Illusory Correlations and Memory Performance

Abstract

Illusory correlations have been shown to create less positive impressions of a minority than of a majority group when the ratio of positive to negative behaviors observed in both groups is constant and greater than 1 (indicating more positive than negative behaviors). This stable phenomenon has been commonly interpreted as due to the distinctiveness and enhanced availability of the rarest event category, negative minority behavior. Theoretically, distinctiveness should lead to enhanced memory for infrequent events and this memory bias is assumed to be the crucial mediator of the illusion. By contrast, an alternative account assumes more information loss or impaired memory for infrequent than for frequent observations (due to the small sample size). The present findings strongly support the latter prediction. After replication of the illusion within the standard paradigm, an examination of judgment accuracy, reliability, and cued-recall performance within the signal-detection framework reveals that the illusion is due to poorer rather than to superior processing of infrequent behaviors. Opposite conclusions based on earlier research are shown to reflect inappropriate analyses rather than conflicting data. Illusory correlations stem from the failure to learn or detect the preponderance of positive behavior in the smaller sample representing the minority, but not enhanced memory for negative minority behavior. This interpretation may also explain the impact of two task manipulations that earlier studies have shown to moderate the effect. When behaviors refer to persons rather than groups, improved memory decreases the illusion. However, the debiasing effect of person coding disappears, and in fact, the strongest illusion occurs when the memory advantage of person coding is prevented by another, interfering encoding tasks.