The effect of imputation procedures on first birth intervals: evidence from five African fertility surveys.

In most African societies there is little motivation to remember dates of demographic events with the level of precision required in demographic surveys. Consequently it is common that the large majority of survey respondents can provide only the calendar year of occurrence or their age at the time of the event. The World Fertility Survey Group decided to handle the problem of poor date reporting by using a computer program to impute the missing information. This article illustrates the effect of these imputation procedures on cross-national differentials in the proportion of premarital first births in Benin, Cameroon, Côte d'Ivoire, Ghana, and Nigeria. The analysis demonstrates that the exceptionally low proportion of premarital first births in Ghana is an artifact of the imputation procedures.