Adult female mortality trends from retrospective questions about maternal orphanhood included in censuses and surveys.

A limitation of the use of retrospective questions about mortality in censuses and surveys in developing countries concerns the time location of the estimates. The estimates represent an average of mortality over a prior period. This paper describes an approach used in field experiments by CELADE in which respondents who report that their mother has died are asked the year of death. This approach allows calculation of the estimated dates to which the mortality figures apply and mortality estimates for periods 0-4 and 5-9 years before the survey. Comparison of results derived from application of the Brass-Bamgboye theoretical procedure to the estimation of probabilities of female survival and results from a national demographic survey in Honduras in which year at death was queried indicates that the empirical approach yields more consistent results than the theoretical one. Although theoretical and empirical estimates were very similar for the 1st 3 age groups (15-19 20-24 and 25-29 years) differences increased with age with the time calculated empirically located further in the past. Moreover use of the year of mothers death measure allowed mortality estimates covering a period of 30 years to be calculated while the traditional procedure estimated mortality only for a period of 7 years.