Comparing Coolness Stability Groups on Spring Aggression
We conducted a 4 (coolness stability group) × 2 (gender) × 5 (ethnicity) multivariate analyses of covariance (MANCOVA) comparing physical, verbal, and relational aggression in the spring. Gender, ethnicity, and spring social preference were included in the analyses as control variables.
The MANCOVA revealed multivariate main effects of coolness stability group (Wilks’ Λ = .97, F (9, 3989) = 3.57, p < .001, η2 = .01), gender (Wilks’ Λ = .97, F (3, 1639) = 15.22, p < .001, η2 = .03), and ethnicity (Wilks’ Λ = .98, F (12, 4336) = 2.16, p < .05, η2 = .01) but no interactions. The between-subjects effects comparing boys and girls revealed that boys scored higher than girls on physical, F (1, 1641) = 36.46, p < .001, η2 = .02, verbal, F (1, 1641) = 25.04, p < .001, η2 = .02 and relational aggression F (1, 1641) = 5.10, p < .05, η2 = .003. The between-subjects effects comparing ethnic groups revealed ethnic group differences on physical aggression, F (4, 1641) = 5.00, p < .01, η2 = .01, verbal aggression, F (4, 1641) = 5.69, p < .001, η2 = .01, and relational aggression, F (4, 1641) = 2.65, p < .05, η2 = .01. Post hoc analyses for each type of aggression revealed that African American students scored higher than all other ethnic groups on physical and relational aggression and they scored higher than all other ethnic groups except for the “other” group on relational aggression. The only other ethnic group difference was that White students scored lower than Latinos on physical aggression.
The tests of between-subjects effects comparing coolness stability groups revealed group differences on all three types of aggression. Table
1 displays the means across the three groups for each type of aggression in the spring and the univariate
F-test associated with that analysis. Post hoc analyses revealed that in the spring, students who maintained their cool status across the school year were perceived by their peers as being more physically, verbally, and relationally aggressive compared to those who were in the stable not cool group for all three types of aggression. For relational aggression, those students who were in the stable cool group scored higher than all three other stability groupings.
Table 1
Coolness stability group differences in average aggression in spring
Aggression | n = 85 | n = 90 | n = 92 | n = 1415 | |
Physical | .33a (.12) | .17ab (.16) | .19ab (.12) | −.04b (.03) | 4.17* |
Verbal | .33a (.12) | .16ab (.16) | .22a (.12) | −.07b (.03) | 5.18* |
Relational | .58a (.13) | .11b (.16) | .10b (.13) | −.08b (.03) | 8.96** |
Changes in Aggression from Fall to Spring
In our second set of analyses, we used Hierarchical Linear Modeling (HLM; Raudenbush et al.
2000), to determine whether coolness stability group status predicted changes in aggression across the year and whether the association between coolness stability and changes in aggression differed as a function of the aggression behavioral norms present in the classrooms. Separately for each type of spring aggression outcome, a two-level model was run. At the student level we included a dummy-coded, uncentered, measure of coolness status stability group where we compared being in the stable cool group, became cool group and became not cool group to the group that was never cool. In addition to fall measures of physical, verbal, and relational aggression, gender and ethnicity (dummy coded as White, Latino, Asian and Other compared to Black) were included as control variables. Each of these variables was group-mean centered. We chose Black students as the comparison group, because Black students were those who differed from the other ethnic groups in the MANCOVAs described earlier. We also included group-mean centered spring social preference as a control variable to examine the unique effects of the stability of coolness on changes in aggression. At the classroom level we tested the effects of the average level of teacher-rated aggression for the classroom on both the intercept of spring aggression and the slope between stable cool group status and spring aggression. This variable was grand-mean centered.
The results of the model predicting spring physical aggression revealed that our control variables were predictive of spring physical aggression. As displayed in Table
2, fall measures of each form of aggression were predictive of spring physical aggression as were gender and Latino and Asian ethnic group membership relative to African American group membership. Beyond these associations, as hypothesized, students who maintained their cool status across the school year demonstrated increases in their physical aggressive reputations from fall to spring relative to students who were in the never cool group. Students who attained cool status by spring also demonstrated increases in their physical aggressive reputations across the school year. See Table
2 for the coefficients and t-tests of their significance for all variables included in this model.
Table 2
Coolness stability group and classroom (fall average norms) predictors of students’ spring aggression controlling for gender, ethnicity, fall aggression and spring social preference
Intercept | −.04 | .02 | −1.64 | −.05* | .02 | −2.04 | .04 | .02 | −1.74 |
Class Agg | −.05 | .04 | −1.20 | −.02 | .04 | −.42 | −.03 | .04 | −.64 |
Girl | −.27*** | .04 | −5.94 | −.24*** | .05 | −5.17 | −.07 | .05 | −1.52 |
White | −.16 | .09 | −1.71 | −.18 | .09 | −1.82 | −.21* | .10 | −2.10 |
Latino | −.21*** | .06 | −3.36 | −.19** | .06 | −2.94 | −.22*** | .06 | −3.37 |
Asian | −.20* | .09 | −2.25 | −.19* | .09 | −2.06 | −.23* | .09 | −2.50 |
Other | −.15 | .09 | −1.65 | −.15 | .09 | −1.60 | −.17 | .09 | 1.81 |
Fall physical Agg | .35*** | .03 | 10.62 | .26*** | .03 | 7.61 | .16*** | .03 | 4.76 |
Fall verbal Agg | .15*** | .04 | 4.09 | .18*** | .04 | 4.74 | .17*** | .04 | 4.57 |
Fall relational Agg | .12*** | .03 | 4.02 | .15*** | .03 | 4.71 | .28*** | .03 | 8.66 |
Stable cool | .35*** | .10 | 3.68 | .32** | .10 | 3.20 | .31** | .10 | 3.07 |
Class Agg | .35 | .19 | 1.77 | .48* | .20 | 2.35 | .17 | .20 | .85 |
Became cool | .28** | .10 | 3.00 | .26** | .10 | 2.60 | .21* | .10 | 2.09 |
Class Agg | .06 | .15 | .41 | .05 | .16 | .29 | .00 | .16 | .01 |
Became not cool | .06 | .09 | .71 | .14 | .09 | 1.49 | .01 | .09 | −.52 |
Class Agg | −.08 | .15 | −.50 | −.26 | .16 | −1.67 | −.05 | .16 | −.16 |
Spring social Pref | .02 | .02 | .86 | .05* | .02 | 2.35 | −.03 | .02 | 1.26 |
The pattern of findings for verbal aggression was similar with fall measures of each form of aggression, gender, and Latino and Asian group membership predictive of spring verbal aggression. Spring social preference was also predictive of spring verbal aggression. As for physical aggression, students who maintained their cool status across the school year demonstrated increases in their verbal aggressive reputations from fall to spring relative to students who were in the never cool group. Moreover, the magnitude of this increase was greater when these students were in classrooms with higher levels of teacher-rated aggression. Students who attained cool status by spring also demonstrated increases in their verbal aggressive reputations across the school year.
For relational aggression, fall measures of each form of aggression were predictive of spring relational aggression, and White, Latino, and Asian students were viewed as less relationally aggressive relative to African American students. Above and beyond these effects, stable cool students and those who became cool by spring displayed increases in relational aggression relative to the group who was never cool (See Table
2).