The accuracy of responses to a question on the frequency of sexual intercourse in the 4-week period prior to an interview is expected to be low. This paper examines the reliability of an alternative question: 'How many days since you last had sexual intercourse?'. This was a part of a larger longitudinal study of fertility dynamics in Matlab, Bangladesh. The response pairs of 61 women who were asked the same question by an interviewer and on the next day by a supervisor, were analysed. Refusals were more common among older women and among women who gave a response at the extremes of the distribution for the other interview. Of the pairs of two responses, 78% had a difference of 3 days or less. Further study and use of the question on days since last intercourse is recommended.
PIP: Reported data reflect coital frequency accurately in a 24-hour period, but data are more questionable for reports over a four-week period. This study compared responses and reinterviews of 51 of about 2000 women in 1979 in the Matlab study area of Bangladesh. Exact correspondence on the last date of sexual intercourse (within one day) was determined. Out of an original 60 women selected, eight refused and one did not have any intercourse within the past month. Results indicate that 15 of cases (30%) coincided exactly. 26 cases (52%) were within one day, and 39 (78%) were within three days. Four cases involved extreme differences of more than 10 days, and these reports also included only one report for the period and were from women aged over 40 years. A comparison of reports of coital activity with those who refused a reinterview showed that women who refused reinterviews reported values at the extremes of the distribution. Either "last night" or "a month ago" were common responses. There was a significant difference in the number of cases reporting coitus the previous night to the supervisor reinterviewer. Care must be taken in wording sensitive questions having to do with the time of last sexual intercourse. Consideration must be given to the sex and social status of respondent, training of interviewers, and privacy for questionnaire administration. The suggestion is made to limit recall time to within the past week for those with frequent sexual intercourse; for those with low levels of sexual activity, a month might be more suitable. A less well-known supervisor might have affected responses in the reinterviews. In conservative settings in Bangladesh, there may be a need to establish a good rapport. Heaping responses for the reinterview did not fit the expected pattern of quadrupling a week's response. Heaping occurred for numbers ending in "0" and "5." The recommendation is made for continued use of the question, because data are reasonably reliable, and this type of information can be useful in estimating monthly coital frequency and fecundability.