“…To reduce the influence of response times that reflect lapses in attention, and to avoid the complete removal of genuine response times that may reflect increased difficulty in the formation of trust or aggression judgments, lengthy response times were Winsorized such that times longer than 4,000 ms were changed to 4,000 ms (greater than 2.5 SDs of the mean, 1.81% of aggression and 2.77% of trustworthiness response times). 1 The process and benefits of Winsorizing are discussed in many articles (e.g., Erceg-Hurn & Mirosevich, 2008;Ruppert, 1988;Wilcox, 2005) and Winsorizing has been used for reaction time (RT) outliers in many recent studies (e.g., Chambers, Swan, & Heesacker, 2014;Lai et al, 2012;Mueller, Makeig, Stemmler, Hennig, & Wacker, 2011;Townsend, Eliezer, Major, & Mendes, 2014;Wilkowski & Meier, 2010). A mixed factorial ANOVA with one within-subjects factor (aggression vs. trustworthiness judgments) and one betweensubjects factor (aggression rated first or second) revealed two main effects: Participants rated aggression faster than trustworthiness (F[1, 128] ϭ 29.70, p Ͻ .001, Cohen's d ϭ 0.96), irrespective of whether they rated aggression or trustworthiness first, and participants provided both judgments faster if they rated aggression first rather than second (F[1, 128] ϭ 9.03, p Ͻ .01, Cohen's d ϭ 0.53; see Figure 3b).…”