Participants made judgments about stimulus materials in which there were 2 possible causes of an outcome. One of these was a common factor, a factor present in all instances presented for judgment, and the other was a positive covariate of the outcome. Instead of interpreting the positive covariate as the cause, participants consistently preferred an interpretation in which the common factor was the cause and the positive covariate enabled the cause to produce its effect. Participants' judgments of both interpretations were predicted by the proportion of instances that were confirmatory for the interpretation and not by ΔP. The findings support a hypothesis that people have a multirole conceptualization of causality including, in addition t...