How indicators of socioeconomic status relate to physical functioning of older adults in three Asian societies

In developed countries, socioeconomic status has been identified as one of the most important demographic and social determinants of older adult health. The relationship has not been well studied or contrasted across much of the developing world. Yet, with population aging occurring rapidly in much of Asia, understanding the factors that distinguish between those in better and worse health becomes important. To this end, the current study has two main aims. It first examines the degree to which two measures commonly used to indicate socioeconomic status, education and income, relate to the physical functioning of older adults in three Asian societies-Taiwan, Thailand, and the Philippines. These three societies are all experiencing population aging, although to varying degrees. They are also characterized by differences in levels of national economic development, a factor that may influence the extent to which health care is available for those in varying socioeconomic positions. Second, the study explores the degree to which these associations are consistent across the three settings. The study results in both expected and unexpected findings. Socioeconomic status indicators are linked to health, but inconsistently across the three settings. Associations are strongest in Taiwan, weaker in Thailand, and almost nonexistent in the Philippines. This leads to questions regarding the universality of the relationship. The paper concludes with a discussion of the possible reasons that socioeconomic status may not influence the health of older adults in consistent ways across different societies, including the effect that economic development can have on health outcomes.

How Indicators of Socioeconomic Status Relate to Adults in Three Asian Societies Zachary Zi] Napaporn Chayovan Josefina Natividad 2003 No. 172 Population CouncilHow Indicators of Socioeconomic Status Relate to Physical Functioning of Older Adults in Three Asian Societies Zachary Zimmer Napapom Chayovan Hui-Sheng Lin Josefina Natividad Support for this research has been provided by the National Institutes of Health-Na­tional Institute on Aging, grant no. 1 R01 AG20063-01. The authors thank Albert Hermalin for comments on an earlier version. Zachary Zimmer is Research Associate, Population Council, New York. Napaporn Chayovan is Professor, Center for Population Studies, Chulalongkorn University, Bangkok, Thailand. Hui-Sheng Lin is Director, Center for Population Research, Taiwan Bureau of Health Promotion, Taichung, Taiwan. Josefina Natividad is Professor, De­partment of Sociology, University of the Philippines, Quezon City, Philippines.Abstract In developed countries, socioeconomic status has been identified as one of the most important demographic and social determinants of older adult health. The relationship has not been well studied or contrasted across much of the developing world. Yet, with population aging occurring rapidly in much of Asia, understanding the factors that distinguish between those in better and worse health becomes impor­tant. To this end, the current study has two main aims. It first examines the degree to which two measures commonly used to indicate socioeconomic status, education and income, relate to the physical functioning of older adults in three Asian societ- ies-Taiwan, Thailand, and the Philippines. These three societies are all experienc­ing population aging, although to varying degrees. They are also characterized by differences in levels of national economic development, a factor that may influence the extent to which health care is available for those in varying socioeconomic posi­tions. Second, the study explores the degree to which these associations are consis­tent across the three settings. The study results in both expected and unexpected findings. Socioeconomic status indicators are linked to health, but inconsistently across the three settings. Associations are strongest in Taiwan, weaker in Thailand, and almost nonexistent in the Philippines. This leads to questions regarding the uni­versality of the relationship. The paper concludes with a discussion of the possible reasons that socioeconomic status may not influence the health of older adults in consistent ways across different societies, including the effect that economic devel­opment can have on health outcomes. This material may not be reproduced without written permission from the authors. For a list of Policy Research Division Working Papers, including those available for down­loading in PDF format, see www.popcouncil.org/publications/wp/prd/rdwplist.html.The current study examines the degree to which two socioeconomic indicators, education and income, are associated with physical functioning among older adults in three Asian societies-Taiwan, Thailand, and the Philippines-and the degree to which these associations are consistent. Population aging occurring in these societies will pro­duce some of the oldest age structures in the world (Kinsella 2000; Knodel, Ofstedal, and Hermalin 2002). Inevitably, population aging will lead to challenges related to popu­lation health and will exert pressures on health care systems (Hermalin 1995; Hermalin and Myers 2002; Phillips 2000). Physical functioning is an important aspect of health and health care, related to the capacity to carry out necessary daily functions in an inde­pendent fashion. In turn, an area of research beneficial for understanding how health and health care will be challenged is an examination of correlates of physical function­ing (Pol and Thomas 1992). Determining what distinguishes between those in better and worse functional health in later life can assist in planning for the aging of Asia in several ways: by determining how population health might change as societies are composed of more and more older adults, ascertaining which groups will be at higher risk for con­tracting functional disorders, and assisting in creating interventions to improve overall population health. Socioeconomic Status and Health Measures of socioeconomic status, such as education and income, have proven to be persistent and robust correlates of health among adults of all ages, as they have an effect on mortality, morbidity, and a host of health disorders (Preston and Taubman 1994). In the Western world, recognition of the association dates back centuries, and can be seen in the concern of Engels and Marx for the living conditions of the working-class poor in London (Turner and Beeghley 1981). Empirical studies go as far back as 1839 with Gavin's examination of the average age of death by occupation of household heads in a London suburb (Antonovsky 1967). Influential works that have fueled the modern- day discourse include Kitagawa and Hauser (1973), Townsend and Davidson (1982), and Marmot, Shipley, and Rose (1984). These studies are still frequently cited, but over the last ten years other studies have reported consistent findings across Western indus­ trialized societies (Dennis et al. 1993; Fox 1989; Hay 1988; Illsley and Svenson 1990; Pappas et al. 1993; Rogers 1992; Rogers, Rogers, and Belanger 1992, to name a few). The association is now considered to be so well established that the National Research Council recently asserted, "[I]n all societies, health and functioning vary ac­cording to socioeconomic position" (National Research Council 2001, p. 221). This statement seems somewhat strong given that very little research has considered non­Western developing societies such as those in Asia. In addition, little is known about the magnitude of the association and how it varies across settings where cultural norms, values, and levels of socioeconomic development differ. Testing for these associations is necessary in order to develop further theoretical perspectives and to affirm the univer­sality of the association. Associations between socioeconomic indicators and health are often attributed to psychological and sociological mechanisms such as health-related behaviors, so­cial support, stress, feelings of self-efficacy, and environmental factors, all of which influence the relationship between status and health (House et al. 1994; Williams 1990). Some researchers have posited a reverse causality whereby those in poor health drift downward on measures of socioeconomic status as their health deteriorates (Smith and Kington 1995). Others have argued that this relationship can only explain a small portion of the association, particularly when education, which is determined early in life, is used as an indicator of status (Fox, Goldblat, and Jones 1986; Haan, Kaplan, and Syme 1989; Winkleby et al. 1992). There is also a clear connection between so­cioeconomic status and access to health care in Western developed societies (Ander­son and Armstead 1995). The association between socioeconomic indicators and health among those at older ages may be somewhat less straightforward (Thorslund and Lundberg 1994). Con­vergence in rates of mortality have been found by Elo and Preston (1995) and Makuc et al. (1990), and House et al. (1994) found narrowing of rates after age 75. But when physical functioning is considered, researchers have found strong associations (Rogers, Rogers, and Belanger 1992; Ross and Wu 1996). Using the Alameda County Study, Camacho et al. (1993) found education to be associated with better functioning, meas­4 ured as the ability to perform 18 self-maintenance, mobility, and physical performance tasks, while Berkman and Gurland (1998) showed similar associations when examining differentials by income. Liang et al. (2000) showed gradients in mortality among older adults in China by socioeconomic indicators such as education, but there is otherwise a dearth of research on Asian societies. There are good theoretical reasons to believe that associations in Asia would be similar to those seen elsewhere. With respect to mediating influences, Preston and Taubman (1994) distinguish between influences that are economic and those that are psychosocial. On the economic side is access to better living and working conditions and the ability to purchase health care, both of which are associated with greater well­being. It is likely that these factors operate in a similar fashion in Asia as health care tends to be purchased, and living conditions are associated with education, income, and occupation. With respect to social and psychological explanations, Link and Phelan (1995) suggest there is persistence in the mechanisms that link socioeconomic status to health across settings as socioeconomic condition is a "fundamental cause" of health. We would thus expect that House et al.'s (1994) empirical test of psychosocial inter­vening mechanisms that connect socioeconomic status to health outcomes would be generalizable to Asian cultures. That is, in Asia as elsewhere, those with higher socio­economic status would tend to have a greater understanding of behaviors that lead to better health, including diet, exercise regimens, and avoidance of risky behaviors. They would also tend to have lower levels of stress, a higher quality of social support, a greater feeling of personal control over their lives, and a higher likelihood of seek­ing medical attention. COMPARATIVE RESEARCH ON THE Health of Older Adults The value of cross-national and cross-cultural research for understanding the fac­tors that determine the health of older adults has been emphasized in a number of recent publications (Albert and Cattell 1994; Bengtson et al. 2000; Chi, Chappell, and Lubben 2001; Hermalin 2002; National Research Council 2001). These have underscored sev­5 eral advantages to comparative research, which include providing a sense of generalizability of observations made in specific national contexts, raising questions about universality versus uniqueness of the aging experience, and revealing variations in the underlying dynamics of aging processes. Jointly, these benefits suggest that such investigations have the potential to highlight the diversity that exists in aging across cultures and geographical locations and to call into question persistent notions in main­stream social science. Given the universality that is often attributed to the association between socioeconomic status and health, comparative research on this topic would clearly be instructive. Yet, in conducting cross-national studies on aging and health in developing re­gions, researchers are faced with a number of obstacles. It is rare to find data that can be used to test for the consistency of associations across geographic settings. Surveys that collect information on the health of older adults in non-Western developing settings are not as abundant as those conducted in the United States and other developed countries, and the comparability of the data that do exist can be questioned. Researchers must also rely on surveys conducted over long spans of time and using different ways of measur­ing health. These limitations exist in addition to the general difficulty of interpreting responses provided by individuals living in a variety of cultural settings. Interpretation problems mean that reliance on a single measure of health generally, and physical func­tioning specifically, may not be sufficient to make confident inferences. We attempt to overcome some of these problems in the current study while rec­ognizing the limitations involved. First, we use recent data collected in Taiwan, Thai­land, and the Philippines. These data were collected in 1995 and 1996 under a single survey program. Second, we examine physical functioning outcomes measured in a va­riety of ways, and we assess the consistency of results across these measures. Third, we conduct two types of analyses. Using survey items that determine the ability to perform similar but not identical physical functioning tasks, we limit the analysis to whether associations exist across settings, and, if so, whether the directions are similar. Then, using outcomes that are more readily comparable, we pool data across the three samples and test for associations that include interaction terms between the country (Taiwan, 6 Thailand, or the Philippines) and the socioeconomic measure (education or income). Only when measures are readily comparable do we attempt to assess consistency. Taiwan, Thailand, and the Philippines, all located in East Asia, are characterized by close familial relations, reciprocity between older and younger family members, and the notion that the health of older adults is a family concern (Asis et al. 1995; Casterline et al. 1991). Yet, while this part of Asia is often described as rapidly developing, there is variation in the amount of development that has occurred. Taiwan's level of develop­ment is the most advanced and that of the Philippines the least. By the same token, life expectancy is about 75 years in Taiwan, 72 in Thailand, and only about 67 in the Philip­pines (Population Reference Bureau 2000). National economic standing can affect health in a number of ways, such as through the availability of health care resources and tech­nology. Some have reported, for instance, a growing number of public health services in tandem with the growth of Taiwan's economy, including the recent inauguration of a universal health care program (Favereau 1995; Chiang 1997). It is likely that these na­tional inputs will affect the micro-level association between socioeconomic status and health-for example, through making access to health services more readily available to those with higher income, or making this access more equitable across status levels. Yet, little research has been conducted to determine the exact mechanisms through which development affects the association in question. The current study may lead to some indication. Methods Data Data are from the Study of Rapid Demographic Change and the Welfare of the Elderly, a series of surveys conducted in Taiwan, Thailand, and the Philippines in 1995 and 1996. Each survey consists of a nationally based sample that, when weighted, is nationally representative. Multi-stage stratified samples were chosen in each country so that appropriate administrative units were first selected (for example, counties in Tai­wan, provinces in Thailand, and barangays in the Philippines), leading to the selection 7 of households within units. (The term "country" for Taiwan is used for purposes of consistency only and is not meant to indicate its political status.) Finally, one eligible respondent was chosen within each household through random selection. Each survey employed proxy respondents when individuals were unable to respond on their own. More detailed information on these surveys can be found in Chang and Hermalin (1997), Chayovan and Knodel (1997), Hermalin (2002), and Natividad and Cruz (1997). The current study considers men and women aged 60 and older at the time of the survey; there were 3,626 such individuals in the Taiwan survey, 4,486 in Thailand, and only 1,311 in the Philippines. Samples were not weighted across countries to be region­ally representative. This can pose a problem in comparative analyses in which data are pooled across samples. Hence, an extra procedure is conducted prior to pooled analyses in which observations are weighted so that sample sizes represent the same proportion of the 60-and-older population in all three countries. Doing this means that sampling fractions after reweighting are equivalent across samples, but the total sample size does not change. Specifically, the additional weighting procedure uses a sampling fraction of one respondent for every 1,186 individuals aged 60 and older living in each of the three countries. Measuring Physical Functioning Physical functioning tasks are often classified either as "activities of daily living" (Katz et al. 1963), which include those necessary for personal care, or as "instrumental activities of daily living" (Lawton and Brody 1969), which are necessary to maintain a living environment. The difficulty in using these types of items in comparative studies is that they may be tied to socially defined roles and environmental circumstances that can make responses culturally dependent and therefore difficult to interpret. Nagi (1976, 1991) advanced a more useful distinction by identifying the loss of ability to perform tasks as a specific "impairment," or physiological abnormality, and "disability," or be­havior patterns that evolve from impairment (see also Kelly-Hayes et al. 1992). Func­tional limitations, defined this way, involve basic body movements and are therefore less tied to and biased by social circumstance (Freedman and Martin 1999). In the cur­8 rent study we adopt this conceptualization and consider the ability to conduct four bodily movements: climbing stairs, lifting objects, walking, and crouching. The self-assessed ability to conduct these movements was measured in the three surveys with the ques­tion, "Do you have any difficulty doing the following without assistance from others and without assistance from a device?" In each survey, responses to each item can be divided into three categories: no difficulty, a little difficulty, and a lot of difficulty (the last includes not being able to conduct the task at all without assistance). Unfortunately, the exact task differed across surveys for three of the four items, and we wish to be particularly sensitive to these differences in our analyses and interpre­tation. For instance, older adults in Taiwan were asked whether they had difficulty climb­ing a greater number of stairs than were those in Thailand or the Philippines, which in turn elicited a higher proportion of affirmative responses (about 18 percent versus about 12 percent and 16 percent). We do not use the un-identically measured items directly in pooled analyses. Instead, we use them only to determine whether the direction and sig­nificance of associations are similar across countries. However, we use the one identi­cally measured item (crouching) and a normalized physical functioning scale in pooled analyses. Specifically, we construct three physical functioning measures. The first is hav­ing at least a little difficulty with at least one of the four bodily movements, where those who had difficulty are coded 1 and those who did not are coded 0. Second, we create a single measure from the one item that is identical across surveys: crouching ability. This is also coded 1 if an individual has difficulty crouching and coded 0 if not. Third, we use all items and response categories to create a composite index of functioning difficulties. Here, we give scores of 0 (have no difficulty), 1 (have a little difficulty), and 2 (have a lot of difficulty) for each item and sum the scores to create a composite scale ranging from 0 to 8. This score will be related to the specific nature of the tasks. For instance, because climbing fewer steps (asked in Thailand) is an easier task than climbing more steps (asked in Taiwan), we expect mean scores for climbing to be higher in Taiwan than in Thailand. To adjust for the specific difficulty of the tasks, we normalize scores so that the measure has a mean of 0 and a standard deviation of 1 in each country. Technically, 9 this original index is a discrete measure, but we treat the normalized scores as continu­ous in our analysis. Table 1 presents distributions for physical functioning outcomes. The current study is not intended as a comparison of the level of functioning across these populations; rather we focus on the nature of specific associations. Thus, we omit any interpretation or analysis of these figures, except to say that the percentage of individuals who had difficulty within specific tasks appears to be related to the degree of difficulty of the task, and that overall the elderly sample in Taiwan appear to be functionally healthier than those in the Philippines and Thailand. Measuring Education, Income, and Other Covariates Socioeconomic status is often represented as education, income, or occupation, and previous research has demonstrated associations employing all three indicators (Anderson and Armstead 1995; Williams and Collins 1995). We do not have compa­rable occupational data across samples, and therefore consider only education and in­come in the analysis. Regarding education, participants in this study were generally born between the turn of the twentieth century and 1936, a period of colonial rule in Table 1 Distribution of physical functioning outcomes by country Percent with Taiwan Thailand Philippines Difficulties crouching8 27.4 37.8 35.2 Difficulties climbingb 25.0 17.2 20.8 Difficulties liftingc 31.8 51.0 30.2 Difficulties walkingd 16.4 35.1 28.4 At least one limitation 42.2 63.8 47.5 Mean physical functioning score 1.58 2.25 1.77 a Crouching question is the same in all three countries: Do you have any difficulties crouching? b Climbing question: Do you have any difficulties . . . Taiwan: . . . walking up two or three flights of stairs? Thailand: . . . walking up two or three steps? Philippines: . . . going up and down the stairs (about 3-5 steps)? c Lifting question: Do you have any difficulties lifting . . . Taiwan: . . . something weighing 11-12 kg (like 2 pecks of rice)? Thailand: . . . about 10 kg in weight? Philippines: . . . something as heavy as a 5 kg bag of rice? d Walking question: Do you have any difficulties walking . . . Taiwan: . . . 200-300 meters? Thailand: . . . one kilometer? Philippines: . . . 200-300 meters? 10 Asia. If they attended school, these individuals were influenced by the educational sys­tems established by the colonial rulers. The Philippines was under American rule after 1898, and Taiwan was under Japanese rule after 1895, while Thailand is the only coun­try in the region that was never occupied. Because the systems of education differed, we use broad divisions and recode years of education into three categories. The first con­sists of those with no formal schooling, the second those with at least some primary schooling. The number of years of schooling considered as completed primary differed across countries at the time that the individuals in our sample were completing their education (four years in Thailand, six years in Taiwan and the Philippines). These speci­fications are used to create the third category, more than primary. Older adults in the Philippines, on balance, have the highest levels of education and those in Taiwan the lowest. This is the result of historical circumstances, as the United States promoted at least a minimum level of schooling in the Philippines during its occupation and even provided teachers (Carroll et al. 1970). During the same period, Taiwan was occupied by Japan, which did not promote higher levels of schooling (Tsurumi 1977). Specifically, the distribution of education (none, primary, more than primary) is 17 percent, 59 percent, and 24 percent in the Philippines; 41 percent, 39 percent, and 21 percent in Taiwan; and 31 percent, 63 percent, and 6 percent in Thailand. While level of education is determined fairly early in life and a good case can be made for the direction of causality from education to health in old age (Ross and Wu 1995; Winkleby et al. 1992), the causal direction between wealth and health is less clear, although some U.S.-based data show that income can be a function of health status (Smith 1999). While recognizing the inherent endogeneity, we consider asso­ciations between income and physical functioning and avoid interpreting results as causal. For those who are married, income is measured as the combined income of husband and wife, while for the unmarried it is measured as individual income. We use a measure of combined income because we are interested in determining the in­come that is available for the "purchase" of health and health care, and we assume that in most cases a couple's income would be pooled. Because of differences in monetary units, economic conditions across countries, and proportions who do not report in­come, we again categorize this measure. Respondents in each country were asked to 11 report their income as a single figure. Those not listing a figure were asked to place themselves within categories that were provided, and the midpoint or median of the category was used to estimate a single income amount. Ignoring missing responses, we divided income into quartiles in each country, although categories do not strictly represent quartiles since the nature of the coding meant there was heaping around certain responses. This quartile division works out very evenly in the Philippines and Thailand, but there is definite imbalance in Taiwan, where most responses were pro­vided within categories rather than as actual amounts; and heaping of responses means that the percentage in one category of income is as low as 7.4 percent of the popula­tion, and, in another category, as high as 39.4 percent. To account for missing income data, we constructed a fifth category for those whose income is not reported, and we include the missing category in multivariate models. Several other covariates are included in the models. These include comparable measures that have been shown, in the past, to be important determinants of functional health. We adjust models for age (measured in five-year categories), sex (1=female) of the respondent, marital status (1=married), and rural/urban residence (1=rural). Hypotheses and Analytical Strategy We test two primary hypotheses: (1) Older adults with higher income and higher levels of education will be less likely to report a functional limitation and will have lower physical functioning diffi­culty scores than will others regardless of whether they live in Taiwan, Thailand, or the Philippines. (2) The magnitude of the relationships between socioeconomic indicators and physical functioning outcomes will be consistent across samples. The first hypothesis refers simply to whether the association exists in the three settings and does not involve comparing exact magnitudes of associations. The second, referring to the consistency of the association across settings, requires equivalent out­come measures, and thus we test this hypothesis using crouching difficulties and nor­malized functioning difficulty scores. 12 Starting with the measure of having at least one of the four functional limitations, we fit separate sample logistic equations predicting having any limitation for each indi­vidual country. Although this represents three independent tests, for convenience we combine results into a single table. Since there is a good deal of overlap between mea­sures of education and income, separate models are constructed for each indicator, and these also control for other covariates. The significance of the socioeconomic status indicator is determined by the change in the -2 X log-likelihood statistic, distributed as %2, which is the difference in the -2 log-likelihood between a model that contains the specific socioeconomic indicator being tested with covariates and one model that con­tains only covariates. This is the preferred method of testing for significance in logistic regression (Hosmer and Lemeshow 1989). In separate procedures, not shown in tabular form, we tested for the sensitivity of these results after using robust variance estimation, clustering cases by the specific stratification unit for each sample (county for Taiwan, province for Thailand, barangay for the Philippines). The results were not substantially different from those reported here. To ascertain whether education and income associations are consistent, we use a pooled sample technique. We fit two logistic models predicting having crouching prob­lems for each indicator. The first assumes no difference in education or income effects across countries. The second assumes that education and income effects differ across country samples, and we fit equations with terms that interact education and income with country. We employ a likelihood ratio test to determine whether the second model provides a better fit. If not, we then assume that influences of education and income are consistent across countries, confirming the second hypothesis. To predict the function­ing difficulty score, we fit OLS equations in a similar fashion, but we use the F-ratio test to determine whether the interaction model provides a better fit. Because interactions can be difficult to interpret intuitively, we also provide figures that plot predicted values across countries by education and income. We also control for age, sex, marital status, and residence, factors that past re­search has identified as determinants of health in adults. We expect, then, that those who are older, are female, and are not married are more likely to have functional problems 13 and are more likely to have higher functioning difficulty scores than are others (Goldman, Korenman, and Weinstein 1995; Murphy, Glaser, and Grundy 1997; Ortmeyer 1974; Sickles and Taubman 1997; Verbrugge 1989). There is little evidence of the effect of rural versus urban residence on health in the countries being studied, but research in developed countries suggests that the elderly in urban areas are healthier (Lee and Cassidy 1985; Palmore 1983). Results We begin by looking at bivariate associations. Table 2 shows the percentage of older adults with at least one difficulty, the percentage with difficulty crouching, and mean normalized functioning difficulty scores across levels of education and income, for each sample separately. For descriptive purposes, the table also includes the per­centage within categories of education and income for each country. Looking at the upper panel, we see that, in general, there are strong associations with education in the expected direction: the elderly with less education are more likely to report one or more limitations and crouching problems and more likely to have higher functioning difficulty scores. Initial indications are that there is some difference in the nature of the association across settings, but it is difficult to confirm or quantify these using this bivariate procedure. Still, it appears as if the association is particularly strong in Tai­wan, where the percentage with at least one limitation and the percentage with prob­lems crouching are more than twice as great for those with no formal education in comparison to those with more than primary education. The association appears weakest in the Philippines. The bivariate association between education and having crouching problems is not significant in the Philippines, and the association between education and normalized functioning difficulty score is less significant in the Philippines than elsewhere. The bottom panel looks at associations with income level. Again, these are gener­ally in the expected directions and nearly all are significant, except for the association between having crouching problems and income in the Philippines, which borders on a 0.05 level of significance. In all three samples, respondents with missing income tend to have a fairly high likelihood of reporting limitations. Those who are most physically 14 Table 2 Bivariate physical functioning distributions by education and income % distribution % with at least one limitation % with problems crouching Mean normalized functioning difficulty score (standard deviation in parentheses) Education Taiwan None 43.7 56.8 37.4 0.216 (1.081) Primary 36.1 35.7 22.0 -0.213 (0.855) More than primary 20.2 25.7 17.7 -0.368 (0.712) P 0.000 0.000 0.000 Thailand None 32.9 75.6 44.1 0.263 (1.047) Primary 59.8 58.9 34.7 -0.241 (0.891) More than primary 7.3 53.7 37.1 -0.223 (1.000) P 0.000 0.000 0.000 Philippines None 16.3 54.7 39.9 0.129 (1.074) Primary 59.4 46.2 34.2 -0.053 (1.009) More than primary 24.4 40.5 30.7 -0.032 (0.994) P 0.010 0.109 0.045 Income Taiwan Lowest 27.7 55.9 39.1 0.198 (1.068) Second 30.7 38.5 25.5 -0.145 (0.901) Third 13.7 28.5 16.6 -0.378 (0.664) Highest 5.8 24.3 15.0 -0.389 (0.719) Missing 22.1 45.7 27.0 0.017 (1.033) P 0.000 0.000 0.000 (0.960) Thailand Lowest 21.1 79.8 44.7 0.340 (1.014) Second 21.2 64.7 38.8 -0.063 (0.948) Third 23.8 56.0 32.4 -0.303 (0.837) Highest 22.9 48.9 32.6 -0.378 (0.841) Missing 11.0 79.2 45.2 0.351 (1.077) P 0.000 0.000 0.000 Philippines Lowest 23.3 59.1 41.9 0.206 (1.046) Second 23.1 45.9 35.3 -0.048 (0.961) Third 23.6 44.4 33.9 -0.095 (0.950) Highest 23.8 41.3 30.2 -0.113 (0.976) Missing 6.3 46.3 33.8 0.061 (1.152) P 0.000 0.054 0.001limited may be less able to report income accurately because they are least involved in paid employment. We now turn to models that control for age, sex, marital status, and rural/urban residence. Table 3 examines logistic regression odds ratios for reporting at least one of four functional problems separately in each of the three samples. Although we avoid a comparison of the magnitude of the association across settings owing to direct incom­patibility of this outcome measure, we are able to make some indirect observations about associations, which exist in Taiwan, exist to some extent in Thailand, and are very weak in the Philippines. In some ways, differences in the bivariate results above become more distinct when controlling for age, sex, marital status, and residence. In Taiwan, for in­stance, the odds of reporting a limitation are almost two and a half times greater for those with no education versus those with more than primary. Both education and in­come result in significant changes in log-likelihood. In Thailand and the Philippines the odds ratio of having a limitation is highest for those without education and lowest for those with more than primary schooling, but the associations are not strong enough to be statistically significant. There is a significant association between income level and having a functional limitation in Thailand. In the Philippines, this association runs in the ex­pected direction, but differences in odds ratios are not great enough to be significant, with the exception of the comparison between the lowest and highest levels of income, which reaches borderline significance. Table 3 also shows that age and sex associations with physical functioning are in the expected directions and are significant in all three samples. Tables 4 and 5 report pooled data results that introduce interaction effects to de­termine whether associations between socioeconomic indicators and physical function­ing outcomes are consistent. Table 4 displays odds ratios for having crouching difficul­ties. Models 1 and 2 show main effects only, that is, before adding the interactions. In these first two models we observe country differences in the likelihood of reporting crouching problems, with fewer problems being reported in Taiwan than elsewhere. The interaction models are 3 and 4. To determine socioeconomic status effects in Thailand and the Philippines it is necessary to view the main education and income odds ratios 16 Table 3 Separate sample odds ratios for having at least one of four functional limitations in Taiwan, Thailand, and the Philippines Taiwan Thailand Philippines Model 1 Model 2 Model 1 Model 2 Model 1 Model 2 Education More than primary 1.00 1.00 1.00 Primary 1.59** 1.13 1.20 None 2.32** 1.22 1.38 A -2LL 58.27** 1.83 2.65 Income Highest 1.00 1.00 1.00 Second 1.13 1.24 a 1.19 Third 1.48* 1.46** 1.13 Lowest 2.38** 2.23** 1.47 a Missing 1.73** 2.30** 0.98 A -2LL 49.25** 70.51** 5.16 Age 60-64 1.00 1.00 1.00 1.00 1.00 1.00 65-69 1.63** 1.63** 1.47** 1.44** 1.02 1.01 70-74 2.53** 2.57** 2.54** 2.32** 1.69** 1.83** 75-79 3.98** 4.00** 4.24** 3.64** 2.81** 2.81** 80 and older 8.33** 8.47** 10.56** 8.68** 5.79** 6.09** Sex Male 1.00 1.00 1.00 1.00 1.00 1.00 Female 2.50** 2.84** 2.95** 2.85** 1.64** 1.69 Marital status Married 1.00 1.00 1.00 1.00 1.00 1.00 Not married 0.93 0.90 1.08 0.95 1.19 1.14 Residence Rural 1.00 1.00 1.00 1.00 1.00 1.00 Urban 1.08 1.06 0.99 1.07 1.19 1.17 LL -2,122.27 -2,126.78 -2,571.76 -2,537.45 -790.11 809.16 A -2 LL (model) 470.34** 459.79** 332.35** 367.96** 89.32** 95.28** ** p < 0.01; * 0.01 < p < 0.05; A 0.05 < p < 0.10 together with the interaction effects, while for Taiwan, which is the comparison cat­egory, the effects can be determined from the main effect odds ratios. Associations be­tween education and having a crouching problem are strong in Taiwan. The odds ratio of 17 having this problem is more than twice as great for those without education as for those with more than primary. But the interaction effects show that impact of education is weaker in Thailand and the Philippines. For instance, having no education in the Philip­pines reduces the main effect odds by a factor of 0.43. If calculated out (various effects can be multiplied to get a total odds ratio), the odds ratios of having crouching problems in the Philippines and Thailand are virtually the same across levels of education. Look- Table 4 Pooled sample odds ratios for having crouching difficulties Main effect models Interaction effect models Model 1 Model 2 Model 3 Model 4 Country Taiwan 1.00 1.00 1.00 1.00 Thailand 1.81** 1.51** 3.07** 3.06** Philippines 1.48** 1.90** 1.94** 2.18** Education More than primary 1.00 1.00 Primary 1.18 1.34* None 1.33* 2.09** Income Highest 1.00 1.00 Second 1.06 1.06 Third 1.19 1.66* Lowest 1.37** 2.75** Missing 1.11 1.61* Age 60-64 1.00 1.00 1.00 1.00 65-69 1.12 1.11 1.12 1.12 70-74 2.16** 1.62** 1.60** 1.64** 75-79 2.19** 2.13** 2.19** 2.14** 80 and older 3.40** 3.35** 3.46** 3.38** Sex Male 1.00 1.00 1.00 1.00 Female 1.70** 1.74** 1.69** 1.73** Marital status Married 1.00 1.00 1.00 1.00 Not married 0.97 0.93 0.97 0.92 Residence Rural 1.00 1.00 1.00 1.00 Urban 1.27** 1.28** 1.27** 1.29** continued 18 Table 4 continued Main effect models Interaction effect models Model 1 Model 2 Model 3 Model 4 Interactions Country x Education Thailand x Primary 0.87 Thailand x None 0.63A Philippines x Primary 0.70A Philippines x None 0.43** A -2LL 22.90** Country x Income Thailand x Second 0.92 Thailand x Third 0.68 Thailand x Lowest 0.42** Thailand x Missing 0.70 Philippines x Second 1.18 Philippines x Third 0.76 Philippines x Lowest 0.49** Philippines x Missing 0.61 A -2LL 29.67** LL -5,677.83 5,713.21 -5,666.38 -5,698.38 A -2LL (model) 335.81** 321.69** 391.62** 392.94** ** p < 0.01; * 0.01 < p < 0.05; A 0.05 < p < 0.10 ing at interaction effects with income, a similar situation exists, that is, the association is significant and strong in Taiwan, while the main effect odds ratios for income are gener­ally lower in Thailand and the Philippines. Changes in the log-likelihood for the educa­tion and income interaction effects are both significant, verifying that the associations are not consistent across the three samples. Table 5 examines OLS estimates predicting the normalized functioning difficulty score, and although the measure is different, the results are largely consistent with those found above. Country effects by themselves are not interpretable across samples be­cause scores have a mean of 0 and a standard deviation of 1 in each. Models 1 and 2 show that across the pooled samples, the elderly with less education and less income tend to have higher scores and thus a greater number of and more severe difficulties. But the interaction effects in Model 3, all of them negative in direction, indicate that the associations with education are weaker in Thailand and the Philippines than in Taiwan. 19 The income interactions show a different picture. Income associations appear almost as strong in Thailand as in Taiwan, but are weaker in the Philippines. Although the ex­plained variance increases by only about 1 percent when introducing the interaction terms, the changes in F-ratio are significant. Tables 4 and 5 also confirm the expected Table 5 Pooled sample OLS estimates for normalized functioning difficulty score Main effect models Interaction effect models Model 1 Model 2 Model 3 Model 4 Country Taiwan - - - - Thailand 0.031 0.073** 0.154* 0.015 Philippines 0.051 0.082* 0.203** 0.148* Education More than primary - - Primary 0.040 0.145** None 0.172** 0.325** Income Highest - - Second 0.029 -0.028 Third 0.108** 0.120* Lowest 0.278** 0.322** Missing 0.245** 0.204** Age 60-64 - - - - 65-69 0.061* 0.054A 0.065* 0.056A 70-74 0.306** 0.307** 0.309** 0.306** 75-79 0.623** 0.609** 0.625** 0.603** 80 and older 1.103** 0.980** 1.015** 0.969** Sex Male - - - - Female 0.316** 0.320** 0.308** 0.315** Marital status Married - - - - Not married 0.015 -0.015 0.018 -0.016 Residence Rural - - - - Urban 0.070* 0.068* 0.069* 0.074* continued 20 Table 5 continued Main effect models Interaction effect models Model 1 Model 2 Model 3 Model 4 Interactions Country x Education Thailand x Primary -0.127A Thailand x None -0.174* Philippines x Primary -0.154* Philippines x None -0.280* A F-ratio 2.35* Country x Income Thailand x Second 0.076 Thailand x Third 0.056 Thailand x Lowest 0.070 Thailand x Missing 0.203* Philippines x Second 0.066 Philippines x Third -0.094 Philippines x Lowest -0.201* Philippines x Missing -0.140 A F-ratio (interactions) 2.11* Constant -0.603 -0.665 -0.703 -0.663 F-ratio (model) 82.93** 66.99** 72.03** 56.34** R-square 0.164 0.173 0.165 0.176 ** p < 0.01; * 0.01 < p < 0.05; a 0.05 < p < 0.10 associations with age and sex, and show some significant influences of residence, with those in urban areas presenting greater limitations than those in rural areas. Figures 1-4 assist in the interpretation of the interaction models by plotting pre­dicted values. Figures 1 and 2 show the predicted probability of having difficulty crouch­ing in each country by education and income. The predicted probability is very similar among those with no education and among those with the lowest level of income in each setting, that is, a probability between roughly 0.35 and 0.40. But this probability differ­ence becomes wide with rising levels of education and income so that those in Taiwan appear to benefit the most from higher socioeconomic status. For example, the probabil­ity of reporting a crouching difficulty declines to about 0.20 in Taiwan for those with more than primary education but remains between roughly 0.30 and 0.40 for those with more than primary in the Philippines and Thailand. 21 Figure 1 Predicted probability of reporting difficulties crouching in the three samples by education -A- Taiwan -B- - Thailand Philippines Figure 2 Predicted probability of reporting difficulties crouching in the three samples by level of income -A- - Taiwan -B- - Thailand -e- - Philippines Figures 3 and 4 show the predicted normalized functioning difficulty scores by edu­cation and income across samples. The mean score is 0 in each country, so the slope of the line in these figures indicates how much better off those with high education and income are in comparison to the mean and how much worse off those with low education and income are in comparison to the mean within each individual sample. The education result is con- 22 Figure 3 Predicted normalized functioning difficulty score in the three samples by education -A- - Taiwan -B- - Thailand Philippines Figure 4 Predicted normalized functioning difficulty score in the three samples by level of income -A- - Taiwan -B- - Thailand -e- Philippines sistent with the results seen in the previous figures. There is a sharp decline in the predicted score as education increases in Taiwan, indicating that those with more education have fewer and less severe limitations. A slight decline by level of education is found in Thailand and very little change in the Philippines. For income, there are sharp declines in the scores in both Taiwan and Thailand, but very little change by level of income in the Philippines. 23 DISCUSSION We conclude by offering five possible explanations as to why the socioeconomic status indicators used in our study might affect physical functioning in different ways among older adults in different settings. These include differences in the way socioeco­nomic status influences health, possible reporting differences, differences in the way individuals perceive health problems, environmental barriers that can diminish the ad­vantage of those of higher status, and differences in stratification systems. A combina­tion of these is quite possible, and we suggest that future research attempt to distinguish between these explanations. First, differences in the way education and income influence physical function­ing of older adults may be accounted for by variations in socioeconomic development across settings. Socioeconomic status may have less of an effect on the physical func­tioning of older adults in settings where there are tangible limitations to medical care and barriers to the attainment of health-related knowledge. Consider again that most of the research that confirms a strong association has taken place in advanced Western societies where those with higher socioeconomic status have access to advanced re­sources that can benefit their health. Of the three societies included in the current study, Taiwan is by far the most advanced economically and is therefore closest to Western nations with respect to development. In turn, results in Taiwan appeared to be most comparable to those of previous studies conducted in the West. In short, it seems reason­able that individuals who have high socioeconomic status benefit most when there is access to resources that may be more conducive to ensuring well-being, such as the availability of better-quality health care. Second, differences in reporting may exist. Lower-status individuals in some set­tings may be more modest about reporting health problems. In addition, reporting of health disorders may vary by cultural circumstance. Angel and Guarnaccia (1989) found that Hispanics who responded to a survey in Spanish rather than English were more likely to report their health in negative terms, suggesting even linguistic differences. If individuals are more likely to report limitations in a particular setting, the influence of education or income on physical functioning may be muted. 24 Third, there may be differences in the way different socioeconomic groups per­ceive and consequently report health problems in different settings. Waidmann, Bound, and Schoenbaum (1995) found that the reporting of health problems increased in the United States during the 1970s as a result of better and earlier diagnosis. In settings where medical access is limited to those in higher-status groups, these individuals are likely to be more aware of and sensitive to health difficulties and therefore may be more likely to report problems. This would explain the lack of association between higher socioeconomic status and reporting of functioning limitations seen in the Philippines. That is, despite having functional problems, those with lower education and income may be less likely to interpret their disorders as something serious enough to report. Fourth, barriers may exist in certain settings that could make some individuals more likely to report functioning disorders. For instance, a number of areas in the Phil­ippines are underdeveloped. Elderly individuals of higher socioeconomic status in these areas may be faced with environmental barriers when conducting functional tasks, such as roads that are difficult to walk or longer distances to walk between locations. Because of this, they may be more likely to perceive a functional disorder as being worth report­ing. Similar barriers might exist in Thailand. Traffic congestion in Bangkok, for in­stance, may make it difficult for older adults to get around, especially as curbs are high and there are few pedestrian crossways. Regardless of levels of socioeconomic status, walking a given distance might present, on average, fewer difficulties in Taipei, which is friendlier to those with functional problems. Consequently, differences in environ­ment can remove some of the advantages normally associated with higher levels of socioeconomic status. Finally, where education and income appeared to have less of an effect on func­tional outcomes, it is possible that our measures of status do not adequately reflect the stratification system in place and therefore do not encompass the access and other psy­chosocial differentials that exist across groups. For example, in Thailand, why did in­come distinguish between some functional outcomes while education did not? Because few older adults in Thailand have more than a low level of education, this measure might not be specific enough. As another example, the elderly in the Philippines would have obtained their schooling during the American occupation. One of the consequences 25 of American rule in the Philippines was an increase in educational attainment and more egalitarian access to education (Hunt 1993). Hence, few older adults in that country in the current sample have no education. Thus, across the three Asian samples, different socioeconomic status indicators might have varying effects on physical functioning if the stratification structures are themselves different. While future cohorts of older adults in Asian societies will have higher levels of education and income and will maintain a higher standard of living, concern has been expressed over the erosion of traditional support for older adults (Mason 1992). Some may point to the decreasing proportion of older adults who live in extended family ar­rangements as evidence of this trend (Asis et al. 1995). In addition, future cohorts of the elderly will have smaller families and therefore smaller network resources regardless of their living arrangements. Our study suggests that changes in socioeconomic character­istics such as education may play a role in offsetting transitions in age structure and available support in some societies, such as Taiwan. Future research should examine further whether there are discrepancies in the effect of socioeconomic status across set­tings and, if so, why these occur. In particular, there is need for research that examines how individuals perceive and report similar health disorders across diverse cultural and geographical settings and how these reporting differences can influence the perceived association between socioeconomic status and health. References Albert, Stephen M. and Maria G. Cattell. 1994. Old Age in Global Perspective: Cross- Cultural and Cross-National Views. New York: G.K. Hall & Co. Anderson, N.B. and C.A. Armstead. 1995. "Toward understanding the association of socioeconomic status and health: A new challenge for the biopsychosocial ap­proach," Psychosomatic Medicine 57: 213-225. Angel, R. and P. Guarnaccia. 1989. "Mind, body and culture: Somatization among His- panics," Social Science and Medicine 28: 1229-1238. Antonovsky, A. 1967. "Social class, life expectancy, and overall mortality," Milbank Memorial Fund Quarterly 45: 31-73. 26 Asis, M.M.B., L. Domingo, J. Knodel, and K. Mehta. 1995. "Living arrangements in four Asian countries: A comparative perspective," Journal of Cross-Cultural Gerontology 10: 145-157. Bengtson, Vern L., Kyong-Dong Kim, George C. Myers, and Ki-Soo Eun. 2000. Aging in East and West: Families, States, and the Elderly. New York: Springer Publishing. Berkman, C.S. and B.J. Gurland. 1998. "The relationship among income, other socio­economic indicators, and functional level in older persons," Journal of Aging and Health 10: 81-98. Camacho, T.C., W.J. Strawbridge, R.D. Cohen, and G.A. Kaplan. 1993. "Functional ability in the oldest old: Cumulative impact of risk factors from the preceding two decades," Journal of Aging and Health 5: 439-454. Carroll, J.J., F. Araneta, C.A. Arnaldo, and J.T. Keane. 1970. Philippine Institutions. Manila: Solidaridad Publishing House. Casterline, J.B., L. Williams, A. Hermalin, M.C. Chang, N. Chayovan, P. Cheung, L. Domingo, J. Knodel, and M.B. Ofstedal. 1991. "Differences in the living ar­rangements of the elderly in four Asian countries: The interplay of constraints and preferences," Comparative Study of the Elderly in Asia Research Report no. 91-10. Ann Arbor, MI: Population Studies Center, University of Michigan. Chang M.C. and A.I. Hermalin. 1997. "Survey of health and living status of the middle aged and elderly in Taiwan," Taiwan Aging Studies Series, Research Report no. 7-2. Taichung: Taiwan Provincial Institute of Family Planning. Chayovan, Napaporn and John Knodel. 1997. A Report on the Survey of the Welfare of the Elderly in Thailand. Bangkok: Institute of Population Studies, Chulalongkorn University. Chi, Iris, Neena L. Chappell, and James Lubben. 2001. Elderly Chinese in Pacific Rim Coun­tries: Social Support and Integration. Hong Kong: Hong Kong University Press. Chiang, T.L. 1997. "Taiwan's 1995 health care reform," Health Policy 39: 225-239. Dennis, B.H., G.S. Zhukovsky, D.B. Shestov, C.E. Davis, A.D. Deev, H. Kim, and H.A. Tyroler. 1993. "The association of education with coronary heart disease mortal­27 ity in the USSR: Lipid research clinics study," International Journal of Epidemi­ology 22:420-427. Elo, I.T. and S.H. Preston. 1995. "Educational differentials in mortality: United States, 1979-85," Social Science and Medicine 42: 47-57. Favereau, F. 1995. Healthcare Handbook: Volume 2.4, Taiwan. Langon: F. Favereau & Associates. Fox, A.J., P.O. Goldblat, and D.R. Jones. 1986. "Social class mortality differentials: Artifact, selection or life circumstances?" in R.G. Wilkinson (ed.), Class and Health. London: Tavistock, pp. 34-49. Fox, J. 1989. Health Inequalities in European Countries. Aldershot: Gower. Freedman, V.A. and L.G. Martin. 1999. "The role of education in explaining and fore­casting trends in functional limitations among older Americans," Demography 36:461-473. Goldman, Noreen, Sanders Korenman, and Rachel Weinstein. 1995. "Marital status and health among the elderly," Social Science and Medicine 40: 1717-1730. Haan, M.N., G.A. Kaplan, and L.S. Syme. 1989. "Socioeconomic status and health: Old observations and new thoughts," in J. Bunker, J.D. Gomby, and D. Kerher (eds.), Pathways to Health: The Role of Social Factors. Menlo Park, CA: Henry H. Kaiser Family Foundation, pp. 76-135. Hay, D.I. 1988. "Socioeconomic status and health status: A study of males in the Canada Health Survey," Social Science and Medicine 27: 1317-1325. Hermalin, A.I. 1995. "Aging in Asia: Setting the research foundation," Asia-Pacific Population Research Report no. 4. Honolulu: East-West Center. ---------. 2002. The Well-Being of the Elderly in Asia: A Four-Country Comparative Study. Ann Arbor, MI: University of Michigan Press. Hermalin, A.I and L.G. Myers. 2002. "Aging in Asia: Facing the crossroads," in A.I. Hermalin (ed.), The Well-Being of the Elderly in Asia: A Four-Country Com­parative Study. Ann Arbor, MI: University of Michigan Press, pp. 1-24. 28 Hosmer, D.W., Jr. and S. Lemeshow. 1989. Applied Logistic Regression. New York: John Wiley and Sons. House, J.S., J.M. Lepkowski, A.M. Kinney, R.P. Mero, R.C. Kessler, and R.A. Herzog. 1994. "The social stratification of aging and health," Journal of Health and So­cial Behavior 35: 213-234. Hunt, C.L. 1993. "The society and its environment," in R.E. Dolan (ed.), Philippines: A Country Study. Washington, DC: Library of Congress, pp. 65-116. Illsley, R. and P.G. Svenson (eds.). 1990. Health Inequalities in Europe, special issue of Social Science and Medicine. New York: Elsevier Science. Katz, S., A.B. Ford, R.W. Moskowitz, B.A. Jackson, and M.W. Jaffe. 1963. "Studies of illness in the aged: The index of ADL, a standardized measure of biological and psychosocial function," Journal of the American Medical Association 185: 914-919. Kelly-Hayes, M., A.M Jette, P.A. Wolf, R.B D'Agostino, and P.M Odell. 1992. "Func­tional limitations and disability among elders in the Framingham study," Ameri­can Journal of Public Health 82: 841-845. Kinsella, Kevin. 2000. "Demographic dimensions of ageing in East and Southeast Asia," in D.R. Phillips (ed.), Ageing in the Asia-Pacific Region. New York: Routledge, pp. 35-50. Kitagawa, E.M. and P.M. Hauser. 1973. Differential Mortality in the United States: A Study of Socioeconomic Epidemiology. Cambridge, MA: Harvard University Press. Knodel, J., M.B. Ofstedal, and A.I. Hermalin. 2002. "The demographic, socioeconomic, and cultural context of the four study countries," in A.I. Hermalin (ed.), The Well-Being of the Elderly in Asia: A Four-Country Comparative Study. Ann Ar­bor, MI: University of Michigan Press, pp. 25-64. Lawton, M.P. and E. Brody. 1969. "Assessment of older people: Self-maintaining and instrumental activities of daily living," The Gerontologist 9: 179-186. Lee, Gary R. and Margaret L. Cassidy. 1985. "Family and kin relations of the rural elderly," in R.T. Coward and G.R. Lee (eds.), The Elderly in Rural Society. New York: Springer Publishing, pp. 151-192. 29 Liang, J., J.F. McCarthy, A. Jain, N. Krause, J.M. Bennett, and S. Gu. 2000. "Socioeco­nomic gradient in old age mortality in Wuhan, China," Journal of Gerontology: Social Sciences 55: 222-233. Link, B.G. and J. Phelan. 1995. "Social conditions as fundamental causes of disease," Journal of Health and Social Behavior (Extra issue): 80-94. Makuc, D.M., J.J. Feldman, J.C. Kleinman, and M.B Pierre. 1990. "Sociodemographic differentials in mortality," in J.C. Cornoni-Huntley, R.R. Huntley, and J. J. Feldman (eds.), Health Status and Well-being of the Elderly. New York: Oxford Univer­sity Press, pp. 155-171. Marmot, M.G., M.J. Shipley, and G. Rose. 1984. "Inequalities in death: Specific expla­nations of a general pattern?" Lancet 8384: 1003-1006. Mason, K.O. 1992. "Family change and support of the elderly in Asia: What do we know?" Asia-Pacific Population Journal 7: 13-32. Murphy, Mike, Karen Glaser, and Emily M.D. Grundy. 1997. "Marital status and long­term illness in Great Britain," Journal of Marriage and the Family 59: 156-164. Nagi, S.Z. 1991. "Disability concepts revisited: Implications for prevention," in A.M. Pope and A.R. Tarlov (eds.), Disability in America: Toward a National Agenda for Prevention. Washington, DC: National Academy Press, pp. 309-327. ---------. 1976. "An epidemiology of disability among adults in the United States," Milbank Memorial Fund Quarterly: Health and Society 54: 439-468. National Research Council. 2001. Preparing for an Aging World: The Case for Cross­National Research. Washington, DC: National Academy Press. Natividad, J.N. and G. Cruz. 1997. "Patterns in living arrangements and familial support for the elderly in the Philippines," Asia-Pacific Population Journal 12: 17-34. Ortmeyer, C.F. 1974. "Variations in mortality, morbidity and health care by marital sta­tus," in C.E. Erhardt and J.E. Berlin (eds.), Mortality and Morbidity in the United States. Cambridge, MA: Harvard University Press, pp. 159-188. 30 Palmore, Erdman. 1983. "Health care needs of the rural elderly," International Journal of Aging and Human Development 18: 39-45. Pappas, G., S. Queen, W. Hadden, and G. Fisher. 1993. "The increasing disparity in mortality between socioeconomic groups in the United States, 1960 and 1986," New England Journal of Medicine 329: 103-109. Phillips, David R. 2000. "Ageing in the Asia-Pacific region: Issues, policies and con­texts," in D.R. Phillips (ed.), Ageing in the Asia-Pacific Region. New York: Routledge, pp. 1-34. Pol, Louis G. and Richard K. Thomas. 1992. The Demography of Health and Health Care. New York: Plenum Press. Population Reference Bureau. 2000. World Population Data Sheet. Washington, DC: Population Reference Bureau. Preston, S.H. and P. Taubman. 1994. "Socioeconomic differences in adult mortality and health status," in L.G. Martin and S.H. Preston (eds.), Demography of Aging. Washington, DC: National Academy Press, pp. 279-318. Rogers, R.G. 1992. "Living and dying in the USA: Sociodemographic determinants of death among blacks and whites," Demography 29: 287-303. Rogers, R.G., A. Rogers, and A. Belanger. 1992. "Disability-free life among the elderly in the United States: Sociodemographic correlates of functional health," Journal of Aging and Health 4: 19-42. Ross, C. and C. Wu. 1996. "Education, age, and the cumulative advantage in health," Journal of Health and Social Behavior 37: 104-120. ---------. 1995. "The links between education and health," American Sociological Re­view 60:719-745. Sickles, Robin C. and Paul Taubman. 1997. "Mortality and morbidity among adults and the elderly," in M.R. Rosenzweig and O. Stark (eds.), Handbook of Population and Family Economics. Amsterdam: Elsevier Science, pp. 559-643. 31 Smith, J.P. 1999. "Healthy bodies and thick wallets: The dual relation between health and economic status," Journal of Economic Perspectives 13: 145-166. Smith, James P. and Raynard Kington. 1995. "Race, socioeconomic status and health in late life," RAND, Labor and Population Program, Working Paper Series 95-10. Los Angeles, CA: RAND. Thorslund, M. and O. Lundberg. 1994. "Health and inequalities among the oldest old," Journal of Aging and Health 6: 51-69. Townsend, P. and N. Davidson. 1982. Social Inequalities in Health: The Black Report. Harmondsworth: Penguin Books. Tsurumi, P.E. 1977. Japanese Colonial Education in Taiwan: 1895-1945. Cambridge, MA: Harvard University Press. Turner, Jonathan H. and Leonard Beeghley. 1981. The Emergence of Sociological Theory. Homewood, IL: Dorsey Press. Verbrugge, Lois M. 1989. "The twain meet: Empirical explanations for sex differences in health and mortality," Journal of Health and Social Behavior 30: 282-304. Waidmann, T., J. Bound, and M. Schoenbaum. 1995. "The illusion of failure: Trends in self-reported health of the U.S. elderly," Milbank Quarterly 73: 253-287. Williams, David R. 1990. "Socioeconomic differentials in health: A review and redirec­tion," Social Psychology Quarterly 53: 81-99. Williams, D.R. and C. Collins. 1995. "U.S. socioeconomic and racial differences in health: Patterns and explanations," Annual Review of Sociology 21: 349-386. Winkleby, M.A., D.E. Jatulis, E. Frank, and S.P. Fortmann. 1992. "Socioeconomic sta­tus and health: How education, income and occupation contribute to risk factors for cardiovascular disease," American Journal of Public Health 82: 816-820. 32 POLICY RESEARCH DIVISION WORKING PAPERS Recent Back Issues 2001_______________________________ 144 John Bongaarts, "Household size and composition in the developing world." 145 John B. Casterline, Zeba A. Sathar, and Minhaj ul Haque, "Obstacles to contraceptive use in Pakistan: A study in Punjab." 146 Zachary Zimmer, Albert I. Herma­lin, and Hui-Sheng Lin, "Whose edu­cation counts? The impact of grown children's education on the physi­cal functioning of their parents in Taiwan." 147 Philomena Nyarko, Brian Pence, and Cornelius Debpuur, "Immuni­zation status and child survival in rural Ghana." *148 John Bongaarts and Zachary Zimmer, "Living arrangements of older adults in the developing world: An analysis of DHS household surveys." 149 Markos Ezra, "Ecological degrada­tion, rural poverty, and migration in Ethiopia: A contextual analysis." 150 Cynthia B. Lloyd, Sahar El Tawila, Wesley H. Clark, and Barbara S. Mensch, "Determinants of educa­tional attainment among adoles­cents in Egypt: Does school quality make a difference?" 151 Barbara S. Mensch, Paul C. Hewett, and Annabel Erulkar, "The report­ing of sensitive behavior among adolescents: A methodological ex­periment in Kenya." 152 John Bongaarts, "The end of the fer­tility transition in the developed world." 153 Mark R. Montgomery, Gebre- Egziabher Kiros, Dominic Agyeman, John B. Casterline, Peter Aglobitse, and Paul Hewett, "Social networks and contraceptive dynamics in south­ern Ghana." *154 Paul C. Hewett and Mark R. Mont­gomery, "Poverty and public ser­vices in developing-country cities." * No longer available 2002 155 Zachary Zimmer, Linda G. Martin, and Ming-Cheng Chang, "Changes in functional limitations and sur­vival among the elderly in Taiwan: 1993, 1996, and 1999." 156 John Bongaarts and Griffith Feeney, "How long do we live?" 157 Zachary Zimmer and Sovan Kiry Kim, "Living arrangements and socio-demographic conditions of older adults in Cambodia." 158 Geoffrey McNicoll, "Demographic factors in East Asian regional inte­gration." 159 Carol E. Kaufman, Shelley Clark, Ntsiki Manzini, and Julian May, "How community structures of time and opportunity shape adolescent sexual behavior in South Africa." *160 Julia Dayton and Martha Ainsworth, "The elderly and AIDS: Coping strategies and health consequences in rural Tanzania." 161 John Bongaarts, "The end of the fertility transition in the developing world." 162 Naomi Rutenberg, Carol E. Kauf­man, Kate Macintyre, Lisanne Brown, and Ali Karim, "Pregnant or posi­tive: Adolescent childbearing and HIV risk in South Africa." 163 Barbara S. Mensch, Wesley H. Clark, and Dang Nguyen Anh, "Pre­marital sex in Vietnam: Is the cur­rent concern with adolescent repro­ductive health warranted?" 164 Cynthia B. Lloyd, Cem Mete, and Zeba A. Sathar, "The effect of gen­der differences in primary school access, type, and quality on the de­cision to enroll in rural Pakistan." 165 Kelly Hallman, Agnes R. Quisum- bing, Marie Ruel, and Benedicte de la Briere, "Childcare, mothers' work, and earnings: Findings from the ur­ban slums of Guatemala City." *166 Carol E. Kaufman and Stavros E. Stavrou, " ‘Bus fare, please': The economics of sex and gifts among adolescents in urban South Africa." *167 Dominic K. Agyeman and John B. Casterline, "Social organization and reproductive behavior in southern Ghana."2003 168 Paul C. Hewett, Annabel S. Erulkar, and Barbara S. Mensch, "The fea­sibility of computer-assisted survey interviewing in Africa: Experience from two rural districts in Kenya." 169 Zachary Zimmer and Julia Dayton, "The living arrangements of older adults in sub-Saharan Africa in a time of HIV/AIDS." 170 Ravai Marindo, Steve Pearson, and John B. Casterline, "Condom use and abstinence among unmarried young people in Zimbabwe: Which strategy, whose agenda?" 171 Sajeda Amin and Nagah H. Al- Bassusi, "Wage work and marriage: Perspectives of Egyptian working women." 172 Zachary Zimmer, Napaporn Chayovan, Hui-Sheng Lin, and Josefina Natividad, "How indica­tors of socioeconomic status relate to physical functioning of older adults in three Asian societies."