The psychometrics of everyday life

Abstract

We examined people's ability to assess everyday life correlations such as the degree of agreement that exists for various kinds of evaluations and the degree of consistency that characterizes social behavior from occasion to occasion. We found substantial accuracy for correlation estimates if two conditions were met: (1) subjects were highly familiar with the data in question and (2) the data were highly "codable," that is, capable of being unitized and interpreted clearly. We generally found extreme inaccuracy if either of these conditions was not met. Subjects were particularly inaccurate about correlations involving social behavior: They severely overestimated the stability of behavior across occasions. In addition, even subjects who were statistically sophisticated showed limited appreciation of the aggregation principle, that is, the rule that the magnitude of a correlation increases with the number of units of evidence on which observations are based.