124 Chapter 6 Second, there are limitations regarding the reliability of baseline testosterone assessments. Endogenous testosterone levels fluctuate during the day and menstrual cycle and are affected by for example levels of activity, contraceptive use and sex. Although we did try to control for these variables (e.g., assessing all participants at the same time of the day and within their menstrual cycle), it is impossible to control for all the variance within the natural context of an exposure session. Some have even argued that the use of salivary testosterone samples is not (yet) justified (Wood, 2009). In previous work, our lab took an average of six samples in the morning and evening to achieve an estimate of baseline testosterone (Giltay et al., 2012). This was not possible in the studies in this dissertation, due to practical and financial constraints. The reliability of our baseline testosterone estimates (based on two samples) may therefore be flawed. On the other hand, more recent work sheds a slightly more positive picture indicating that saliva samples of testosterone form a moderately stable measure over time (good test-retest reliability), and contraceptive use did not affect testosterone levels (Dabbs, 1990; Liening, Stanton, Saini, & Schultheiss, 2010; Sellers, Mehl, & Josephs, 2007). Therefore, salivary testosterone can be regarded as a suitable marker for individual differences, but multiple samples (at least two) are recommended. More research on its reliability and the relationship between testosterone and exposure outcomes is encouraged. Third, the proposed mechanism of action described in this dissertation stated that testosterone reactivity reduces subtle avoidance and safety behaviors in individuals with SAD. In other words, it stimulates within-session approach behavior and engagement, resulting in better learning during exposure as assessed by retention of learning in the following session. In our design we used fear levels as a marker for learning and retention effects and we assessed automatic avoidance tendencies prior and post exposure. However, we did not directly assess differences in within session approach/ avoidance behaviors between the two treatment groups. Therefore, it remains unclear whether testosterone indeed affects within session approach behavior. Due to time constrains, we did not analyze the videotaped speeches of the participants as a measure of avoidance-approach behavior. It would be worthwhile to analyze these in the future. Although, these video ratings could be a valuable addition to the current results, they are not the answer to all our remaining questions. For example, it is questionable if video ratings are a reliable measure of avoidance and safety behavior since safety behaviors are a complex construct. For example, they frequently take a cognitive form (e.g., rehearsing text in your mind over and over) and therefore cannot be observed (Cuming et al., 2009). Moreover, the individual may not even be aware that their cognitive strategies are considered a safety behavior. As such, using videos or self-report questionnaires may not be a reliable measure of subtle avoidance and safety behavior during exposure.
RkJQdWJsaXNoZXIy MTk4NDMw