maladaptive or harmful purposes for the individual, group, or organization. For example, following anger expressions, in one study 69% of respondents reported no subsequent impact on the relationship, 10% said it was worse, and 21% said it was better (Glomb and Hulin, 1997). It is also possible for anger episodes to simultaneously produce both positive and negative outcomes such as solving a problem while also damaging a relationship (Geddes and Callister, 2007, Glomb and Hulin, 1997).
Second, our data suggested that anger episodes generated two levels of outcomes: individual outcomes (e.g., physiological reactions such as elevated stress, harbored grudges, or desires for revenge) and interpersonal, group or work unit level outcomes (e.g., lowered morale, reduced productivity, solutions to problems or new policies, improved attitudes, and increased mutual understanding). We refer to these as social outcomes. These were also consistent with descriptions of outcomes in the extant literature (e.g., Averill 1982, Frost 2003, Glomb and Hulin 1997, Tavris 1989). For example, in negotiations, anger can both increase the expresser’s bargaining power and elevate the importance of critical issues (Schoomaker 1989, Thompson, Nadler and Kim 1999).
With these two dimensions (valence and level) in mind, we used MDS to see if these factors or others we had not anticipated could be distinguished in our data. This process involved the following steps: (1) The 68 categories identified earlier were typed on cards; (2) 58 senior undergraduate and graduate students (mean age = 24.7 years and blind to the purpose of the study) sorted the cards into categories of similar items using as many categories as they believed were relevant; (3) we analyzed these categories with multidimensional scaling (MDS). We determined how similar two items were by calculating the proportion of students who placed the two items in the same group. We used nonmetric MDS to find a low-dimensional derived space; that is, two items that are nearby in the derived space are grouped together more often than items that are far apart in the derived space. The findings of the MDS analysis are reported in the results section.
Documentation of Results