The aim of this thesis was to examine and further develop an expanded groupthink model (Granstrom & Stiwne, 1998), and also to develop methods to capture the phenomenon in ordinary, authentic working groups. The groupthink model was first proposed by Janis (1972, 1982) and incorporates an omnipotent stance as something highly cohesive groups use handle difficult situations. The model was derived from studying powerful groups mainly in American politics. To be able to better use the concept on more ordinary groups an expanded groupthink model was hypothesised incorporating both an omnipotent and a depressive groupthink version based on the work of Klein (1977) and Bion (1961).

This thesis builds on four studies of the bipolar groupthink phenomenon. In the first study questionnaire data from 308 participants in 29 groups where factor analysed resulting in three factors. The questionnaire could differentiate between groups and also between different organisations in the study. The use of organisations as basis for comparisons was based on the concept group propensity (Granström,1989), which describes reoccurring patterns of basic assumptions in one and the same organisation. In the second study the main focus was to refine the way data from the questionnaire is interpreted. Questionnaire data from 409 participants in 39 groups was used. As an alternative to the method used in the first study a circumplex analysis was made. It showed bipolar characteristics of the data, but also that the third factor from the previous factor analysis, although important for the groupthink concept, could not differentiate between the two types of groupthink. It was suggested that questionnaire data should be interpreted using only two factors. In the third study the focus was on how bipolar groupthink could be manifested in religious groups and how they had chosen to organise their work. Data from seven religious groups (n=64), categorised into one of three ways of organising work, was used. The results showed that more hierarchical groups also displayed more depressive groupthink tendencies, and more team-oriented groups displayed more omnipotent groupthink tendencies. However, the religious groups did not differ from groups in general. In the fourth study the connection between groupthink and group cohesiveness was investigated. First, a questionnaire on group cohesiveness was created and tested (n=202) giving a possibility to capture two aspects of cohesiveness: commitment to group relations and commitment to task. In the second part of the study data from five groups (n=35) on both the questionnaires showed a strong connection between commitment to task and the morale aspects of bipolar groupthink. Commitment to group relations did not differentiate between the groups indicating possible problems using it as a sole basis for group cohesiveness when examining groupthink - something that many previous studies have done.