Martijn van Teffelen

Provoked aggression, psychopathy and narcissism 57 3 Procedure The Ethical Review Committee Psychology and Neuroscience at Maastricht University provided ethical approval to carry out the study (ECP-145 07_10_2014). At sign-up, people were randomly assigned to either social exclusion or insult. When participants arrived at the lab, the experimental procedure was explained, followed by agreement of informed consent. The procedural explanation depended on the relevant condition. After agreeing upon participation, demographic variables and negative affect were assessed. Next, participants engaged in one of the two provocation procedures, followed by measurement of NA and threat perception. Then, participants played the noise blasting game (CRTT) to measure aggression. The participants were told that the current study was being run in several labs spread across universities. In addition, they were told that they would play the game against an unknown player on our network. To support the believability of our manipulation, we told the participants that every experimenter at the different labs was tracking their participants’ progress on a mobile phone application, so that every experimenter had an online overview of available participants. This (bogus) application was shown to the participants. Lastly, psychopathic, and narcissistic personality traits were assessed (in random order). Afterwards, participants were extensively debriefed about the nature of the experiment. In addition, an exit-interview was performed. Using open-ended questions, the participants were asked what they thought the experiment was about, what they thought of the student-assistant/other participant (i.e., played by the same confederate), and what they thought of their opponent during the reaction time game (i.e., CRTT). Responses were scored by the interviewer on a three-point scale (i.e., whether participants believed, doubted, or did not believe the manipulations). People were excluded from data analysis when they indicated that they totally did not believe the manipulations (e.g., by indicating that the CRTT opponent was a computer, or by indicating that the assistant/confederate was playing a role). Lastly, participants were thanked for their participation, and received a gift voucher or course credit if they were students. One experimental session lasted 75 minutes on average. Statistical Analysis SPSS version 23 was used for inferential statistical analysis. JASP (JASP Team, 2018) was used for Bayesian statistical analysis, as we had no a priori expectations on group differences between our social exclusion and insult conditions. By using Bayesian statistics for this part of the analysis we wanted to examine whether the data fit to a null-hypothesis (i.e., no group difference) or to an alternative hypothesis (i.e., a group difference). First, a Wilcoxon signed- rank test was run to examine change in negative affect from pre- to post-provocation. Second, Bayes’ factors were calculated to examine group means in terms of the primary outcomes, i.e., unprovoked, and provoked aggression, and the secondary outcomes, i.e., change in negative affect and threat perception. The use of Bayes factors provides an alternative way to hypothesis testing. In Bayesian statistical reasoning, one starts with a prior degree of belief