A model comparison approach shows stronger support for economic models of fertility decline Mary K. Shenka,1, Mary C. Townerb, Howard C. Kressc, and Nurul Alamd a Department of Anthropology, University of Missouri, Columbia, MO 65211-1440; bDepartment of Zoology, Oklahoma State University, Stillwater, OK 74078;
[email protected]; and dHealth and Demographic Surveillance Unit-Dhaka, International Center for Diarrhoeal Disease Research, Dhaka 1000, Bangladesh
Edited by Karen L. Kramer, University of Utah, Salt Lake City, UT, and accepted by the Editorial Board March 25, 2013 (received for review October 31, 2012)
T
he demographic transition, in which high fertility and mortality rates decline to low levels, is a global phenomenon with significant ramifications for both global population and modern social organization (1). Beginning in late 18th century Europe, the demographic transition spread during the 19th and 20th centuries, until much of the world experienced major reductions in both mortality and fertility (2–4). Although most literature links the transition to the economic, social, and technological changes associated with development (2), the causal mechanisms underlying it remain the subject of intense debate (4–6). The literature on the demographic transition, especially the remarkable decreases in fertility that characterize it, is crowded with competing theories, making comprehension difficult for academics and policymakers alike. Scholars working on this topic often call for more comprehensive, better-controlled studies that will allow us to tease apart different theoretical explanations (2, 5, 7). However, the data demands for systematic comparative analysis are heavy, and only limited work has been done (8–10). In this article we address this gap by explicitly comparing three prominent classes of models to determine which produces the most robust explanation of a rapid, recent demographic transition in rural Bangladesh. To compare models rigorously, we use an evidence-based statistical approach using model selection techniques derived from likelihood theory (11, 12) and data collected explicitly for this type of comparative analysis. Although this approach is ideal for comparative analysis, it is not frequently used in the social sciences (12) and has not been applied in a comprehensive way to the demographic transition (10). Demographic Transition Theory Theoretical approaches to the demographic transition come from several disciplines, notably demography, economics, and www.pnas.org/cgi/doi/10.1073/pnas.1217029110
evolutionary anthropology, but often share key predictions (Table 1) that can be organized into three classes. Risk and Mortality Models. These models derive from Classic Demographic Transition Theory (13), which proposes that as infant mortality rates fall parents will change their reproductive behaviors to match the increased survival of their children. Rapid population growth will occur during an adjustment period, but once parents recognize that more children will survive childhood, fertility rates will rapidly decline. Recent approaches based on life history theory also stress the importance of decreasing risk as a primary factor in decreasing optimal fertility. Such research examines the relative risk of mortality or high levels of stress for either children (14) or adults (15, 16) as a primary factor in fertility decisions. More recently, Unified Growth Theory suggests that increases in adult life span and child survival rates allow greater payoffs to investments in self and individual children (17, 18). Economic and Investment Models. These models examine the costs and benefits of investing in self and children. For example, Caldwell’s Wealth Flows approach (19) suggests that in traditional agricultural societies, children provide their parents with significant wealth through labor, favoring high fertility, whereas in modern economies children consume wealth, resulting in low fertility. Other researchers have argued that children are always costly but that children’s work can subsidize parental reproduction, leading to higher fertility (20, 21). In contrast, Human Capital Theory (22) suggests that fertility declines with increasing payoffs to investment in human capital (primarily education) in modern labor markets. This approach has also been combined with life history theory (23) and the study of longevity (8). In modern economies, if child quality is a function of the investments made in children, then parents should raise fewer, high-quality children (23–25). This relates to a broader finding (24, 25) that when wealth is heritable, the costs of raising children increase, and fertility levels drop. Opportunity costs of raising children also increase in modern labor markets, especially for women (26–29), who may intentionally reduce fertility to take advantage of new labor market opportunities [including those offered by microcredit or other programs designed to increase women’s market participation (30)] or delay reproduction so long that their fertility is reduced by biological limits (31). Cultural Transmission Models. Cultural transmission theories of the demographic transition were proposed in demography by Cleland and Wilson (32) and have since been used extensively (33, 34). Fertility reductions are thought to result from changes
Author contributions: M.K.S., M.C.T., H.C.K., and N.A. designed research; M.K.S., H.C.K., and N.A. performed research; M.K.S. and M.C.T. analyzed data; and M.K.S., M.C.T., and H.C.K. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. K.L.K. is a guest editor invited by the Editorial Board. 1
To whom correspondence should be addressed. E-mail:
[email protected].
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1217029110/-/DCSupplemental.
PNAS | May 14, 2013 | vol. 110 | no. 20 | 8045–8050
ANTHROPOLOGY
The demographic transition is an ongoing global phenomenon in which high fertility and mortality rates are replaced by low fertility and mortality. Despite intense interest in the causes of the transition, especially with respect to decreasing fertility rates, the underlying mechanisms motivating it are still subject to much debate. The literature is crowded with competing theories, including causal models that emphasize (i) mortality and extrinsic risk, (ii) the economic costs and benefits of investing in self and children, and (iii) the cultural transmission of low-fertility social norms. Distinguishing between models, however, requires more comprehensive, bettercontrolled studies than have been published to date. We use detailed demographic data from recent fieldwork to determine which models produce the most robust explanation of the rapid, recent demographic transition in rural Bangladesh. To rigorously compare models, we use an evidence-based statistical approach using model selection techniques derived from likelihood theory. This approach allows us to quantify the relative evidence the data give to alternative models, even when model predictions are not mutually exclusive. Results indicate that fertility, measured as either total fertility or surviving children, is best explained by models emphasizing economic factors and related motivations for parental investment. Our results also suggest important synergies between models, implicating multiple causal pathways in the rapidity and degree of recent demographic transitions.
Table 1. Key citations and related predictions for three classes of models of fertility decline Risk/mortality models*† 1. Classic Demographic Transition Theory (13): Fertility declines with reductions in infant mortality, increases in development 2. Childhood Environment (15): Fertility declines with decreases in local mortality rates or chronic stress 3. Extrinsic Risk (14): Fertility declines with decreases in extrinsic mortality, especially in infancy and childhood 4. Variance Compensation (16): Fertility declines with decreasing mortality rates, variance in mortality 5. Unified Growth Theory (17, 18): Fertility declines with increasing adult lifespan, increasing child survival rates
Economic/investment models*†
Cultural transmission models*†
1. Wealth Flows (19, 21): Fertility declines with the reduction in child productivity, following a shift away from agriculture 2. Human Capital (22)/Embodied Capital (23)/Unified Growth Theory (18): Fertility declines with increasing payoffs to investment in human capital in modern labor markets 2a. Women’s Opportunity Costs (27–29): Increasing investment in women’s education and careers produce a tradeoff with children and/or delays in childbearing 2b. Investment in Child Quality (22, 23): Increasing payoffs to investments in children motivate parents to have fewer children and increase tradeoffs between children 2c. Rising Costs of Children (22, 25): Fertility declines with increasing costs of raising children, especially when wealth is heritable
1. Diffusion (32, 34): Fertility declines with critical mass of low fertility innovators or mass media technology 2. Social Network Effects (35, 36): Fertility declines with changes in social network structure that foster transmission of new information or adoption of new fertility behaviors 3. Cultural Evolution (37, 38): Fertility declines with an increasing number of high prestige adopters of low fertility 4. Kin Influence (39): Fertility declines with decreasing interactions with kin, increasing interactions with nonkin
*Given space constraints, we focus here on some of the best-known models from several disciplines. † The models discussed often have numerous predictions. We focus on those amenable to modeling with our sample.
in the perception of the value of children, ideal family size, or the acceptance of modern family planning methods. Change begins with adoption of low fertility norms or behaviors (e.g., using contraception or delaying childbearing) by elites and then spreads through society via media or social contact with relatives, neighbors, friends, or partners in social programs (30, 31, 33). A related approach (35, 36) applies social network analysis to fertility decline, focusing on both the transmission of new information as well as the influence of social network members on each other’s behavior. Heterogeneous and sparse networks seem to facilitate the flow of information, whereas homogeneous and dense networks strengthen the effects of social influence. Another set of models comes from cultural evolution theory and proposes that humans seek to increase prestige and have evolved learning biases that lead them to adopt behaviors that aid in this aim (37, 38). If low fertility is characteristic of high prestige members of a society, the rest of the society may emulate low fertility as a means of achieving higher prestige. Related models (39) suggest that kin may help maintain a cultural preference for high fertility, whereas nonkin may introduce low-fertility behaviors or ideals. As societies become more mobile and gain new communication technologies, interactions with kin may decrease and nonkin increase, reducing pronatal social pressures. Cultural transmission models can be seen as either mechanisms of how fertility decline spreads, or as causal models that posit why individuals adopt low fertility. In this article we address only their interpretation as causal models, seeking to test their efficacy as predictors of the adoption of low fertility behavior alongside other potential motivators of change. Comparative Approaches. Many authors have called for comparative research on the demographic transition (5–7, 40), but such work has been limited (8–10). Comparative analyses are challenging for several reasons. Different models often emphasize similar variables or do not produce unique predictions, making them difficult to distinguish using conventional hypothesistesting methods. Datasets often lack the many variables needed to adequately compare multiple models at once. Finally, the ubiquity of standard regression methods means that comparative testing often consists simply of assessing models according to how well one or two key predictors perform (41). Although this method can be useful, model selection methods, which are 8046 | www.pnas.org/cgi/doi/10.1073/pnas.1217029110
becoming the standard in fields such as ecology, are more appropriate for direct comparisons (12). Existing comparative literature supports economic and/or investment models (10, 42), infant mortality reductions (9, 43), and cultural transmission as the primary causes of demographic transitions (44–46). These studies have limitations, however: they only compare a small number of models or variables, and/or their methods are not well-suited to model comparison. Furthermore, the inconsistent methods and variables used by different researchers means we do not know the relative strengths of different predictors or which motivation(s) are more likely to serve as an impetus for large-scale demographic change. In this article we seek to overcome these limitations by using data collected and methods designed explicitly for comparative analysis. To do this, we use an evidence-based statistical approach using model selection techniques derived from likelihood theory (11, 12) and data collected explicitly for comparative analysis. Traditional statistical analyses are poorly suited to model comparison because only the null hypothesis is in a position to be rejected; moreover, P values are poor indicators of the weight of evidence in the data for a particular hypothesis (47). In contrast, an evidence-based approach uses measures such as Akaike Information Criterion (AIC) based on likelihoods to quantify the relative evidence for multiple alternative models (11). Using these methods, we are better able to interpret the support the data give for alternative models, even when models are not nested or have overlapping predictions. Study Setting Data were collected in rural Matlab, Bangladesh, an area known for demographic and public health research conducted by the International Center for Diarrheal Disease Research, Bangladesh (ICDDR,B). The primary economy of the Matlab area is farming of rice and other crops, followed by fishing (48, 49). Villagers may participate in agriculture even if they own no land themselves. Extended patrilocal families live together in a bari or neighborhood containing several small houses. Women practice a limited form of purdah or seclusion and usually spend most of their time in the bari engaged in agricultural processing work, cooking, and childcare. Income is generated from agriculture, fishing, day labor, handicraft production, small businesses, and remittances from family members working in cities or abroad (49). Average annual income was an estimated $1,584 in 2010 purchasing power Shenk et al.
parity dollars (50). Education levels vary considerably, but 30% of the population has no schooling (49). Since the 1990s education has become more widely available, and a small but growing number of men have obtained education-based employment. Education has also become more acceptable for women, a fraction of whom have entered the labor market (51). Labor migration, primarily by men, and remittances have become increasingly important in the Matlab economy (44, 51). These shifts are thought to be linked to decreasing land ownership due to rising population (52) and increasing access to national and international markets for labor and goods (44, 53). The demographic transition in Bangladesh has been studied since the early 1980s (SI Text). Between 1966 and 2010, total fertility rates have fallen from 6.7 to 2.6 children per woman. Conversely, life expectancy at birth has risen from 53 to 69.3 y for men and from 51 to 73.2 y for women, owing mainly to decreases in infant and child mortality (54). Results We analyzed data for two outcome variables: (i) total fertility— the total number of children born to a woman, and (ii) surviving children—the number of a woman’s children currently surviving or having lived past age 10 y. Demographers most commonly use total fertility, whereas evolutionary anthropologists and biologists often use surviving children because it is a better proxy for reproductive success. These different variables may yield different insights into the demographic transition, because total fertility has declined sharply with the demographic transition, whereas surviving children has shown a more modest decline (6). Table 2 shows the best (most parsimonious) model for each model class and outcome variable based on model selection among all potential predictor variables. SI Text provides descriptions of all variables analyzed; Methods describes the process of variable selection. We use AIC to compare alternative models (11). For a given model, AIC = −2log(L) + 2K, where L is the likelihood of the model given the data, and K is the number of parameters in the model. With n = 799, we do not need to use AICc to correct for small samples. Among a specific set of alternative models, the relative likelihoods can be normalized, such that the values sum to 1. These are termed Akaike weights (wi) and are interpreted as the relative likelihood that model i is the best model among those being compared (11). For a single model with a given number of variables, one can also calculate Akaike weights for submodels with all permutations of those variables. The sum of the Akaike weights for each model that a given variable appears in is defined as the relative importance of that variable compared with the others in the model (11).
Risk/Mortality Variables. We analyzed 22 indicators of risk or mortality levels, including measures of food insecurity, length and severity of illness, life expectancy, mortality in local neighborhoods, population mortality rates, residence in an area receiving healthcare interventions, and perception of several types of risk and mortality. Consistent with predictions, more child deaths and higher infant mortality rates were associated with higher fertility, whereas living in a health intervention area was associated with lower fertility. Counter to some expectations (18), higher life expectancy was also associated with higher fertility, although only after adjusting for measures of mortality. Economic/Investment Variables. We analyzed 18 indicators of economic or investment motivations for fertility reduction, including measures of income, occupation, level of education, costs of education, time spent with children, marriage costs, and microcredit use. Consistent with predictions, land ownership and engagement in agriculture were both associated with higher fertility, whereas higher education and nonagricultural occupations were associated with lower fertility. Although it had a negative bivariate correlation with fertility, consistent with other findings (23, 40) income was associated with higher fertility in the presence of other variables. Cultural Transmission. We analyzed 25 indicators of cultural trans-
mission of low fertility norms among women and their close kin, such as education, travel, labor migration, media exposure, and access to and attitudes about contraception. Consistent with predictions, the general fertility rate has a positive effect on a woman’s fertility, whereas higher levels of education, residence in an intervention area where free contraceptives and promotion of family planning were introduced in 1978, and the husband’s labor migration to cities or abroad have negative effects on fertility. Exposure to modern media has a negative bivariate relationship with fertility, but contrary to predictions the relationship becomes positive with other variables in the model. Model Comparison. Table 3 shows the results of a comparison across model classes. For total fertility, the economic/investment model performs best, having the lowest AIC value and taking 0.738 of the weight; the risk/mortality model has the next lowest AIC value and garners 0.234 of the weight (the Δ of 2.3 suggests that the risk/mortality has moderate support in comparison with the economic/investment model, which has substantial support); the cultural transmission model is better than a base model but receives very little of the weight (0.029). Likewise, for surviving children the economic/investment model performs best, with by far the lowest AIC value and the highest weight (0.996). In
Table 2. Most parsimonious model for each model class
Variable*† Total fertility Child deaths in bari Woman in intervention area Infant mortality rate‡ Life expectancy at birth‡
Sign Importance + − + +
1.00 0.86 0.83 0.74
Surviving children Child deaths in bari + Infant mortality rate‡ + Woman in intervention area −
0.95 0.81 0.69
Economic/investment model
Cultural transmission model
Variable*†
Variable*†
Total fertility Woman’s level of education Whether family owns land Husband’s primary occupation Family engaged in agriculture Household income Surviving children Family engaged in agriculture Whether family owns land Woman’s level of education Household income
Sign Importance − + − + +
0.95 0.91 0.87 0.73 0.66
+ + − +
0.99 0.98 0.67 0.64
Sign Importance
Total fertility Woman’s level of education Husband’s location Woman in intervention area General fertility rate‡
− − − +
0.98 0.85 0.78 0.69
Surviving children Woman in intervention area Woman’s level of education Husband’s location General fertility rate‡ Exposure to modern media
− − − + +
0.78 0.70 0.67 0.67 0.61
*Variables are listed in order of importance. † Woman’s age and age at marriage are included as control variables in all models. ‡ Figure given for the year of the woman’s marriage, when childbearing is likely to begin.
Shenk et al.
PNAS | May 14, 2013 | vol. 110 | no. 20 | 8047
ANTHROPOLOGY
Risk/mortality model
contrast, essentially no weight goes to either the risk/mortality or cultural transmission models. Finally, Table 4 shows the best inclusive models that result when all independent variables from Table 2 are modeled simultaneously. The inclusive models, which are allowed to draw independent variables from across model classes, are superior to the best model within any single model class. The inclusive models also have the lowest AIC values (for total fertility AIC = 2936.4, for surviving children AIC = 2793.8) and receive all of the weight (1.000) in a model comparison with the other model classes. For the inclusive model of total fertility, R2DEV,adj,1 = 0.49 and R2adj = 0.47, whereas for the inclusive model of surviving children, R2DEV,adj,1 = 0.49 and R2adj = 0.46. These measures suggest that the models discussed have relatively high explanatory value. A table showing the importance values for all variables modeled in Table 2 is included in Supporting Information. Discussion The demographic transition is a global phenomenon with significant ramifications for worldwide population levels and resource availability in the 21st century and beyond (1). High birth rates are a key deterrent to economic development in less developed nations, whereas population aging combined with belowreplacement fertility undermines social safety nets and contributes to social tensions in more developed nations. This article sheds light on both theoretical and methodological concerns in the study of fertility decline, with implications for understanding demographic transitions ongoing in rural Bangladesh and globally. Although studies of fertility decline abound, comparative research has been limited. Our study shows the utility of model selection methods in weighing relative evidence for alternative models. Distinguishing between different classes of models has several important implications. First, it leads to a more nuanced empirical understanding of the demographic transition, answering previous calls in the literature for research of this type (4, 5, 7). Second, it raises the important theoretical questions of why economic changes that affect motivations for investment in self and children should have stronger effects than changes in other social domains, and whether this phenomenon is universal or varies regionally. Third, identifying the strongest drivers of fertility decline greatly improves our ability to design the policies and interventions most likely to have an impact on fertility decisions. Given the very large expenditures on family planning programs globally, this is far from a trivial concern. Methodological Implications. Model selection approaches avoid the limitations that can arise when focusing tests on individual models or extrapolating from a limited number of variables. Our work is unique in using a dataset with enough detail (number and types of variables) to provide a strong test of the comparative strength of different models of fertility decline, and our findings demonstrate
Table 3. Model comparison Model Total fertility Economic/investment model Risk/mortality model Cultural transmission model Base model Surviving children Economic/investment model Risk/mortality model Cultural transmission model Base model
K
AIC
Δi
wi
7 7 7 2
2950.1 2952.4 2956.6 2978.5
0 2.3 6.5 28.4
0.738 0.234 0.029 0.000
6 6 8 2
2800.4 2811.8 2815.2 2825.4
0 11.4 14.8 25.0
0.996 0.004 0.001 0.000
K refers to the number of fitted parameters for each model; Δi refers to the change in AIC between the lowest value and the variable of interest; wi refers to the Akaike weight.
8048 | www.pnas.org/cgi/doi/10.1073/pnas.1217029110
Table 4. Most parsimonious inclusive models Variable*† Total fertility Child deaths in bari (R) Husband’s primary occupation (E) Whether family owns land (E) Woman in intervention area (R, C) Woman’s level of education (E, C) Infant mortality rate‡ (R) Life expectancy at birth‡ (R) Surviving children Whether family owns land (E) Family engaged in agriculture (E) Child deaths in bari (R) Woman in intervention area (R, C) Household income (E) Woman’s level of education (E, C)
Sign
Importance
+ − + − − + +
0.99 0.93 0.93 0.81 0.75 0.67 0.62
+ + + − − −
0.98 0.98 0.87 0.85 0.58 0.53
R refers to a risk/mortality variable, E refers to an economic/investment variable, and C refers to a cultural transmission variable. *Variables are listed in order of importance. † Woman’s age and age at marriage are included as control variables in all models. ‡ Figure given for the year of the woman’s marriage, when childbearing is likely to begin.
the value of model comparison with the study of the demographic transition and thus potentially to other complex, multicausal social phenomenon. Our findings corroborate those of previous studies, especially the work of Kabeer (52), who emphasizes land saturation due to rising population as a key predictor of fertility decline, as well as previous findings on the efficacy of Matlab fertility and health interventions (55, 56). However, we are also able to determine the relative importance of each of these sets of predictors, both in relation to each other and to other variables and models, yielding results that suggest (i) the importance of economic/investment models compared with other models, and (ii) that analyses focused on limited sets of variables may miss significant relationships, such as that between risk and total fertility, which have not been emphasized in the Bangladeshi context. Theoretical Implications. There are three key theoretical implications of these findings: (i) the primacy of economic models over other types of models in predicting fertility decline, (ii) a possible threshold for mortality risk, and (iii) the multicausal nature of fertility decline. What the best models and predictors therein have in common is that they reflect meaningful changes in the ecological conditions (i.e., the tangible costs and benefits) faced by individuals. Economic circumstances almost always shape individual costs and benefits—as exemplified by the importance of variables such as whether the family owns land or is involved in agriculture, the woman’s level of education, and her husband’s occupation. Risk variables can also affect individual costs and benefits, especially when risks are high, as reflected in variables such as the infant mortality rate, the number of child deaths in the marital bari, and the effect of residing in a health intervention area. In contrast, cultural transmission models as a group may be comparatively weak because many transmission variables do not alter the individual cost and benefit calculus appreciably. Our results thus call into question the importance of models of cultural transmission that focus on variables such as media exposure or contact with foreigners or modern ideas (32, 57), and suggest that the importance of more generalized, anonymous forms of transmission such as media exposure is very limited in comparison with intensive, individualized forms of transmission, such as contraceptive interventions or the location of the husband, which have more tangible effects on costs and benefits. Model comparison results also suggest a stronger role for economic/investment models over other models, although this Shenk et al.
Shenk et al.
Methods Data. The Matlab study area includes ∼250,000 people on whom detailed demographic data have been collected since 1966 as part of the ICDDR,B Health and Demographic Surveillance System (HDSS). The region is divided into an ICDDR,B Area, where many health and family planning interventions have been made, and a Government Service Area, functioning as a control area where basic health services are available from the Bangladeshi government. The majority of demographic research on Matlab uses the HDSS sample. Because the intensive data demands of our analysis required more detailed individual information than was available, we drew a random subsample of women from the HDSS for a tailored survey. Our sample included even numbers of women from (i) the ICDDR,B Area and the Government Service Area, and (ii) each of three 15-y age categories (20–34, 35– 49, and 50–64), allowing for better representation of older women and 45 y of time depth regarding fertility and its correlates. Our final survey sample size was 944 women. Our survey was based on a broad review of the demographic transition literature and designed to address predictions from causal models such as those in Table 1. Survey data contain sufficient information to allow the simultaneous comparison of multiple models and have enough power to provide robust tests. A few variables not taken from the survey come from published data on the Matlab population and linked to survey respondents on the basis of their age or age at marriage. Because not all women in the sample had completed fertility, woman’s age and woman’s age at marriage were used as control variables in all models (SI Text). A key advantage of our dataset is that we have numerous measures for each class of model, and sometimes multiple proxies for key variables. To identify appropriate independent variables from the pool of candidate variables, we used the following criteria. First, the variable had to have been suggested in the literature as an indicator of a particular class of model. Second, the meaning of the variable in the local Bangladeshi context had to match the meaning suggested by theory so that the interpretation of the variable would be consistent with model predictions. Finally, each variable was screened for data entry problems and completeness. Some theoretical models of fertility assume that the system is at equilibrium, a condition that is rarely met in human samples (59). Fertility in Matlab is clearly not at equilibrium because the phenomenon of interest is the result of change. Our analysis, however, addresses this concern in several ways (SI Text). Such concerns are also equally true for risk, economic, and cultural transmission variables, and thus should not affect the interpretation of one set of variables compared with the others. Analysis. We focused our analyses on all women in our survey married for at least 5 y with or without children (n = 810) and for whom data were available on all variables used to test all models (n = 799) (SI Text). We constructed models separately for the outcome variables total fertility (total number of births) and surviving children (number of children surviving past the age of 10 y). Because these outcome variables are count data with a limited range (0–11 children) and no evidence of overdispersion, we estimated generalized linear models with a Poisson error distribution and log link function. Our primary analytical goal was to evaluate the relative evidence for alternative models within and across model classes; analyses were completed using R functions (60), including the glmulti package (61). Our modeling approach was structured as follows. First, we conducted model selection analyses within each of the three major classes of models (Table 2). For each model class, we wanted a set of independent variables that, when put together, produced a strong contender for that class. The initial set of variables was first reduced by removing (i) variables that had no apparent relationship with the outcome variable, and (ii) variables that were conflated with others in the model. We then used glmulti to systematically draw combinations of variables from the narrower list and find the model with the lowest AIC value. For the best model within each class, we report the direction of the relationship between the independent variables and the outcome variables, and also each independent variable’s importance value (Table 2). Second, for each outcome variable, we did a formal comparison of the models that resulted from step 1 by determining AIC values for each model, finding the difference between each model’s AIC value and the lowest AIC value among the models compared, and using the AIC differences to calculate Akaike weights (Table 3). Finally, for each outcome variable, all independent variables resulting from step 1 were analyzed together using glmulti to find an inclusive model that retained the independent variables of highest importance (Table 4). A variety of pseudo-R2 measures have been developed for comparing nonlinear models, although they are not measures of explained variance in an ordinary least-squares sense. Mittlebock and Waldhor (62) propose several such measures for Poisson
PNAS | May 14, 2013 | vol. 110 | no. 20 | 8049
ANTHROPOLOGY
result is more moderate in conditions of higher risk, suggesting a possible threshold effect for mortality risks on fertility decisions. This is especially clear when comparing the results for total fertility with those for surviving children. Economic/investment models are by far the best predictors of fertility for surviving children; they are also the best predictors of total fertility, although here risk/mortality models also retain some weight. Risk/ mortality models gain ground with the total fertility outcome variable because of the 25% of women in the sample who have lost children; women who have lost multiple children strengthen this effect. Older women faced exceptionally high levels of mortality and risk during their reproductive years, experiencing first Bangladesh’s 1971 Liberation War, followed by devastating floods and famines in 1974–1975. When we remove these women and limit our analyses to the 725 women younger than 60 y we find that the risk/ mortality model of total fertility becomes much less competitive against the economic/investment model. In particular, infant mortality rate is no longer a salient variable. This suggests that once mortality rates become low enough in a population, they may cease to exert a strong influence on fertility decisions. This balance between economics and risk matches the predictions of models that focus on the interaction of economics and mortality constraints, predicting that mortality will be a key determinant of fertility as long as mortality remains high, but that economic or parental investment factors will become primary once mortality rates or variance in mortality fall (8, 16, 23). Once mortality rates become low enough, it may pay for parents to make larger expenditures on schooling, or a high-status modern occupation. To benefit from larger investments, however, parents must limit the number of children they have to begin with or restrict inheritance to a particular child or children (6, 58). High economic growth is also known to reduce fertility to very low levels (18, 22, 23), however, which mortality reduction itself has not been shown to do. Our findings echo those of previous studies comparing theoretical models from historical (18), cross-national (9), and evolutionary perspectives (10) that human capital investments deriving from the demand for well-trained workforces have driven demographic transitions in many parts of the world. In the Bangladeshi case, economic/investment models may be particularly powerful owing to the interaction of land saturation (decreasing per capita land ownership due to population growth) with the increasing integration of villagers into a wage-labor economy (52). This situation creates a feedback loop in which children become relatively more expensive because they no longer subsidize themselves through agricultural labor and, at the same time, payoffs to alternate forms of investment such as education are increasing, prompting parents to reduce fertility in favor of more intensive educational or skills investment. Finally, our findings make it clear that fertility decline in Matlab is multicausal. The best inclusive models contain variables from each class of causal model, suggesting that a full explanation of the demographic transition is likely to include changes in economic conditions, risk profiles, and more individualized forms of cultural transmission either acting separately or in feedback with each other. Moreover, several of the strongest predictors are variables that tap into more than one causal pathway. Variables such as a woman’s education, residence in the intervention area, and the location of the respondent’s husband may exhibit a combination of economic and cultural transmission effects. Other variables, including residence in the intervention area, suggest a potential interaction between risk and cultural transmission factors. SI Text provides further discussion of this point. Such variables may be especially important drivers of fertility decline. In summary, although we find that when compared head to head, models emphasizing economic and investment variables are by far the best predictors of fertility, our results also corroborate previous research regarding important predictors and indicate that multiple causal pathways are needed to explain the rapid, recent fertility declines in modern Bangladesh and much of the developing world.
ACKNOWLEDGMENTS. We thank the International Center for Diarrheal Disease Research, Bangladesh, especially Peter Kim Streatfield, and Taslim
Ali, our field assistants (Shifat Khan, Nargis Sultana, Latifun Nahar, Akterun Naher, Lutfa Begum, Mouloda Aziz, and Farhana Akand), research assistants (Roslyn Fraser and Kathrine Starkweather), and the people of Matlab, Bangladesh for their unfailing helpfulness during this project; and Barney Luttbeg, Curtis Atkisson, and several others for statistical and methodological advice. This research was funded by US National Science Foundation Award BCS-0924630 and a Research Leave from the University of Missouri.
1. Lam D (2011) How the world survived the population bomb: Lessons from 50 years of extraordinary demographic history. Demography 48(4):1231–1262. 2. Caldwell JC, Caldwell BK, Caldwell P, McDonald PF, Schindlmayr T (2006) Demographic Transition Theory (Springer, Dordrecht, The Netherlands). 3. Chesnais J-C (1992) The Demographic Transition: Stages, Patterns, and Economic Implications: A Longitudinal Study of Sixty-Seven Countries Covering the Period 17201984 (Clarendon Press, Oxford). 4. Mason KO (1997) Explaining fertility transitions. Demography 34(4):443–454. 5. Borgerhoff Mulder M (1998) The demographic transition: Are we any closer to an evolutionary explanation? Trends Ecol Evol 13(7):266–270. 6. Hobcraft J (2006) The ABC of demographic behaviour: How the interplays of alleles, brains, and contexts over the life course should shape research aimed at understanding population processes. Popul Stud (Camb) 60(2):153–187. 7. Clarke AL, Low BS (2001) Testing evolutionary hypotheses with demographic data. Popul Dev Rev 27(4):633–660. 8. Galor O (2011) Unified Growth Theory (Princeton Univ Press, Princeton). 9. Sanderson S, Dubrow J (2000) Fertility decline in the modern world and in the original demographic transition: Testing three theories with cross-national data. Population Environment 21(6):511–537. 10. Shenk MK (2009) Testing three evolutionary models of the demographic transition: Patterns of fertility and age at marriage in urban South India. Am J Hum Biol 21(4): 501–511. 11. Burnham KP, Anderson DR (2002) Model Selection and Multi-Model Inference: A Practical Information-Theoretic Approach (Springer, New York). 12. Towner MC, Luttbeg B (2007) Alternative statistical approaches to the use of data as evidence for hypotheses in human behavioral ecology. Evolutionary Anthropol 16(3): 107–118. 13. Coale A (1972) The demographic transition. IUSSP Liege International Population Conference (International Union for the Scientific Study of Population, Liege, Belgium), pp 53–72. 14. Quinlan RJ (2007) Human parental effort and environmental risk. Proc Biol Sci 274(1606):121–125. 15. Chisholm JS, et al. (1993) Death, hope, and sex: Life-history theory and the development of reproductive strategies. Curr Anthropol 34(1):1–24. 16. Leslie P, Winterhalder B (2002) Demographic consequences of unpredictability in fertility outcomes. Am J Hum Biol 14(2):168–183. 17. Cervellati M, Sunde U (2005) Human capital formation, life expectancy, and the process of development. Am Econ Rev 95:1653–1672. 18. Galor O (2012) The demographic transition: Causes and consequences. Cliometrica 6: 41–28. 19. Caldwell JC (1982) Theory of Fertility Decline (Academic Press, London). 20. Sear R, Coall D (2011) How much does family matter? Cooperative breeding and the demographic transition. Popul Dev Rev 37(Suppl 1):81–112. 21. Kramer K (2005) Children’s help and the pace of reproduction: Cooperative breeding in humans. Evol Anthropol 14(6):224–237. 22. Becker GS (1992) Fertility and the economy. J Popul Econ 5(3):185–201. 23. Kaplan H (1996) A theory of fertility and parental investment in traditional and modern human societies. Yearb Phys Anthropol 101(23):91–135. 24. Luttbeg B, Borgerhoff Mulder M, Mangel M (2000) To marry again or not: A dynamic model for demographic transition. Adaptation and Human Behavior: An Anthropological Perspective, eds Cronk L, Chagnon N, Irons W (Aldine de Gruyter, Hawthorne, NY). 25. Mace R (1998) The coevolution of human fertility and wealth inheritance strategies. Philos Trans R Soc Lond B Biol Sci 353(1367):389–397. 26. Becker G (1991) A Treatise on the Family (Harvard Univ Press, Cambridge, MA), 2nd Ed. 27. Budig M, England P (2001) The wage penalty for motherhood. Am Sociol Rev 66: 204–225. 28. Low BS, Simon CP, Anderson KG (2002) An evolutionary ecological perspective on demographic transitions: Modeling multiple currencies. Am J Hum Biol 14(2): 149–167. 29. Turke PW (1989) Evolution and the demand for children. Popul Dev Rev 15:61–89. 30. Schuler SR, Hashemi SM (1994) Credit programs, women’s empowerment, and contraceptive use in rural Bangladesh. Stud Fam Plann 25(2):65–76. 31. Kaplan HS, Hill KR, Lancaster JB, Hurtado AM (2000) A theory of human life history evolution: Diet, intelligence, and longevity. Evol Anthropol 9:156–185. 32. Cleland J, Wilson C (1987) Demand theories of the fertility transition: An iconoclastic view. Population Studies 41(1):5–30.
33. Basu A (1993) Cultural influences on the timing of first births in india: Large differences that add up to little difference. Population Studies 47:85–95. 34. Bongaarts J, Watkins SC (1996) Social interactions and contemporary fertility transitions. Popul Dev Rev 22(4):639–682. 35. Kohler HP (2001) Fertility and Social Interactions: An Economic Perspective (Oxford Unive Press, Oxford). 36. Kohler H-P, Behrman JR, Watkins SC (2001) The density of social networks and fertility decisions: Evidence from South Nyanza district, Kenya. Demography 38(1): 43–58. 37. Boyd R, Richerson PJ (1985) Culture and the Evolutionary Process (Univ of Chicago Press, Chicago). 38. Richerson PJ, Boyd R (2005) Not by Genes Alone: How Culture Transformed Human Evolution (Univ of Chicago Press, Chicago). 39. Newson L, et al. (2007) Influences on communication about reproduction: the cultural evolution of low fertility. Evol Hum Behav 28(3):199–210. 40. Bock J (2002) Introduction: Evolutionary theory and the search for a unified theory of fertility. Am J Hum Biol 14(2):145–148. 41. Heuveline P (2001) Demographic pressure, economic development, and social engineering: An assessment of fertility declines in the second half of the twentieth century. Popul Res Policy Rev 20(5):365–396. 42. Palloni A, Rafalimanana H (1999) The effects of infant mortality on fertility revisited: new evidence from Latin America. Demography 36(1):41–58. 43. Reher DS (2004) The demographic transition revisited as a global process. Popul Space Place 10:19–41. 44. Asfar R (2009) Unraveling the Vicious Cycle of Recruitment: Labor Migration from Bangladesh to the Gulf States (International Labor Organization, Geneva, Switzerland). 45. Bravo JH (1996) Theoretical views on fertility transitions in Latin America: What is the relevance of a diffusionist theory? The Fertility Transition in Latin America, eds Guzman JM, Singh S, Rodriguez G, Pantelides EA (Clarendon, Oxford). 46. Bryant J (2007) Theories of fertility decline and the evidence from development indicators. Popul Dev Rev 33(1):101–127. 47. Anderson DR, Burnham KP, Thompson WL (2000) Null hypothesis testing: Problems, prevalence, and an alternative. J Wildl Manage 64(4):912–923. 48. Holman DJ, O’Connor KA (2004) Bangladeshis. Encyclopedia of Medical Anthropology, eds Ember CR, Ember M (Kluwer Academic, New York). 49. ICDDR,B (2007) Health and Demographic Surveillance System—Matlab. 2005 Socioeconomic Census (International Center for Diarrheal Disease Research, Bangladesh, Dhaka, Bangladesh). 50. International Monetary Fund World Economic Outlook Database(2012). 51. Razzaque A, Streatfield PK, Evans A (2007) Family size and children’s education in Matlab, Bangladesh. J Biosoc Sci 39(2):245–256. 52. Kabeer N (2001) Ideas, economics and the ‘sociology of supply’: Explanations of fertility decline in Bangladesh. J Dev Stud 38(1):29–70. 53. Nowak JJ (1993) Bangladesh: Reflections on the Water (Indiana Univ Press, Bloomington, IN). 54. ICDDR,B (2012) Health and Demographic Surveillance System—Matlab. Registration of Health and Demographic Events 2010 (International Center for Diarrheal Disease Research, Bangladesh, Dhaka, Bangladesh). 55. Baqui AH, et al. (2009) Effectiveness of home-based management of newborn infections by community health workers in rural Bangladesh. Pediatr Infect Dis J 28(4):304–310. 56. Haines A, et al. (2007) Achieving child survival goals: Potential contribution of community health workers. Lancet 369(9579):2121–2131. 57. Barkat-e-Khuda, Hossain MB (1996) Fertility decline in Bangladesh: Toward an understanding of major causes. Health Transit Rev 6(Suppl):155–167. 58. Hrdy SB, Judge DS (1993) Darwin and the puzzle of primogeniture: An essay on biases in parental investment after death. Hum Nat 4(1):1–45. 59. Low B, Hazel A, Parker N, Welch K (2008) Influences on women’s reproductive lives: Unexpected ecological underpinnings. Cross-Cultural Res 42:201–219. 60. R Development Core Team (2012) R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, Vienna). 61. Calcagno V (2012) glmulti: GLM model selection and multimodel inference made easy. R package version 0.6-3. Available at http://CRAN.R-project.org/package=glmulti. 62. Mittlbock M, Waldhor T (2000) Adjustments for R2-measures for Poisson regression models. Comput Stat Data Anal 34(4):461–472.
distributions, including R2DEV,adj,1. We also calculated R2adj for our inclusive models based on ordinary least squares regression (i.e., as if our count data were continuous). All variables analyzed are listed in SI Text, as are details of variables included in final models.
8050 | www.pnas.org/cgi/doi/10.1073/pnas.1217029110
Shenk et al.
Supporting Information Shenk et al. 10.1073/pnas.1217029110 SI Text Fertility Decline in Bangladesh. The demographic transition in
Bangladesh has been studied since the early 1980s, often in terms of the efficacy of family planning programs. Early research in Matlab focused on cultural transmission (1), proposing that fertility reductions occurred through the Community Health Research Workers, local women who visited households regularly to deliver family planning services, do health assessments, and actively work to change people’s perceptions of ideal family size and acceptance of contraception. Other studies have found diffusion effects for contraceptive use rates (2), changes in ideal family size (3–5), changes in women’s roles (6), and changes in cultural values (7). Collectively, these studies suggest that fertility in Matlab and Bangladesh has been reduced through diffusion processes. In contrast, other studies have found economic factors including women’s education and the presence of schools to be significant predictors of fertility reduction (8–13). Additionally, Kabeer (12) suggests that many of the social changes observed in cultural transmission studies are the result of economic pressures from Bangladesh’s exploding population, which has reduced the availability of arable land, thus motivating parents to reduce the number of heirs. Risk reduction models have not been commonly used to explain the demographic transition in Bangladesh, although Hossain et al. (14) find that infant deaths reduce birth spacing and may increase overall fertility levels in six regions. Comparative studies (15, 16) find that the majority of observed reduction in fertility levels can be attributed to the effects of female education, female employment, and access to media on contraceptive use. Neither study can point to which set of factors is primary, however, nor do they take into account the effects of mortality or risk. Explanation of Variables. General notes on sample and variables. Data were collected from April to August 2010. Respondents were women drawn from a full list of all members of the eligible study population, including all women aged 20–64 y in the International Center for Diarrheal Disease Research, Bangladesh (ICDDR,B) Health and Demographic Surveillance System, a population of more than 65,000 women. Women were drawn with equal probability from within the ICDDR,B Area and the Government Service Area, and also from within each of three 15-y age categories (20–34, 35–49, and 50–64), allowing for better representation of older women who would otherwise be underrepresented owing to rapid population growth. Our data are censored because younger women in our sample may not yet have commenced or completed fertility. Survival analysis was not feasible owing to data constraints with respect to the timing of women’s births, thus we deal with potential censorship in three ways. First, we include controls for age and age at marriage in all models to adjust for primary effects of fertility timing. Second, we limit the sample to married women. There is very little nonmarital fertility in our study population owing to early marriage, strong taboos against premarital sex, and social segregation of the sexes (especially before marriage). We thus consider unmarried women not at risk for fertility and exclude them from these analyses. Five divorced or widowed women with no children are also excluded from the sample. Third, we deal with the potential effects of censorship by limiting the sample to women who have been married 5 y or more. In other words, we exclude newlyweds who are the least likely to have completed (or even commenced) fertility. Finally, given the current fertility trend for women to have two to three children in the first several Shenk et al. www.pnas.org/cgi/content/short/1217029110
years of their marriage before stopping altogether, it is likely that the majority of women older than 30 y have completed their fertility. Thus, only a small fraction (5%) of the women in our sample are likely to be at high risk of further childbearing because they are younger than 30 y and have fewer than two children. Our survey included questions on a broad range of topics related to fertility, and women were asked to answer questions relating to their childhoods as well as their life after marriage. All categorical variables use scales suggested by open-ended interviews and adjusted by researchers after extensive pretesting of survey questions and examination of survey data. Descriptions of all variables included in the final models, as well as lists of variables investigated but excluded, follow. Summary statistics for included variables are shown in Tables S1 and S2. Outcome and control variables. (i) Total Fertility: Count variable, given by the respondent (i.e., the woman included in our survey) then checked against existing demographic data. The woman’s total number of live births, a commonly used proxy for fertility in the demographic literature (Fig. S1). (ii) Surviving Children: Count variable, given by respondent, checked against existing demographic data. The woman’s total number of children currently surviving or having lived past age 10, a commonly used proxy for fertility in the evolutionary anthropology literature (Fig. S1). (iii) Age: Continuous variable, calculated from existing demographic data and confirmed via interview with respondent. Used as a control variable in all models including the base model. (iv) Age at Marriage: Continuous variable, given by respondent and checked against date of marriage (if known) in existing demographic data. Used as a control in all models, including the base model. This is a key control because fertility risk begins at marriage in this culture, thus the length of time since marriage is a key correlate of both fertility and the risk of fertility. Although some older women in this sample were married as young as age 7 y, because the purpose of this variable was to adjust for time at risk for fertility, age at marriage was adjusted to 11 y for all women married younger than 11 y because this is the youngest age at which pregnancy is likely to be possible. As discussed above, newlyweds (women married less than 5 y) were excluded from the sample. Risk/mortality variables in model. (i) Number of Child Deaths in the Marital Bari: Continuous variable, given by respondent. The number of child deaths occurring in the respondent’s marital bari (patrilineal neighborhood), or the bari/area where she lived after marriage, since the time of her marriage other than the deaths of her own children. Higher number of child deaths in the marital bari are associated with higher fertility. (ii) Infant Mortality Rate in the Respondent’s Year of Marriage: Continuous variable, determined from publicly available ICDDR,B annual reports. The infant mortality rate is calculated annually out of 1,000 children born. Higher infant mortality rates are associated with higher fertility. (iii) Woman in Intervention Area: Categorical variable, determined from sample based on respondent’s location. This variable was recorded as a 1 if the respondent resided in the ICDDR,B area and as a 0 if she did not. A program providing free maternal and child health care, free access to frequently needed treatments such as oral rehydration therapy, and hospital care for severe illness was introduced in the ICDDR,B intervention area, comprising roughly half the study population, in 1978. At this time our oldest respondents would have been in their early 30s, and many of them would still have been bearing children. Similar, although often more limited, programs were introduced in the nonintervention area by the Bangladeshi government or other nongovernmental organization in the 1980s and 1990s—thus 1 of 5
women in the intervention area have experienced longer and more reliable exposure to better health care than have women in the nonintervention area. Residence in the intervention area is associated with lower fertility. This variable is also a proxy for access to contraception and family planning messages (details in Cultural transmission variables in model, below). (iv) Life Expectancy at Birth in the Respondent’s Year of Marriage: Continuous variable, determined from publicly available ICDDR,B annual reports. Higher life expectancy at birth is associated with lower fertility before controls and higher fertility when mortality controls are included in the model. Risk/mortality variables not retained. Additional risk/mortality variables examined but eliminated from final models based on model selection criteria include the following: child and adult deaths in the respondent’s natal bari; adult deaths in the respondent’s marital bari; woman’s level of childhood food insecurity; woman’s frequency and severity of recent and childhood illnesses; woman’s perception of several different types of risks, including child morbidity and mortality, adult morbidity and mortality, and the frequency of accidents and violence in the community; and whether the woman was in her childbearing years during any of four major mortality shocks (the 1971 Liberation War, a devastating flood and famine in 1974–1975, a large shigellosis epidemic in 1984, and a cholera epidemic in 1991–1992). Economic/investment variables in model. (i) Whether the Family Owns Land: Categorical variable, given by respondent. This variable was recorded as a 1 if the respondent and her husband owned any land and as a 0 if they did not. Amount of land owned was also collected but did not improve the fit of the model. Land ownership is associated with higher fertility. Several authors have linked land ownership to fertility (12, 17, 18). (ii) Family Engaged in Agriculture: Categorical variable, given by respondent. This variable was recorded as a 1 if the respondent’s husband or the respondent herself were engaged in agriculture as a primary or secondary occupation and recorded as a 0 if they were not. Engagement in agriculture is associated with higher fertility. (iii) Woman’s Level of Education: Continuous variable, given by respondent. Level of education was collected as years of education; most women have no education (n = 313) or primary education only (n = 268), with smaller numbers of women having secondary education (n = 199) and only a few women having more than secondary education (n = 17). Although a categorical version of the variable was tried, the continuous version was more parsimonious. Many economic models of fertility emphasize the increasing costs and importance of investment in education in reducing fertility. Higher levels of education are associated with lower fertility even when age at marriage is controlled. This variable is also included in Cultural transmission variables in model, below. (iv) Husband’s Occupation: Categorical variable, given by respondent. Primary occupation of the respondent’s husband, categorized using data from qualitative interviews to reflect the degree of engagement with a modern market economy: 0 = farmer/fisherman (limited engagement) or laborer (moderate engagement), 1 = business owner (moderate to high engagement) or salaried worker (high engagement). A four-category version of the coding was tried but found to be less parsimonious. More market-engaged types of occupations are associated with lower fertility. (v) Household Income: Logged continuous variable, given by respondent. Combined monthly income of husband and respondent from all sources in Bangladeshi Taka, logged to adjust for high variance and right skew. Higher income is associated with higher fertility once education is controlled. Economic/investment variables not retained. Additional economic/investment variables examined but eliminated from final models based on model selection criteria include the following: husband’s education; costs of child education; whether the woman worked Shenk et al. www.pnas.org/cgi/content/short/1217029110
outside the home after marriage (women are rarely used outside the home in this sample, so a more detailed measure was not feasible); marriage costs of the woman and her husband (a measure of both wealth and direct investment in the new family by the couple’s parents); the time the woman spent with her own mother or father when she was a child (a proxy for direct investment by parents); the time the woman and her husband spend with their oldest son and daughter; the amount of land owned by the woman and her husband; and whether the woman’s household participated in a microcredit program. Cultural transmission variables in model. (i) Woman’s Level of Education: Continuous variable, given by respondent. Many models of fertility decline argue that information or social norms learned in school may encourage reduced fertility. Higher levels of education are associated with lower fertility. This variable is also included in Economic/investment variables in model, above. (ii) Woman in Intervention Area: Categorical variable, determined from sample based on respondent’s location. This variable was recorded as a 1 if the respondent resided in the ICDDR,B area and as a 0 if she did not. A program providing free access to contraceptives, education on how to use them, and messages from health workers that family limitation was important for family happiness, well-being, and child health was introduced in the ICDDR,B intervention area, comprising roughly half the study population, in 1978. At this time our oldest respondents would have been in their early 30s, and many of them would still have been bearing children. Similar, although often more limited, programs were introduced in the nonintervention area by the Bangladeshi government or other nongovernmental organization in the 1980s and 1990s—thus women in the intervention area have had longer and more reliable exposure to both contraception and family planning messages than have women in the nonintervention area. Residence in the intervention area is associated with lower fertility. This variable is also a proxy for greater access to health care (details in Risk/mortality variables in model, above). (iii) Location of Husband: Categorical variable, given by respondent. Husband’s location was used as a proxy for exposure to potential low-fertility social norms associated with professional occupations, high status, and urban residence. The variable was coded as follows: 1 = in Matlab (limited exposure), 2 = in Dhaka or another Bangladeshi city (moderate exposure), and 3 = abroad (high exposure). Residence of the husband in cities or abroad was associated with increasingly low levels of fertility. Although part of this effect may be due to physical absence, several lines of data suggest that low fertility among labor migrants is intentional. Labor migrant grooms marry early and plan long visits home with the goal of getting their wives pregnant, and in qualitative interviews respondents often discussed the fact that higher levels of education, higher status jobs, and better marriages for children are made possible by labor migration and fertility reduction. (iv) General Fertility Rate in Respondents’ Year of Marriage: Continuous variable, determined from publicly available ICDDR,B annual reports. The general fertility rate (GFR) is a count of the number of births per year per 1,000 women, used here as a proxy for prevailing social norms regarding fertility. A high GFR is associated with higher individual fertility. (v) Index of Exposure to Modern Media: Continuous variable, calculated based on data given by respondent. This variable draws on questions about exposure to modern media including television, radio, newspapers, books, posters, or billboards for both the respondent and her husband using an eight-category temporal scale from never to daily. Categories were collapsed into statistically meaningful units, then entered into a factor analysis. The first principal component was taken and used as a measure of media exposure in model comparison analyses; this version of the variable had moderately better predictive power than versions using an additive index or including the second principal component. Although consistent 2 of 5
with predictions a bivariate correlation suggests greater media exposure is associated with lower fertility, in the presence of the other variables (notably woman’s age) higher exposure to media is associated with higher fertility. Cultural transmission variables not used. Additional cultural transmission variables examined but eliminated from final models based on model selection criteria include the following: number of people in the natal or marital bari living abroad; whether the woman, her husband, or her mother or father had visited Dhaka or Matlab bazaar (frequency of visits was also tried); the woman’s frequency of conversations about contraception with husband, all kin, or anyone; whether the woman’s mother used contraception; the woman’s attitude toward contraception; the number of the woman’s and husband’s siblings; the woman’s and husband’s religion (Muslim, Hindu, or other); the distance between the woman’s natal and marital bari (a proxy for the potential influence of her natal kin); and whether the woman’s household participated in a microcredit program.
sample range from 5 to 15 y, the amount of change from the start to end of a woman’s reproductive period is much less than that occurring over a lifetime or over the research period as a whole. This means that individual women’s reproductive spans are much closer to being at equilibrium than the sample is as a whole. Women need to rely on some cues to behavior, and although these may change over time, cues chosen very close to the start of reproduction are less likely to be subject to problems related to temporal change. Finally, most temporally adjusting predictors in our sample show clear directional trends. Although the values for particular years may fluctuate, most annual values will be clearly associated with the directional change as a whole. This reduces the problems with prediction associated with temporally changing variables.
Temporal Change and Equilibrium. Some theoretical models that make predictions about fertility make the simplifying assumption that the system is at equilibrium, a condition that is rarely met in human samples (20). This is true of some risk models (21), some economic models (22, 23), and even of the more formal cultural models (24). The Matlab system is clearly not at equilibrium—in fact the phenomenon of interest is the result of change. Consequently, some variables may not explain as much as others because women cannot use them to make predictions about future environments. This issue has been noted specifically with mortality risk, but it is also true of economic and social change and thus in our case should not systematically affect the interpretation of one set of variables compared with the others. Moreover, although economic models may prevail in our model comparison results, variables from all model classes are represented in the best inclusive model. Our analysis also addresses temporal changes in several ways. First, we include age and age at marriage in all models as controls for temporal differences. Second, values of temporally changing predictors are based on each respondent’s year of marriage, increasing the specificity of such predictors for each individual woman. Third, because the reproductive spans of women in the
Feedback and Interactions Between Model Classes. Several of the strongest predictors in our models are variables that tap into more than one causal pathway. Variables such as a woman’s education, residence in the intervention area, and the location of the respondent’s husband may exhibit a combination of economic and cultural transmission effects. Women’s education is consistently one of the strongest predictors of fertility cross-culturally as well as a key component of both economic and cultural models of fertility decline. Qualitative data from our study suggest that the primary motivations for educating daughters are economic, including providing employment options in case of divorce, improving their position on the marriage market, and helping them better educate their own children. It is also clear, however, that women gain knowledge of low fertility social norms and information on health and child welfare when they are at school. Kaplan (23) argues that although increased investment in human capital may reduce fertility, cultural transmission of low fertility ideals could follow and exaggerate this effect creating a more dramatic fertility decline; a related argument has been made by Boyd and Richerson (24). Our data are consistent with such interpretations. Other variables suggest a potential interaction between risk and cultural transmission factors. An important example is residence in the intervention area, associated with key aspects of cultural transmission, including access to free contraception accompanied by family planning information and training. However, living in the intervention area is also associated with risk reduction, because intensive health-related programs were introduced in this area during the same period and were often communicated through the same individuals. These health programs have changed the cost/ benefit analysis of women living in the area as mortality reductions made them less likely to experience child mortality (25). The effects of these two types of interventions are difficult to disentangle from one another: although they have short-term effects that may counteract each other—contraception reducing fertility and health interventions increasing the number of surviving children—in the longer term both effects have likely decreased fertility through risk and cultural transmission mechanisms.
1. Cleland J, Phillips JF, Amin S, Kamal GM (1994) The Determinants of Reproductive Change in Bangladesh: Success in a Challenging Environment (The World Bank, Washington, DC). 2. Arends-Kuenning M (2001) How do family planning workers’ visits affect women’s contraceptive behavior in Bangladesh? Demography 38(4):481–496. 3. Adnan S (1998) Fertility under absolute poverty: Paradoxical aspects of demographic change in Bangladesh. Econ Polit Wkly 33:1337–1348. 4. Bongaarts J, Watkins SC (1996) Social interactions and contemporary fertility transitions. Popul Dev Rev 22(4):639–682. 5. Kincaid DL (2000) Social networks, ideation, and contraceptive behavior in Bangladesh: A longitudinal analysis. Soc Sci Med 50(2):215–231. 6. Jeffery P, Basu A (1998) Introduction. Appropriating Gender: Women’s Activism and Politicized Religion in South Asia, eds Jeffrey P, Basu A (Routledge, New York). 7. Basu AM, Amin S (2000) Conditioning factors for fertility decline in Bengal: History, language identity, and openness to innovations. Popul Dev Rev 26(4):761–794.
8. Bates LM, Maselko J, Schuler SR (2007) Women’s education and the timing of marriage and childbearing in the next generation: Evidence from rural Bangladesh. Stud Fam Plann 38(2):101–112. 9. Caldwell B, Barkat-e-Khuda (2000) The first generation to control family size: A microstudy of the causes of fertility decline in a rural area of Bangladesh. Stud Fam Plann 31(3):239–251. 10. Caldwell JC (2005) On net intergenerational wealth flows: An update. Popul Dev Rev 31(4):721–740. 11. Caldwell JC, Barkat-E-Khuda, Caldwell B, Pieris I, Caldwell P (1999) The Bangladesh fertility decline: an interpretation. Popul Dev Rev 25(1):67–84. 12. Kabeer N (2001) Ideas, economics and the ‘sociology of supply’: Explanations of fertility decline in Bangladesh. J Dev Stud 38(1):29–70. 13. Razzaque A, Streatfield PK, Evans A (2007) Family size and children’s education in Matlab, Bangladesh. J Biosoc Sci 39(2):245–256. 14. Hossain MB, Phillips JF, Legrand TK (2007) The impact of childhood mortality on fertility in six rural thanas of Bangladesh. Demography 44(4):771–784.
Full Inclusive Model Results. As described in Methods, we used the glmulti package to model each outcome variable against all combinations of the 12 independent variables and control variables. An exhaustive search of more than 16,000 models (drawing all possible subsets of variables) found the models with the lowest AIC values, which we report in Table 4. In Table S3, we summarize the global results, including importance values for all 12 variables. We also report the number of times each variable appeared in one of the 50 models with the lowest AIC values; these models all had an AIC difference of 10 or less compared with the best model. [Models with AIC differences over 10 have essentially no support from the data (19).]
Shenk et al. www.pnas.org/cgi/content/short/1217029110
3 of 5
15. Barkat-e-Khuda, Hossain MB (1996) Fertility decline in Bangladesh: Toward an understanding of major causes. Health Transit Rev 6(Suppl):155–167. 16. van Ginneken J, Razzaque A (2003) Supplyand demand factors in the fertility decline in Matlab, Bangladesh in 1977-1999. Eur J Popul 19:29–45. 17. Low B, Clarke A (1990) Family patterns in nineteenth-century Sweden: Impact of occupational status and landownership. J Fam Hist 16:117–138. 18. Wrigley E, Schofield R (1989) The Population History of England 1541-1871 (Cambridge Univ Press, Cambridge, UK). 19. Burnham KP, Anderson DR (2002) Model Selection and Multi-Model Inference: A Practical Information-Theoretic Approach (Springer, New York). 20. Low B, Hazel A, Parker N, Welch K (2008) Influences on women’s reproductive lives: Unexpected ecological underpinnings. Cross-Cultural Res 42:201–219.
21. Chisholm JS (1999) Death, Hope, and Sex: Steps to an Evolutionary Ecology of Mind and Mortality (Cambridge Univ Press, Cambridge, UK). 22. Becker G (1991) A Treatise on the Family (Harvard Univ Press, Cambridge, MA), 2nd Ed. 23. Kaplan H (1996) A theory of fertility and parental investment in traditional and modern human societies. Yearb Phys Anthropol 101(23):91–135. 24. Boyd R, Richerson PJ (1985) Culture and the Evolutionary Process (Univ of Chicago Press, Chicago). 25. ICDDR,B (2012) Health and Demographic Surveillance System—Matlab. Registration of Health and Demographic Events 2010 (International Center for Diarrheal Disease Research, Bangladesh, Dhaka, Bangladesh).
Fig. S1. Boxplots of total fertility (dark gray bars) and surviving children (light gray bars) by mother’s age (n = 799).
Table S1. Summary statistics for continuous variables (n = 799 women) Variable Dependent variables Total fertility (births) Surviving children (children) Control variables Woman’s age (y) Woman’s age at marriage (y) Risk/mortality variables Number of child deaths in bari Infant mortality rate (per 1,000) Life expectancy at birth (y) Economic/investment variables Household income (Taka per year logged) Woman’s education (y) Cultural transmission variables Woman’s education (y) General fertility rate (births per 1,000) Index of exposure to modern media
Shenk et al. www.pnas.org/cgi/content/short/1217029110
Mean
SD
Median
Range
4.00 3.56
2.04 1.69
4 3
0 to 11 0 to 9
43.67 16.25
11.27 3.05
44 16
21 to 67 11 to 35
+ −
1.56 103.0 52.5
1.87 30.51 4.61
1 112.80 52.29
0 to 15 39.1 to 146.8 43.7 to 62.1
+ + +
11.07
1.05
11.18
6.7 to 14.9
+
3.63
3.69
3.00
0 to 16
−
3.63 152.50 −0.06
3.69 35.44 0.98
3.00 157.60 −0.18
0 to 16 88.7 to 219.2 −1.60 to 3.08
Direction
− + +
4 of 5
Table S2. Summary statistics for categorical variables, n = 799 women Parameter Risk/mortality and cultural transmission variable Woman in intervention area 0 = No 1 = Yes Cultural transmission variable Location of husband 1 = in Matlab 2 = Dhaka or other city 3 = Abroad Economic/investment variables Whether family owns land 0 = No 1 = Yes Family engaged in agriculture 0 = No 1 = Yes Husband primary occupation 0 = Agriculture or day labor 1 = Business or salaried
n
Direction
466 333
Reference −
571 138 90
Reference − −
205 594
Reference +
409 390
Reference +
390 409
Reference −
Table S3. Inclusive models with all independent variables Variable*† Total fertility Child deaths in bari (R) Whether family owns land (E) Woman in intervention area (R, C) Woman’s level of education (E, C) Husband’s occupation (E) Household income (E) Life expectancy at birth‡ (R) Family engaged in agriculture (E) Infant mortality rate‡ (R) Husband’s location (C) General fertility rate (C) Exposure to modern media (C) Surviving children Whether family owns land (E) Family engaged in agriculture (E) Child deaths in bari (R) Woman in intervention area (R, C) Household income (E) Infant mortality rate‡ (R) Woman’s level of education (E, C) General fertility rate (C) Husband’s occupation (E) Exposure to modern media (C) Husband’s location (C) Life expectancy at birth‡ (R)
Sign
Importance
n in top 50
+ + − − − + + + + − + +
1.00 1.00 1.00 0.98 0.94 0.63 0.62 0.51 0.48 0.40 0.28 0.01
50 50 50 49 46 30 31 26 24 20 17 1
+ + + − + + − + − + + +
1.00 1.00 1.00 0.85 0.63 0.52 0.39 0.39 0.25 0.14 0.09 0.05
50 50 50 41 31 27 20 20 13 8 5 3
R refers to a risk/mortality variable, E refers to an economic/investment variable, and C refers to a cultural transmission variable. *Variables are listed in order of importance. † Woman’s age and age at marriage are included as control variables in all models. ‡ Figure given for the year of the woman’s marriage, when childbearing is likely to begin.
Shenk et al. www.pnas.org/cgi/content/short/1217029110
5 of 5
