geronb      J Gerontol B Psychol Sci Soc Scigeronb      The Journals of Gerontology Series B: Psychological Sciences and Social Sciences      J Gerontol B Psychol Sci Soc Sci      1079-5014      1758-5368              Oxford University Press                    3610.1093/geronb/62.1.S36                        Journal of Gerontology: Social Sciences                            Functional Limitations and Changes in Levels of Depression Among Older Adults: A Multiple-Hierarchy Stratification Perspective                                          Schieman            Scott                                                          Plickert            Gabriele                                    Department of Sociology, University of Toronto, Ontario, Canada.                    1        2007            62      1      S36      S42                        20          7          2006                          15          12          2005                            Copyright 2007 by The Gerontological Society of America        2007                          Objectives. This study examined the effects of functional limitations on changes in levels of depression over time. A multiple-hierarchy stratification perspective framed the analyses of potential stratification-based contingencies of race, gender, and socioeconomic status.        Methods. We derived data from a longitudinal study of adults aged 65 and older in the Washington, DC, metropolitan area over a 3-year period (2001–2003). We used ordinary least squares regression models in order to assess the focal associations across a set of social status interaction terms.        Results. Changes in functional limitations were associated with changes in depression, but the patterns depended on interactions among race, gender, and socioeconomic status.        Discussion. These observations contribute to research by blending ideas from double and triple jeopardy and differential vulnerability perspectives. Although analyses of multiple contingencies create analytical challenges, this complexity is needed to accurately specify the mental health effects of functional limitations.                              hwp-legacy-fpage          S36                          hwp-legacy-dochead          RESEARCH ARTICLE                                      THE stress process perspective identifies limitations in activities of daily living as among the most pernicious stressors in late life (Pearlin & Skaff, 1996). Levels of functional limitations are higher among older age groups (Long & Pavalko, 2004) and are associated positively with depression (Kelley-Moore & Ferraro, 2005; Yang & George, 2005). Moreover, limitations are distributed unequally across statuses such as gender, race, and socioeconomic status (SES; House, Lantz, & Herd, 2005). Although most research has focused on status variations in the levels of limitations, for this study we asked the following question: Do dimensions of stratification modify the effects of changes in limitations on changes in depression? In order to frame our analyses, we applied Clark and Maddox's (1992) multiple-hierarchy stratification perspective, which asserts that “it is useful in understanding health outcomes to consider income, education, gender, and race as independent but potentially interactive influences on changes in functional status over time” (p. S223).      Social scientists have long been interested in the ways in which ascribed statuses intersect to influence health over the life course (McLeod & Nonnemaker, 1999). Of those statuses, race and gender are among the most influential because of their association with racism, sexism, and socioeconomic disadvantages (Williams, 2005). For example, the double jeopardy thesis posits that African Americans have worse health outcomes at all ages (Ferraro & Farmer, 1996)—but being African American and old has compounding effects (Clark & Maddox, 1992). This draws upon the cumulative disadvantage thesis, which posits that health disparities increase with age (Dannefer, 2003; O'Rand, 1996). Alternatively, some scholars contend that age is a leveler because it erodes the harmful effects of disadvantages, reducing the race-based health gap with age or even creating a reversal among the oldest-old population (Dowd & Bengtson, 1978). In terms of health, some scholars have identified elderly African Americans as a more robust group than their younger peers (Gibson, 1991) or a survival elite (Kelley-Moore & Ferraro, 2004). Others have described a disability crossover in which African Americans in old age surpass similarly aged White peers in health and physical functioning (Clark & Maddox, 1992; Johnson, 2000).      According to Ferraro and Farmer (1996), “Another feature of the literature that merits further investigation is the extension of the concept of double jeopardy to triple jeopardy due to sexism” (p. 29). That is, inequalities associated with gender extend these ideas to the triple jeopardy hypothesis: African American women are most likely to experience the sharpest health declines during late life (Clark & Maddox, 1992). Yet, Clark and Maddox found that African American women had more modest physical health declines than African American men. Evidence on this matter remains inconclusive. For example, Manton (1980, 1988) contended that African American men who survive to old age do not have a lower life expectancy than White men. After age 75, in fact, African American men may have better life expectancy. Despite these advances in knowledge, the role of SES in triple jeopardy processes remains unclear.      Researchers have attributed race-linked health disparities partly to SES disadvantages (Williams, 2005). We examined the extent that SES—in combination with gender and race—influences the association between changes in limitations and depression. In order to frame our ideas, we propose the resource vulnerability versus resource erosion hypotheses. The resource vulnerability hypothesis is derived from the differential vulnerability perspective, which implies that lower SES groups tend to experience more deleterious effects of stressors because they have fewer resources to avoid or manage stressors (McLeod & Nonnemaker, 1999). These ideas suggest that limitations are associated with greater increases in depression among low-SES elders. Moreover, the multiple-hierarchy stratification perspective includes race and gender variants of the SES-based resource vulnerability hypothesis. Clark and Maddox (1992) asserted that “the greater probability of entering old age in poverty and with fewer years of formal education exemplifies and may add to the impact of minority status on the experience of aging, it has been argued, particularly for minority women” (p. S222). Applying these ideas, the gender variant predicts that limitations are associated with the largest increase in depression among low-SES women; the race variant predicts that increased limitations have the strongest effect among low-SES African Americans.      In contrast to the resource vulnerability view, it is plausible that high-SES groups have the most to lose with respect to health status. For example, Pampel and Rogers (2004) contended that “high socioeconomic status groups are harmed the most by unhealthy behaviors because, given their greater potential for good health, they have the most to lose from damaging lifestyles” (p. 307). African Americans and low-SES groups tend to experience lower life expectancy, higher morbidity and mortality, and worse health across the life course (Kelley-Moore & Ferraro, 2004). If Whites and high-SES groups are more likely to have experienced health advantages over the life course, then late-life exposure to limitations may be unexpected, unfamiliar, and challenging. By extension, although socioeconomic resources may enhance coping and reduce vulnerability to depression, limitations could undermine these resource benefits (the resource erosion hypothesis). Moreover, the resource erosion view includes race and gender variants, which researchers must consider simultaneously because White women have the highest life expectancy (80.0 years), followed by African American women (74.9), White men (74.8), and African American men (68.2) (Centers for Disease Control, 2002). Although women report poorer health than men on some measures, men die at younger ages from some conditions (Rieker & Bird, 2005). And although high-SES Whites tend to have better health than other race/ethnic groups over the life course, race may modify health patterns in late life. If elderly African Americans are indeed a healthier group than their White peers, then increased limitations are likely to be more depressing for high-SES Whites and least depressing for low-SES African American men.      In sum, we expected that increases in limitations would be associated with increases in depression. The differential vulnerability hypothesis proposes that individuals with low status are more vulnerable to limitations, whereas higher status is protective against the distressing effects of increased limitations. The multiple-stratification perspective underscores the intersections of race, gender, and SES. For example, race and gender variants of the differential vulnerability thesis would predict that the depressing effects of limitations would be strongest among low-SES elders, especially African Americans and women; we labeled this the resource vulnerability hypothesis (the converse is the resource benefit, in which higher statuses have protective effects). By contrast, the resource erosion hypothesis would contend that increased limitations would erode the resource benefits of those who have traditionally enjoyed higher status advantages (i.e., high-SES Whites).              Methods              Sample        The data in this sample derived from in-person interviews conducted in 2001–2002 with people 65 years and older residing in the District of Columbia and two adjoining Maryland counties (Prince George's and Montgomery). Sample selection and recruitment began with the Medicare beneficiary files for the three areas. In addition to the names of all people 65 years and older who are entitled to Medicare, the files provided information about the race and gender of each beneficiary. The next step entailed selection from the large pool of potential participants. To maximize the social and economic diversity, we randomly selected a total of 4,800 names equally divided among the three locales, African Americans and Whites, and women and men; this strategy created 12 groups, each containing 400 names. The goal was to enlist a sample of 1,200 people living independently, with approximately 100 in each of the 12 groups. Approximately 65% of eligible respondents (1,741) contacted agreed to participate, yielding 1,167 cases; eligible participants spoke English, lived in the community, and were able to independently complete the interview. Data collection occurred in three waves, each separated by approximately 12 months. Wave 1 interviews occurred during 2001–2002. For this article, we analyzed data from Waves 1 and 3 because of insufficient change in levels of depression between Waves 1 and 2. At Wave 3, which occurred approximately 2 years after the first interview, we reinterviewed 925 individuals (79%). We present results for individuals who were in Waves 1 and 3 and for whom we had complete responses to focal measures (N = 898).                    Measures        The Appendix presents the specific items used to measure depression, functional limitations, and diseases. Depression items asked about symptoms in the past 7 days. We averaged the items to create the index (αT1 =.770; αT3 =.764). We selected these items from the longer version of the widely used Hopkins Symptom Checklist (Derogatis, Lipman, Rickels, Uhlenhuth, & Covi 1974). Studies document that the depression score is correlated with major depression as defined by Diagnostic and Statistical Manual of Mental Disorders (American Psychiatric Association, 1994). We subtracted T1 scores from T3 scores in order to create the change in depression score.        The functional limitations items asked participants about the extent to which they experienced difficulties performing nine different activities of living (Katz, Ford, Moskowitz, Jackson, & Jaffee, 1963). Three additional questions, selected from well-known indices (see McDowell & Newell, 1996), asked about scenarios that required physical effort. We standardized and averaged all items to create the limitations index; higher scores indicate more limitations (αT1 =.920, αT3 =.898). We subtracted T1 scores from T3 scores in order to assess change in limitations.        In order to assess the joint effects of gender and race, we created three dummy variables (African American women, White women, and White men) with African American men as the contrast code because, as we described above, the literature identifies the experience of African American men in late life as the most different (Clark & Maddox, 1992; Johnson, 2000; Manton 1980, 1988).        We created an SES index by averaging standardized education and household income (α =.700). We coded education from low to high: eighth grade or less (1), some high school but did not graduate, high school graduate or general equivalency diploma, specialized training, some college but no degree, and college graduate or more (6). Household income (before taxes in the past year) included the participant's best estimate of all sources of income for household members from salaries, money market funds, Social Security, pensions, real estate, or government entitlements. Eleven categories ranged from low (less than $10,000) to high ($100,000 or more).        We coded marital status such that divorced, widowed, and never married persons were contrasted with married persons. We coded current/recent smoker as 1 if participants currently smoked or had smoked in the past 10 years, and 0 otherwise. We calculated body mass index as weight (kg) divided by height squared (cm2) based on the self-reported measures of weight (in pounds) and height (in inches). The disease index summed the number of health problems in the 5 years prior to the Wave 1 interview (see Appendix). In addition, the new diseases index summed the number of new diseases that occurred after the Wave 1 interview.                    Plan of Analysis        Using ordinary least squares regression techniques, we tested a series of models: (a) the stability model, which included only Wave 1 depression as the sole predictor of change; (b) the additive model, which included gender–race groups, baseline and change in limitations, and controls (excluding the interaction terms); (c) the model including the Gender–race × Change in limitations terms; and (d) the inclusion of other health measures. We included the fourth model to ensure that other health statuses and conditions did not account for our focal associations. In addition, we examined these models separately for low (< median) and high (> median) SES groups. Separate analyses (not shown) examined three-way interactions between gender–race groups, limitations, and SES to confirm their statistically significant effects; however, presenting all possible interactions in one table is logistically complex. Thus we present results only for the low- versus high-SES groups.        Following the procedures outlined by Mirowsky and Ross (2001), we used the difference score as the dependent variable (T3 depression minus T1 depression score) and adjusted for T1 depression in all models to control for regression to the mean. A common problem of longitudinal data analysis is that sample attrition may bias the results, especially if depression at baseline affects attrition. In order to account for this, we adjusted for the likelihood of attrition in all of our regression models. This adjustment took into account the probability that the baseline depression or any other of the T1 independent variables predicted attrition (Winship & Mare, 1992). Using logistic regression techniques, we found that baseline depression was unrelated to the odds of attrition. Limitations, low SES, and age, however, were associated positively with the odds of attrition; body mass index and diseases, paradoxically, were associated negatively with attrition. None of the other baseline variables were statistically significant. The model yielded a pseudo r-square of.05, indicating that almost all attrition was random with respect to the model's baseline measures (results available upon request).                    Results      Table 1 presents summary statistics for all of the focal measures across SES and gender–race groups. Table 2 presents the findings for the low-SES group. Model 1 presents the stability model, showing that baseline levels of depression were associated negatively with change in depression. Model 2 indicates that baseline levels and changes in functional limitations were associated with increases in depression, and levels of depression increased more for White women compared to African American men. However, we observed significant Gender–race × Change in limitations interactions in Model 3, indicating that increases in limitations were associated with increases in depression. That pattern was stronger among White women and men compared to African American men; tests for model improvement showed that the inclusion of these interactions improved model fit (F = 2.99, p <.05). In sum, increases in limitations were associated with concurrent increases in depression for White women and men only. Moreover, as Model 4 shows, controlling for other health measures only slightly altered the interactions.      Table 3 presents findings for the high-SES group. Model 1 indicates that baseline depression was associated negatively with change in depression over time. Model 2 indicates that baseline levels and changes in limitations were associated with increases in depression. In Model 3, we observed significant Gender–race × Change in limitations interactions that were different from those among low-SES elders. The negative White women × Change in limitations coefficient indicates that increases in limitations were associated with increases in depression more strongly among African American men compared to White women; however, tests for model improvement showed that the inclusion of these interactions marginally improved model fit (F = 2.17, p =.09). Adjustments for other health measures in Model 4 had little effect.      In sum, we observed that increases in limitations were associated with increases in depression; these patterns depended on gender, race, and SES. Among low-SES elders, increases in limitations were associated with increases in depression among White men and women only; among high-SES elders, increases in limitations were associated with increases in depression among African American women, African American men, and White men only. We have presented observations separately by SES for ease of interpretation; however, analyses (not shown) of three-way interactions supported the findings presented in Tables 2 and 3 (White men: t = 2.83, p <.01; White women: t = 2.14, p <.05; full analyses available upon request).              Discussion      Although the patterning of levels of functional limitations across dimensions of social stratification is well established, we expanded the scope of research by documenting the status-contingent effects of limitations on changes in depression. Drawing upon stress process theory, we proposed that social strata yield differential vulnerabilities with respect to the effects of limitations. We also used the multiple-hierarchy stratification perspective to examine race, gender, and SES contingencies. This allowed us to identify resource vulnerability and resource erosion variants of the SES-based differential vulnerability thesis across race–gender groups.      Among low-SES elders, we observed that increases in limitations were associated with concurrent increases in depression among White men and, to a lesser extent, White women. These observations are consistent with the resource erosion view in that Whites, irrespective of SES, may enjoy better health and fewer limitations over the life course than African Americans. With advancing age and increases in limitations, individuals may perceive threats to well-being as unfamiliar and unexpected. Moreover, the fact that limitations did not influence levels of depression among African Americans of low SES reinforces and extends the notion that elderly African Americans may be an especially robust group in terms of health and may reflect a survival elite. Our observations are also consistent with the idea of the disability crossover, in which African Americans in old age surpass similarly aged White peers in terms of health and physical functioning. According to Ferraro and Farmer (1996), “The advantages that majority persons held over minority persons may cease to be as important when all persons are confronted with the basic challenges to health and functional ability” (p. 28). Here, we identify the advantage with respect to the fact that, at the same level of increase in limitations, African Americans (especially men) reported lower levels of a concurrent increase in depression than White men and women. For lower SES individuals of minority status, being a member of the survivor elite may yield psychosocial benefits that counter the depressing consequences of impairment.      We also found evidence consistent with the resource erosion view of SES. This view implies that high SES should buffer against the distressing effects of increasing limitations. The resource erosion dimension of increasing levels of limitations, however, may overwhelm the protective benefits of high SES. Sooner or later, even people who have enjoyed health advantages over the life course experience health problems or impairment (House et al., 2005). The question becomes: At the same level of limitations, what are the mental health effects for individuals who are familiar with prior health adversities (normative) compared to those who are unfamiliar with health stressors (non-normative)? The stability of health inequalities idea implies that the mental health impact of limitations is worse for people who have enjoyed health advantages over the life course—in this case, people with high SES. Thus, there may be an advantage resource margin that compresses during late-life in terms of morbidity and limitations (House et al., 2005), as well as the deleterious mental health correlates of limitations. Our findings among high-SES African American women, African American men, and White men are consistent with the resource erosion view: people with higher SES tend to experience more advantages with respect to health over the life course. For them, the sudden exposure to limitations may be unexpected, unfamiliar, and challenging. By contrast, our observations for high-SES White women are more consistent with the resource benefits view: higher SES is protective against the distressing effects of limitations.      Several limitations of our study deserve brief mention. First, the small sample sizes of specific subgroups suggest caution in the interpretation of our estimates. Moreover, there may be reporting biases across these subgroups that we are unable to assess in the present study. The regional nature of the sample is also a potential limitation. On the one hand, the metropolitan District of Columbia area likely contains a slightly higher number of well-educated, wealthier individuals. This is particularly the case for elderly African American women and men in Prince George's County. On the other hand, this is also a potential strength of the data set, because it allows us to examine health and well-being processes among a group of African American elders that represents a solid cross-section of the socioeconomic spectrum. In addition, the low r-square suggests that much of the variance in changes in depression remains unexplained. This is hardly unique to this study, but it deserves brief mention here and attention in future investigations. Finally, the short duration of time between interviews limited the degree of change in focal measures. Future plans for additional interviews of these participants, however, should help to expand the scope of the present study to include growth-curve models of changes in functioning over a longer time span.              Conclusion      The stress process framework identifies functional limitations as a stressor in late life. We drew upon an array of theoretical and empirical views to propose that the increased limitations have different psychological effects that vary across social strata. Our observations underscore the necessity of using a multiple-hierarchy stratification perspective to examine race, gender, and SES contingencies to refine and extend current knowledge about the effects of limitations.                                      Decision Editor: Kenneth F. Ferraro, PhD                          Table 1.                      Means or Proportions for All Study Variables Across Socioeconomic Status and Race–Gender Groups.                                                                              Socioeconomic Status < Median                                                                Socioeconomic Status > Median                                                                                            Variable                African American Women (n = 156)                African American Men (n = 119)                White Women (n = 112)                White Men (n = 54)                African American Women (n = 62)                African American Men (n = 97)                White Women (n = 117)                White Men (n = 181)                                                                    Change in depression                −0.018 (0.552)                −0.075 (0.634)                0.061 (0.472)                0.089 (0.440)                0.044 (0.512)                0.065 (0.365)                0.015 (0.532)                0.015 (0.442)                                            Baseline depression                1.480 (0.526)                1.459 (0.603)                1.515 (0.610)                1.349 (0.416)                1.390 (0.476)                1.253b (0.375)                1.480c (0.504)                1.324 (0.403)                                            Baseline functional limitations                0.186c (0.900)                −0.057b (0.803)                0.283c (0.956)                −0.231 (0.756)                −0.037 (0.741)                −0.345a (0.752)                −0.215 (0.556)                −0.294 (0.584)                                            Change in functional limitations                0.088 (0.694)                0.160 (0.831)                0.132 (0.646)                0.132 (0.589)                0.081 (0.368)                0.061 (0.589)                0.052 (0.462)                −0.011 (0.508)                                            Age                73.750 (6.802)                73.403 (5.419)                76.152 (6.921)                75.556 (5.971)                72.661 (5.572)                72.082c (5.063)                73.632 (5.757)                74.376 (6.551)                                            Married                0.244c                0.555a,b                0.330c                0.685                0.435c                0.814a,b                0.538c                0.790                                            Divorced/separated                0.224c                0.168                0.107                0.056                0.177c                0.062                0.120                0.050                                            Widowed                0.481c                0.244a,b                0.509c                0.204                0.371c                0.113a                0.248c                0.088                                            Never married                0.051                0.034                0.054                0.056                0.016                0.010                0.094                0.072                                            Current/recent smoker                0.141                0.319a,b,c                0.143                0.130                0.113                0.082                0.103                0.072                                            Body mass index                29.295b (5.264)                27.200a (4.551)                26.697 (6.525)                27.346 (3.852)                28.549b,c (6.182)                27.451b (3.826)                24.890 (4.581)                25.983 (3.342)                                            Baseline diseases                2.256 (1.463)                2.176 (1.516)                2.357 (1.361)                2.148 (1.535)                2.274 (1.473)                1.918 (1.441)                2.068 (1.400)                1.796 (1.413)                                            New diseases                1.686 (1.463)                1.387 (1.491)                1.616 (1.390)                1.444 (1.436)                1.290 (1.311)                1.165 (1.336)                1.350 (1.347)                1.155 (1.255)                                                                        Notes: Standard deviations are shown in parentheses.                                      aSignificantly different from African American women (p <.05).                                      bSignificantly different from White women (p <.05).                                      cSignificantly different from White men (p <.05).                                                Table 2.                      Regression of Change in Depression Among Individuals in the Low-SES Group (n = 441).                                                              Variable                Model 1                Model 2                Model 3                Model 4                                                                    Focal associations                                                                                                                Baseline depression                −.557*** (.038)                −.611*** (.039)                −.622*** (.039)                −.641*** (.040)                                                African American womena                                .044 (.055)                .028 (.055)                .025 (.057)                                                White womena                                .138* (.059)                .115 (.060)                .111 (.060)                                                White mena                                .129 (.072)                .089 (.073)                .086 (.073)                                                Baseline functional limitations                                .111*** (.027)                .117*** (.027)                .087** (.029)                                                Change in functional limitations                                .133*** (.031)                .047 (.048)                .045 (.049)                                                African American women × Change in functional limitationsa                                                .085 (.070)                .072 (.070)                                                White women × Change in functional limitationsa                                                .162* (.081)                .142 (.081)                                                White men × Change in functional limitationsa                                                .298*** (.112)                .258* (.112)                                            Basic control measures                                                                                                                Age                                −.001 (.004)                −.002 (.004)                −.001 (.004)                                                Divorcedb                                .081 (.064)                .090 (.063)                .107 (.064)                                                Widowedb                                .012 (.051)                .013 (.050)                .025 (.051)                                                Never marriedb                                .112 (.101)                .124 (.101)                .135 (.101)                                            Health measures                                                                                                                Previous or recent smoker                                                                −.004 (.056)                                                Body mass index                                                                −.003 (.004)                                                Baseline diseases                                                                .019 (.017)                                                New diseases                                                                .037* (.017)                                            Constant                .817                .780                .807                .814                                                              R                  2                                .325                .382                .395                .409                                                                        Notes: Data are presented as unstandardized regression coefficients with standardized coefficients in parentheses. Models show regression of change in depression on baseline levels of depression (1); race–gender groups, functional limitations, and basic controls (2); interactions (3); and other health statuses and conditions (4).                                      aCompared to African American men.                                      bCompared to currently married.                                      *p <.05; **p <.01; ***p <.001 (two-tailed).                                                Table 3.                      Regression of Change in Depression Among Individuals in the High-SES Group (n = 457).                                                              Variable                Model 1                Model 2                Model 3                Model 4                                                                    Focal associations                                                                                                                Baseline depression                −.398*** (.045)                −.496*** (.045)                −.499*** (.045)                −.511*** (.046)                                                African American womena                                −.040 (.068)                −.039 (.069)                −.034* (.068)                                                White womena                                .021 (.057)                .036 (.057)                .029 (.058)                                                White mena                                −.017 (.051)                −.013 (.051)                −.016 (.051)                                                Baseline functional limitations                                .174*** (.033)                .173*** (.033)                .172*** (.035)                                                Change in functional limitations                                .212*** (.040)                .295*** (.070)                .280*** (.071)                                                African American women × Change in functional limitationsa                                                −.028 (.156)                −.016 (.158)                                                White women × Change in functional limitationsa                                                −.259* (.106)                −.259*(.107)                                                White men × Change in functional limitationsa                                                −.060 (.091)                −.069 (.091)                                            Basic control measures                                                                                                                Age                                .005 (.004)                .005 (.003)                .005 (.004)                                                Divorcedb                                .198** (.069)                .213** (.069)                .202** (.070)                                                Widowedb                                .012 (.054)                .005 (.054)                .004 (.054)                                                Never marriedb                                .004 (.083)                .008 (.083)                −.001 (.083)                                            Health measures                                                                                                                Previous or recent smoker                                                                .039 (.068)                                                Body mass index                                                                −.002 (.005)                                                Baseline diseases                                                                −.016 (.015)                                                New diseases                                                                .039* (.016)                                            Constant                .570                .731                .730                .794                                                              R                  2                                .146                .266                .277                .288                                                                        Notes: Data are presented as unstandardized regression coefficients with standardized coefficients in parentheses. Models show regression of change in depression on baseline levels of depression (1); race–gender groups, functional limitations, and basic controls (2); interactions (3); and other health statuses and conditions (4).                                      aCompared to African American men.                                      bCompared to currently married persons.                                      *p <.05; **p <.01; ***p <.001 (two-tailed).                                                            Appendix                                Survey Items.                                                              Item Wording                Response Categories                                                                    Depression Index                                                                Lack enthusiasm for doing anything                No days (1)                                                Feel bored or have little interest in things                1 or 2 days (2)                                                Cry easily or feel like crying                3 or 4 days (3)                                                Feel downhearted or blue                5 or more days (4)                                                Feel slowed down or low in energy                                                                Blame yourself for everything that goes wrong                                                                Have your feelings hurt easily                                                            Functional Limitations Index                                                                Level of difficulty in:                                                                    Dressing and undressing                Without difficulty (1)                                                    Getting in and out of bed                With difficulty, but without help (2)                                                    Taking a bath or shower                With a little help from someone (3)                                                    Getting to and using the toilet                Unable to do this without complete help from someone or special equipment (4)                                                    Climbing up stairs                                                                    Keeping balance while walking                                                                    Going food shopping                                                                    Getting from your home to where you need to go                                                                    Figuring out your own monthly bills                                                                Let's suppose that you had to reach over your head to lower a bag of sugar. What is the heaviest bag of sugar you could lower?                15 or more pound bag (1)                                                            10 pound bag (2)                                                            5 pound bag (3)                                                            1 pound bag (4)                                                            Not at all able to lower the bag (5)                                                Let's suppose that you had to stand, without help, in a long line. About how long could you stand?                1 hr or more (1)                                                            45 min (2)                                                            30 min (3)                                                            15 min (4)                                                            5 min (5)                                                            Not at all (6)                                                How long are you able to walk without stopping to rest?                1 hr or more (1)                                                            45 min (2)                                                            30 min (3)                                                            15 min (4)                                                            5 min (5)                                                            Not at all (6)                                            Disease Index                                                                Asthma or emphysema                No (0)                                                Arthritis                Yes (1)                                                Diabetes                                                                High blood pressure                                                                Heart disease and/or heart attack                                                                Stomach disorders                                                                Stroke                                                                Cancer of any kind                                                                Osteoporosis                                                                High cholesterol                                                                Painful joints                                                                Cataracts, glaucoma, detached retina, or any other condition of the retina                                                                                    An NIA grant award AG17461 (Leonard I. Pearlin, P.I.) supports this work. Address correspondence to Scott Schieman, PhD, University of Toronto, Department of Sociology, 725 Spadina Avenue, Toronto, Ontario M5S 2J4, Canada. E-mail: scott.schieman@utoronto.ca.              References              American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.                    Centers for Disease Control. (2002). United States life tables, 2000. National Vital Statistics Reports, 51(3).                    Clark, D. O., Maddox, G. L. (1992). Racial and social correlates of age-related changes in functioning. Journal of Gerontology: Social Sciences, 47B,S222-S232.                    Dannefer, D. (2003). Cumulative advantage/disadvantage and the life course: Cross-fertilizing age and social science theory. Journal of Gerontology: Social Sciences, 58B,S327-S357.                    Derogatis, L. R., Lipman, R., Rickels, K., Uhlenhuth, E. H., Covi, L. (1974). The Hopkins Symptom Checklist (HSCL): A self-report symptom inventory. Behavioural Science, 19,1-15.                    Dowd, J. J., Bengtson, V. L. (1978). Aging in minority populations: An examination of the double jeopardy hypothesis. Journal of Gerontology, 33,427-436.                    Ferraro, K. F., Farmer, M. M. (1996). Double jeopardy to health hypothesis for African Americans: Analysis and critique. Journal of Health and Social Behavior, 37,27-43.                    Gibson, R. C. (1991). Age-by-race differences in the health and functioning of eldely persons. Journal of Aging and Health, 3,335-351.                    House, J. S., Lantz, P. M., Herd, P. (2005). Continuity and change in the social stratification of aging and health over the life course: Evidence from a nationally representative longitudinal study from 1986 to 2001/2002 (Americans' Changing Lives Study). Journal of Gerontology: Social Sciences, 60B,S15-S26.                    Johnson, N. E. (2000). The racial crossover in comorbidity, disability, and mortality. Demography, 37,267-283.                    Katz, S., Ford, A. B., Moskowitz, R. W., Jackson, B. A., Jaffee, M. W. (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.                    Kelley-Moore, J. A., Ferraro, K. F. (2004). The Black/White disability gap: Persistent inequality in late-life? Journal of Gerontology: Social Sciences, 59B,S34-S43.                    Kelley-Moore, J. A., Ferraro, K. F. (2005). A 3-D model of health decline: Disease, disability, and depression among Black and White older adults. Journal of Health and Social Behavior, 46,376-391.                    Long, J. S., Pavalko, E. K. (2004). Comparing alternative measures of functional limitations. Medical Care, 42,19-27.                    Manton, K. (1980). Sex and race specific mortality differentials in multi-cause of death data. The Gerontologist, 20,480-493.                    Manton, K. (1988). A longitudinal study of functional change and mortality in the United States. Journal of Gerontology: Social Sciences, 43,S153-S161.                    McDowell, I., Newell, C. (1996). Measuring health: A guide to rating scales and questionnaires. New York: Oxford University Press.                    McLeod, J. D., Nonnemaker, J. M. (1999). Social stratification and inequality. In C. S. Aneshensel & J. C. Phelan (Eds.), Handbook of the sociology of mental health (pp. 321–344). New York: Kluwer Academic/Plenum.                    Mirowsky, J., Ross, C. E. (2001). Age and the effect of economic hardship on depression. Journal of Health and Social Behavior, 42,132-150.                    O'Rand, A. M. (1996). The precious and the precocious: Understanding cumulative disadvantage and cumulative advantage over the life course. The Gerontologist, 36,230-238.                    Pampel, F. C., Rogers, R. G. (2004). Socioeconomic status, smoking, and health: A test of competing theories of cumulative advantage. Journal of Health and Social Behavior, 45,306-321.                    Pearlin, L. I., Skaff, M. M. (1996). Stress and the life course: A paradigmatic alliance. The Gerontologist, 36,239-247.                    Rieker, P. P., Bird, C. E. (2005). Rethinking gender differences in health: Why we need to integrate social and biological perspectives. Journal of Gerontology: Social Sciences, 60B,S40-S47.                    Williams, D. R. (2005). The health of U.S. racial and ethnic populations. Journal of Gerontology: Social Sciences, 60B,S53-S63.                    Winship, C., Mare, R. D. (1992). Models for sample selection bias. Annual Review of Sociology, 18,327-350.                    Yang, Y., George, L. K. (2005). Functional disability, disability transitions, and depressive symptoms in late life. Journal of Aging and Health, 17,263-292.