|We have two main messages in
this paper. The first is that parametric fertility models may be estimated from
open-interval (DLB) birth data in straightforward fashion. Differences between
open-interval estimation and standard methods lie mainly in the construction of the data
sets, not in the application of statistical methods.
The second message is that use of DLB data can make a critical difference to the quality of statistical results. This is particularly true when analyzing populations at highly disaggregated levels. In fertility estimates from small samples of women, sampling noise can drown out signal. This is a familiar problem, of course, but using more of the information inherent in DLB data can greatly improve the precision of statistical estimators. More precise estimators make for better analysis and stronger conclusions. Our examples with Brazilian census data make this point in several ways: maps of demographic parameters are more coherent, spatial statistics have more power, and regressions provide clearer answers to questions about fertility's relations with other social and demographic variables.
DLB data are often available to researchers, but they are seldom used to their full potential. When fertility data are collected in last-birth or open-interval form, the methods elaborated in this paper can significantly improve the demographic analysis of small samples.
|Estimating Parametric Fertility Models
with Open Birth Interval Data
Carl P. Schmertmann
André Junqueira Caetano
© 1999 - 2000 Max-Planck-Gesellschaft ISSN 1435-9871