norms (suppressing, contingent and legitimating) and form of expression (authentic, controlled and silent) on anger outcomes. Since our qualitative methodology does not enable us to determine whether these effects are separate or interactive, we can only offer provisional conclusions about these effects. For example, there may be a direct effect of work unit anger norms on outcomes, or norms may moderate the relationship of anger expression on outcomes, and/or form of expression may exert a separate direct effect on outcomes. We indicate these possibilities with dotted lines in the model. Our data strongly suggest the possibility of an interaction effect. In any case, it is important for both work unit anger norms and form of expressions to be taken into account, in order to understand the outcomes of anger episodes in work units. We also suggest that feedback from anger outcomes influences subsequent anger episodes through both work unit norms and form of anger expressions (Barley and Tolbert 1997, Giddens 1979, Stern and Barley 1996) as indicated by the dashed arrows from anger outcomes.
Limitations and Future Research
Our work is limited by our methodological approach. Our approach afforded us a rich dataset that spanned several organizations and over 150 anger episodes. However, by using qualitative tools and recall rather than observational methods, our ability to draw causal inferences is limited. Our methods are appropriate for theory building, but future work should extend our findings using other methodological approaches to test our theory and extend our findings.
Another limitation of this study is that with our data we were not able to disentangle status and occupation. For example, it was the doctors in the surgical unit to whom anger discretion was granted despite generally negative views of anger expression in the unit. It may be that high status members in work units are permitted to violate the norms without repercussions (see Keltner Gruenfeld and Anderson 2003). In cases where anger norms were legitimating, the anger expression had a clear “collective purpose that either accompanied status (as in the case of coaches, who expressed anger for motivational purposes) or conferred it through its expression (Tiedens et al. 2001) (as in the case of the union organizers and social workers whose expressed anger in order to take an advocacy position on behalf of