Abstract
This study used two measures of cohesion for the process analysis of 12 short-term, time-limited groups for complicated grief. The measures had similar theoretical definitions but differed in terms of rater source (member vs. observer), measurement score (mean of items vs. global rating), and rating unit (individual vs. group). We examined the relationship between the measures, assessed the development of cohesion over the life of the group, and evaluated each measure’s relationship to outcome. A principal components analysis with each measure yielded one cohesion component, which supported a unidimensional model; however, the two cohesion components were independent of each other, which supported a multidimensional model. Repeated measures analyses indicated that observer-rated cohesion developed in a quadratic manner (v pattern) across sessions, while member-rated cohesion developed in a linear manner. The object focus (the group, other members, the therapist) of the members’ ratings determined whether cohesion increased or decreased across sessions. No significant relationships between cohesion and outcome were identified. Implications of the findings for the understanding of group cohesion are considered.