Cover Image

PAPERBACK
$51.75



View/Hide Left Panel

6
Selection Processes in the Study of Racial and Ethnic Differentials in Adult Health and Mortality

Alberto Palloni and Douglas C. Ewbank


This chapter examines the potential influence exerted by selection processes in the estimation of racial and ethnic differentials in health and mortality. Selection is important because it may influence the direction and magnitude of observed racial and ethnic differentials. In addition, selection processes may exaggerate (attenuate) estimates of effects of membership in a racial or ethnic group that occur due to the existence of intervening mechanisms. We do not undertake this task with the presumption that selection processes are the only or even the most important mechanisms that generate observed racial and ethnic disparities in health and mortality. Instead, we argue that the formulation of sensible inferences and a richer understanding of these disparities require that we consider them explicitly, on an equal explanatory footing, with other possible interpretations. Giving short shrift to or dismissing selection processes on the grounds that they are of trivial importance or because they have been invoked at times in ill-advised applications of social Darwinism only obfuscates the problem. Indeed, some selection processes at least involve mechanisms through which social and economic disparities within racial or ethnic groups are reproduced over time and across generations. To the extent that these mechanisms are empirically relevant for health and mortality, selection processes become an integral part of the production of racial and ethnic disparities. Therefore, they should be treated adroitly instead of being portrayed as a nuisance. We will show that, far from negating the role that material, cultural, social, and behavioral factors have in the production of health and mortality inequalities, explanations that invoke selection mechanisms identify alternative



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life 6 Selection Processes in the Study of Racial and Ethnic Differentials in Adult Health and Mortality Alberto Palloni and Douglas C. Ewbank This chapter examines the potential influence exerted by selection processes in the estimation of racial and ethnic differentials in health and mortality. Selection is important because it may influence the direction and magnitude of observed racial and ethnic differentials. In addition, selection processes may exaggerate (attenuate) estimates of effects of membership in a racial or ethnic group that occur due to the existence of intervening mechanisms. We do not undertake this task with the presumption that selection processes are the only or even the most important mechanisms that generate observed racial and ethnic disparities in health and mortality. Instead, we argue that the formulation of sensible inferences and a richer understanding of these disparities require that we consider them explicitly, on an equal explanatory footing, with other possible interpretations. Giving short shrift to or dismissing selection processes on the grounds that they are of trivial importance or because they have been invoked at times in ill-advised applications of social Darwinism only obfuscates the problem. Indeed, some selection processes at least involve mechanisms through which social and economic disparities within racial or ethnic groups are reproduced over time and across generations. To the extent that these mechanisms are empirically relevant for health and mortality, selection processes become an integral part of the production of racial and ethnic disparities. Therefore, they should be treated adroitly instead of being portrayed as a nuisance. We will show that, far from negating the role that material, cultural, social, and behavioral factors have in the production of health and mortality inequalities, explanations that invoke selection mechanisms identify alternative

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life paths through which these factors may influence health and mortality. Thus, interpretations based on selection arguments can serve to identify social and economic processes that perpetuate social stratification in societies at large as well as within racial and ethnic groups. In this chapter we introduce terminological clarifications and examine some examples of selection processes. We provide a precise definition of selection processes that pertain to health and mortality inequalities, and introduce a simple taxonomy to classify them. We examine strategies for conceptualizing selection processes within the literature on health and mortality. We review a broad array of arguments regarding selection, from those that promote it as a universal cause of all social and economic inequalities in mortality and health, to those that consider it as the intellectual debris of genetic determinism. We examine in some detail three classes of selection processes that are relevant in the area of racial and ethnic health disparities. Using a mixture of simulated and empirical data, we estimate the potential magnitude of their effects and show that, in some cases at least, the impact of selection processes can be quite large—large enough to lead to misinterpretation of observable data and to erroneous policy prescriptions. The chapter ends with a brief discussion of alternative approaches that address conceptual and empirical problems associated with the identification of selection processes. CONCEPTUAL CLARIFICATION: NATURE OF SELECTION PROCESSES Conceptualization and Examples The observed association between an individual’s social class or position and health status and mortality risks can be due to two different processes. The first is one whereby influences on health and mortality result from the action of characteristics intrinsic to the social position. Individuals are endowed with these characteristics only by virtue of having attained the social position. For example, members of higher social classes may experience lower mortality because they command more wealth or have attained higher educational levels, and either of these traits is conducive to better health and lower mortality risks. The second process occurs because individuals have traits or attributes that simultaneously increase their likelihood of accessing (leaving) social positions and exert an influence on their health status and mortality risks. For example, attributes that enhance an individual’s health status during adulthood may also contribute to more advantageous earning profiles and to higher educational attainment. In this case the observed association between social position or class, on the one hand, and health status and

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life mortality risks, on the other, is at least partially the result of the mechanics of accession processes, not the consequence of endowments of the social position or class conferred to individuals when they reach it. The observed association between social class and health or mortality is a consequence of a “health selection process” whereby those sharing a particular social position are disproportionately “selected” from among members of the population who also share particularly low (high) values of the health-relevant traits or attributes. The literature on social class and racial and ethnic health and mortality differentials has conventionally focused on mechanisms of the first type and seeks to quantify the direct effects of social stratification on health and mortality differentials. However, because of the presence of the second mechanism alluded to earlier, inferences regarding the direct effects of membership in certain social positions cannot be based solely on observed correlations. First, identification of selection mechanisms and estimation of their contribution to observed correlations is an important endeavor that leads to more precise estimation of the direct effects of social classes or social positions. In statistical jargon, accounting for selection is necessary to obtain consistent estimates of the effects of social class on health status and mortality. Second, except in the case when traits relevant for both health and social stratification are allocated at random, mechanisms involving health selection are part of the overall process whereby social stratification generates health and mortality inequalities and thus should also be a focus of study for researchers interested in the genesis of health and mortality differentials. We now review examples of selection processes. The “Healthy Worker” Effect The “healthy worker” effect refers to cases when individuals are able to occupy a place in an occupational hierarchy by virtue of their superior health status. For example, a study relating work activities and heart disease (Paffenbarger, Laughlin, and Gima, 1970) showed that the optimal job allocation strategy among stevedores required careful matching of physical demands associated with job duties and individuals’ characteristics. As a result individuals in superior health were more likely to be assigned to the most demanding and risky jobs. This could lead to the paradoxical situation whereby individuals who occupy more demanding, stressful, and risky positions are in better health. In this case, the observed association between, say, a measure of occupational risk or exposure to stress, on the one hand, and health and mortality, on the other, is less than the true association. The fact that incumbents of occupational positions with higher exposure to illnesses and disability are drawn from among the healthiest mem-

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life bers of the population can lead to an attenuation and even reversal of the true association between risk exposure and prevalence of illness or disability. Clearly, one cannot infer from observation that more stressful and physically strenuous jobs are more beneficial for individuals’ health. Although in this example the allocation of positions according to health status takes place through an explicit decision-making process, this rarely will be the case. Invariably it will occur via the operation of mechanisms that are latent, complex, influenced by time lags, and hardly ever explicitly manifested or justified. In another study focusing on the association between mortality and exposure to radiation hazards, it was shown that mortality was lower among employees working within a nuclear facility than among the population living in the surrounding area (Voelz et al., 1978). But this observation cannot be construed to suggest that exposure to higher levels of radiation is immaterial for mortality risks. This is because individuals who worked in the nuclear facility may have been highly selected for characteristics or traits that affect their overall exposure to the risk of cancers and other chronic conditions. Thus, like the first example, possession of traits that enable accession to the occupation also influence their health status, but were part of the individuals’ endowment before accession to the social position. But, unlike the first example, there is an additional mechanism that could create the observed association between occupation and cancer incidence. This mechanism operates via the existence of traits or behaviors acquired after accession to the position that contribute to reducing mortality risks. Behavioral modifications adopted to minimize exposure to alternative carcinogens (smoking) may be the direct consequence of occupancy, part of a conscious deployment of individual behaviors to offset increased exposure to known risks. Healthier profiles that result from behavioral management designed to compensate for increased exposure at work will create an observed association between (lower) mortality risks and health status that is genuinely produced by occupancy of the social position itself. These relations contribute to the observational correlation between social position and mortality, but are not selection effects as defined earlier. Instead, they should be genuinely attributable to the occupancy of the position. The “Healthy Migrant” Effect Migration is an action through which some individuals living in one residential area accede to another area. As other acts of accession to social positions, individual migration requires the possession of individual traits, some of which may be personal (e.g., intelligence, risk tolerance, time pref-

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life erences) and others may involve membership in a group (e.g., social connections, social support etc.). Except for some types of migration such as forced relocation and refugee flows, migration requires decision making that is heavily dependent on the previously mentioned individual attributes. Some, though not all of them, may be connected to health status. Selection through migration goes beyond the fact that migration rates among the disabled, mentally ill, or other population categories with obviously impaired health status are lower than among the rest of the population. Individuals who migrate may be more educated, less risk averse, more aggressive and entrepreneurial, more resilient, with low discount rates of the future, and better prepared to face stressful situations. The net result could be that the distribution of health status in the migrant population at destination will look quite different from the health status distribution of a random sample of the population of origin. A healthy migrant effect leads to the same potential misinterpretation identified previously in the case of the healthy worker effect. A comparison between members of the migrant group and the population at origin will reveal health status disparities. But this cannot be used to infer effects associated with adoption of traits and behaviors at destination or, alternatively, with the act of migration itself. Similarly, comparisons between the migrant group and the population at destination do sometimes reveal unexpectedly low differences. But these should not automatically lead one to infer the importance of cultural advantages (disadvantages) of the migrant over the native group. The healthy migrant effect has been examined in a number of contexts. It is a prime target of epidemiological studies seeking to isolate the effects of environment on health status. One of the earliest and best studied examples was the case of Japanese migrants in California who experienced lower incidence of gastric cancers than the Japanese in Japan (Dunn and Buell, 1966). The most obvious interpretation is that reduced rates of gastric cancers are associated with a newer diet in the place of destination; thus, an environmental effect might be inferred (see also Kasl and Berkman, 1983). But this may overlook the fact that Japanese migrants to California were not a random sample of the Japanese population, either in terms of social class or in terms of region of residence or their own ethnicity. A second study found that Japanese living in Japan as well as those living in Hawaii and Los Angeles displayed serum cholesterol levels that were directly related to the percentage of calories supplied by fats in their diet; this provides added evidence for the environmental hypothesis (Keys et al., 1957). A third study, also among Japanese living in the United States and in Japan (Marmot, Adelstern, and Bulusu, 1994; Marmot and Syme, 1976), shows that those living in the United States (particularly in California) displayed higher rates of coronary heart disease than those living in Japan. Although

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life this finding could be associated with different intensity of exposure to key risks factors such as stress, it also can be attributed to characteristics of the “new position” (e.g., stress of being a migrant in the United States, newer and more deleterious lifestyle). But it could just as easily be explained by the fact that migrants are more likely to be drawn from a population that experiences higher risk of coronary diseases and blood pressure anywhere, regardless of migration status—namely, type A personalities (Graham and Graham-Tomasi, 1985; Rosenman, Friedman, and Strause, 1964). An important finding that tilts the balance toward the environmental interpretation is that coronary heart disease is lower among Japanese migrants who adhere more strongly to Japanese culture, a behavioral strategy that may offset some of the added risks imposed by increased environmental stress (Marmot and Syme, 1976; Marmot et al., 1994). A number of studies of migrants’ health status show the recurrent finding that migrants to an area display lower mortality rates than those in the origin population. This is a distinctive marker of migrant selection, and though it does not prove its existence, it certainly suggests its presence (Kasl and Berkman, 1983; Marmot et al., 1994; Swallen, 1997b). An important exception to this regularity is the classic study of Irish migrants living in Boston and their siblings living in Ireland. This study found no important differences in death rates due to cardiovascular disease across the groups (Trulson et al., 1964). Finally, the “Hispanic paradox” in the United States refers to the fact that Mexican and some non-Mexican Hispanics experience similar or better health status and lower adult mortality rates, and their infants are born at higher weights than African Americans and non-Hispanic whites (Palloni and Morenoff, 2001). These regularities have been attributed to a number of factors, all the product of traits and endowments that migrants may bring with them or acquire during their stay at a destination. These include more favorable behavioral profiles in terms of diet, smoking, and alcohol consumption (Abraido-Lanza et al., 1999; Markides and Coreil, 1986; Sorlie et al., 1993); more cohesive social networks; and superior social support (Frisbie, Cho, and Hummer, 2001). But it is just as likely to be the result of superior health status of migrants that preceded and facilitated the act of migration. Social Stratification and Health Status There are mechanisms other than the “healthy worker” effect that facilitate or impede individuals’ accession to positions in the social stratification system. Even if, for example, earnings and income differentials were largely explained by educational attainment (and they are not), the question remains about the degree to which educational attainment, and more gener-

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life ally, cognitive abilities and other market-related skills commanding higher salaries and wages are influenced by health status experienced by individuals early in their lives. The possibility that health status may play a nontrivial role in the allocation of individuals across the social stratification system is strengthened by findings that suggest that earnings are tightly related to unconventional skills, those that are not part of the bundle of labor inputs in a standard production function (Bowles and Gintis, 2000). Some of these factors are related to health status early in life. Relatively recent work on the effects of early childhood on lifecycle trajectories (see Chapter 5, this volume) suggests new insights into the mechanisms that may link health status and earnings potential. Although some of these relations have been suspected for a long time (Goldberg and Morrison, 1963; Harkey, Miles, and Rushing, 1976; Illsley, 1955), efforts to incorporate them as an integral part of the study of social class mortality and health differentials are of more recent origins (Case, Fertig, and Paxson, 2003; Goldman, 2001; Palloni and Milesi, 2002; Power, Fogelman, and Fox, 1986; Power, Manor, and Fox, 1991; Power and Matthews, 1997; Power, Matthews, and Manor, 1996; Stern, 1983; West, 1991). Because the relevant processes may be spread out over a lifetime, involve long time lags, and are mediated by a number of intervening mechanisms, the resulting effects are referred to in the epidemiological literature with the rather unfortunate label of “indirect selection effects.” This is apparently to distinguish from “direct selection effects,” which are more akin to reverse causality (to be reviewed). Dilution of Mortality Excesses and Closing Ethnic Mortality Gaps: Heterogeneity It has been observed that comparisons of mortality levels between two racial or ethnic groups could lead to different inferences depending on the age interval to which they refer. This occurs when age-specific mortality rates of the groups being compared either converge toward each other or cross over at some point. The explanations for this pattern of differentials are diverse and include the possibility of data artifacts, the presence of mechanisms with age-specific effects, and selection processes. We will review three examples of this phenomenon. Not all of them are pure examples of selection processes, but all were, at one time or another, attributed to selection processes. The first example involves the comparison of black and white mortality rates in the United States. What intrigued most researchers was the fact that mortality rates for blacks converged toward that of whites at older ages (Manton, Poss, and Wing, 1979; Manton and Stallard, 1984). As we know now, most of the convergence is due to a data artifact produced by poor

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life quality of age declaration (overstatement) among blacks both in death certificates and in Censuses or other population registers used to calculate the rates (Coale and Kisker, 1990; Preston et al., 1998). But for some time, the idea that the convergence was real and not the product of data errors dominated the discussion. One explanation proposed at the time that did not receive much support was that factors that determine mortality in each group are not invariant with age, and a few of them that could have been harmful (beneficial) early on become protective (deleterious) later in life. Factors may include lifestyle and affluence. The most influential explanation for the apparent convergence is that, as a result of mortality differentials early in life, the composition by health status of surviving members changes more drastically in one racial group than in the other. Thus, because blacks are exposed to a more severe mortality regime early in life, the survivors to older ages may be healthier, or less frail, than their white counterparts, and their average mortality risk closer to that of whites than earlier in the life course. The underlying mortality differentials between the two races remain unchanged, while the one we observe suggests that differentials change across the age span. A second example of convergence involves mortality risks of Hispanics and non-Hispanic whites in the United States. These tend to converge at older ages even though Hispanic mortality is considerably lower than non-Hispanic white mortality at younger ages (Palloni and Arias, 2003). Here, too, there is the possibility that convergence is an artifact of age misstatement, but, unlike the case of the black-white crossover, the evidence to support this conjecture is not strong. As argued by researchers trying to explain a similar convergence of infant and child health status, the pattern could be an outcome of assimilation and adoption of harmful behavioral profiles among Hispanics or due to the cumulated effects of lower quality health care that are seen as duration of residence in the United States (and, with it, age of incumbents) increases (Morenoff, 2000; Rumbaut and Weeks, 1991; Scribner, 1996). Another possibility we will explore as an example of reverse causality is that at older ages, there is substantial return migration to Mexico by Mexican immigrants with poor health status. Alternatively, it could just as easily be the result of a selection process similar to that invoked to explain the black-white mortality convergence. The only difference in this case is that it is the relatively more severe early mortality regime to which non-Hispanic whites are exposed that could drive the convergence toward the lower mortality rates of that of Hispanics.1 A final example of heterogeneity involves comparisons by risk groups. Recent work on disease and mortality in a regional sample of Hispanics reveals that obesity—a risk factor for diabetes, cardiovascular diseases and circulatory problems, among other chronic conditions—is associated with lower, not higher, odds of mortality at older ages (Markides et al., 2001).

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life Similar findings have been associated with cancer morbidity (Woodbury and Manton, 1977). The robustness of these findings remains to be decided and obvious competing hypotheses—such as the fact that the contrast group includes individuals with very low body mass due to the presence of severe illnesses—need to be eliminated, but these outcomes could be the result of selection processes analogous to those invoked to explain mortality crossover. Individuals who are obese do indeed experience higher mortality risks at some ages. But those who escape higher mortality and survive to older ages may be selected for factors that are protective: Survivors from the pool of individuals with higher mortality risks associated with obesity, cancer, or other co-morbidities are disproportionately drawn from a subpopulation with a more beneficial health profile that confers them protection against higher risks associated both with obesity (or cancer or other morbidity), diabetes, and other conditions. In these examples, the observed differentials between groups (race, ethnic, risk group) shift as individuals age. Except for the case of the black-white crossover, the shift is real and observed, not an illusion. However, the shift cannot be interpreted as a consequence of changes in the differentials across groups (which, in all examples given earlier, may have been fixed). The variability in the magnitude and direction of differentials is an outcome of processes of selective survival that are different across the groups being compared. The commonality in all these examples is this: As in the illustrations of selection posed earlier, there are traits (underlying individual health) that affect the likelihood of individuals’ promotion to positions (surviving to older ages) and that simultaneously influence the mortality risks experienced in those positions (underlying risks at older ages). The key phenomenon common to all these examples is selection on some health-relevant traits among individuals who survive to an older age. This type of selection effect is referred to in the literature as heterogeneity. Reverse Causality, ‘Drift’ or Social Mobility Through Ill Health There are situations in which severe health limitations and impairments constrain individuals affected by them to occupy a much narrower range of occupations or social positions than the general population. The best known examples involve physical limitations such as blindness or psychiatric conditions such as schizophrenia, both of which severely limit possible jobs, occupations, and social positions. But the relatively worse health status of these individuals is a cause rather than a consequence of their social positions. A subtler and more commonly found class of reverse causation includes situations in which deterioration of health status directly leads to erosion of an individual’s social and economic positions. Individuals make decisions

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life about labor supply that partially depend on their current or anticipated health status. Thus, those who retire and leave the labor market due to health reasons will forego income and may endure fast erosion of savings to keep desired standards of living. Disinvestments and asset dilution are phenomena that can occur as a result of health deterioration. Similarly, large out-of-pocket expenses to defray caring costs of chronic illnesses and disability may be more or less common depending on illness, disability, and type of insurance coverage. A third example of reverse causality appears in the analysis of health and mortality differentials involving migrants when return migration flows are possible and of some demographic relevance. This is because immigrants who return to the place of origin may be disproportionately drawn from among those who are affected by ill health or disability. As in the cases described previously, the change of social status (from immigrant to outmigrant) is a direct consequence of actual or anticipated health status. Admittedly this type of reverse causation is relevant for only some migrant groups, not all. The common feature in all three examples of reverse causality described previously is that the observed total correlation between social position and health and mortality will be partially influenced by a subset of individuals who occupy positions as a result of their preexisting health status. Thus, it may be that the overall negative association between, say, occupational prestige and mortality is attenuated once we take into account the fact that individuals with physical or psychiatric impairments can only occupy low-prestige occupations by virtue of their impairments. Or, the observed strong and negative association between wealth and mortality will diminish considerably once we account for the fact that chronic illnesses result in wealth dilution. Or, finally, the observed advantageous health status of migrant relative to native population at older ages may disappear altogether once we account for return migration of the most frail among return migrants. In all these examples, the observed association between health and social position is a result of the direct effect of health status on social mobility. Unlike the healthy migrant or healthy worker effect, the observed association is not a product of traits that simultaneously influence both health status and social mobility. Like the healthy migrant and healthy worker effect, the observable association does not reflect the influence of characteristics or traits of social positions on the health status or mortality risks of individuals. In the epidemiological literature, these relations and their observational features are referred to as “direct selection effects” or “drift” (West, 1991) and have been treated mostly by enhancing the observational plan (Fox, Goldblatt, and Jones, 1985). In the social sciences and economic literature, these relations are referred to as reverse causality or endogenous effects and have been treated with a combination of more powerful and well-grounded

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life theoretical models and better observational plans. Thus, work on savings motives, on the dynamic of wealth accumulation over the life course, and on their relation to health status is becoming better integrated in economists’ lifecycle models (Lilliard and Weiss, 1996; Smith, 1999). Recent empirical work confirms that there are important effects of individuals’ antecedent health status on subsequent wealth accumulation or dilution. The relation operates through a variety of intervening mechanisms, including, but not limited to, saving decisions (Adams et al., 2003; Smith, 1999). Selection Effects and Their Relevance for Assessment of Racial and Ethnic Disparities Taxonomy In the examples described previously there is a fundamental distinction between effects associated with characteristics of a social position (or age) and those associated with individual traits that influence both individuals’ health status and subsequent mortality risks and his or her ability or potential to occupy the position. These are all processes that can be represented with a causal diagram such as the one in Figure 6-1a. Although in the more FIGURE 6-1a Relations involved in health selection effects.

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life or intensification of a given social class gradient (on health or mortality). Thus, for example, it is believed that if health selection into a social class is an important contributor to the relation, the social class gradient would then attenuate as individuals became older. As we indicated before, this type of pattern will occur even in the absence of health selection effects. All it takes is additional selection through survival. This problem is ubiquitous: In most cases, a particular outcome for the gradient will be consistent with multiple interpretations. Unless we have an explicit identification procedure to select from among alternative interpretations, we should not delude ourselves into believing that we produce evidence for or against a particular hypothesis. This is corroborated by simulation models. For example, Goldman (1994) has shown that the range of patterns for mortality gradients generated by different health selection processes is much broader than what researchers have speculated. Although Goldman’s simulations were tailored to deal with the problem of marriage selection, her results are equally valid for social class gradients. This suggests what not to do: Unless we have a clear model about what selection processes are at work and how they are operating, we should refrain from interpreting simple patterns of social class gradients as evidence for or against them. Addressing Reverse Causality Far from being insoluble, the two examples of reverse causality discussed in this chapter are quite tractable. In all cases tractability requires satisfying two conditions. The first is that there should be an explicit model reflecting expected theoretical relations. The second is that there should be adequate data to test the model. A good example of these two conditions is available in the work by Adams et al. (2003) as well as in the brief analyses presented by Smith (1999). In both cases the researchers impose structure on the data from theoretical expectations about how reverse causality can work, for example, by identifying the mechanisms that make it happen. Then they use the data to test the presence of such mechanisms. One may disagree with the cogency of these models or quarrel with a few or all of the assumptions made. But because these are in the open, one will not be led astray. The only problem with these models is that they require panel data, and are simply not identifiable through cross-sectional information. The example of return migration among Hispanics suggests alternative methodologies to identify reverse causation. The test described in this chapter is just one among many other candidates. One of them is to retrieve information on the health status and mortality patterns of Mexicans who recently returned to Mexico from the United States, then compare these

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life patterns with those of migrants who stay in the United States. The test demands a binational data set, one that is more expensive and complex than the more commonly available data on migrants and migration to the United States. Addressing Health Selection In some cases microsimulation models can be of great help by enabling us to test for the existence of effects and to verify relations expected from theory. Seldom, however, will they be useful for more than establishing boundary conditions and ranges of values for parameter estimates. This is exactly how these models were used when dealing with the Hispanic paradox (Palloni and Morenoff, 2001), or with the issue of health selection and mortality differentials by marital status (Goldman, 2001). In these two cases, microsimulation models were used to estimate the magnitude of uncertainty associated with a process with a given set of observable relations. This is a first and, in some cases, the only, line of attack when health selection effects are suspected. A second strategy is to directly model the health selection process of interest, that is, to explicitly represent it as a set of relations between measurable quantities. Although this is possible, the strategy is rarely a feasible one. The reason is that these models are plagued by identification problems that make estimation of relevant parameters from observable relations a difficult endeavor, and that frequently require assumptions that stifle interpretations. The identification problem could be minimized if the information available is appropriate. The data demands to deal with health selection effects addressed in this chapter are high because the researcher requires information over the life course of individuals. Although a growing number of research projects are being designed to collect and make available data for birth cohorts, the required information continues to be a scarce commodity. But one should not necessarily be confined to waiting for a new cohort study to begin or for those already in the field to reach a more mature stage. As argued by Ewbank (1998), it is possible to estimate parameters of complex models employing different and independent data sets, each of them informing a different aspect or set of relations among the total set of relations the researcher would like to represent. For example, in a recent review document (Palloni and Milesi, 2002), different data sets are used to estimate the amount of variance in earnings that could be explained by early child health status. The procedures are not yet completely worked out and many difficulties remain to be resolved, but they are promising and worth exploring further.

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life Addressing Unmeasured Heterogeneity There are two known characteristics of models for heterogeneity or selection via survival. The first is that estimates of parameters are somewhat sensitive to the specification of distributional properties of the unmeasured traits. Recall that selection through survival is attributable to individual traits that increase (decrease) chances of surviving from one point in time to the next and, simultaneously, affect subsequent health and mortality. In the examples examined in this chapter we employed the assumption that unmeasured traits or attributes could be represented by a gamma-distributed random variable. The problem is that the choice of distribution is not as mundane as we made it out to be. Indeed, inferences about the process we are studying may not always be very robust to violation of assumptions about the distributions of the trait causing heterogeneity in the population (Heckman and Singer, 1982; Trussell and Richards, 1985; Vaupel and Yashin, 1985). Although lack of robustness may be limiting in some problems and in data sets, not all is lost. In some cases it is possible to establish with some degree of confidence the type of distribution that is more appropriate for the data, and even venture a range of values for the associated parameters of such distributions. An interesting example in which this is convincingly demonstrated is discussed by Ewbank and Jones (2001) in their research on the APOE-4 gene. The second, and perhaps more devastating, property of these models is that even if one were to adopt a conservative stance and impose no parametric form on the unmeasured trait (Heckman and Singer, 1982), thus escaping the first problem, estimates could still be sensitive to the form of the underlying risk (Trussell and Richards, 1985). This would not be an important issue if we knew well the age pattern of progression of events at older ages. But, contrary to what most demographers thought, this is not even true for the case of mortality, let alone for the study of other, much lesser understood, phenomena such as disability or morbidity associated with chronic illness. An important limitation of both parametric and nonparametric approaches to unmeasured heterogeneity is that they rest the questionable assumption that factors or traits that account for heterogeneity are fixed over time and may not change regardless of what health events an individual may experience throughout his or her life. Needless to say, this is a somewhat unsatisfactory assumption. Recent progress on parametric models suggests that there are ways of relaxing the assumption of invariance, although more experimentation is needed to understand the properties, advantages, and shortcomings of these models (Weitz and Fraser, 2001). In summary, a model-based approach combined with conventional longitudinal data will not solve the problem of inferences in the presence of

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life selection via survival. However, a combination of data expansion and model formulation offers potential advantages. Data Expansion Conventional information on events of interest, such as timing and occurrence of deaths or disability, refer to individuals and, at best, to the household they belong or to the larger community where they reside. The fact that some traits responsible for heterogeneity are frequently shared by related individuals makes it possible to at least neutralize their effects. This requires information on survival (or disability or illness), not just for the individuals who are targets of investigation, but for those related to them who may share some of the unmeasured traits. For example, information on spouses can be used to attenuate the effects of unmeasured conditions shared by spouses. Information on siblings can be used to neutralize the effects of traits shared by siblings who grew up together or apart, including some shared genetically related frailty. Information on twins has been conventionally used to draw inferences about environmental effects net of genetic endowments. The methods for estimating parameters with these kinds of expanded data sets exist and have been tried many times, some with more success than others. An interesting illustration of how estimates of unmeasured heterogeneity derived from twin studies can be applied to solve problems posed by heterogeneity in a completely different setting is discussed by Ewbank (2000) and Caselli and colleagues (2000). Another example is the estimation of unmeasured heterogeneity using spouse data and its application to the assessment of mortality differentials by social class (Mare and Palloni, 1986). Model-Based Approaches Ewbank and Jones (2001) suggest a number of model-based and data expansion approaches to measure individual variance or heterogeneity of mortality risks. One of them is tantamount to data reduction and yields estimates of the distribution of individual mortality risks attributable to factors known to affect mortality included as measured characteristics in the data set. Estimates of the variability under model specifications that differ in terms of the covariates included can produce information on how the distribution of unmeasured traits behaves. This information could conceivably be used in other data sets when some of the covariates known to affect mortality are not available.

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life An Important Recommendation Investing efforts to identify selection processes in the study of racial, ethnic, and social-class mortality and health disparities is not tantamount to denying the importance of race or social class. As we argued before, if they are relevant at all, these processes may reveal rather than obscure a number of mechanisms that perpetuate racial, ethnic, and social-class inequalities. The first condition for identification is to recognize that some selection processes may be at work and that they could be relevant for the problem being investigated. The second condition is to explicitly include a description of the relations implied by the selection process. The third and final condition is to estimate, reveal, and discuss pertinent measures of uncertainty. We do not refer to conventional assessment of variability of estimates given the appropriateness of a model, but to measures that reflect the range of possible estimates when competing models are equally plausible. ENDNOTES 1.   Duration effects that are invariant to controls for age at entry into the United States are not as easy to interpret as a partial product of selection effects and are probably attributable to health differences across migrant cohorts. 2.   For a review of pertinent studies, see Palloni and Milesi (2002). 3.   Although the term “unmeasured heterogeneity” mostly refers to issues of selection through survival, it often has been used to refer to the so-called mover-stayer problem, or “state-dependent heterogeneity.” This is a situation in which a subset of individuals under observation can never experience the event of interest or their risk of experiencing it is low. To simplify terminology and unless otherwise stated, we use the term “unmeasured heterogeneity” to refer to selection processes through survival only. However, research that requires the formulation of multistate hazard models must face up to both “heterogeneity” problems. 4.   This section of the chapter is a summary of findings reported by Palloni and Morenoff (2001). 5.   Our initial intention was to study the mortality experience at ages above 40. Because individuals who are younger than 40 at baseline will contribute variable amounts of exposure at ages 40 and above during the follow-up period, we opted for a compromise solution and included individuals who at baseline were aged 35 and older. 6.   In particular, one can test for differences in the slope of the hazard rates. Slope differences are a natural way in which gradual shifts in the composition of population by health status are manifested in mortality data. 7.   The literature on heterogeneity suggests that a Gamma-distributed frailty captures well the effects on unmeasured traits that influence mortality (Ewbank, 2000; Ewbank and Jones, 2001; Vaupel et al., 1979; Yashin et al., 1999) but other specifications are equally plausible. 8.   The absence of evidence for unmeasured heterogeneity does not mean there is none. It simply tells us that within the age range we are examining, it is immaterial. Its presence may be felt more heavily at older ages. 9.   We will evaluate expressions above age 40 and ignore the effects of mortality dynamics prior to this age.

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life 10.   It should be noted that, as a rule, the examination of period instead of cohort life tables is likely to underplay the role of selection via heterogeneity in frailty. 11.   Arguments analogous to these can be invoked to justify attention to health selection via reverse causality, or processes whereby health status directly impacts the risk of social mobility. 12.   This data set is not ideal for our purposes. As mentioned already, it has the important shortcoming of only including men who were in the labor force at the outset of the study. The effect will be to underestimate the impact of reverse causality. REFERENCES Abraido-Lanza, A.F., Dohrenwend, B.P., Ng-Mak, D.S., and Turner, J.B. (1999). The Latino mortality paradox: A test of the “salmon bias” and health migrant hypotheses. American Journal of Public Health, 89(10), 1543-1548. Adams, P., et al. (2002, October 10-11). A re-examination of the Hispanic mortality paradox. Paper presented at the Center for Demography and Ecology 40th Anniversary Symposium, Madison, WI. Adams, P., Hurd, M.D., McFadden, D., Merrill, A., and Ribeiro, T. (2003). Healthy, wealthy, and wise? Tests for direct causal paths between health and socioeconomic status. Journal of Econometrics, 112, 3-56. Barker, D.J.P. (1991). The fetal and infant origins of inequalities in health in Britain. Journal of Public Health Medicine, 13(2), 64-68. Ben-Shlomo, Y., and Kuh, D. (2002). A life course approach to chronic disease epidemiology: Conceptual models, empirical challenges and interdisciplinary perspectives. International Journal of Epidemiology, 31, 285-293. Blane, D., Smith, G.D., and Bartley, M. (1993). Social selection: What does it contribute to social class differences in health? Sociology of Health and Illness, 15(1), 1-15. Bowles, S. and Gintis, H. (2000). The inheritance of economic status: Education, class, and genetics . Working paper no. 01-01-005, University of Massachusetts, Department of Economics. Bowles, S., Gintis, H., and Osborne, M. (2000). The determinants of earnings: Skills, preferences, and schooling. University of Massachusetts, Department of Economics. Unpublished manuscript. Case, A., Lubotsky, D., and Paxson, C. (2002). Socioeconomic status and health in childhood: Origins of the gradient. American Economic Review, 92(5), 1308-1334. Case, A., Fertig, A., and Paxson, C. (2003). From cradle to grave? The lasting impact of childhood education and circumstance. Center for Health and Wellbeing, Princeton University. Unpublished manuscript. Caselli, G., Vaupel, J., and Yashin, A. (2000). Longevity, heterogeneity and selection. Atti della XL Riunione Scientifica della Società Italiana di Statistica, 49-72. Coale, A.J., and Kisker, E.E. (1990). Defects in data on old age mortality in the United States: New procedures for calculating mortality schedules and life tables at the highest ages. Asian and Pacific Population Forum, 4(1), 1-31. Dunn, J.E., Jr., and Buell, P.E. (1966, September 19). Gastro-intestinal cancer among the ethnic groups in California. Paper presented at the meeting of Third World Congress of Gastroenterology, Tokyo, Japan. Ewbank, D. (1998). APOE and the risks of Alzheimer’s disease and ischemic heart disease: A demographic meta-analytic model. Philadelphia: University of Pennsylvania Population Studies Center. Ewbank, D.C. (2000). Mortality among the least frail: Lessons from research on the APOE gene. Philadelphia: University of Pennsylvania Population Studies Center.

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life Ewbank, D.C., and Jones, N. (2001, March 29). Preliminary observations on the variations in the risk of death. Paper presented at the meeting of the Population Association of American, Washington, DC. Forsdahl, A. (1978). Living conditions in childhood and subsequent development of risk factors for arteriosclerotic heart disease. Journal of Epidemiology and Community Health, 32, 34-37. Fox, A.J., Goldblatt, P.O., and Jones, D.R. (1985). Social class mortality differentials: Artifact, selection or life circumstances? Journal of Epidemiology and Community Health, 39, 1-8. Frisbie, W.P., Cho, Y., and Hummer, R.A. (2001). Immigration and the health of Asian and Pacific Islander adults in the United States. American Journal of Epidemiology, 153(4), 372-380. Goldberg, E.M., and Morrison, S.L. (1963). Schizophrenia and social class. British Journal of Psychiatry, 109, 785-791. Goldman, N. (1994). Social factors and health: The causation-selection issue revisited. Proceedings of the National Academy of Sciences, 91(February), 1251-1255. Goldman, N. (2001). Social inequalities in health: Disentangling the underlying mechanisms. In M. Weinstein, A.I. Hermalin, and M.A. Stoto (Eds.), Population health and aging: Strengthening the dialogue between epidemiology and demography. New York: New York Academy of Sciences. Graham, S., and Graham-Tomasi, R. (1985). Reviews and commentary: Achieved status as a risk factor in epidemiology. Journal of Epidemiology (Formerly American Journal of Hygiene), 122(4), 553-558. Grossman, M. (1972a). On the concept of health capital and the demand for health. Journal of Political Economy, 80(2), 223-255. Grossman, M. (1972b). The demand for health—a theoretical and empirical investigation. New York: National Bureau of Economic Research. Guend, A., Swallen, K.C., and Kindig, D. (2002). Exploring the racial/ethnic gap in healthy life expectancy: United States, 1989-1991 (Center for Demography and Ecology Working Paper 2002-02). Madison: University of Wisconsin Press. Harkey, J., Miles, D.L., and Rushing, W.A. (1976). The relation between social class and functional status: A new look at the Drift Hypothesis. Journal of Health and Social Behavior , 17, 194-204. Harrison, R.M., and Taylor, D.C. (1976). Childhood seizures: A 25 year follow-up. Lancet, 1, 948-957. Heckman, J., and Singer, B. (1982). Population heterogeneity in demographic models. In K. Land and A. Rogers (Eds.), Multidimensional mathematical demography. New York: Academic Press. Hurd, M. (1987). Savings of the elderly and desired bequests. American Economic Review, 77, 298-312. Hurd, M., and Wise, D. (1989). Wealth depletion and life-cycle consumption by the elderly. In D. Wise (Ed.), Topics in the economics of aging. Chicago: University of Chicago Press. Illsley, R. (1955). Social class selection and class differences in relation to stillbirths and infant death. British Medical Journal, 2, 1520-1526. Illsley, R. (1986). Occupational class, selection and the production of inequalities in health. The Quarterly Journal of Social Affairs, 2(2), 151-165. Kasl, S.V., and Berkman, L. (1983). Health consequences of the experiences of migration. Annual Review of Public Health, 4, 69-90. Kestenbaum, B. (1986). Mortality by nativity. Demography, 23(1), 87-90. Keys, A. (1957, April 10). Lessons from serum cholesterol studies in Japan, Hawaii and Los Angeles. Paper presented at the Symposium on the Pathogenesis of Coronary Heart Disease, Boston.

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life Kuh, D., and Wadsworth, M. (1989). Parental height: Childhood environment and subsequent adult height in a national birth cohort. International Journal of Epidemiology, 18, 663. Lilliard, L., and Weiss, Y. (1996). Uncertain health and survival: Effect on end-of-life consumption. Journal of Business and Economic Statistics, 15(2), 254-268. Lundberg, O. (1986). Class and health: Comparing Britain and Sweden. Social Science and Medicine, 23, 511-517. Lundberg, O. (1991). Childhood living conditions, health status, and social mobility: A contribution to the health selection debate. European Sociological Review, 7(2), 149-162. Manton, K., and Stallard, E. (1984). Recent trends in mortality analysis. New York: Academic Press. Manton, K.G., and Woodbury, M.A. (1983). A mathematical model of the physiological dynamics of aging and correlated mortality selection. Application to the Duke Longitudinal Study. Journal of Gerontology, 38(4), 406-413. Manton, K.G., Poss, S.S., and Wing, S. (1979). The black/white mortality crossover: Investigation from the perspectives of the components of aging. Gerontologist, 19, 291-299. Manton, K.G., Stallard, E., and Vaupel, J.W. (1981). Methods for comparing the mortality experience of heterogeneous populations. Demography, 18(3), 389-410. Mare, R.D., and Palloni, A. (1986). Selection bias and program assessment in the job training Longitudinal Survey, Part II: Design for modeling and statistical analysis. Paper prepared for U.S. Bureau of the Census, Survey Methods Division. Markides, K.S., and Coreil, J. (1986). The health of Hispanics in the southwestern United States: An epidemiologic paradox. Public Health Reports, 101(3), 253-265. Markides, K.S., et al. (1997). Health status of Hispanic elderly. In L.G. Martin and B.J. Soldo (Eds.), Racial and ethnic differences in the health of older Americans (pp. 285-300). Committee on Population, Commission on Behavioral and Social Sciences and Education, National Research Council. Washington, DC: National Academy Press. Markides, K.S., Black, S.A., Ostir, G.V., Angel, R.J., Guralnik, J.M., and Lichtenstein, M. (2001). Lower body function and mortality in Mexican American elderly people. The Journals of Gerontology: Series A: Biological Sciences and Medical Sciences, 56(4), M243-M247. Marmot, M.G., and Syme, L.S. (1976). Acculturation and coronary heart disease in Japanese-Americans. American Journal of Epidemiology, 104(3), 225-247. Marmot, M.G., Adelstein, A.M., and Bulusu, L. (1994). Lessons from the study of immigrant mortality. Lancet, 1, 1455-1458. Meadows, S.H. (1961). Social class migration and chronic bronchitis. British Journal of Preventative Social Medicine, 15, 171-176. Morenoff, J.D. (2000). Unraveling paradoxes of public health: Neighborhood environments and racial/ethnic differences in birth outcomes (Ph.D. dissertation). Chicago: University of Chicago Press. Mulatu, M.S., and Schooler, C. (2002). Causal connections between socio-economic status and health: Reciprocal effects and mediating mechanisms. Journal of Health and Social Behavior, 43, 22-41. Nam, C.B., and Okay, A. (1977). Factors contributing to the mortality crossover pattern. Paper presented at the XVII General Conference of the International Union for the Scientific Study of Population, Mexico City. National Center for Health Statistics (NCHS). (2000). Healthy people 2000. Hyattsville, MD: Author. Nystrom Peck, A.M. (1992). Childhood environment, intergenerational mobility and adult health: Evidence from Swedish data. Journal of Epidemiology and Community Health, 46, 71-74.

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life Nystrom Peck, A.M., and Vagero, D.H. (1987). Adult body height and childhood socioeconomic group in the Swedish population. Journal of Epidemiology and Community Health, 41. Nystrom Peck, M., and Lundberg, O. (1995). Short stature as an effect of economic and social conditions in childhood. Social Science and Medicine, 41(5), 733-738. Paffenbarger, R.S., Jr., Laughlin, M.E., and Gima, A.S. (1970). Work activity of longshoremen as related to death from coronary heart disease and stroke. New England Journal of Medicine, 282, 1190-1214. Palloni, A., and Arias, E. (2003). The Hispanic paradox of adult mortality revisited. (Center for Demography Working Paper 2003-01). Madison: University of Wisconsin Press. Palloni, A., and Milesi, C. (2002, October 24-27). Social classes, inequalities and health disparities: The intervening role of early health status. Ethnic variations in intergenerational continuities and discontinuities in psychosocial features and disorders. Paper presented at the Jacobs Conference, Zurich, Switzerland. Palloni, A., and Morenoff, J.D. (2001). Interpreting the paradoxical in the Hispanic paradox: Demographic and epidemiologic approaches. In M. Weinstein, A.I. Hermalin, and M.A. Stoto (Eds.), Population health and aging: Strengthening the dialogue between epidemiology and demography. Demography and Epidemiology: Frontiers in Population Health and Aging. New York: New York Academy of Sciences. Perrott, G. St. J., and Collins, S.D. (1935). Relation of sickness to income and income change in 10 surveyed communities. Health and Depression Studies No. 1. Public Health Reports, 50, 595-622. Persico, N., Postlewaite, A., and Silverman, D. (2001). The effect of adolescent experience on labor market outcomes: The case of height. (PIER working paper). Philadelphia: Department of Economics, University of Pennsylvania. Power, C., and Matthews, S. (1997). Origins of health inequalities in a national population sample. Lancet, 350, 1584-1589. Power, C., Fogelman, K., and Fox, A.J. (1986). Health and social mobility during the early years of life. Quarterly Journal of Social Affairs, 2(4), 397-413. Power, C., Manor, O., and Fox, A.J. (1990). Health in childhood and social inequalities in health in young adults. Journal of the Royal Statistical Society, Series A, 153(1), 17-28. Power, C., Manor, O., and Fox, A.J. (1991). Health and class: The early years. London: Chapman and Hall. Power, C., Matthews, S., and Manor, O. (1996). Inequalities in self rated health in the 1958 birth cohort: Lifetime social circumstances or social mobility. British Medical Journal, 313, 449-453. Power, C., Matthews, S., and Manor, O. (1998). Inequalities in self-rated health: Explanations from different stages of life. Lancet, 351, 1009-1014. Preston, S.H., Elo, I.T., Foster, A., and Fu, H. (1998). Reconstructing the size of the African-American population by age and sex: 1930-1990. Demography, 35, 1-21. Preston, S.H., Elo, I.T., and Stewart, Q. (1999). Effects of age misreporting on mortality estimates at older ages. Population Studies, 53, 165-177. Rogot, E. (1992). A study of 1.3 million persons by demographic, social and economic factors: 1979-1985 follow-up (National Institutes of Health Pub. No. 92-3297). Bethesda, MD: National Institutes of Health. Rosenman, R.H., Friedman, M., and Strause, R. (1964). A predictive study of coronary heart disease: The Western Collaborative Group Study. Journal of the American Medical Association, 189, 15-22. Rosenwaike, I. (1987). Mortality differentials among persons born in Cuba, Mexico and Puerto Rico residing in the United States, 1979-1981. American Journal of Public Health, 77(5), 603-606.

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life Rosenwaike, I. (1991). Mortality of Hispanic populations. New York: Greenwood Press. Rumbaut, R.G., and Weeks, J.R. (1991). Perinatal risks and outcomes among low-income immigrants. Final Report for the Maternal and Child Health Research Program. Rockville, MD: U.S. Department of Health and Human Services. Scribner, R.A. (1996). Paradox as paradigm—the health outcomes of Mexican Americans. American Journal of Public Health, 86(3), 303-305. Smith, J. (1999). Healthy bodies and thick wallets: The dual relation between health and economic status. Journal of Economic Perspectives, 13(2), 145-166. Smith, J.P., and Kington, R.S. (1997a). Demographic and economic correlates of health in old age. Demography, 34(1), 159-170. Smith, J.P., and Kington, R.S. (1997b). Race, socioeconomic status, and health in late life. In L.G. Martin and B.J. Soldo (Eds.), Racial and ethnic differences in the health of older Americans (pp. 105-162). Committee on Population, Commission on Behavioral and Social Sciences and Education, National Research Council. Washington, DC: National Academy Press. Sorlie, P.D., Backlund, M.S., Johnson, N.J., and Rogat, F. (1993). Mortality by Hispanic status in the United States. Journal of the American Medical Association, 270(20), 2464-2468. Stern, J. (1983). Social mobility and the interpretation of social class mortality differentials. Journal of Social Policy, 12(1), 27-49. Stevenson, T.H.C. (1923). The social distribution of mortality from different causes in England and Wales. Biometrika, 15, 382-400. Strehler, B.L. (1977). Time, cells, and aging. New York: Academic Press. Swallen, K.C. (1997a, March). Cross-national comparisons of mortality differentials: Immigrants to the US and stayers in common countries of origin. Paper presented at the meeting of the Population Association of America. Washington, DC . Swallen, K.C. (1997b). Do health selection effects last? A comparison of morbidity rates for elderly adult immigrants and US-born elderly persons. Journal of Cross-Cultural Gerontology, 12(4), 317-339. Townsend, P., and Davidson, N. (1982). Inequalities in health: The black report. London: Pelican. Trulson, M.F., Clancy, R.E., Jessop, W.J.E., Childers, R.W., and Stare, F.J. (1964). Comparisons of siblings in Boston and Ireland. Journal of the American Dietetic Association, 4, 225-229. Trussell, J., and Richards, T. (1985). Correcting for unmeasured heterogeneity in hazard models using the Heckman-Singer procedure. Sociological Methodology, 15, 248-276. Trussell, J., et al. (1985). Determinants of birth-interval length in the Philippines, Malaysia, and Indonesia: A hazard-model analysis. Demography, 22, 145-168. Vaupel, J., Manton, K., and Stallard, E. (1979). The impact of heterogeneity in individual frailty on the dynamics of mortality. Demography, 16(3), 439-454. Vaupel, J.W., and Yashin, A.I. (1985). Heterogeneity’s ruses: Some surprising effects of selection on population dynamics. The American Statistician, 39(3), 176-185. Voelz, G.L., et al. (1978, March 13-17). International symposium of the late biological effects of ionizing radiation. International Atomic Energy Agency, Vienna. {Paper presented]. Wadsworth, M.E.J. (1986). Serious illness in childhood and its association with later-life achievement. In R.G. Wilkinson (Ed.), Class and health (pp. 50-74). London: Tavistock Institute. Wadsworth, M. E. J. (1991). The imprint of time: Childhood, history and adult life. Oxford, England: Clarendon Press.

OCR for page 171
Critical Perspectives on Racial and Ethnic Differences in Health in Late Life Wadsworth, M.E.J., and Kuh, D.J.L. (1997). Childhood influences on adult health: A review of recent work from the British 1946 National Birth Cohort Study, the MRC National Survey of Health and Development. Paediatric and Perinatal Epidemiology, 11, 2-20. Weitz, J.S., and Fraser, H.B. (2001). Explaining mortality rate plateaus. Proceedings of the National Academy of Sciences, 98(26), 15383-15386. West, P. (1991). Rethinking the health selection explanation for health inequalities. Social Science and Medicine, 32(4), 373-384. Wilkinson, R.G. (1986). Socioeconomic differentials in mortality: Interpreting the data on size and trends . In R.G. Wilkinson (Ed.), Class and health: Research and longitudinal data (pp. 1-20). London: Tavistock. Woodbury, M.A., and Manton, K.G. (1977). A random walk model of human mortality and aging. Theoretical Population, Biology, 11, 37-48. Yashin, A.I., et al. (1999). Genes, demography, and life span: The contribution of demographic data in genetic studies on aging and longevity. American Journal of Human Genetics, 65, 1178-1193.