In their well-designed study described earlier (see page 13), Cole et al. (1999) found that, despite the high levels of stability across time reported for each of their measures, levels of negative affect predicted changes in perceived academic competence for both boys and girls. That is, middle school students who self-reported relatively high levels of negative affect at the beginning of the study were subsequently more likely to experience decreases in perceived academic competence over the two years of the study than were students who initially reported less negative affect. Interestingly, however, Cole and colleagues found that the causal direction appeared to be reversed for perceived social competence. Low levels of perceived social competence predicted increases in negative affect across assessments, but negative affect did not uniquely predict changes in perceived social competence.