(Griffith Feeney)
PDF image file of complete paper, 533 KB. PDF file reconstructed from image by OCR (optical character recognition). Excludes table 1 and French Summary.
A method of population projection based on parity progression is developed. First births are projected by applying suitably defined age-specifIc first birth rates to an initial age distribution of zero parity women, second births by applying duration-in-parity-specific second birth rates to an initial open birth interval distribution of parity one women, and so on for third and higher order births.
A proportional hazards model for projection with changing schedules of progression to first birth and an indirect procedure for estimating these schedules are proposed.
Procedures for obtaining parity progression data retrospectively from fertility survey birth history data are given. When two or more surveys are available, comparison of the overlapping portions of retrospectively estimated trends provides a powerful check on data quality.
A method for estimating parity progression data slmultaniously from fertility survey birth, history data and vital registration data on births by order is presented. This provides another way of checking data quality, reconciles the two sources of data, and alleviates the small number and age selection bias problems present in most survey data.
When registered births by order are the only source of data, but are available for a reasonably long period, parity progression data may be estimated indirectly by Inverse projection.
The conclusion identifies applications and potential advantages of parity progression projection In analytical and forecasting applications.