Although all informants were separately able to significantly predict ASPD, our results do support the inclusion of multiple informants in ASPD prediction models to increase the predictive utility. The relatively low correlations of aggression ratings between informants, a typical finding in the literature [38], indicated the ability to add more than one informant rating into the same model. With multiple informants in one model, AUC values rose. Thus, the more information available from multiple informants, the more accurate the ASPD prediction. In line with other studies encouraging multiple raters [34, 36, 37], our results offer yet more motivation to include more than one informant when collecting behavioral ratings.
