Mortality rates mortality events and the number of births.
提出新方法分离和估计生育替代中的直接替代、囤积和生物替代效应，利用马来西亚家庭生活调查数据展示方法，发现校正虚假相关后替代率仅为0.21。
According to demographic transition theory a decline in infant mortality is viewed as a precondition for a fertility decline. This discussion describes some new methods that can be applied to cross sectional data in transition theory--the replacement hypothesis. The objective is to show how the effects of direct replacement hoarding and biological replacement can be separated and then estimated. Focus is on the methodology although data from the Malaysian Family Life Survey are used to illustrate the method. The methods employed use statistical techniques more fully described elsewhere. To estimate the fertility hoarding component of replacement in cross sectional data it is necessary to relate variations in the child mortality rate across families. The true child mortality rate for a family is not observable; only the realized mortality rate for a family which measures the true rate with error is observable. The child mortality rates for families may differ because of actions taken by the family. If conscious actions concerning inputs of time to child care are correlated with conscious actions to have children the familys observed mortality rate may be related to fertility not because of hoarding but because parents who desire more children also like to spend more time with them and so suffer a lower rate of child mortality. To avoid this source of contamination it is necessary to calculate the family mortality rate net of those factors which affect child survival and are possibly subject to parental choice. To do this a model of waiting time to the death of a child over the first 10 years of life for each child in the family was estimated. The method involves the estimation of a regression equation with the length of life of a child as the dependent variable. Once the family specific component of the child mortality rate has been estimated the number of births can be regressed on the number of deaths and the family specific mortality rate. The family size regressions given cannot reveal the effects of breastfeeding nor can they indicate the speed with which couples replace dead children. Because the Malaysia data used here provide detailed information on the dates of birth weaning and death of children it is possible to directly observe the effects of breastfeeding or the death of a child on the probability that another conception leading to a live birth results. The best way to exploit this data is to estimate a waiting time model. The model developed with Wolpin is again used to estimate the conception interval waiting time model. When the number of births was regressed on the number of deaths and corrected for spurious correlation the estimate replacement rate was 0.21. The uncorrected least squares replacement rate was 1.4 demonstrating the large effect of the spurious correlation.