106 Chapter 5 terone. Participants held the liquid under their tongues for 1 minute. This dose yields a sharp increase in plasma testosterone concentrations within 15 minutes and declines to baseline within 90 minutes in women (van Rooij et al., 2012). Pharmacodynamic effects can be assayed 4-6 hours after intake (Bos, Panksepp, et al., 2012; Tuiten et al., 2000). Until completion of the primary outcome analyses, therapists, participants and researchers were blind to the group allocation. Exposure intervention The participants engaged in two 90-minute public-speaking exposure sessions delivered one week apart in accordance with the protocols developed by Rodebaugh and colleagues (Rodebaugh et al., 2013; Smits, Rosenfield, Davis, et al., 2013). Sessions were standardized with respect to preparation time (max. 5 minutes), exposure length (6-8 minutes), the availability of notes and speech topic as well as reaction of the experimenter (neutral). On the morning of the first day, participants received psychoeducation about SAD and exposure, with the first session starting after four hours. In both sessions, psychoeducation was repeated and personalized goals and harm expectancies were assessed. Then, the participants presented their prepared speech in front of two confederates, a therapist and a video camera. Participants reviewed their videotaped performance afterwards with the therapist. The latter being psychology students in their last year of training (BA and MA level) trained and supervised by board-certified psychologists (authors, M.H. and M.K). Deviations in data analytic strategy We deviated from the pre-registered data analyses at the Open Science Framework (see https://osf.io/3cxsv) in two minor ways. First, in the pre-registration we stated that for our first preparatory analyses we would conduct a one sample t-test to check if AAT effect scores for facial expressions at baseline significantly differ from zero, followed by an ANOVA with valence (happy, angry, neutral) as factor on AAT effect scores (Push – Pull). However, we ran a two-way repeated measures ANOVA on the raw reaction times which we consider an improvement. Secondly, we did adhere to the exploration of adding baseline testosterone to our models, but only for the main effects. Details results preparatory analyses The two-way repeated measures ANOVA showed a main effect of picture type: F(3, 159) = 12.95, p = .023. Post-hoc comparisons with a Bonferroni adjustment showed that the Reaction times of the checkerboards were significantly slower compared to all facial stimuli (all p < .001). Reaction times of the facial expressions did not significantly differ
RkJQdWJsaXNoZXIy MTk4NDMw