The Validity of Retrospective Reports of Unwanted Births in the United States: Effects of Preference for Children of a Given Sex

Abstract

The effects of preference for children of a given sex upon the validity of retrospective reports of unwanted births were investigated. Ryder and Westoffs (1975) definition of unwanted birth as the outcome of an unwanted pregnancy in the 1965 National Fertility Study was used. Preference for children of a given sex varies with the number of previous births and the sex of present children. Births that do not satisfy sex-of-child preferences can be expected to be reported retrospectively as unwanted more often than births satisfying preferences. A Fortran computer program was prepared to classify births as unwanted timing failure timing success or nonfailure. After being checked the program was used to classify single births for the 4810 women in the 1965 National Fertility Study sample under age 45. 50.3% of the births of this sample were males. As wanted and unwanted births are defined according to contraceptive practice and fertility desires before conception and since the same biological process determines the sex of either a wanted or unwanted pregnancy the sex distribution of unwanted births should be the same as the sex distribution of all births except for random variation. The comparison of the percent of unwanted male births with the percent of all male births is the basic analytic technique. The data indicate that preference for children of a given sex moderately distorts retrospective reports of unwanted births. For 7 of the 9 cases in which direction of bias was predicted the expected bias was observed. When there is no preference for children of a given sex the sex distribution of retrospectively reported unwanted births is not distorted. The total patterns of evidence support the hypothesis that preference for children of a given sex moderately distorts retrospective reports of unwanted births.