Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 1
INTRODUCTION Breastfeeding and contraception are the two key dis- cretionary variables affecting fertility; that is, they are the primary ways in which women may, through their own behavior, influence when, if ever, they have a next birth. Breastfeeding and contraceptive practice are themselves influenced by socioeconomic characteristics. In addition, a child death may truncate breastfeeding or alter contraceptive behavior. Recent work has shown the impor tance of these intermediate variables in accounting for differences among populations in aggregate fertility levels (Bongsarts, 1976, 1982). Biometric models and clinical studies give us quite precise estimates of the contribution of an additional month of breastfeeding or contraception to the length of the birth interval (for a summary of some of this work, see Leridon, 1977; Sheps and Menken, 1973; Bong Marts, 1983; or Bongaarts and Menken, 1983). However, the biometric models have relied on mathematical simplifications, while the clinical studies have been restricted to small samples, usually based on local populations for which detailed prospective data could be gathered. In this paper, a model of the dynamics of childbearing for the birth histories of individual women is developed and applied to World Fertility Survey (WFS) data from Colombia and Costa Rica. The analysis focuses on the determinants of breastfeeding and contraception, and on the ways they, in turn, influence fertility. The sophisticated and precise biometric models are extended so that they are applicable to the gross level of mea- surement and heterogeneous samples of the retrospective birth histories available from surveys. To the extent that this analysis produces empir ical results comparable to those expected from biometric models and clinical 1
OCR for page 1
2 studies, it both lends credibility to the use of so p is- ticated modeling techniques with survey data and extends the validity of the biometric mcdele to broader popula- tions. To the extent that breastfeeding and contraceptive use can be modeled, new insights into some of the behsv- ioral aspects and the dynamics of the fertility decision- making process are gained. The statistical techniques used rely on estimation procedures recently developed for the analysis of survival data and event histories (Cox, 1972; Kalbfleish and Prentice, 1980). m ese techniques have previously been applied to the analysis of labor force dynamics (Flinn and Beckman, 1982, Tuma et al., 1979), marriage dissolution (Menken et al., 1981) , contraceptive discontinuation (Potter and Phillips, 1980) and child survival (Truseell and Ha_ralough , 1983), as well ~ to a wide variety of biomedical data. -their application to fertility is a natural one (see, for example, Singer and Beckman, 1982, Braun and Boem, 19797. Previous work has segmented the birth interval into a waiting time to conception, a period of gestation, and a period of postpartum amenorrhea. Typically, each of these segments teas been modeled separately. The length of ache waiting time to conception has been assumed to be dependent upon fecundity, the monthly risk of conception. The lengths of gestation and postpartum amenor~nea were taken as either constant or Weir distributions were modeled separately. The modeling strategy adopted here calculates conception rates (which may be equal to zero) by duration, disregarding segmentation of the birth interval. mat is, it replaces the idea of fecundability with a conception rate dependent on bre~tfeeding and contraceptive use, and on a biologically determined propensi~cy to conceive that is dependent on the time elapsed since the last birch. We question is whether, given the reported data on duration of breastfeeding and duration of contraceptive use from retrospective surveys like the WFS, we can estimate ache impact of breastfeeding and contraceptive practice on fertility as measured through birth interval length and parity progression. The basic idea is that at each point in time since the previous birth, there is some risk of conception. This risk is influenced primarily by breastfeeding and contraception, as well as by socioeconomic factors . I n addition, it is well known that infant mortality can influence the length of the birth interval, either because breastfeeding stops when the child dies or because contraceptive practice changes. A similar ~ ~ _ _
OCR for page 1
3 strategy is adopted for Analyzing termination rates for breastfeeding and contraceptive use. Just as there is some risk of conception at each point in time from the previous birth, there is some risk of terminating breast- feeding or contraceptive use at each point in time after these behaviors begin. Infant mortality influences the conception rate by changing these risks. Similarly, the principal impact of the socioeconomic variables on fer- tility can be expected to operate through the decision to breastfeed or contracept and through the duration of these practices. In what follows the data from the World Fertility Surveys for Colombia and Costa Rica are described along w ith the social, economic, and demographic settings of these two countries for the period 1960 to 1976. A three-stage scheme for the analysis of breastfeeding, contraception, and fertility is also described. The stages of the model include: first, defining a set of background covariates that predict whether or not a woman breastfed and whether she used contraception; second, modeling durations of brea~tfeeding and of contraceptive use; third, development of a biometrically-based model of the interval between births. Since the theory for the last stage is most fully developed, the scheme in described in reverse order in the text. Estimation procedures for descriptive statistics and for the equa- tions specified are given next, along with a description of the birth intervals sampled and definitions of the variables used. This is followed by a set of descriptive r esults, the f inal results, and a sugary.