Summary
In a conventional impression formation paradigm, five sets of descriptions of an anonymous person were developed with use of adjectives selected from positive and negative extremes of Anderson's list of personality trait words. Effects of sex of S, proportions of the descriptions that were consistent, and valence (positive or negative evaluative sign) of the words listed by Ss were studied in a 2 × 4 × 2 factorial design using 176 Ss (88 of each sex). Following the descriptions and a brief interpolated activity, Ss were asked to recall the exact descriptive words included in the original description. Both words correctly recalled and words listed, independent of correctness, were employed as dependent measures. Predictions of a positivity bias for words listed (i. e., more positive adjectives listed than negative) were supported for the measure not requiring correct recall (p < .01). Also not considering word correctness, females produced more positive or fewer negative adjectives than males (p < .05). No similar effects were found for accurate recall. Results were related to the frequently observed phenomenon of negative salience effects.