The primary aim of this study was to identify variables which may attenuate or eliminate the effect of the correspondence bias. A secondary aim was to identify aspects of the research paradigm which may encourage subjects to commit the bias. The quiz game paradigm developed by Ross, Amabile & Steinmetz (1977) was employed. The study examined the effect of origin of questions and number of questions correct on observers' ratings of a questioner and contestant's general knowledge ability. The subjects were 100 university students randomly assigned to one of four conditions. The combination of questions supplied and six questions correct condition eliminated the bias. Open ended questions were also employed to identify factors that encouraged subjects to commit the bias. The results of the study indicate that subjects utilised a number of normally appropriate strategies to judge general knowledge ability. Weaknesses of the study are outlined, as are implications of the study and recommendations for future research.