Victimisation T1 Coping T1 Loneliness T1 Peer- Victimisation T2 Coping T2 Loneliness T2 Peer- Victimisation T3 Coping T3 Loneliness T3
182 Table 7.5 Final Unstandardised Estimates for Longitudinal Cross-Lagged Model for the
Verbal Victimisation and Expected Internalising Coping Model
Regression Path Estimate (S.E) 95% Confidence Interval Autoregressive Paths Verbal T1 → Verbal T2 1.03 (0.21)*** 0.63, 1.44 Verbal T2 → Verbal T3 0.97 (0.13)*** 0.71, 1.22 Internalising T1 → Internalising T2 1.86 (0.14)*** 1.60, 2.13 Internalising T2 → Internalising T3 0.68 (0.05)*** 0.58, 0.79 Loneliness T1 → Loneliness T2 0.94 (0.12)*** 0.89, 1.56 Loneliness T2 → Loneliness T3 0.81 (0.14)*** 0.61, 1.11 Cross-Lagged Paths Internalising T1 → Verbal T3 0.51 (0.11)*** 0.71, 1.22 Note: *** p<.001
The final estimates for the social victimisation and expected internalising coping longitudinal cross-lagged model can be found in Table 7.6. The autoregressive paths for social victimisation and school loneliness were significant predictors of later time points, however the autoregressive pathways for expected internalising coping resulted in model convergence issues and were therefore removed from the model. There were no significant cross-lagged pathways and therefore no model comparison was conducted. Subsequently, this model did not provide support for Hypothesis 7.2a.
Table 7.6 Final Unstandardised Estimates for Longitudinal Cross-Lagged Model for the
Social Victimisation and Expected Internalising Coping Model
Regression Path Estimate (S.E) 95% Confidence Interval Autoregressive Paths Social T1 → Social T2 1.03 (0.17)*** 0.69, 1.38 Social T2 → Social T3 0.57 (0.24)* 0.11, 1.03 Internalising T1 → Internalising T2 -- -- Internalising T2 → Internalising T3 -- -- Loneliness T1 → Loneliness T2 0.73 (0.13)*** 0.48, 0.98 Loneliness T2 → Loneliness T3 0.79 (0.13)*** 0.53, 1.05 Note: * p<.05 *** p<.001
183 The final estimates for the attack on property victimisation and expected internalising coping longitudinal cross-lagged model can be found in Table 7.7. All specified autoregressive paths were significant predictors of later time-points. There were no significant cross-lagged pathways and therefore no model comparison was conducted. Similar to the verbal and social victimisation models, no support was found for Hypothesis 7.2a.
Table 7.7 Final Unstandardised Estimates for Longitudinal Cross-Lagged Model for the
Attack on Property Victimisation and Expected Internalising Coping Model
Regression Path Estimate (S.E) 95% Confidence Interval
Autoregressive Paths
Attack on Property T1 → Attack on Property T2
1.03 (0.17)*** 0.31, 1.12
Attack on Property T2 → Attack on Property T3 0.57 (0.24)* 0.13, 1.36 Internalising T1 → Internalising T2 1.31 (0.11)*** 1.10, 1.52 Internalising T2 → Internalising T3 0.97 (0.08)*** 0.81, 1.12 Loneliness T1 → Loneliness T2 0.73 (0.13)*** 0.63, 1.12 Loneliness T2 → Loneliness T3 0.79 (0.13)*** 0.55, 1.08 Note: * p<.05 *** p<.001
The final estimates for the physical victimisation and expected internalising coping longitudinal cross-lagged model can be found in Table 7.8. All specified autoregressive paths were significant predictors of later time-points. There were no significant cross- lagged pathways and therefore no model comparison was conducted. Furthermore, no support was found for Hypothesis 7.2a.
184 Table 7.8 Final Unstandardised Estimates for Longitudinal Cross-Lagged Model for the
Physical Victimisation and Expected Internalising Coping Model
Regression Path Estimate (S.E) 95% Confidence Interval Autoregressive Paths Physical T1 → Physical T2 0.71 (0.11)*** 0.49, 0.92 Physical T2 → Physical T3 0.62 (0.11)*** 0.42, 0.84 Internalising T1 → Internalising T2 1.67 (0.11)*** 1.45, 1.88 Internalising T2 → Internalising T3 0.81 (0.09)*** 0.59, 0.92 Loneliness T1 → Loneliness T2 0.89 (0.12)*** 0.64, 1.10 Loneliness T2 → Loneliness T3 0.81 (0.13)*** 0.56, 1.06 Note: *** p<.001
7.7.1.2 Expected Retaliation Coping Models
Expected retaliation coping behaviour was examined as a mediator between peer- victimisation and school loneliness across time, whereby expected retaliation coping at Time 2 was hypothesised to mediate the relationship between peer-victimisation at Time 1 and School Loneliness at Time 3 (H7.2a). A model was specified for each peer- victimisation type.
The final estimates for the verbal victimisation and expected retaliation longitudinal cross-lagged model can be found in Table 7.9. All specified autoregressive paths were significant predictors of later time-points. However, there were no significant cross-lagged pathways and therefore no model comparison was conducted. Again, no support for Hypothesis 7.2a was observed.
185 Table 7.9 Final Unstandardised Estimates for Longitudinal Cross-Lagged Model for the
Verbal Victimisation and Expected Retaliation Coping Model
Regression Path Estimate (S.E) 95% Confidence Interval Autoregressive Paths Verbal T1 → Verbal T2 1.12 (0.15)*** 0.83, 1.40 Verbal T2 → Verbal T3 0.76 (0.14)*** 0.49, 1.02 Retaliation T1 → Retaliation T2 2.13 (0.21)*** 1.73, 2.54 Retaliation T2 → Retaliation T3 0.83 (0.06)*** 0.72, 0.94 Loneliness T1 → Loneliness T2 0.91 (0.13)*** 0.66, 1.67 Loneliness T2 → Loneliness T3 0.83 (0.06)*** 0.56, 1.09 Note: *** p<.001
The final estimates for the social victimisation and expected retaliation longitudinal cross-lagged model can be found in Table 7.10. All specified autoregressive paths were significant predictors of later time-points. There were no significant cross-lagged pathways and therefore no model comparison was conducted. Subsequently, the model did not provide any support for Hypothesis 7.2a.
Table 7.10 Final Unstandardised Estimates for Longitudinal Cross-Lagged Model for the
Social Victimisation and Expected Retaliation Coping Model
Regression Path Estimate (S.E) 95% Confidence Interval Autoregressive Paths Social T1 → Social T2 0.91 (0.12)*** 0.68, 1.13 Social T2 → Social T3 1.23 (0.14)*** 0.95, 1.50 Retaliation T1 → Retaliation T2 1.83 (0.14)*** 1.56, 2.11 Retaliation T2 → Retaliation T3 0.89 (0.09)*** 0.72, 1.05 Loneliness T1 → Loneliness T2 0.87 (0.06)*** 0.60, 1.14 Loneliness T2 → Loneliness T3 0.83 (0.14)*** 0.57, 1.10 Note: *** p<.001
The final estimates for the physical peer-victimisation and expected retaliation coping cross-lagged longitudinal model can be found in Table 7.11. All specified autoregressive paths were significant predictors of later time-points. The only significant cross-lagged pathway was the relationship between expected retaliation coping at Time 1 and physical victimisation at Time 3 (b= 0.31, p<.01), whereby higher levels of expected
186 retaliation at Time 1 predicted higher levels of physical victimisation at T3. The final model fit the data significantly better than the auto-regressive only model whereby ∆𝜒2=
-7.26 (1), p<.01. However, given that physical peer-victimisation did not predict future expected retaliation coping, no support for Hypothesis 7.2a was found.
Table 7.11 Final Unstandardised Estimates for Longitudinal Cross-Lagged Model for the
Physical Victimisation and Expected Retaliation Coping Model
Regression Path Estimate (S.E) 95% Confidence Interval Autoregressive Paths Physical T1 → Physical T2 0.67 (0.15)*** 0.38, 0.97 Physical T2 → Physical T3 0.72 (0.16)*** 0.40, 1.04 Retaliation T1 → Retaliation T2 1.59 (0.21)*** 1.18, 1.99 Retaliation T2 → Retaliation T3 0.84 (0.11)*** 0.62, 1.05 Loneliness T1 → Loneliness T2 0.30 (0.07)*** 0.17, 0.43 Loneliness T2 → Loneliness T3 0.83 (0.13)*** 0.57, 1.08 Cross-Lagged Paths Retaliation T1 → Physical T3 0.31 (0.12)** 0.07, 0.55 Note: ** p<.01 *** p<.001
The final estimates for the attack on property peer-victimisation and expected retaliation coping cross-lagged longitudinal model can be found in Table 7.12. All specified autoregressive paths were significant predictors of later time-points. There were no significant cross-lagged pathways, therefore no model comparison was conducted and nor was there any support for Hypothesis 7.2a.
187 Table 7.12 Final Unstandardised Estimates for Longitudinal Cross-Lagged Model for the
Attack on Property Victimisation and Expected Retaliation Coping Model
Regression Path Estimate (S.E) 95% Confidence Interval
Autoregressive Paths
Attack on Property T1 → Attack on Property T2
0.79 (0.10)*** 0.60, 0.97
Attack on Property T2 → Attack on Property T3 0.43 (0.09)*** 0.27, 0.60 Retaliation T1 → Retaliation T2 1.17 (0.15)*** 0.88, 1.45 Retaliation T2 → Retaliation T3 1.45 (0.08)*** 1.29, 1.62 Loneliness T1 → Loneliness T2 0.86 (0.13)*** 0.61, 1.10 Loneliness T2 → Loneliness T3 0.82 (0.13)*** 0.56, 1.08 Note: *** p<.001
7.7.1.3 Expected Avoidance Coping
Expected avoidance coping behaviour was examined as a mediator between peer- victimisation and school loneliness across time, whereby expected avoidance coping at Time 2 was hypothesised to mediate the relationship between peer-victimisation at Time 1 and School Loneliness at Time 3 (H7.2a). A model was specified for each peer- victimisation type.
The final estimates for the verbal peer-victimisation and expected avoidance coping cross-lagged longitudinal model can be found in Table 7.13. All specified autoregressive paths were significant predictors of later time-points. The only significant cross-lagged pathway was the relationship between verbal victimisation at Time 1 and School Loneliness at Time 3 (b=0.28, p<.01), whereby higher levels of verbal victimisation at Time 1 predicted higher levels of School Loneliness at Time 3. The final model fit the data significantly better than the auto-regressive only model whereby ∆𝜒2=
-13.61 (1), p<.001. However, given that there was no indirect effect between verbal victimisation and school loneliness via expected avoidance coping the final model did not support Hypothesis 7.2a.
188 Table 7.13 Final Unstandardised Estimates for Longitudinal Cross-Lagged Model for the
Verbal Victimisation and Expected Avoidance Coping Model
Regression Path Estimate (S.E) 95% Confidence Interval Autoregressive Paths Verbal T1 → Verbal T2 1.04 (0.12)*** 0.81, 1.28 Verbal T2 → Verbal T3 0.77 (0.10)*** 0.59, 0.96 Avoidance T1 → Avoidance T2 1.58 (0.27)*** 1.06, 2.11 Avoidance T2 → Avoidance T3 0.61 (0.07)*** 0.48, 0.75 Loneliness T1 → Loneliness T2 0.87 (0.12)*** 0.62, 1.12 Loneliness T2 → Loneliness T3 0.77 (0.14)*** 0.50, 1.03 Cross-Lagged Paths Verbal T1 → Loneliness T3 0.28 (0.09)** 0.12, 0.45 Note: * p<.05 ** p<.01 *** p<.001
The final estimates for the social peer-victimisation and expected avoidance coping cross-lagged longitudinal model can be found in Table 7.14. All specified autoregressive paths were significant predictors of later time-points. There were no significant cross- lagged pathways and therefore no model comparison was conducted.
Table 7.14 Final Unstandardised Estimates for Longitudinal Cross-Lagged Model for the
Social Victimisation and Expected Avoidance Coping Model
Regression Path Estimate (S.E) 95% Confidence Interval
Autoregressive Paths Social T1 → Social T2 0.78 (0.11)*** 0.55, 1.10 Social T2 → Social T3 0.87 (0.09)*** 0.70, 1.11 Avoidance T1 → Avoidance T2 1.83 (0.17)*** 1.49, 2.16 Avoidance T2 → Avoidance T3 0.44 (0.04)*** 0.35, 0.52 Loneliness T1 → Loneliness T2 0.85 (0.13)*** 0.57, 1.18 Loneliness T2 → Loneliness T3 0.71 (0.06)*** 0.59, 0.86 Note: * p<.05 ** p<.01 *** p<.001
In summary Section 7.7.1 tested Hypothesis 7.2a, whereby it was found that there was a lack of support to suggest that expected maladaptive coping behaviour mediates the relationship between peer-victimisation and school loneliness over time. On the basis of these findings no further models were specified to test whether friendship moderates
189 the relationship between peer-victimisation and expected maladaptive coping, and the relationship between expected maladaptive coping and school loneliness (Hypothesis 7.2b). However, the findings did indicate that expected internalising coping behaviour predicted future verbal victimisation and expected retaliation coping behaviour predicted future physical coping and therefore the next section will explore the role of friendship within these models.
Post-Hoc Analyses: Expected Maladaptive Coping Behaviour and the Role of Friendship
Although no indirect effects via expected coping were identified in any of the models, two models (verbal victimisation and expected internalising coping, and physical victimisation and expected retaliation coping) indicated a relationship between expected coping behaviour and future peer-victimisation and therefore the role of friendship was examined in these models as additional analyses. On the basis of findings from Chapter 6 reciprocity (proportion) and friendship conflict (as an indicator of friendship quality) were examined as moderators. It was therefore hypothesised that:
H7.4: Friendship (reciprocity (proportion) and friendship conflict) will moderate the
pathway between expected coping at Time 1 and future peer-victimisation at Time 3.
7.7.2.1 Expected Internalising Coping, Verbal Victimisation, and the Role of Friendship
A model was specified to examine the moderating effect of friendship conflict and reciprocity (proportion score) on the relationship between expected internalising coping at Time 1 and verbal victimisation. For the sake of parsimony, school loneliness was excluded from the model.
The final estimates for the interaction between friendship and expected internalising coping as a predictor of future verbal peer-victimisation can be found in Table 7.15. All specified auto-regressive paths were significant predictors of later time– points. The final model yielded one significant main effect of reciprocity at Time 1 as a predictor of future verbal peer-victimisation at Time 3 (b=-0.25, p<.05), whereby higher levels of reciprocity (proportion) predicted lower levels of verbal victimisation. There was a significant interaction between friendship conflict and expected internalising coping (b=0.51, p<.01), however there was no significant interaction between reciprocity (proportion) and expected internalising coping. The final model was compared to the main effects only model and found to significantly improve overall model fit ∆𝜒2= 6.42 (2) p<.05.
190 The interaction between expected internalising coping at Time 1 × friendship conflict at Time 3 was probed at the mean, 1 SD above the mean, and 1 SD below the mean to indicate comparatively low, medium, and high levels of friendship conflict. The simple slopes analyses suggest that the use of expected internalising coping at Time 1 is strongly and significantly associated with verbal peer-victimisation at Time 3 in those children who have high conflict with their friends (b=0.69, p<.001). There was no significant association between expected internalising coping and future verbal victimisation in those children with average and low friendship conflict levels (b= 0.18,
p>.05; -0.32, p>.05; See Fig 7.3). Subsequently, partial support was found for Hypothesis