All statistical analyses were performed using Stata 13.0 (StataCorp, College Station, TX, USA) and all were adjusted for the clustered nature of the twin data. Statistical significance for all analyses was considered to be p < 0.05. Total aggression mean scores of the participants from the different informants were compared for those with and without ASPD, both with the sexes combined and separated. Additionally, direct and indirect aggression score mean differences between the sexes were compared. Mean differences were tested for significance using the adjusted Wald test. Effect sizes were calculated as Cohen’s d. Pearson correlations were calculated between the total aggression scores of different informants, as well as between direct and indirect aggression scores from each informant. Additionally, to obtain an initial correlation of ASPD and aggression, a continuous ASPD behavior count was used for correlations with total aggression scores from different informants.
