In the Opponent's Shoes: Increasing the Behavioral Validity of Attackers’ Judgments in Counterterrorism Models