Modelling Marital Fertility by Age and Duration: An Empirical Appraisal of the Page Model

Abstract

Demographers studied data from 38 completed World Fertility Surveys by using the Page model of marital fertility by both age and duration of 1st marriage. Researchers measured goodness of fit of the Page model with generalized linear models (GLIM) and found a significant lack of fit at the 1% level in 11 countries and at the 5% level in 2 countries. All of these countries lie in Africa or Asia. On the other hand the Page model fits the data very well for Latin American and Caribbean countries. Haiti was found to have the worst fit. The model tends to overestimate fertility in the 1st year of marriage and to underestimate it the following year however. Researchers also judged its ability to replicate observed total marital fertility. When including all durations up to 36-37 years the fitted sums of marital fertility rates were close to observed rates (average absolute percentage error = 1.46% over 38 countries; maximum = 7.62% Nigeria). The highest natural fertility level was in Jordan at 12.6% < the maximum. The lowest level was in Haiti at 48.6% < the maximum. Additionally natural fertility increases by 1 percentage point for every additional month of breastfeeding and also by 1 point for every increase of 2% in use of contraception for spacing. Nepal had the lowest degree of control of marital fertility with an index of 1% which means that after 10 years of marriage the fertility of women is on 1% < the natural fertility which corresponds to their age at the time. Portugal had a value of 69.8. A clear monotonic relationship exists between the marital fertility index and contraceptive use (linear regression coefficient = .856). This index may be considered roughly equivalent to the contraceptive prevalence rate.