Estimation of fecundability from survey data

The estimation of fecundability from survey data is plagued by methodological problems such as misreporting of dates of birth and marriage and the occurrence of premarital exposure to the risk of conception. Nevertheless, estimates of fecundability from World Fertility Survey data for women married in recent years appear to be plausible for most of the surveys analyzed here and are quite consistent with estimates reported in earlier studies. The estimates presented in this article are all derived from the first interval, the interval between marriage or consensual union and the first live birth conception.
PIP: The estimation of fecundability from survey data is plagued by methodological problems such as misreporting of dates of birth and marriage and the occurrence of premarital exposure to the risk of conception. The availability of data collected with a standard interview schedule from over 40 countries in the World Fertility Survey (WFS) is an invaluable resource for assessing the potential utility of measures of fecundability derived from single-round surveys as well as for comparing estimates across countries and regions of the world. In this article, data are used from 5 WFSs in Latin America (Colombia, Costa Rica, Panama, Mexico and Paraguay) and 3 in Asia (Korea, Malaysia and Sri Lanka) to determine the general usefulness of single-round survey data for the estimation of fecundability from survey data, given the limited information on contraceptive use available from many surveys and the data quality problems associated with reports of dates of marriage and dates of birth. Explored in the process are several different procedures for estimation and variations in estimates of fecundability by country, time period, and women's age. For most of this analysis, the median waiting time to conception in the absence of contraception is used as a measure of fecundability. All of the estimates presented are derived from the 1st birth interval. The estimates are based on data collected in both the birth and the marriage histories in the WFS individual interviews. The 8 surveys chosen for this analysis are characterized by relatively complete reporting of dates of birth and marriage. The primary conclusion of this exercise is that reasonable estimates of fecundability can be derived from WFS data only if one is careful to avoid numerous methodological pitfalls. The most plausible estimates appear to be for women married in the period from about 2 to 10 years before the survey. The average waiting times to 1st conception range from about 4 to 7 months; the corresponding monthly probabilities of conception lie between 0.17 and 0.26. The effect of age at marriage on fecundability is most apparent for ages below 16; differences between women married at ages 16-17 and at ages 18 and above are more modest. Suggestions for improvement of the estimation of fecundability by including a number of questions in survey questionnaires are presented.