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.
44 are such that the interactive effects of boy sufficiency and socioeconomic status are negative for DCU and positive for LF, unambiguous sign expecta- tions for four of seven socioeconomic variables in the semireduced form model cannot be derived. These points comprise the sixth major conclusion of our reduction exercise. Discussion The conclusions we have reached here about the implied and actual semireduced form coefficients depend on certain assumptions, especially those involving the effects of sex preference. Expectations about the signs of many of the semireduced form coefficients depend on the effects of sex preference, especially in transitional settings. It therefore appears that these preferences are crucial to understanding the effects of soazoecenomic variables on CM, LCU, and CEB. m us, if SOB in inter- action with socioeconomic status is expected to decrease LCU and increase OF, there Is no point in regressing LCU and LF on the socioeconomic variables because we could not interpret the effects. On the other hand, if we think alternative assumptions about sex preference are appropriate, we can reach conclusions of varying firmness about the socioeconomic effects. ffl ese conclusions concern functional form, as well as expecta- tions about the nature of covariabilty between implied and actual semireduced form socioeconomic effects and the degree of transitionality. This exercise offers several payoffs. First, we have been able to show what expectations about socioeconomic effects depend on. Second, we have shown that given key assumptions, it is possible to derive expecta- tions for socioeconomic effects that not only are quite specific, but also differ from usual expectations not explicitly based on a structural model such as ours. Third, we have pinpointed areas within the model which deserve highest priority, in particular the boy and girl sufficiency variables. If the effects of these variables turn out to be minimal or follow the patterns leading to the expectations delineated in Table 1.3, Part A, there will be grounds for using the estimation of LCU and CEB based on insights developed in this section to motivate our empirical analyses. Finally, we have shown that contextual differences in the semireduced form model of CM will be concentrated in intercept differ- ences, rather than in variation in socioeconomic effects. 1.5 CONCLUSION We have proposed, justif led, and explored a model that organizes and synthesizes hypotheses about the determinants of fertility and related behaviors . m is model is designed for use with cross-sectional survey data sets, especially those collected by the World Fertility Survey. Although the discussion has focused primarily on the assumptions and hypotheses embodied in the model, the major operational issues have received attention as well. In developing the model, we have concentrated on two main problems: the establishment of a set of micro-level struc- tural equations and the formulation of hypotheses about structural
45 c:oeff icients in the model. These concluding remarks indicate ma jor considerations affect) ng these two areas of concern. The Micro Equations The key to understanding the structural equations is that this model is motivated by (1) a concern for those aspects of fertility most likely to show changes if total cohort reproduction Is dropping in developing countries; (2) a need for generality sufficient to make the estimation of the model meaningful across the broad spectrum of countries participating in the WFS; and (3) a need to limit the formulation to those variables which can be defined in each WFS data set. Point (1) has led us to distinguish between three components of cohort reproduction: onset, early fertility, and later fertility. We have hypothesized that later fertility and (probably to a lesser extent) onset are the more discretionary phases of the reproductive cycles Our division of the fertility process permits testing of this hypothesis while elucidating the ways in which reproductive behavior is changing in the developing countries. m is threefold distinction has additional advantages. In particu- lar, it provides leverage on three important relationships: between fertility and child mortality; between fertility and contraceptive behavior; and between child mortality and contraceptive behavior. If we used the conventional counts of children ever born and child mortality, we would have to posit simultaneities between fertility, child mortality, and contraceptive behavior. This would be problematic for two reasons. First, it is unclear how these simultaneities could be identified in a structural equations model. In particular, it is theoretically uncertain whether each of these three endogenous variables is affected by a pre- determined variable that does not affect the other two. Second, even granting that there may be such predetermined variables, we do not consider them to be operationalized in the WFS data sets or other similar surveys. Separating children ever born into early and later fertility, and similarly dividing child mortality, helps resolve this problem. It permits later fertility and contraceptive behavior (which is primarily later in the developing countries) to be treated as endogenous with respect to early child mortality and early fertility, and later child mortality as endogenous with respect to early fertility The model we have developed is appropriate for developing countries, not the wealthy and industrialized developed countries. The need for a single model with which to compare these countries (point t2)) inevitably requires procrustean decisions, such as the placement of husband's educa- tion and occupation within the model. Because these characteristics are defined as of the time of interview in the WFS core and similar question- naires, status histories for each husband are lacking. mis omission, combined with the fact that many survey respondents have been married more than once, means that any placement of husband education and occupa- tion in the structural equations will be more appropriate for some respondents than for others. Moreover, the degree of suitability of a particular placement will vary across countries. Our strategy for
46 addressing this problem has been to allow for the possibility of changes in husband's status, either through remarriage or through the intragener- ational mobility of the husband in couples in which the female respondent has been married only once. This has no consequences for the magnitudes of the direct effects of the husband's characteristics on later child mortality, later fertility, and (later) contraceptive behavior, but does have certain consequences for the indirect effects. In particular, although the directions of the hypothesized effects remain the same, the functional forms of the LOU and CEB reduced-form equations change if husband's occupation and education are treated as exogenous. mere is no immediate solution to this kind of problem. Our strategy has been to minimize ambiguity about causal ordering by treating any given pair of variables as contemporaneous unless a strong case to the contrary could be presented . It appears to be characteristic of comparative analysis generally that a decision made to permit comparisons can also mitigate the meaningfulness of those comparisons. This is the case for husband's occupation and education since in some societies lifetime monogamy and status stability are common, while in others this is much less so. Allowing for variability makes the model more suitable for societies with relatively high rates of remarriage, and less suitable for societies with low rates of remarriage. Although this is a weakness of the model, it must be seen in perspective. In fact, the same problem exists within societies. To address it, we must specify different models for different life-course patterns. However desirable this may be in principle, it is well beyond what available data will allow. Our model is designed for use with WFS data. Moreover, with the exception of ethnicity, it is limited to only those variables present in every WFS data set (point (3~. This seems entirely appropriate since it is the WFS data that we are primarily interested in here. Theories and general conceptions, which are inherently broader than models, rest not on one or another specific data set, but on the entire range of salient knowledge. Our view is that models are best linked to specific data. The WFS core questionnaire was developed nearly a decade ago; it reflects contemporary, competing theoretical and political perspectives on the need for particular kinds of information. The formulation of our present model is based on research and theorization conducted over the past decade. Thus, unlike the WFS data, the model Is not constrained by decisions made ten years ago. The model is tailored to available data in an attempt to fit current theoretical understanding and empirical knowledge to the variables provided in the WFS data sets. It is through such attempts that the scientific value of the WFS will become apparent. The problem of comparability is present at every phase of model development, and can easily be seen to penetrate points (1) to (3~. His is a problem of degree that is present within as well as between coun- tries. It by no mans fatally defeats the meaningfulness of analyzing multiple data sets. Hermalin and Mason (1980 :104-106) discuss this problem In detail in connection with the meaning of educational attain- ment for comparative analyses of reproductive behavior. AS these authors show, a sensitivity to the problem has ramifications for empirical analysis.
47 This problem of comparability is brought into sharp focus by the development of a set of structural equations. The reasoning used in the placement of variables in our equations is intricate and has not been discussed in a cultural and societal void. Likewise, our hypotheses about the signs of coeff icients take explicit account of potential cultural and societal variation. On occasion, this renders single-sign hypotheses within a traditional or transitional setting untenable. Meaningful use of the equations requires that the analyst inquire about the comparability problem as it affects the equations. In specifying a set of structural equations that depends on an explicit body of reasoning, we can pinpoint particular areas for special attention. In essence, the approach we take shifts the comparability problem from the general and abstract to the particular and concrete. His seems to be a constructive way to handle a problem that will never go away. Hypotheses About Coeff icients The establishment of a set of micro-level structural equations that can be estimated in different countries is but the first step in developing a comparative model. The second, and no less critical, step is to develop an understanding of how and why the parameters can vary. The combination of empirical knowledge and reasoning that led to the hypotheses summarized in Tables 1.2 and 1.3 is based on our distinction between traditional and transitional settings. This distinction is the key to understanding the derivations of the various hypotheses about coefficient signs. To review, in an ideal-typic traditional setting, nobody contracepts. In a transitional setting, some portion of the population does contracept. If the great majority of the population contracepts, and does so effec- tively, then by definition the transition is complete, or nearly 50. We have not provided a label for this possibility, nor do we expect to analyze data for counts ies in this category. The traditional/transitional distinction does not connote as much as might at first be inferred. It is not equivalent to the traditional/ modern distinction, or to the gemeinschaft/gesellechaft distinction. Indeed, as must be obvious, we have deliberately avoided these more familiar distinctions. There are a number of reasons for this. The more familiar dichotomies suggest a stopping point to social change. Hey suggest a view concerning means-ends reasoning in traditional settings that has long been known to be incorrect. Hey do not appear to provide a productive vocabulary for the study of the kinds of changes now taking place in developing countries. In our thinking, the traditional and transitional labels apply to a continuum of a single operational variable, pervasiveness of contracep- tion, with traditional settings defined as the zero point. There is no such specificity for placement of transitional settings on the contracep- tive pervasiveness dimension. A transitional setting is simply one in which there is some contraception. In a sense, this is a blend of two distinct epistemologies, referring to the traditional/transitional dis- tinction as ideal-typic, but at the same time mapping it onto the range of a variable. Our reason for invoking ideal types is to connote rela-
48 Lively concrete images of soc fetal settings . me empir ical research we plan to conduct, and the hypotheses we have advanced in planning that research, all assume the methodology and conceptual framework of variables, as distinguished from ideal types. We view traditional settings as those in which the combined force of social structure, culture, the polity, and the economy are consistent with or even cause a lack of contraception. These populations are characterized in the main by natural fertility, with fertility levels determined by taboos' infant feeding practices, rules of residence, etc. (Henry, 1961~. We envision these societies as having no industrializa- tion and minimal contacts with the weal thy and industrialized nations. In these settings, we expect that people perceive no compelling reasons for limiting family size. Indeed, they may not be aware that family size can be limited. Although barely a sketch, this view of traditional settings is akin to an ideal type. AS we perceive them, transitional settings are not only those in which some portion of the population sees reasons for limiting family size and acts on them, but also those in which other aspects of traditionality can be changing. For example, these settings include societies in which there has been little industrialization, but some portion of the population has begun limiting family size because of the influence of family planning programs. Also included are societies in which there may be little family planning program effort, or ineffective organized efforts, but in which some industrialization has taken place. There can be much or only a little contact with the wealthy nations. There may or may not be a record of colonization. The society may differ from the traditional ideal type in a number of respects. What is crucial is that for a substantial portion of the population, the traditional structural, cultural, economic, and political constraints still come together to provide a barrier against family limitation. However, for the society as a whole, this barrier is permeable since some people have begun to limit family size. It is perhaps a semantic fine point whether this treatment of transitional settings is in fact ideal-typic. What matters for our purposes is that we can distinguish traditional and transitional settings such that we can develop a realistic model, as we have done. In sum, we use the distinction between traditional and transitional settings to develop hypotheses about the nature of effects within the structural model, and more particularly, about the ways the structural coefficients can be expected to change (or exhibit stability) over the range of variability represented in WFS countries. Thus, the traditional/ transitional distinction is used to generate a multi-level model. At the micro ~ ever, there is a set of structural equations, for which the choices of variables and cause' ordering have been discussed at length. me coefficients of this model are expected to vary across settings, and the macro component of the model is devoted to understanding this variability. AS. noted earlier, we did not develop the macro structural equations, which would explain between-country variability in the micro-structure' coefficients, because of the priority need to set forth the micro equations in detail. To complete the statement of the multi-level model, the next phase of theory development will have to include work on the
49 macro model. We begin this phase of the work in Chapter 2. In addition empirical research based on the model as developed thus far is both practical and desirable. Thus the following chapter presents the results of an initial, tentative empirical application of the micro-macro model to 15 countries and illustrates the micro specification for two countries--Peru and Korea. Future research will further extend this effort.
