Talk:Recall bias
From Wikipedia, the free encyclopedia
[edit] Choice of examples
I didn't give any real-life examples, because the example from the ABC hypothesis page seemed difficult and unworthwhile to explain simply, and I couldn't find any other good examples. (Google pulled up various pages that discussed it, but none seemed to give a good, illustrative connection to a specific study; the pages all either spoke in generalities, or stated that the results of the studies they described were not skewed by recall bias. This, incidentally, is probably a result of publication bias; scientists are less likely to publish a study whose results they think might be skewed by recall bias.) At any rate, I decided hypothetical examples would illustrate the point better; but I'm definitely open to better examples, especially illustrative real-life examples, if anyone has them. Ruakh 07:47, 4 Jan 2005 (UTC)
[edit] One study to include
One study that I wanted to include, but didn't remember enough details of, is one I read about once in U.S. News (I think), that actually measured the latter type of recall bias. The researchers took three groups of women and surveyed them about their sexual desires, or something to that effect (maybe it asked whether they ever masturbated?). The difference in the three groups was in how guaranteed the anonymity of their answers seemed; I don't remember exactly how they all were, but it was something like:
- (least anonymous-seeming) gave survey directly to researcher, with guarantee of confidentiality (so the researcher would know who it was).
- folded their surveys and gave them to the researcher (so the researcher might know who it was).
- (most anonymous-seeming) folded their surveys and put them in a slot in a locked box, with all the other women's survey results (so the researcher could not know who it was).
What they found was the percentage of women in group #3 who admitted to certain kinds of sexual desires was greater than that in group #2, which was greater than that in group #1. I think the overall point was that the percentage of women in group #3 who admitted to these things was about the same as the percentage of men who admitted to these things, even though prior surveys had seemed to show that fewer women than men had these desires. So the survey demonstrated the effect of recall/response bias. Ruakh 07:47, 4 Jan 2005 (UTC)
