Approach temperament, anger, and evaluation: resolving a paradox.

Factor analytic investigations into the structure of naturalistically observed self-reported mood suggest that anger loads together with avoidance temperament markers, such as fear and anxiety. However, when anger is examined following experimental manipulation, it appears to relate more to approach temperament markers, such as determined and active. We explored 1 potential reason why there is a discrepancy between naturalistically collected self-reported mood versus experimental manipulation of affect with regard to the location of anger in the mood space. We propose that evaluation--endorsing items of similar valence regardless of their semantic content--confounds the self-reported structure of mood. The evaluative dimension of self-reported mood does not appear to represent actual behavior, because it combines items with contradictory semantic content as long as they have similar valence. For example, someone with a positive self-view may endorse both calm and excited while denying sluggish and manic, and so forth, even though these items describe opposite traits. Isolating evaluation across 4 inventories and samples showed that anger clustered together with approach temperament markers. We conclude that isolating evaluation generates a self-reported structure of mood that aligns more closely with experimental investigations.