123 Summary and General Discussion fundamental insights to practice has value as it could improve treatment for individuals who do not profit sufficiently from current evidence based treatments (Lema, Gamo, Yang, & Ishizuka, 2018; Pratt & Hall, 2018), Third, as regards the participants in our studies, we only included individuals who met the criteria for SAD, providing a good representation of individuals within an outpatient treatment center. Fourth, related to the study designs, the correlational studies were complemented with a more causal pharmacological intervention using a well-established testosterone administration method (Tuiten et al., 2000) tested in a randomized triple blind design. Participants, research assistants who administered the drug and researchers who analyzed the effects were blind to the allocated treatment condition. Therefore, we minimized the influence of expectations from participants as well as the researchers on the results of the study. Fifth, the studies were preregistered and the results of all studies were analyzed with linear mixed models, which take into account random effects, are better suited to handle missing data (Baayen, 2008) and critically, the chance of a type I error is much lower compared to repeated measures ANOVA, because they take into account non-independence of the repeated measures within the studies (Barr, Levy, Scheepers, & Tily, 2013; Matuschek, Kliegl, Vasishth, Baayen, & Bates, 2017). Limitations In addition to the strengths of this dissertation there are also several limitations that should be addressed. First, our randomized proof-of-concept trial only included females due to the fact that the administration method we used has as yet only been applied in females (Tuiten et al., 2000). Therefore, we cannot say whether our findings will generalize to males. However, recent single dose administration studies in males do show that testosterone administration affects different social approach and avoidance behaviors (see Carré & Robinson, 2020; Geniole & Carré, 2018 for reviews). Based on these findings similar results for males may be expected. Relatedly, due to inclusion restrictions such as hormonal birth control types and pregnancy, it remains to be tested whether findings generalize to a broader group of individuals with SAD. This limitation is especially important since there are large intra and inter-individual differences in testosterone levels (e.g., depending on time of the day, menstrual cycle and age) and these differences should be taken into account in future studies by including a larger and more varied sample to improve the ecological validity of the current findings. 6
RkJQdWJsaXNoZXIy MTk4NDMw