Table 4.
Multilevel logistic regression models for frequent fighting, average rule of law index and sex
| Model 1 Rule of Law Random intercept | Model 2 Rule of Law Random intercept | Model 3 Rule of Law Random slope | Model 4 Rule of Law Cross‐level interaction | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| OR | 95% CI | p | OR | 95% CI | p | OR | 95% CI | p | OR | 95% CI | p | |
| Intercept | 0.22 | (0.17, 0.29) | 0.000 | 0.13 | (0.10, 0.17) | 0.000 | 0.12 | (0.09, 0.16) | 0.000 | 0.12 | (0.09, 0.17) | 0.000 |
| Sex | 2.63 | (2.55, 2.72) | 0.000 | 2.74 | (2.36, 3.19) | 0.000 | 2.70 | (2.31, 3.16) | 0.000 | |||
| Country level: Av. Rule of Law | 1.28 | (1.03, 1.59) | 0.024 | 1.28 | (1.04, 1.59) | 0.022 | 1.26 | (1.02, 1.56) | 0.031 | 1.32 | (1.02, 1.70) | 0.037 |
| Cross‐level interaction | 0.95 | (0.81, 1.13) | 0.570 | |||||||||
| Random effects | ||||||||||||
| Intercept variance | 0.47 | (0.31, 0.71) | 0.47 | (0.31, 0.70) | 0.67 | (0.45, 1.01) | 0.67 | (0.45, 1.01) | ||||
| L1 coeff. variance | 0.27 | (0.18, 0.42) | 0.27 | (0.18, 0.41) | ||||||||
| Cov (coefficient) | −0.25 | (‐0.40, −0.10) | −0.25 | (−0.39, −0.10) | ||||||||
| ICC | 12.57% | (8.72, 17.78) | 12.43% | (8.62, 17.61) | 16.98% | (11.95, 23.56) | 16.96% | (11.94, 23.52) | ||||
| Log likelihood | −63897.45 | −61997.52 | −61627.35 | −61627.19 | ||||||||
| Likelihood ratio test | M1 vs. Empty model χ 2 (1 df): 4.89; p < 0.05 | M3 vs. M2: χ 2 (1 df): 3799.86; p < 0.001 | M4 vs. M3: χ 2 (2 df): 740.34; p < 0.001 | M5 vs. M4: χ 2 (1 df): 0.32; p = 0.571 | ||||||||
Rule of Law is the average Rule of Law Index score for 2005–2015, z‐standardised with a mean of zero and standard deviation of 1.
n‐sizes for all models. Pupil level (n = 222,547), average pupils per country (n = 4,280), countries (n = 52).
All models include t‐1 dummies for year to account for time‐trends (not tabled).