This study was designed as a longitudinal study of 80 participants in cognitive group therapy (RCT, n = 40) and interpersonal group therapy (RIPT, n = 40) for social phobia during ten weeks residential therapy. The aim was to investigate the patterns of group climate development and its impact on treatment outcome. Data was collected using MacKenzie's Group Climate Questionnaire (GCQ) four times during treatment, and a multilevel (mixed) model approach was used in the analyses. Engagement in

more »

... groups showed a linear increase during treatment in contrast to a linear decline among patients in RIPT groups. This divergence might be explained by the focus on extragroup and intragroup relationships in RCT and RIPT respectively. Neither conflict nor avoidance followed the expected pattern nor did their mean levels influence outcome. However, when six extreme values of conflict were removed, there was support for a low-high-low pattern of conflict. In general, these results do not support MacKenzie's generic model of group climate development but suggest that sample characteristics, the treatment models and setting can play major roles in determining the group climate. Of the group climate variables, only the mean level of engagement predicted a change in social anxiety over the course of treatment.