The proposed research will evaluate the performance of the two methods that are used for the study of fertility control in historical and developing- country populations for which there is no direct information on the use of birth control. The research seeks to demonstrate how well each method measures what it purports to measure, to delineate the circumstances under which the two methods produce answers that differ in useful and important ways, and thereby to identify strategies for optimal use of the two methods. No previous research has evaluated the ability of these methods to describe either levels or trends in fertility behaviors in a simulated population in which the investigator knows precisely the fertility control strategies being implemented. Neither has there been extensive research using survey data to compare the results about levels or trends in fertility control obtained from these methods with direct information on the levels or trends in prevalence of contraceptive use. This research will fill both gaps. The Coale-Trussell model (M&m) and Cohort Parity Analysis (CPA) approach the measurement of fertility control in different ways and produce parameters that correspond to different notions of the dynamics underlying fertility transitions. They share, however, a common point of departure -- each method compares the populations under study (the target population) to some model population (the standard population) that is thought not to be practicing fertility control. The methods differ in the way they make the comparison and in the types of fertility behaviors that are thought to reflect deliberate control. Testing and clarifying these techniques is important for further progress in understanding the dynamics of fertility transitions. We will use simulation techniques to address several questions, including: How well does CPA distinguish deliberate controllers from non-controllers? How successful is CPA in distinguishing the extent of control from amount of control? Can unobserved population heterogeneity of the type addressed by CPA lead the M&m technique to conclude that fertility control is absent? Estimating both the CPA and M&m parameters from survey data collected by the DHS, we will address the following questions. How well do the estimates of the extent of fertility control implied by CPA correspond to reports of contraceptive use? How do the estimated parameters M and m correlate with breastfeeding behavior, the extent of contraceptive use for spacing births (among women who want more children) and the extent of contraceptive use for stopping further childbearing (among women who want no more children)? In the context of both simulated and real data (from countries in which the DHS was preceded by a WFS), we will also address whether the methods can identify trends in fertility control. Even if the methods cannot identify well levels of control, they may perform well when assessing trends because the population at the earlier time becomes (implicitly or explicitly) the standard for the target population at the later time. If, however, the methods cannot identify trends in control, then they will have little value in the quest to understand fertility transitions.
|Guinnane, T W; Okun, B S; Trussell, J (1994) What do we know about the timing of fertility transitions in Europe? Demography 31:1-20|