Double Stratification: By Age and Baseline Aggression.

CI = confidence interval; OASMa = Overt-Aggression-Scale-Modified–Aggression-Subscale; SE = standard error. T-tests hold for the larger samples due to the Central Limit theorem, but for those with baseline values of zero, might be deemed problematic for these smaller, under 40 samples. Focusing on findings significant (or near significant) in the table above, however, ordinal logistic regression upholds the basic findings: All excluding outliers: Statin: β = -0.98 (SE = 0.40) (95%CI = -1.8, -0.19) P = 0.015. Simvastatin: β = -1.0 (SE = 0.47) (95%CI = -1.9, -0.10) P = 0.030. Pravastatin: β = -0.82 (SE = 0.43) (95%CI = -1.7, 0.02) P = 0.055. OASMa = 0: Statin: β = -1.9 (SE = 0.72) (95%CI = -3.3, -0.51) P = 0.008. Simvastatin: β = -2.1 (SE = 0.83) (95%CI = -3.7, -0.47) P = 0.01. Pravastatin: β = -1.6 (SE = 0.87) (95%CI = -3.3, 0.13) P = 0.069.
Double Stratification: By Age and Baseline Aggression.