Recall Bias: A Proposal for Assessment and Control

Abstract

Probably every major text on epidemiology offers atleast some discussion of recall bias in retrospectiveresearch. A potential for recall bias exists wheneverhistorical self-report information is elicited fromrespondents. Thus-, the potential for its occurrence isgreatest in case-control studies or cross-sectionalstudies which include retrospective components.Recall bias is said to occur when accuracy of recallregarding prior exposures is different for cases versuscontrols.The current paper will describe the process and con-sequences of recall bias. Most critically, it will suggestmethods for assessment and adjustment of its effects.WHAT IS RECALL BIAS?Generally, recall bias has been described in terms of'embroidery' of personal history by those respondentswho are cases. For example, Mausner and Kramer