The last area, group cohesion, has had mixed results and has frequently not been found to be associated with groupthink. One reason for the mixed results is there are varying approaches to operationalizing cohesion (Hogg & Hains, 1998, p. 325), as Janis did not provide the aspects of cohesion considered. As such, experimenters have tested cohesion from a diversity of perspectives. For example, Bernthal and Insko (1993) evaluated cohesion from a social emotional perspective, whereas Hogg and Hains (1998) evaluated a friendship basis. Different still, Tetlock et al. (1992) defined cohesion as well-defined and shared goals.
Test of the full model. Table 5 lists five studies attempting to test a large portion of the model. Park (2000) conducted the most ambitious investigation, attempting to assess all 24 variables of the groupthink model (p. 873). The test assumed a sequential relationship among the elements of the model and measured the relative contribution of the various elements (p. 875). Sixty-four four-person teams completed a 50-minute decision-making exercise designed to simulate a ―complex non-routine dilemma‖ (p. 875). As noted, there was ―no real consequence‖ of the group‘s decision-making (p. 885). The study provides only partial support of Janis‘ model (p. 883), with predictions ―confirmed in only two of twenty-three cases‖ (p. 873).