistically controlled, resulting in poor models for studying the relation between peer victimization and cognition, and larger trends cannot be tracked over time. A longitudinal study is needed to see whether (a) cognitions can worsen or be repaired with varying levels of peer victimization over time, (b) how stable peer victimization is for individuals and for the cohort through the course of middle childhood to early adolescence, and (c) whether changes between subtypes of peer victimization occur. Existing longitudinal studies have not been able to answer these questions, as they either ignore the subtype of peer victimization, duration of the victimization, grade level of the victims, or look at depression rather than self-cognitions (Sinclair et al., 2012; Tran et al., 2012). The specific goals of this study are (1) to track the trajectories of peer victimization over the course of the three wave longitudinal study, (2) to track the trajectories of negative selfcognitions over the period of three years, (3) to evaluate their relation to each other over time, and (4) to evaluate the possible moderating effects of gender and grade level on the formation TRACKING VICTIMIZATION AND COGNITION 6 and maintenance of negative self-cognitions. In light of recent research on the harmful effects of peer victimization on self-esteem, negative self-cognitions are predicted to be more prevalent, stronger, and more stable across the three time points in children who are experiencing relational peer victimization than in those experiencing physical peer victimization and their nonvictimized peers. Once established, negative self-cognitions are expected to remain stable or worsen if peer victimization becomes more frequent or intense over time. However, if peer victimization diminishes between any of the waves of data, negative self-cognitions may be replaced by healthier self-cognitions. Children experiencing peer victimization at a prior time point will, on average, continue to experience peer victimization for the following time points. In addition, we also expect moderating effects of gender and grade level. As boys are less likely to seek social support when stressed and girls tend to be more invested in social networks (Rose & Rudolph, 2006), we anticipate that girls will be more likely to suffer overtly when their social status declines and support system weakens, perhaps reporting more frequent negative self-cognitions. However, the problems of boys might be compounded by their hesitance to seek help or support from others, perhaps causing them to self-report more persistent negative selfcognitions. A gender difference has been shown between the types of peer victimization most commonly experienced; boys are more likely to experience physical victimization than girls, but gender differences in relational victimization are found to be less consistent, though it is generally agreed that girls experience more relational peer victimization than do boys (Cole et al., 2010; Crick & Grotpeter, 1995). Not many studies have been done on the moderator of grade level or age on the experience and effects of peer victimization. This study hopes to find a sensitive period for occurrence of peer victimization as well as the development and maintenance of negative self- TRACKING VICTIMIZATION AND COGNITION 7 cognitions. Peer victimization is expected to peak around fifth or sixth grade then decrease by eighth grade, with a peak occurring slightly earlier for girls than for boys. Negative selfcognitions are anticipated to be persistent if peer victimization consistently occurs by the sixth grade. These hypotheses will be tested through a three-wave longitudinal study with self-report measures of peer victimization and self-report measures of cognitions. Method Participants We recruited participants from two suburban elementary schools and one middle school in middle Tennessee. Parental consent forms and letters describing the project were distributed to 626 third, fourth, fifth, and sixth graders. 421 returned completed consent forms giving us permission to include them in the study; of these students, 95% (400) were present on the day of data collection and were able to participate. At Time 2 (the next year), we returned to the same schools and sent home 656 consent forms to parents of children in grades four through seven. 470 students returned consent forms, and 88% (412) of them participated in the study. The consent form process was repeated at Time 3 (the third consecutive year); however, we were only permitted to consent the children who had participated in a previous year, greatly reducing our sample size. We distributed consent forms to 261 students in grades five through eight, 55% (144) of whom returned consent forms and were present to participate. In this three-wave longitudinal study, our total sample size of 956 contained two primary patterns of missing data: those who participated at one or both of the first two years but not all three time points (dropouts, 15%) and those who missed one of the first two years but participated in a subsequent time point (joiners, 12%). The primary reason for drop-outs was essentially the reverse of the primary reason for joiners, as families moved out of and into the school district. Overall, participants did TRACKING VICTIMIZATION AND COGNITION 8 not vary significantly from nonparticipants in comparisons of ethnicity, gender, and grade level. At all time points, participants were evenly distributed among genders (53.9% female) and grade levels (24.4% 3rd grade, 23.75% 4th grade, 27.63% 5th grade, 24.63% 6th grade, 24.95% 7th grade, and 21.5% 8th grade). We will use full information maximum likelihood statistics to avoid unnecessary sampling bias and to enhance the fidelity of parameter estimation, allowing us to include all participants. Measures Targeted peer victimization. The self-report measure was a six-item questionnaire designed to assess the occurrence and frequency of relational and physical peer victimization (created by expanding on the items used by Ladd & Kochenderfer-Ladd, 2002). Children were asked to rate the following relational items on a four-point scale [1 (never), 2 (rarely), 3 (sometimes), and 4 (a lot)]: (a) Tell others to stop being your friend, (b) Say you can’t play with them, and (c) Say mean things to other kids about you. The same process was used for the following physical items: (d) Kick you, (e) Hit you, and (f) Push you. Cognition. We used the Cognitive Triad Inventory for Children (Kaslow, Stark, Printz, Livingston, & Tsai, 1992), a 36-item self-report questionnaire evaluating children’s views of themselves (e.g., "I'm a failure"), their world (e.g., "The world is a very mean place"), and their future (e.g., "Nothing is likely to work out for me"). Responses are scored on three-point scales, as children mark yes, no, or maybe for each item. However, the CTI-C is a triad in name only; recent factor analysis indicates that a two-factor solution tends to emerge during middle childhood: an overall positive cognitions factor and an overall negative cognitions factor, which simplifies the measure to two subscales (Cole et al., 2010). The Children’s Automatic Thoughts Scale (Schniering & Rapee, 2002) assesses negative TRACKING VICTIMIZATION AND COGNITION 9 self-cognitions in youth by asking children to rate the frequency in which they’ve had fifty-six different negative thoughts in the previous week on a scale from “not at all” (1) to “all the time” (5). CATS includes four subscales to gain a more comprehensive view of the possible range of children’s self-cognitions: physical threat (such as “I’m going to get hurt”), social threat (e.g., “I’m afraid I will make a fool of myself”), personal failure (“It’s my fault that things have gone wrong”), and hostility (“I won’t let anyone get away with picking on me”). Procedure The time interval between each of the three waves of data collection was approximately one year. Prior to all three data collection times, we distributed consent forms to children in each participating classroom and asked them to take the forms home to their parents. To compensate participating classrooms, we offered a $100 donation to each classroom if 90% of the students returned their signed consent forms, either granting or denying permission for participation. Research assistants gathered consented students during regular school hours, separating them into small groups to administer the questionnaires. For the third and fourth grade students, a research assistant read the items aloud while allowing each participant time to answer the questions on their individual forms. Older students were allowed to work at their own pace after a research assistant introduced the measures. At all grade levels, research assistants circulated through the groups to answer questions throughout the process. At the end of the survey, participating students were given a decorated pencil and snacks for their participation. Results Preliminary Analyses Table 1 contains descriptive statistics for all study variables. Means and standard deviations are similar to those reported in previous research with community samples. TRACKING VICTIMIZATION AND COGNITION 10 Table 1: Descriptive statistics Variable Mean SD Min Max Wave A Relational PV 5.92 2.54 3 12 Physical PV 4.62 2.17 3 12 CATS 72.22 30.19 40 196 CTI-N 26.5 6.89 18 54 Wave B Relational PV 6.06 2.29 3 12 Physical PV 4.6 2.23 3 12 CATS 69.45 29.44 40 188 CTI-N 25.99 7.29 18 52 Wave C Relational PV 4.67 1.91 3 11 Physical PV 3.66 1.56 3 12 CATS 54.55 19.23 40 143 CTI-N 23.71 6.85 18 51 Note: PV=peer victimization; CATS=Children’s Automatic Thoughts Scale; CTI-N=Cognitive Triad Inventory for ChildrenNegative subscale Data analysis overview The specific goals of this proposed study are (1) to track the trajectories of peer victimization over the course of the three wave longitudinal study, (2) to track the trajectories of negative self-cognitions over the period of three years, (3) to evaluate their relation to each other over time, and (4) to evaluate the possible moderating effects of gender and grade level on the formation and maintenance of negative self-cognitions. The first two goals were addressed through a series of analyses including correlations and one-way ANOVA. The third and fourth goals were jointly addressed through cross-sectional and longitudinal tests of interaction. TRACKING VICTIMIZATION AND COGNITION 11 Goal 1 The aim of Goal 1 was to examine the stability of peer victimization over time. We took two approaches toward this end. First was to examine the over-time correlations of different types of peer victimization. Second was to examine the changes in mean-level victimization over time. Correlations. Pearson correlations appear in Table 2. Cross-wave stability correlations for relational victimization ranged from .605 to .528. Cross-wave stability correlations for physical victimization were somewhat lower, ranging from .393 to .291. TRACKING VICTIMIZATION AND COGNITION 12 Table 2: Pearson correlations. Measure 1. 2. 3. 4. 5. 1. Physical PV (A) 1.00 2. Relational PV (A) .380 1.00 3. Physical PV (B) .393 .245 1.00 4. Relational PV (B) .209 .528 .475 1.00 5. Physical PV (C) .291 .316 .551 .400 1.00 6. Relational PV (C) .237 .605 .462 .536 .431 6. 1.00 Note: A, B, and C denote the first, second, and third waves, respectively. PV=peer victimization. Means. The means for physical and relational victimization appear in Tables 3 and 4, respectively, broken down by grade and gender. We conducted one-way ANOVA tests to evaluate the mean differences in each subtype of victimization by grade level. After a series of these analyses, we determined that the mean differences between victimization scores for each grade level were generally nonsignificant at the 0.05 level with the notable exception of the data obtained from third grade students. The mean difference in both physical and relational victimization scores between grades 3 and every other grade were significant at the 0.05 level. The remaining mean differences between grade levels for physical victimization were not significant. However, the mean difference between grades 4 and 6 (p-value: 0.045) and grades 5 and 6 (p-value: 0.022) for relational victimization were significant at the 0.05 level, suggesting a need for additional analyses with closer attention to potential moderators. Figure 1 depicts the mean physical victimization scores by cohort, and Figure 2 shows the mean relational victimization scores by cohort. We further distilled the data by obtaining the TRACKING VICTIMIZATION AND COGNITION 13 mean victimization scores for each grade level averaged across cohorts. See Figure 3 for physical peer victimization and Figure 4 for relational peer victimization. In both sets of analyses, there is a generally decreasing trend in both physical and relational victimization scores across the genders and all four cohorts. These results support the first hypothesis. TRACKING VICTIMIZATION AND COGNITION 14 Table 3: Mean Physical Peer Victimization Score by Cohort Cohort 1 boys Cohort 1 girls Cohort 2 boys Cohort 2 girls Cohort 3 boys Cohort 3 girls Cohort 4 boys Cohort 4 girls Grade 3 Grade 4 Grade 5 Grade 6 Grade 7 Grade 8 5.7841 5.1111 3.9231 4.7692 4.3137 3.6071 4.9362 4.4603 3.4615 3.8776 4.3036 3.8824 4.6304 5.2857 4.3529 3.9455 3.8182 3.4 4.6389 4.5854 3.6429 4.57 5.1364 3.1765 Figure 1: Mean Physical Peer Victimization by Cohort Scatterplot Mean Physical Peer Victimization by Cohort 8 7 6 Mean PPV Score Cohort 1 boys Cohort 1 girls 5 Cohort 2 boys Cohort 2 girls 4 Cohort 3 boys 3 Cohort 3 girls Cohort 4 boys 2 Cohort 4 girls 1 0 3 4 5 6 Grade level 7 8 TRACKING VICTIMIZATION AND COGNITION 15 Table 4: Mean Relational Peer Victimization Score by Cohort Grade 3 Grade 4 Grade 5 Grade 6 Grade 7 Grade 8 6.5444 5.8889 4.1538 7.8173 6.68 4.6429 5.2128 6.0317 4.2308 6.6327 6.7143 5.8235 4.8913 5.8182 4.8824 5.6 5.5091 4.44 4.9167 5.4146 3.6429 5.82 6.2955 5.2941 Cohort 1 boys Cohort 1 girls Cohort 2 boys Cohort 2 girls Cohort 3 boys Cohort 3 girls Cohort 4 boys Cohort 4 girls Figure 2: Mean Relational Peer Victimization by Cohort Scatterplot Mean Relational Peer Victimization by Cohort 8 7 Mean RPV Score 6 Cohort 1 boys Cohort 1 girls 5 Cohort 2 boys Cohort 2 girls 4 Cohort 3 boys 3 Cohort 3 girls Cohort 4 boys 2 Cohort 4 girls 1 0 3 4 5 6 Grade level 7 8 TRACKING VICTIMIZATION AND COGNITION 16 Figure 3: Mean Physical Peer Victimization Score Mean Physical PV Score Mean Physical Peer Victimization Score 8 7.5 7 Physical Peer Victimization (Self-report) Means Male 6.5 6 Female Physical Peer Victimization (Self-report) Means Female 5.5 5 4.5 4 3.5 3 Grade 3 Grade 4 Grade 5 Grade 6 Grade 7 Grade 8 Overall gender mean A A and B A, B, and A, B, and B and C C C C Grade Level and Waves Involved A, B, and C TRACKING VICTIMIZATION AND COGNITION 17 Figure 4: Mean Relational Peer Victimization Score Mean Relational PV Score 8 Mean Relational Peer Victimization Score 7.5 Relational Male Peer Victimization (Self-report) Means Male 7 6.5 Relational Female Peer Victimization (Self-report) Means Female 6 5.5 5 4.5 4 3.5 3 Grade 3 Grade 4 Grade 5 Grade 6 Grade 7 Grade 8 Overall gender mean A A and B A, B, A, B, B and C C A, B, and C and C and C Grade Level and Waves Involved TRACKING VICTIMIZATION AND COGNITION 18 Goal 2 The aim of Goal 2 was to examine the stability of negative self-cognitions over time. We took two approaches toward this end. First was to examine the over-time correlations of different types of negative self-cognitions. Second was to examine the changes in mean-level negative self-cognitions over time. Correlations. Pearson correlations appear in Table 5. Cross-wave stability correlations for the Children’s Automatic Thoughts Scale (CATS) ranged from .421 to .351. Cross-wave stability correlations for the Cognitive Triad Inventory for Children-Negative subscale (CTI-N) were less variable, ranging from .452 to .457. Table 5 Pearson correlations Measure 1. 2. 3. 4. 5. 1. CATS (A) 1.00 2. CTI-N (A) .741 1.00 3. CATS (B) .421 .347 1.00 4. CTI-N (B) .403 .452 .745 1.00 5. CATS (C) .351 .421 .602 .646 1.00 6. CTI-N (C) .356 .457 .596 .689 .789 6. 1.00 Note: A, B, and C denote the first, second, and third waves, respectively. CATS=Children’s Automatic Thoughts Scale; CTI-N=Cognitive Triad Inventory for Children- Negative subscale TRACKING VICTIMIZATION AND COGNITION 19 Means. The means for CATS and CTI-N appear in Tables 6 and 7, respectively, broken down by grade and gender. We conducted one-way ANOVA tests to evaluate the mean differences in each type of negative self-cognition by grade level. After a series of these analyses, we determined that the mean differences between cognition scores for each grade level were generally nonsignificant at the 0.05 level with the notable exception of the data obtained from third grade students. The mean difference in both CATS and CTI-N scores between grades 3 and every other grade were significant at the 0.05 level. The remaining mean differences between grade levels for both CATS and CTI-N were not significant. Figure 5 depicts the mean CATS scores by cohort, and Figure 6 shows the mean CTI-N scores by cohort. We further distilled the data by obtaining the mean negative self-cognition scores for each grade level averaged across cohorts. See Figure 7 for CATS and Figure 8 for CTI-N. In both sets of analyses, there is a slight decreasing trend in both CATS and CTI-N scores across the genders and all four cohorts. These results support the second hypothesis. TRACKING VICTIMIZATION AND COGNITION 20 Table 6: Mean Children’s Automatic Thoughts Scale Score by Cohort Cohort 1 boys Cohort 1 girls Cohort 2 boys Cohort 2 girls Cohort 3 boys Cohort 3 girls Cohort 4 boys Cohort 4 girls Grade 3 Grade 4 Grade 5 Grade 6 Grade 7 Grade 8 75.5619 66.8431 53.3846 89.7027 73.8952 50.8608 62.3731 67.4827 50.0907 71.7134 71.9133 64.9412 68.4938 74.44509876 59.4118 65.9446 65.0099 49.2746 67.0681 61.8937 48.9595 78.3079 73.3269 61.4133 Figure 5: Mean Children’s Automatic Thoughts Scale Score by Cohort Scatterplot Mean Children's Automatic Thoughts Scale Score by Cohort 100 90 Cohort 1 boys Cohort 1 girls Cohort 2 boys Cohort 2 girls Cohort 3 boys Cohort 3 girls Cohort 4 boys Cohort 4 girls 80 Mean CATS Score 70 60 50 40 30 20 10 0 3 4 5 6 Grade level 7 8 TRACKING VICTIMIZATION AND COGNITION 21 Table 7: Mean Cognitive Triad Inventory-Negative Score by Cohort Cohort 1 boys Cohort 1 girls Cohort 2 boys Cohort 2 girls Cohort 3 boys Cohort 3 girls Cohort 4 boys Cohort 4 girls Grade 3 Grade 4 Grade 5 Grade 6 Grade 7 Grade 8 28.4015 24.4941 22.3627 29.4508 25.8047 23.5948 25.3031 25.6120 22.2308 25.6783 26.6744 24.4118 25.5946 27.0252 25.6471 25.3995 25.4267 22.6471 25.6340 25.2250 21.4286 27.1635 27.3222 26.7889 Figure 6: Mean Cognitive Triad Inventory-Negative Score by Cohort Scatterplot Mean Cognitive Triad Inventory-Negative Score by Cohort 35 30 Cohort 1 boys Cohort 1 girls Cohort 2 boys Cohort 2 girls Cohort 3 boys Cohort 3 girls Cohort 4 boys Mean CTI-Neg Score 25 20 15 10 5 0 3 4 5 6 Grade level 7 8 TRACKING VICTIMIZATION AND COGNITION 22 Figure 7: Mean Children’s Automatic Thoughts Scale Score Mean Reported CATS Score 95 90 CATS Means Male 85 Female Mean CATS Score 80 CATS Means Female 75 70 65 60 55 50 45 Grade 3 Grade 4 A A and B Grade 5 Grade 6 Grade 7 A, B, and C A, B, and C B and C Grade Level and Waves Involved Grade 8 Overall gender mean C A, B, and C TRACKING VICTIMIZATION AND COGNITION 23 Figure 8: Mean Cognitive Triad Inventory-Negative Subscale Score Mean Reported CTI-Negative Score 30 29 28 Male CTI-Negative Means Male Mean CTI-Negative Score 27 26 CTI-Negative Female Means Female 25 24 23 22 21 20 Grade 3 Grade 4 Grade 5 Grade 6 Grade 7 Grade 8 Overall gender mean A A and B A, B, and A, B, and B and C C C Grade Level and Waves Involved C A, B, and C TRACKING VICTIMIZATION AND COGNITION 24 Goals 3 and 4 The overarching aim of Goals 3 and 4 was to examine the relation of negative selfcognitions to peer victimization over time. Toward this goal, we analyzed correlations and conducted a series of regression analyses. In each analysis, a cognition measure was regressed onto that measure at a previous time point, a measure of peer victimization, grade level, and the PV x grade level interaction. These analyses were then repeated, replacing grade level with gender. Regressions. Regressions appear in Tables 8-11, with Pearson correlations appearing in the first column of each table. The correlations for the Cognitive Triad Inventory-Negative subscale ranged from .146 for Physical PV to .255 for Relational PV. The correlations for the Children’s Automatic Thoughts Scale ranged from .181 for Physical PV to .322 for Relational PV. Results indicated that relational victimization was more closely associated with negative cognitive outcomes than was physical victimization. In the regression analyses, however, when we statistically controlled for prior levels of each cognitive variable, the effect of both relational and physical victimization on cognition was no longer significant. Tables 8-11 also show the effect of grade level on the cognitive measures; however, grade did not have a significant effect for either cognitive measure or type of victimization. In analyses not shown, we replaced grade with gender. No main effects for gender were significant. In other analyses not shown here, we also discovered that neither the gender x PV nor the grade x PV interactions were significant, contrary to our hypotheses. TRACKING VICTIMIZATION AND COGNITION 25 Table 8: Cognitive Triad Inventory-Negative subscale and Relational Victimization Regression Predictor r B Intercept 11.458 CTI-N (A) .452 .434 Rel PV (A) .255 .343 Grade level .050 .527 Dependent variable: CTI-N (B) SE (B) 2.566 .073 .512 .395 Beta .430 .049 .084 t 4.465 5.982 .670 1.333 Sig .000 .000 .504 .184 Table 9: Children’s Automatic Thoughts Scale and Relational Victimization Regression Predictor r B Intercept 48.346 CATS (A) .421 .318 Rel PV (A) .322 2.091 Grade level -.070 -.828 Dependent variable: CATS (B) SE (B) 8.575 .071 2.335 1.565 Beta .370 .076 -.034 t 5.638 4.479 .896 -.529 Sig .000 .000 .372 .597 Table 10: Cognitive Triad Inventory-Negative subscale and Physical Victimization Regression Predictor r B Intercept 10.981 CTI-N (A) .452 .464 Phys PV (A) .146 -.072 Grade level .050 .460 Dependent variable: CTI-N (B) SE (B) 2.554 .066 .451 .385 Beta .459 -.010 .073 t 4.299 7.036 -.161 1.195 Sig .000 .000 .873 .234 Table 11: Children’s Automatic Thoughts Scale and Physical Victimization Regression Predictor r B Intercept 47.307 CATS (A) .421 .349 Phys PV (A) .181 .970 Grade level -.070 -1.102 Dependent variable: CATS (B) SE (B) 8.460 .058 1.789 1.528 Beta .406 .036 -.045 t 5.592 6.055 .542 -.721 Sig .000 .000 .588 .472 TRACKING VICTIMIZATION AND COGNITION 26 Discussion Three major findings emerged from this study. First, in a sample of children and adolescents, levels of both peer victimization and negative self-cognitions showed high rates of stability over time. Second, although both genders experience higher levels of relational victimization than physical victimization, females experienced more relational victimization than males, and males experienced more physical victimization than females. Third, relational victimization was more related to changes in negative self-cognitions than was physical victimization; however, neither type of PV was significant after controlling for prior levels of the cognitive variable. Each of these results will be further discussed below. First, the data supported our hypotheses that peer victimization and negative selfcognitions would be fairly stable over time. Relational peer victimization was more stable over time than physical victimization. The two measures of negative self-cognition, CATS and CTIN, were highly correlated with each other (ranging from .789 to .741) and did not significantly differ in cross-wave stability. Second, multiple analyses lent support to the hypothesis that females would experience higher levels of relational victimization than males would and males would experience more physical victimization than females would. This result was found in the correlations, means obtained in the cohort analyses, distilled means averaged across grade levels, and in the regression analyses. The data also indicate that, on average, children and adolescents experience more relational than physical victimization overall, regardless of their grade level or gender. These results add longitudinal support to previous research (Cole et al., 2010; Tran et al., 2012). A related result shows that at each time point, correlations between victimization and self-cognition were consistently larger for the relational peer victimization measure, suggesting TRACKING VICTIMIZATION AND COGNITION 27 that relational victimization has a stronger and more pervasive effect on measures of negative self-cognition than does physical victimization. These findings provide additional support for the harmful effects of relational victimization on a child or adolescent’s sense of self (Sinclair et al., 2012). However, the effect of both types of peer victimization on cognition disappeared after we statistically controlled for prior levels of the cognitive variable. While the correlations between peer victimization and cognitions were significant, their significance dropped off when we controlled for prior levels of the dependent variable. The patterns of nonsignificance in this study are noteworthy. Contrary to our hypotheses, grade level did not have a significant moderating effect on the interaction between either type of peer victimization and measures of negative self-cognition. While there was a significant difference between the scores of third graders and those of the remaining grades, this result may be better explained by the limitations of using self-report in a sample of third grade students than by any psychological principle. The overall nonsignificance of grade level is also related to the high mean level stability of each of the variables, as both victimization and cognition were strongly correlated over time. In a similar vein, the same patterns of nonsignificance were found for the gender variable. Implications and Future Research The implications for these findings are varied and interdisciplinary: school systems could benefit from the data on the rates and effects of peer victimization, and clinical psychology could benefit from the data on possible precursors to childhood and adolescent depressive symptoms. While this study yields results with interdisciplinary significance, there are a number of limitations. Our longitudinal study did not collect information on the aggressors or bullies themselves; therefore, we cannot distinguish between subtypes of aggressors, identify their TRACKING VICTIMIZATION AND COGNITION 28 motivations, or target specific interventions to stop peer victimization. In addition, this is not an experimental study. Without random assignment to treatment and control conditions, it is difficult to determine cause and effect, as there are many unanticipated variables which can affect outcomes. While strong correlations between peer victimization and negative selfcognitions were found, correlation does not imply causation. The current study also relies heavily upon self-report for measures of cognitions, which can result in under-reporting (due to stigma, worries about confidentiality, or biased recall) or over-reporting (due to negative emotions at the time of the study, biased recall, or a desire to maximize support or sympathy). Finally, we did not collect information on the children’s support system and resiliency factors. The children who are experiencing peer victimization may not develop negative selfcognitions if they have other support systems such as family members, friends in their neighborhood, or teachers. Certain children may have other resiliency factors such intelligence, a sense of humor, or hobbies which diminish the impact of peer victimization. A future study might ask children to complete a questionnaire with a Likert scale ranking the quality of their friendships at school, their neighborhood, and among their family members. These subjects could then complete a Likert scale ranking the importance of these peer and non-peer relationships to them, allowing researchers to understand the value that victimized and nonvictimized children place on these relationships. In addition, subjects could rank the importance of nonsocial domains such as academics, sports, and hobbies; children who value their academics, athletics, or other hobbies more than peer relationships may not be affected as severely by peer victimization, regardless of the type or duration. TRACKING VICTIMIZATION AND COGNITION 29 References Burt, K. B., Obradovic, J., Long, J. D., & Masten, A. S. (2008). The interplay of social competence and psychopathology over 20 years: Testing transactional and cascade models. Child Development, 79, 359–374. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, N.J.: L. Erlbaum Associates. Cole, D. A. (1991). Change in self-perceived competence as a function of peer and teacher evaluation. Developmental Psychology, 27, 682–688. Cole, D. A., Martin, J. M., Powers, B., & Truglio, R. (1996). Modeling causal relations between academic and social competence and depression: A multi-trait multi-method longitudinal study of children. Journal of Abnormal Psychology, 105, 258–270. Cole, D. A., Maxwell, M. A., Dukewich, T. L., & Yosick, R. (2010). Targeted peer victimization and the construction of positive and negative self-cognitions: Connections to depressive symptoms in children. Journal of Clinical Child and Adolescent Psychology, 39(3), 421 436. Cooley, C. H. (1902). Human nature and the social order. New York, NY: Scribner’s. Crick, N. R., Casas, J. F., & Ku, H. (1999). Physical and relational peer victimization in preschool. Developmental Psychology, 35, 376–385. Harter, S. (1985). Manual for the self-perception profile for children. Denver, CO: University of Denver. Hawker, D. S. J, & Boulton, M. J. (2000). Twenty years' research on peer victimization and psychosocial maladjustment: a meta-analytic review of cross-sectional studies. Journal of Child Psychology Psychiatry, 41, 441–455. TRACKING VICTIMIZATION AND COGNITION 30 Hunter, S. C., & Boyle, J. M. E. (2002). Perceptions of control in the victims of school bullying: The importance of early intervention. Educational Research, 44, 323–336. Kaslow, N. J., Stark, K. D., Printz, B., Livingston, R., & Tsai, S. L. (1992). Cognitive Triad Inventory for Children: Development and relation to depression and anxiety. Journal of Clinical Child Psychology, 21, 339–347. Ladd, G. W., & Kochenderfer-Ladd, B. (2002). Identifying victims of peer aggression from early to middle childhood: analysis of cross informant data for concordance, estimation of relational adjustment, prevalence of victimization, and characteristics of identified victims. Psychological Assessment, 14, 74–96. Malti, T., Perren, S., & Buchmann, M. (2010). Children's peer victimization, empathy, and emotional symptoms. Child Psychiatry and Human Development, 41, 98-113. Rigby, K. (2000). Effect of peer victimization in schools and perceived social support on adolescent well-being. Journal of Adolescence, 23, 57-68. Rose, A. J., & Rudolph, K. D. (2006). A review of sex differences in peer relationship processes: Potential trade-offs for the emotional and behavioral development of girls and boys. Psychological Bulletin, 132, 98–131. Schniering, C. A., & Rapee, R. M. (2002). Development and validation of a measure of children’s automatic thoughts: The children’s automatic thoughts scale. Behaviour Research and Therapy, 40, 1091–1109. Sinclair, K., Cole, D., Dukewich, T., Felton, J., Weitlauf, A., Maxwell, M., et al. (2012). Impact of physical and relational peer victimization on depressive cognitions in children and adolescents. Journal of Clinical Child and Adolescent Psychology, 41(5), 570-583. TRACKING VICTIMIZATION AND COGNITION Tran, C. V., Cole, D. A., & Weiss, B. (2012). Testing reciprocal longitudinal relations between peer victimization and depressive symptoms in young adolescents. Journal of Clinical Child and Adolescent Psychology, 41(3), 353-360. 31