Flipbook

CHAPTER 6 152 information (i.e., main effect of Information type), and that highly fearful participants would give higher danger ratings than participants with low fear (i.e., main effect of Group; cf. Arntz et al., 1995; Engelhard et al., 2001, 2002; Gangemi et al., 2012; van den Hout et al., 2014). Model 1 (see Table 1) contained these two main effects. Model 2 contained these two main effects and the hypothesized effect that, compared to participants with low fear, danger ratings of highly fearful participants would be lower when approach behavior was described than when no approach behavior was described. The BF for model 2 compared to model 1 (i.e., BF 2,1 : BF model 2 / BF model1 ) gives the amount of evidence for the hypothesized Group x Behavior type interaction effect. Additionally, we explored whether this interaction effect was larger for scenarios with objective danger versus safety information (Model 3; i.e., a Group x Information type x Behavior type interaction effect), was similar for scenarios with objective danger and safety information (Model 4), or was larger for scenarios with objective safety versus danger information (Model 5). RESULTS Groups We were unable to create a low and high anxious group based on the DASS anxiety subscale, because few participants ( n = 9) had scores above the moderate range (>14; Lovibond & Lovibond, 1995b). A median split resulted in a low ( M = 0.96, SD = .77, median = 2) and high ( M = 6.56, SD = 3.74) anxious group that both fell, on average, in the normal range (0–7). Therefore, we created a low ( M = 1.64, SD = 2.21, n = 169) and high ( M = 37.42, SD = 26.71, n = 156) spider fearful group with a median split on the FSQ score (median = 7). The mean FSQ score in the spider fearful group is lower than the mean of a clinical sample (e.g., M = 89.10, SD = 19.60; Muris & Merckelbach, 1996), but higher than that of non-phobic individuals ( M = 3.00, SD = 7.80).