Taken generally, recall bias occurs when the way a survey respondent answers a question is affected not just by the correct answer, but also by the respondent's memory. [1] [2] This can affect the results of the survey. As a hypothetical example, suppose that a survey in 2005 asked respondents whether they believed that O. J. Simpson had killed his wife. Respondents who believed him innocent might be more likely to have forgotten about the case, and therefore to state no opinion, than respondents who thought him guilty. If this is the case, then the survey would find a higher-than-accurate proportion of people who believed that O.J. did kill his wife.

Relatedly but distinctly, the term might also be used to describe an instance where a survey respondent intentionally responds incorrectly to a question about the respondent's personal history. [3] As a hypothetical example, suppose that a researcher conducts a survey among women of group A, asking whether they have had an abortion, and the same survey among women of group B. If the results are different between the two groups, it might be that women of one group are less likely to have had an abortion, or it might simply be that women of one group who have had abortions are less likely to admit to it. If the latter is the case, then this would skew the survey results; this is a kind of response bias. (It is also possible that both are the case: women of one group are less likely to have had abortions, and women of one group who have had abortions are less likely to admit to it. This would still affect the survey statistics.)

See also

• Statistics under Bias