point out that residualized scores contain a higher proportion of error variance, which implies that the true relationship between anexternal variable and a type of aggression is statistically harder to detect. Therefore, residualized scores should be considered conservative lower limit estimates of the true effect. Because we were interested in the relations between PD traits and the unique aspects of reactive and proactive aggression, and we want to be cautious in estimating true effects, we reported correlates with both raw and residualized aggression types scores, but we attribute most value to the residualized correlates.