As always, the results of any one study are never
perfect. This study has at least two important limitations
to consider. The first is that we did not directly measure
attributions for conflict in this study, despite the fact that
this is an important part of our theorizing. It is possible,
therefore, that although our hypotheses were supported,
that our explanations for them are incorrect. For ex-
ample, we found that negative performance feedback
predicts greater task conflict at time 2. We suggest that
this happens because group members engage in a search
for understanding about why performance was poor.
However, it is possible that this occurs for some other
reason (e.g., a search for justification for why the source
of the feedback may be wrong and the group ultimately
correct). We have no direct way to address such ques-
tions. Testing such questions by measuring attributions
remains for future research. However, the longitudinal
design of this study does, at least, eliminate some im-
portant alternative explanations for the effects we found
(e.g., controlling for both task and relationship conflict
at time 1 when predicting task conflict at time 2 elimi-
nates the alternative notion that group composition is
the only significant predictor of task conflict).