Table 2.
Crude and adjusted multilevel regression models to predict self-esteem and SDQ total behavior difficulties from victim type
| Self-esteem | Behavior difficulties (SDQ) | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Crude model | Adjusted modela | Crude model | Adjusted modelb | ||||||
| B [95% CI] | p | B [95% CI] | p | B [95% CI] | p | B [95% CI] | p | ||
| Intercept | 19.40 [19.17, 19.63] | <.001 | 11.33 [6.57, 16.09] | <.001 | 11.14 [10.87, 11.40] | <0.001 | 24.73 [19.26, 30.21] | <.001 | |
| Victim type | |||||||||
| Pure DV | −2.62 [−3.35, −1.90] | <.001 | −2.79 [−3.51, −2.07] | <.001 | 3.97 [3.14, 4.79] | <.001 | 4.00 [3.17, 4.82] | <.001 | |
| Pure RV | −1.60 [−2.45, −0.76] | <.001 | −1.56 [−2.40, −0.73] | .005 | 3.14 [2.17, 4.11] | <.001 | 2.95 [1.98, 3.93] | <.001 | |
| Pure CV | −2.69 [4.54, −0.84] | .004 | −2.19 [3.95, −0.44] | .004 | 4.63 [2.50, 6.75] | <.001 | 4.13 [2.08, 6.18] | <0.001 | |
| DV & RV | −4.64 [−5.39, -3.89] | <.001 | −4.58 [−5.31, −3.84] | <.001 | 6.28 [5.42, 7.13] | <.001 | 5.96 [5.11, 6.81] | <.001 | |
| DV & CV | −3.03 [−5.13, −0.92] | .004 | −2.89 [−4.88, −0.87] | .014 | 4.95 [2.57, 7.32] | <.001 | 4.59 [2.30,6.88] | <.001 | |
| RV & CV | −4.48 [−6.54, −2.42] | <.001 | −2.87 [−4.87, −0.87] | <.001 | 7.46 [5.14, 9.79] | <.001 | 5.95 [3.65, 8.24] | <.001 | |
| DV, RV, & CV | −6.10 [−6.99, −5.21] | <.001 | −5.34 [−6.22, −4.47] | <.001 | 8.37 [7.36, 9.38] | <.001 | 7.54 [6.53, 8.55] | <.001 | |
Non-victims were the reference category. Crude models include the predictor (victim type) on each outcome variable. Adjusted models controlled for level 1 child and family variables (sex, ethnicity, parent education, pupil premium status (an indicator of deprivation) and percentage attendance) and included school as a level 2 (nested), random factor
DV direct victims, RV relational victims, CV cyber-victims. Values in brackets are 95% confidence intervals
aAll level 1 control variables were significant: higher self-esteem was predicted by sex (boys), age (younger), attendance (higher) (p < .001), ethnicity (minority) (p = .002), pupil premium (no) (p = .035) and parent education (12–13 years; college level) (p = .011). The level 2 control variable (school) was not significant (p = .236)
bExcept for parent education (p = .073) all level 1 control variables were significant (p < .001): higher total difficulties were predicted by sex (female), age (older), ethnicity (White British), attendance (lower), and pupil premium status (yes). The level 2 control variable (school) was not significant (p > .250)