The CBGT followed the protocol developed by Heimberg and Becker with the addition of two group sessions. The first two group sessions were aimed at teaching participants the role and components of anxiety and how to identify and challenge negative automatic thoughts. Sessions 4�C14 focused primarily on individually tailored in-session exposure in combination with cognitive restructuring. Prior to exposure exercises, participants identified and disputed negative automatic thoughts, developed rational alternatives, and behavioral goals were set. Following the exposure exercises, additional cognitive restructuring was conducted and goal attainment was reviewed. Participants were also given homework to continue exposure exercises in the same fashion in their home environment. Session 14�C15 were devoted to assessing the progress of the participant and setting goals for the future. A detailed plan was created for each participant to ensure that goals and methods to achieve them were clear. The therapists facilitating the CBGT sessions were six clinical psychologists with 2 to15 years experience in treating patients with SAD using CBT. Statistical analyses were conducted using PASW version 18.0. The non-inferiority margin of the primary outcome measure LSAS was set at D10 points, which was based on clinical judgment and a review of the evidence of CBGT compared to credible control conditions for SAD. Meta-analytic reviews, adopting random-effects models, have estimated the lower bound of the 95% confidence interval of the between group effect size to 0.39. Assuming a standard variance of LSAS scores in our sample, this supported the use of 10 LSAS points as a non-inferiority margin. Test criterion for noninferiority was that the lower bound of the 95% CI of the mean difference should fall within D. With 95% probability, the mean difference between ICBT and CGBT had to be smaller than 10 LSAS points. As this was a non-inferiority trial, this criterion did not apply for the upper bound of the CI, meaning that the CI could exceed 10 LSAS points if in favor of ICBT. For the other continuous measures, the non-inferiority margin was set at D Cohen��s d= 0.5. Test criterion for non-inferiority for these measures was that the lower bound of the 95% CI of between group effect sizes should fall within this range. This criterion was judged acceptable as it has been proposed that an effect size of 0.5 marks the border between a mild and Y-27632 ROCK inhibitor moderate effect. Thus, this criterion meant that mild effects up to the border of moderate effects were acceptable. Main outcome continuous variables were analyzed using a Nilotinib linear mixed effects model because of its superior qualities regarding missing data as well as in reducing the risk of committing type I errors. We employed the restricted maximum likelihood method assuming a compound symmetry model as covariance structure since it provided the best model in an information criteria comparison.