117 Dialectical behavior therapy and cognitive behavior therapy in individuals with binge eating disorder: what works for whom? 5 The models included main effects for treatment, moderator (test for prediction), and baseline objective binge episodes, as well as a treatment-by-moderator interaction (test for moderation). By including the baseline covariate for objective binge episodes, we are analyzing residualized change in binge episodes from baseline to end-of-treatment and from baseline to follow-up. This method is preferable to change scores because the residualized change score is uncorrelated with the baseline score (in contrast to calculated change score) (Cohen et al., 1983). Alpha correction for multiple comparisons was controlled using the false discovery rate (FDR; Benjamini & Hochberg, 1995), with an overall FDR of 0.05. Effect size information for predictors and moderators is expressed as the incidence rate ratio (IRR), the increase in incidence rate for each one unit change in the predictor/moderator. IRR values of 1.22, 1.86, and 3.00 can be interpreted as small, medium, and large, respectively (Olivier et al., 2017). Graphs of significant interactions were created using regression coefficients to generate predicted scores across the observed range of the moderator. Data and study materials are available from the first author (ML) upon reasonable request. Results Descriptive statistics Demographic characteristics and baseline assessments of OBE frequency and moderators are provided in Table 1. Demographic variables did not differ between treatment groups except for education, with the CBT+ group obtaining a near significant (p = .058) higher percentage university-level education (10.6% vs 2.8%). In the DBT-BED group, a significantly higher percentage (p = .007) was living with a partner/spouse (71.8% vs 51.5%). Baseline OBE frequency did not significantly differ between participants in DBT-BED or CBT+. Also, pretreatment levels of moderator variables did not differ between treatment groups (see Table 1).
RkJQdWJsaXNoZXIy MjY0ODMw