Reliability of reports of sexual behavior: a study of married couples in rural west Africa.

ABSTRACT

A heterosexual partners survey in rural Senegal, Africa, was carried out in 1992 to evaluate the reliability of self-reported answers about sexual practices. The authors followed 62 married couples weekly during a 5-week period to assess the level of agreement 1) between answers made by members of the same couple but collected separately, 2) between retrospective reporting of sexual activity during the last 4 weeks versus weekly reporting of sexual activity during 4 weeks, and 3) between answers to a set of identical retrospective questions asked 5 weeks apart. Reports over recent and short periods of time such as 7 days are reliable The dates of sexual acts with spouse reported during the weekly interviews were concordant between members of a couple at 0-day or 1-day intervals in 72% of cases. The concordance of weeks reported with or without intercourse was also high. Reports over longer periods of time are less reliable The comparison of retrospective reports versus weekly reports regarding mean number of sexual acts during the last 4 weeks shows a clear overreporting that was higher among men (4.5 vs. 2.7 sexual acts) than among women (3.7 vs. 2.7).

ABSTRACT

PIP Accurate measures of sex behavior are very important for interventions, the subsequent evaluation of behavior change, and for the development of epidemiologic models of sexually transmitted diseases. Measures of sex behavior are usually based upon self-reports of sexual practices. The reliability of such information is, however, questionable. The authors report their findings from a 1992 study in the region of Ziguinchor, Senegal, to evaluate the reliability of self-reported answers from married couples about sex practices. 62 married couples were followed weekly during a 5-week period to assess the level of agreement between answers made by members of the same couple but collected separately, between retrospective reporting of sexual activity during the last four weeks versus weekly reporting of sexual activity during four weeks, and between answers to a set of identical retrospective questions asked five weeks apart. Reports over recent and short periods of time such as seven days are reliable; the dates of sex acts with spouse reported during the weekly interviews were concordant between members of a couple at 0-day or 1-day intervals in 72% of cases. The concordance of weeks reported with or without intercourse was also high. Reports over longer periods of time are, however, less reliable. Comparison of retrospective reports versus weekly reports on the mean number of sex acts during the last four weeks found a clear overreporting that was higher among men, 4.5 vs. 2.7 sex acts compared to 3.7 vs. 2.7 among women.