Validity of questionnaire information on frequency of coitus

A total of 91 women provided reproductive histories, including usual frequency of coitus, at their enrollment into prospective studies conducted by the National Institute of Environmental Health Sciences in Research Triangle Park, North Carolina, in 1984-1986. Those data were compared with coital data recorded during study participation. Overall, women reported a significantly higher frequency of coitus on the interviewer-administered questionnaire than they recorded daily, by an average of 0.8 episodes per week. The size of this difference did not vary significantly for subgroups of women defined by demographic and other covariates. Excluding days of menses from the prospective records reduced the difference by 25%. The authors attribute the overestimate on the questionnaire to a tendency to report a coital frequency that might exist in the absence of travel, illness, and other transient factors that are likely to decrease frequency. This nondifferential information bias is unlikely to produce misleading comparisons or erroneous associations in epidemiologic studies of reproduction.