Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
DESCRIPTIVE RESULTS In this section, the distributions of the variables to be used in the final set of analyses in the next section are given and the quality of the data evaluated, with particular attention paid to the dependent variables. THE PROPENS ITY TO BREASTFEED AND THE PROPENS ITY TO CONTRACEPT Recall that the dependent variable for the analysis of the propensity to breastfeed and the propensity to contracept is in each case an indicator variable; the first is defined to be one if the woman breastfeeds in either the open or the closed birth interval and to be zero otherwise, while the other is defined to be one if the woman contracepted in either the open or the closed birth interval and to be zero otherwise. This variable definition was chosen to avoid any biases that might arise from using information from the open interval only thereby eliminating currently pregnant women. The results below show that behavior in the closed interval is a good predictor of behavior in the open interval. Table 2 gives cross-tabulations of breastfeeding and contraceptive use for the last closed and the open birth intervals for Colombia and Costa Rica. The top panel shows that in each of the last closed and the open birth intervals, 91 percent of Colombian women breastfed. In Costa Rica, 77 percent reported breastfeeding in the last closed birth interval and 81 percent in the open interval Women who breastfed in one interval are likely to do so in the next: in Colombia, 95 percent of women who reported breastfeeding in the closed interval also reported breastfeeding in the open interval; in Costa 31 .
32 TABLE 2 Cross-Tabulation of 8reastfeeding and of Contraceptive Use for the Last Closed and the Open Interval BREASTFEED ING Colomb is Open Int e rval yes no Closed Interval yes 1220 64 no 56 67 91: 9Z Colombia Open Interval Costa Rica Open Interval yes no 91: yes 898 52 9% no 114 174 1407 81: 19: CONTRACEPTION Costa Rica Open Interval yes no yes no Closed Interval yes 391 65 31Z yes 518 51 no 461 556 69: no 535 232 58X 42: 1473 79% 211 77Z 23: 1243 43: 57: 1336 Rica ~ the f igure was 94 percent. The lower panel shows that, although the prevalence of contraception is con- siderably greater in Costa Rica, contraceptive use is near ly twice as likely in the open as in the closed bir th interval in both counts ies . Even so, behavior in the closed interest is quite a good predictor of behavior in the open interval: in Colombia, 86 percent of when who used in the closed also used in the open; in Costa Rica, 91 percent did so. SURVIVOR FUNCTIONS FOR BIRTH INTERVALS, DURATION OF BREASTFE~DING, AND DURATION OF CONTRACEPTIVE USE The survivor function was formally defined earlier: for birth intervals, at any time t after the last birth, it
33 in the proportion of women who have not yet had a next conception. Figure 3 gives the survivor function for birth intervals for Colombia in the upper panel, and for Costa Rica in the lower panel. The dotted lines chow 95 percent confidence bands. Notice that the survivor f unc- tions are quite smooth, which suggests that there are no strong preferences for reporting particular birth dates in either country. If the two graphs were superimposed, the survivor function for Costa Rica would lie above that for Colombia, and the 95 percent confidence bands for the two countries would not overlap after intervals about two years long, suggesting somewhat longer birth intervals in Costa Rica than in Colombia. Figure 4 gives the survivor functions for breastfeeding for the two countries. Both show relatively sharp drops at 12, 18, and 24 months, probably reflecting a combination of digit preference and actual behavior resulting from social norms dictating how long a child ought to be breastfed. for Colombia lies above that for Costa Rica, except for very long durations. The estimates at these long durar tions are somewhat suspect since they rely on a very small number of cases, particularly in Costa Rica. Figure 5 gives the survivor functions for contraceptive use for the two countries. These, like the survivor functions for births, are quite smooth, although some jumps at multiples of 12 months are evident. The survivor function for Costa Rica is far above that for Colombia, revealing considerably longer periods of use in the former country. The survivor function MEANS AND VARIANCES OF THE INDEPENDENT VARIABLES U SED IN THE ANALYS IS Table 3 gives descriptive statistics for some of the independent variables used in the analysis. Colombian and Costa Rican women are about equally likely to have worked since they were married or to have worked outside the home. In both countries the women have slightly less education than their husbands, but educational achievement is considerably higher in Costa Rica than in Colombia. However, the Colombian women are more likely than the Costa Rican women to be living in an urban area. As already noted, the level of fertility is about the came in both countries, possibly slightly higher in Colombia than in Costa Rica. Infant mortality is also somewhat higher in Colombia than in Costa Rica. The incidence of contraceptive use, particularly for coitus-dependent
34 8 ISTh.5. COLONS IR - RCL wOMEN i i O . B , O . 6 - Z ~ ° fir it. · \- ~. A. ~ B - x. A. O . 6 ~ 2' ·~ ~- \ a.,, 'I.... 9 1 M O N ~ H 5 l 3 2 B!RIMS. CO5TR ~ [C~ - RLL DOYEN \ .` . .~: on O' . 9 50.75 92.5 ~ O N ~ ~ 5 131.25 176 FIGURE 3 Survivor Functions for Time to Next Live-Birth Co nception
35 BREqSlf'EO ING. COLOMB lq - RtL WoMEN 1 O. , ! a: 0. 5 'A 0. 2 ~- o ~ ! ~ O l 0.6\ I \ C.6: \ ! 0 . ~ - 0 2 - o~ , O M Q N I H 5 BRERSTfCEO ING. .051R ~ ICE - ALL WOMEN 60 \ t3. 5 .~ 27 HO. 5 O ~ 1 ~ S 5q F IGURE 4 Survivor Functions for Duration of Breastfeeding
36 CONlRRCEPT TON. COLOMB 1R - RLL WOMEN O . O . 6 a: so ~ 0.2g a AL i ]L,~ l ~ ! G 0. 6 1- U~ I Z ° ~ 41- _ 1 O I 1 G 0. 2 O I o 1< Hi'. .N 'A "A ' 01 24 M O N ~ ~ 5 CONIR9CERT 10~. COSTR R ICR - RLL WoMEN 23.75 97.5 71.25 ~ O N ~ ~ 5 FIGURE 5 Survivor Functions for Duration of Co ntracepti~,e Use g6 ss
37 TABLE 3 Means and Var lances of Some Independent Variables Used in the Analysis Colombia Costs Rica s t andard ~ t andard mean deviat ion mean deviat ion work ~ ince marr i age . 34 .48 . 36 .48 work away from home . 18 . 39 . 24 .43 woman ' s education 3.2 2.7 4.6 3.6 husbands educ at ion 3 . 5 3 . 1 4 . 8 4 . 1 urban res idence .58 .49 .43 .49 parity . 5.7 3.0 5.6 3.1 births surviving to age one .94 .24 .95 .22 coitus independent contracept ton in a birth interval .21 .41 .24 .43 co i t us dependent contraception in a birth interval .20 .40 .32 .47 b irths surviving t 0 age one among children who are bread t fed .97 . 1 8 .98 . 1 5 methods, is higher in Costa Rica. Finally, the last figure in the table provides information on infant mortality among children who are breastfed. A comparison with the same figure for all births shows lower infant mortality among this group; however, thin difference cannot be attributed to breastfeeding since breastfeeding might never be started if a child is in poor health or dying.
