A Conceptual Matrix of the Temporal and Spatial Dimensions of Socioeconomic Status and Their Relationship with Health

A Conceptual Matrix of the Temporal and Spatial Dimensions of Socioeconomic Status and Their... Abstract Objectives In this study, we (a) draw on fundamental cause theory, the life course perspective, and neighborhood effects to develop conceptual matrix of socioeconomic status (SES) by temporal and spatial dimensions in order to highlight the multidimensional ways in which SES relates to general health, and then (b) assess the multidimensional ways in which income (as a measure of SES) is related to disability in adulthood. Methods Data from the Panel Study of Income Dynamics were linked with Census data to assess (a) which temporal and spatial dimensions of income were associated with disability in adulthood, and (b) whether the various components of income interact with each other when predicting disability. Results Negative binomial regression results indicated both 1970 and 2013 household income were associated with lower levels of disabilities in adulthood, as was 2013 neighborhood-level income, but 1970 neighborhood-level income was not associated with disability in adulthood. Further, 4 of the 6 possible interactions between the multiple dimensions of income were associated with significant reductions in adult disability. Discussion These findings provide several important empirical insights, but also help inform a framework for thinking about the multidimensional ways in which SES relates to health. Socioeconomic status, Disability, Life course, Neighborhood Background According to fundamental cause theory (Link & Phelan, 1995), SES-related resources are important when they may be accessed to help prevent health risks, and reduce the consequences of poor health (Link & Phelan, 1995). Therefore, those with higher SES are better positioned to avoid health problems through the deployment of knowledge, money, support networks, and psychosocial coping resources (Phelan, Link, & Tehranifar, 2010). For instance, those with less education are at greater risk of earlier onset of disability, and this increased risk is attenuated by variations in income (House et al., 1994). Thus, SES clearly represents a primary social determinant of health (Wadsworth, Butterworth, Marmot, & Wilkinson, 2006). However, SES represents a multidimensional construct involving both time and space, and the relationship between SES and health likely varies across different dimensions of SES. Little is known, however, about the combined way the temporal and spatial dimensions of SES relate to health. If the association between SES and health does indeed vary by time and space, this is important for social scientists and policy makers developing prevention–intervention efforts aimed at improving population health. Therefore, in the following sections, first, we review the literature regarding how the temporal and spatial dimensions of SES relate to adult health, and we articulate a conceptual matrix of SES in order to highlight the multidimensional ways in which SES relates to health. Second, we apply the conceptual matrix of SES to the empirical case of the relationship between household income (as a measure of SES) and disability (as a measure of health). Conceptual Matrix of SES by Temporal and spatial Dimensions In this section, we offer a conceptual matrix of the multidimensional ways in which SES may relate to health. The purpose of this conceptual matrix is to provide a guide to (a) understand the multidimensional ways in which SES relates to health in general, and (b) to specifically situate the present study, in which we are assessing how income (as an indicator of SES) is related to disability (an example of physical health). The goal here is to combine insights from fundamental cause theory, the life course perspective, and neighborhood effects to develop a unified typology for how SES relates to health. Link and Phelan (1995:p 87) argue “the essential feature of fundamental social causes, is that they involve access to resources that can be used to avoid risks or to minimize the consequences of disease once it occurs.” To this, we add a caveat, informed by the life course perspective. Although we agree that one way SES relates to health is through the ability to access current economic resources, we think it is additionally important to acknowledge individuals’ past circumstances relating to SES (i.e., their socioeconomic history). Socioeconomic history is distinguished from current SES because historical resources are no longer accessible, yet they continue to shape health. Indeed, according to life course sociology, current health conditions are a product of both proximal and distal antecedents (Elder, 1999; Ferraro, 2011; Ferraro & Shippee, 2009). One’s current circumstances are certainly important for health (proximal), but current circumstances are also shaped by one’s history of prior circumstances (distal). Furthermore, according to cumulative inequality theory, childhood is part of the past that tends to influence people’s health as adults (Ferraro, 2011; Ferraro & Shippee, 2009; Ferraro, Shippee, & Schafer, 2009). Socioeconomic status during childhood/adolescence provides the foundation for early life course development and circumstances that are related to a wide array of adult health conditions (Haas, 2008; Poulton et al., 2002). As shown in Figure 1, in our temporal dimension of SES, SES may be delineated into both (a) socioeconomic present and (b) socioeconomic history. Socioeconomic present represents SES one currently has at their disposal. Past resources, however, are qualitatively difference than current resources—in large part because they are no longer accessible to use to help reduce health risks or mitigate the consequences of poor health. Thus, we define socioeconomic history, as representative of an individual’s SES history (or part of their SES history). Socioeconomic history is importantly different from an individual’s socioeconomic present, because the conditions of socioeconomic history (a) are no longer accessible, but (b) nevertheless play an important role in adult health because they influence people’s health behaviors and health trajectories. Figure 1. View largeDownload slide Conceptual matrix of SES by temporal and spatial dimensions. NOTE: The examples of SES (i.e., income) in the grey boxes are just that: examples that fit within the typology defined by the matrix. These particular examples were selected because they are the variables used in the present study. We believe this matrix would be useful for studying other forms of SES as well (e.g., parental education and individual education along the temporal dimension, or city income per capita and state income per capita along the spatial level). Figure 1. View largeDownload slide Conceptual matrix of SES by temporal and spatial dimensions. NOTE: The examples of SES (i.e., income) in the grey boxes are just that: examples that fit within the typology defined by the matrix. These particular examples were selected because they are the variables used in the present study. We believe this matrix would be useful for studying other forms of SES as well (e.g., parental education and individual education along the temporal dimension, or city income per capita and state income per capita along the spatial level). Taken together, our model offers the following assumption about the temporal relationship between SES and health: Assumption #1: Current health is a function of both (i) an individual’s socioeconomic present, because these accessible resources may help mitigate health risks and poor health, but also (ii) an individual’s socioeconomic history, because this history shapes people’s past opportunities, behaviors, and decisions in such a way that matter for current health. In addition to the temporal dimension, the relationship between SES and health also involves a spatial dimension. Investigations into neighborhood effects (Sampson, Morenoff, & Gannon-Rowley, 2002), have identified four overarching mechanisms linking neighborhood context to health behaviors and outcomes. These include (a) institutional mechanisms; (b) social interactive mechanisms; (c) geographical mechanisms; and (d) environmental mechanisms (Galster, 2012). Within these overarching mechanisms, Galster (2012) articulates a total of 15 specific mechanisms linking neighborhoods to health. Examples of institutional mechanisms include the extent to which the neighborhood provides access to institutions like good schools and medical centers, and the extent to which there are public stereotypes held about the residents. Examples of social-interactive mechanisms include the degree of social control within the neighborhood, neighborhood-based social networks that individuals can call upon, and the collective norms and values of the neighborhood. An example of a geographical mechanism is access to employment opportunities (e.g., the geographic feasibility of transportation to and from work). Finally, examples of environmental mechanisms include the extent to which a neighborhood exposes residence to violence, deteriorated structures, and environmental toxins (for a full list of the 15 mechanisms, see Galster (2012)). In short, neighborhood socioeconomic conditions may both expose people to health risks but also provide access to health related resources. On one hand, economically-disadvantaged neighborhoods might expose individuals to health risks like high rates of poverty, unemployment, and crime (Aneshensel & Sucoff, 1996; Cutrona, Wallace, & Wesner, 2006; Sampson et al., 2002). They may also expose people to social conditions that negatively impact health, for example, by limiting the opportunities for social integration or limiting access to resource-dense social networks (Taylor, Repetti, & Seeman, 1997). On the other hand, wealthier neighborhoods may provide people with access to important health resources like health care options, quality transportation, and places to exercise (Ellen, Mijanovich, & Dillman, 2001). Wealthier neighborhoods may also promote more health enhancing social norms, better public services because of a larger per capita tax base, and include powerful social actors who are positioned and knowledgeable enough to advocate on behalf of the neighborhood (Galster, 2012). Thus, neighborhood-level SES may have both positive and negative associations with health. As shown in Figure 1, following Ross and Mirowsky (2008), here we delineate between what we are calling individual resources and contextual resources. Individual resources refers to the SES resources and circumstances possessed by an individual (or an individual’s household), while contextual resources represents resources and circumstances shaped by the environment in which a person resides. Although our focus is on neighborhoods, people find themselves embedded in a variety of spatial contexts (e.g., states, counties, or countries) that shape their circumstances and resources, which subsequently impact their health (Marmot, 2015; Ross & Mirowsky, 2008). This spatial distinction is generally buttressed by empirical evidence. For instance, low neighborhood-level SES is associated with poor physical functioning, independent of individual-level SES (Pickett & Pearl, 2001; Tomey, Diez Roux, Clarke, & Seeman, 2013). However, individual-level SES is associated with physical functioning, net of neighborhood SES (Stafford, Gimeno, & Marmot, 2008). Taken together, our model offers the following assumption about the spatial relationship between SES and health: Assumption #2: Current health is a function of both (i) individual resources, because these individual resources are important for mitigating one’s health risks and poor health, but also (ii) an individual’s contextual resources, because the context within which an individual resides provides access to health opportunities and exposes individuals to health risk factors. In addition to direct relationships between socioeconomic history, socioeconomic present, individual resources, and contextual resources with current health, we further suspect that these four domains interact with one another to shape current adult health. For instance, childhood circumstances do not preordain specific later life health. While many adults grow up to have similar class backgrounds to those in which they grew up—this is not case for all adults (Bloome & Western, 2011; Hout & Beller, 2006). Similarly, the correlation between neighborhood SES in childhood and neighborhood SES in adulthood is modest enough to suggest there is variation in the kinds of neighborhoods people live in during childhood and adulthood (Clarke et al., 2014; Wheaton & Clarke, 2003). In this context, adult socioeconomic status may be thought of as a resource that may be mobilized to offset the exposure to poor socioeconomic conditions in childhood (Ferraro & Shippee, 2009). Similarly, although there is a nontrivial amount of income segregation in the United States, there is also a considerable amount of within neighborhood income variation as well (Reardon & Bischoff, 2011). Thus, the health benefits of living in a wealthy neighborhood are likely different for someone with relatively lower personal income, and, conversely, the health consequences of living in a poorer neighborhood are likely different for someone with relatively greater personal income. Moreover, there is evidence the relationship between individual-level SES and neighborhood-level SES on health outcomes are not merely additive, but combine in a multiplicative fashion. For instance, the association between childhood neighborhood disadvantage and mental health varies across levels of parental education, such that effect of neighborhood disadvantage is greatest among adults whose parents had lower levels of education (Wheaton & Clarke, 2003). Thus, our model offers a final assumption about the spatial relationship between SES and health: Assumption #3: The relationship within and between the temporal and spatial dimensions of SES and health are not just additive, but rather, interact with each other in a multiplicative fashion to shape adult health. To evaluate our conceptual matrix of SES, in the following sections, we bring our model to bear on the empirical case of the ways that household and neighborhood-level income in both childhood and adulthood relate to level of disability in adulthood. Assessing the Conceptual Matrix of SES: The Empirical Case of Household Income and Disability A significant share of the U.S. adult population experiences physical disability during their lives. According to Census estimates, 17% of nonelderly adults live with some form of physical disability (Brault, 2012). Moreover, an estimated 12% of the federal budget is spent on nonelderly adults with disabilities (Brault, 2012; Livermore, Stapleton, & O’Toole, 2011). Therefore, both in terms of personal and societal costs, physical disability is an important feature in the lives of many adults in the United States. Disability can be conceptualized as “the gap between a person’s intrinsic capabilities and the demands created by the social and physical environment—a product of the interaction of the individual with the environment” (Jette, 2006). That is, disability represents how a person’s physical condition limits their ability to perform the social role requirements necessary for their daily lives. The typical measurement of disability involves asking respondents how much they are limited in doing a particular task (e.g., using the restroom, preparing a meal, etc.) (Beauchamp et al., 2015). In order to understand and limit the prospects of developing a disability, and to potentially identify interventions to reduce these prospects, it is important to identify the factors that are associated with elevated risk of becoming disabled (Seeman, Bruce, & McAvay, 1996). Although there are numerous risk factors associated with disability, one of the primary risk factors for becoming disabled is low socioeconomic status (SES) (Antonovsky, 1967; Link & Phelan, 1995). In the following sections, we consider how SES may relate to disability across temporal and spatial dimensions according to our conceptual matrix of SES. SES and disability: The temporal dimension According to the conceptual matrix of SES, both (a) socioeconomic history and (b) socioeconomic present should shape the extent to which one experiences disability in adulthood, and this is largely borne out by the empirical evidence. On one hand, early life SES is related to a wide array of adult health conditions, including disability (Haas, 2008; Link & Phelan, 1995; Montez, 2013; Poulton et al., 2002). Moreover, childhood and adolescence are periods where nutrition and physical activity are important for the development of bone mass, and related to subsequent “risk of fracture and osteoporosis” (Cohen, Janicki-Deverts, Chen, & Matthews, 2010)—factors relevant to disability in adulthood. Because the socioeconomic circumstances to which one is exposed in childhood are related to late life disability, we developed our first hypothesis: Hypothesis 1: There will be a negative association between childhood income and level of disability in adulthood, net of income in adulthood. On the other hand, adult SES is clearly related to level of disability in adulthood (Haas, 2008; House et al., 1990; Montez, 2013). Lower levels of SES in general, and income in particular, are associated with greater levels of disability in adulthood (House et al., 1990). Inasmuch as income in adulthood is robustly linked to disability, we developed our second hypothesis: Hypothesis 2: There will be a negative association between adulthood income and level of disability in adulthood, net of income in childhood. SES and disability: The spatial dimension According to the conceptual matrix of SES, both (a) individual resources and (b) contextual resources should shape the extent to which one experiences disability in adulthood. In general, this is supported by the empirical evidence. On one hand, household-level SES can both protect against disability, but the lack of SES resources can be a risk factor for disability. Those with higher SES can use more available resources to manage and overcome life hardships, as well as access higher quality medical treatment and preventative care (Freedman & Martin, 1999; Phelan et al., 2010). Conversely, having lower household-level SES can be a financial stress which, in turn, can have negative consequences for poor physical health (Robert, 1998; Skinner, Zautra, & Reich, 2004; Wheaton, 1994). The link between household SES and disability is well established (Adler et al., 1994; Kail & Taylor, 2014; Nagi, 1976; Taylor, 2010). For example, lower individual-level SES is associated greater disability (Gayman, Turner, & Cui, 2008). Similarly, individual-level SES is associated with greater risk of becoming functionally limited, initial level of functional limitation, and subsequent growth of disability over time (Kail & Taylor, 2014). Clearly, individual-level SES is robustly linked with physical health. As such, based on this empirical evidence and our conceptual matrix of SES, we developed our third hypothesis: Hypothesis 3: There will be a negative association between household-level income and adult disability, net of neighborhood-level income. On the other hand, at the neighborhood-level, socioeconomic position shapes the very conditions in which one lives (Phelan et al., 2010). Indeed, neighborhoods represent the physical and social spaces in which adults spend much of their nonworking lives. Because different neighborhoods provide access to resources and expose residents to various risk and protective factors, neighborhoods represent another important spatial dimension of SES. Indeed, neighborhood-level SES is widely linked to physical health and disability (Diez Roux & Mair, 2010; Robert, 1998; Ross & Mirowsky, 2001). For example, poor neighborhoods are associated with declining physical functioning (Balfour & Kaplan, 2002; Schootman, Jeffe, Baker, & Walker, 2006). Neighborhoods involving high levels of economic strain may have negative health consequences by reducing the availability of health promoting resources and behaviors like physical exercise (Diez Roux & Mair, 2010; Phelan et al., 2010). Living in economically deprived neighborhoods may also produce stress which increases the risk of illness by threatening the immune system and accelerating the aging process (Segerstrom & Miller, 2004). Although the particular linking mechanisms are beyond the scope of the current study, there is evidence neighborhood SES has important implications for disability. Thus, based on our conceptual matrix of SES and existing empirical evidence, we developed our fourth hypothesis: Hypothesis 4: There will be a negative association between neighborhood-level income and disability, net of household-level. Finally, there is limited evidence to suggest that various dimensions of SES interact to shape physical functioning among adults (Montez, 2013). Therefore, based on this evidence, and the assumptions of the interactive nature of SES upon which our matrix of SES is predicated, we derive our final hypothesis, specifically: Hypothesis 5: The association between any one dimension of income and disability in adulthood will be conditioned by the other dimensions of income. In summary, based on insights from fundamental cause theory, the life course perspective, and neighborhood effects we have provided a Conceptual Matrix of SES by Temporal and Spatial Dimensions. We believe (1) the association between SES and health (in this case, the association between income and disability) involves (a) a temporal dimension involving (i) socioeconomic history and (ii) socioeconomic present, and (b) a spatial dimension involving (iii) individual resources and (iv) contextual resources, and (2) these four dimensions interact with one another to help shape adult health. Data And Method This study uses data from the Panel Study of Income Dynamics (2013) (PSID) for the years 1970 and 2013. The PSID is a nationally-representative, longitudinal study of U.S. families beginning in 1968 and has continued to follow sample members and their descendants over time. Our empirical objective is to evaluate the impact of adult and childhood income on disability in adulthood. Thus, for our analysis, we selected respondents who were under age 18 in 1970, and who were also in the sample as adults in 2013. Our final sample includes 2,344 individuals, who ranged in age from 1 to 17 in 1970, and 42 to 63 in 2013. Of the 7,590 children in the 1970 PSID sample, approximately 33% remain in the sample in 2013. The remainder dropped out of the sample, mostly because of refusal/nonresponse, or because of a PSID sample reduction occurring in 1997. The main sources of attrition in our sample are refusal/nonresponse (77% of those who dropped out), a PSID sample reduction occurring in 1997 (22%), death (0.6%), and jail/prison (0.2%). Although the PSID has maintained a reinterview rate between 95% and 98% across virtually all survey waves, even small attrition from wave to wave accumulates over time (Panel Study of Income Dynamics, 2013). Assessments of the representativeness of the PSID find that income and health estimates align closely with other cross-sectional surveys, despite accumulating attrition (Schoeni, Stafford, McGonagle, & Andreski, 2013). We conducted our own investigation of respondents who drop out of our sample. We found nonrespondents have 1970 family incomes that are about 20% lower than respondents who remain in the sample; childhood health among these two groups is roughly equivalent (based on two-tailed t tests comparing 1970 family income and reports of poor health in childhood among respondents and nonrespondents). We further restrict our analysis to respondents designated as household heads or spouses in 2013, because questions on disability were only asked of these household members. This eliminates 91 sample members, all of whom were residing in the home of other relatives or nonrelatives. Most commonly, they were residing the home of their parents. Second most commonly, they were residing in the home of their adult children. Further investigation reveals these sample members have family incomes that were about 18% lower in 1970 and 40% lower in 2013 than those who are retained in our sample (based on two-tailed t tests comparing 1970 and 2013 family incomes among 2013 heads/wives and non-heads/wives). Given the selectivity of our sample, it is important to note that results may not be representative of families with low incomes. Our final sample includes 2,344 individuals, who ranged in age from 1 to 17 in 1970, and 42 to 63 in 2013. Discrepancies between 1970 and 2013 age arise based on the timing of interviews. Ages of individuals are asked and reported in each wave of the study. Because interviews are seldom taken exactly 1 year apart for the same individual wave to wave, age gaps can arise. Also, individuals’ ages or birthdates can be misreported. The PSID conducts internal consistency checks for age discrepancies, but they do no alter ages if it cannot be determined which age is correct. Dependent Variable Our outcome variable is disability in adulthood. Disability is measured in 2013 with six items asking whether respondents had any difficulty doing the following activities by themselves and without special equipment: using the toilet; getting outside; walking; getting in and out of bed; eating; and dressing. Responses were coded as either “yes” or “no,” and then summed into an index of “yes” responses. Thirty-seven respondents were missing values on some or all of the disability items; their values were multiply imputed. Study Variables Socioeconomic status (SES) is measured at two time points—childhood and adult—and at two levels—household and neighborhood. SES is a construct representing an individual’s economic and social position, generally based on income, education, and occupation. Because income, education, and occupation are highly correlated, we rely on a single indicator—income—to represent SES in our study. Household SES is measured in 1970 (childhood) and in 2013 (adulthood) as a continuous measure representing the total income of all family members in the previous year including all taxable, transfer, and social security income. Income is self-reported by the household head for all family members. In 1970, negative values (i.e., net losses) were bottom coded at $1. To be consistent, we also recode negative values in 2013 to $1. All results are reported in real 1970 and 2013 dollars, and are not adjusted for inflation. Data exploration revealed a significant right skew produced by a few families with exceedingly high incomes. To adjust for non-normality and improve the fit of our predictive models, we use the natural log of childhood and adult family income in our regression analyses. One respondent is missing the value for 1970 family income, and was multiply imputed. Neighborhood SES was captured with a continuous measure representing the average income in the neighborhood surrounding the respondent in childhood and adulthood. To identify neighborhoods, we use the PSID’s supplemental Geospatial Match Files to link addresses of PSID respondents in 1970 and 2013 to corresponding codes for census tracts. Tract data come from the 1970 U.S. decennial census and the 2013 American Community Survey 5-Year Estimates, with tract boundaries normalized to 2010 (GeoLytics, 2014). Tract average family income was calculated by dividing the total aggregate income in the previous year of all families in the tract by the total number of families in the tract in 1970 and 2013. Neighborhood family income is also non-normally distributed and exhibits a right skew. We therefore used the natural log of neighborhood family income in our regression models. Missing values for neighborhood SES required a special strategy. In 1970, many less-populated areas of the United States had not yet been divided into census tracts. Consequently, some tracts have missing data in 1970, including 635 respondents in our sample. For respondents with missing 1970 tract values, we substitute the overall tract mean and incorporate a dichotomous variable indicating this substitution into all models. We are precluded from employing a more sophisticated imputation strategy for these tracts because all 1970 tract variables are missing (not only income), leaving us with minimal information to impute from. However, we felt it was important to retain individuals with missing 1970 tract values in the analysis to preserve the representativeness of our sample. In sensitivity analyses, we used a measure of tract level poverty rate instead of tract level income. Those results were substantively similar to those presented below and are available upon request. Independent Variables Several time-invariant covariates of disability are also included in the analyses. They include the respondent’s sex (male = 1, female = 0) and race/ethnicity (dummy coded as black, white, and other). Childhood health is included to account for the possibility respondents with poor health in childhood are more likely to develop disability in adulthood. Childhood self-reported health is based on a 2013 survey question asking respondents to retrospectively recall their health before they were 17 years old. Responses were recorded as “excellent,” “very good,” “good,” “fair,” or “poor.” In our analyses, we dichotomize the responses so a value of “0” represents excellent/very good/good, and a value of “1” represents fair/poor self-reported childhood health. Finally, age is included as a covariate in adulthood, since older individuals may be more likely to experience disability. In our sample, 10 respondents are missing values for race/ethnicity and 55 respondents are missing values for childhood health. These values are multiply imputed. Following White et al., (2011), all of our model’s covariates and the outcome variable (i.e., functional limitations) are included in the multiple imputation models predicting race/ethnicity and childhood health. According to the imputed data, respondents who were missing values for race are slightly less likely to be identified as white and slightly more likely to be identified as black or other, compared to respondents without missing values. There is no substantial difference in childhood health in our imputed and nonimputed samples. Analysis Plan Our empirical goal is to assess how the temporal and spatial dimensions of income impact disability in adulthood. Thus, we rely on a negative binomial regression equation to predict the number of disability items with which an individual expediencies difficultly. The negative binomial model is specifically designed to predict overdispersed count variables like our dependent variable. It is more appropriate than standard OLS for count variables (which are based on integers and cannot be negative) rather than continuous values. Indeed, model fit statistics reveal a negative binomial model is preferred to OLS in our data estimates. All models utilize Huber-White clustered standard errors to account for the clustering of siblings who grew up in the same 1970 household. All analyses were conducted in STATA/MP 14.2 (StataCorp, 2015). Because we were concerned about the extent selection into 2013 neighborhood might be biasing our results, we ran additional robustness checks. Those are available as an online supplement to this manuscript, but in short, findings were comparable, and so we present the more parsimonious results below. Results Univariate Results As shown in Table 1, the average person in this sample experienced 0.29 of a disability in adulthood (and 13% had one or more disability). Average household income in 1970 was $10,270 and $87,680 in 2013, and average neighborhood income was $10,469 in 1970 and $79,374 in 2013. Table 1. Summary Statistics   Mean/Proportion  SD  Minimum  Maximum  Functional Limitations  0.29  0.90  0.00  6.00  HH Income 1970  10,269.84  7,492.34  1.00  91,660.00  HH Income 2013  87,680.46  147,396.00  1.00  3,316,000.00  Neighborhood Income 1970  10,468.69  3,620.94  3,664.92  47,720.74  Neighborhood Income 2013  79,373.95  40,936.13  1.00  458,681.00  Male  0.42  --  1.00  1.00  Black  0.43  --  0.00  1.00  White  0.54  --  0.00  1.00  Other  0.03  --  0.00  1.00  Childhood health  0.04  0.19  0.00  1.00  Age 2013  51.94  4.99  42.00  63.00  N. Inc. 1970 Missing  0.27  --  0.00  1.00  N. Inc.2013 Missinga  0.00  --  0.00  1.00    Mean/Proportion  SD  Minimum  Maximum  Functional Limitations  0.29  0.90  0.00  6.00  HH Income 1970  10,269.84  7,492.34  1.00  91,660.00  HH Income 2013  87,680.46  147,396.00  1.00  3,316,000.00  Neighborhood Income 1970  10,468.69  3,620.94  3,664.92  47,720.74  Neighborhood Income 2013  79,373.95  40,936.13  1.00  458,681.00  Male  0.42  --  1.00  1.00  Black  0.43  --  0.00  1.00  White  0.54  --  0.00  1.00  Other  0.03  --  0.00  1.00  Childhood health  0.04  0.19  0.00  1.00  Age 2013  51.94  4.99  42.00  63.00  N. Inc. 1970 Missing  0.27  --  0.00  1.00  N. Inc.2013 Missinga  0.00  --  0.00  1.00  Note: n = 2,344; This is reported based on the 1st of 10 imputed datasets (summary statistics for all 10 available upon request). aThis value is less than 0.01 because of rounding. View Large Table 1. Summary Statistics   Mean/Proportion  SD  Minimum  Maximum  Functional Limitations  0.29  0.90  0.00  6.00  HH Income 1970  10,269.84  7,492.34  1.00  91,660.00  HH Income 2013  87,680.46  147,396.00  1.00  3,316,000.00  Neighborhood Income 1970  10,468.69  3,620.94  3,664.92  47,720.74  Neighborhood Income 2013  79,373.95  40,936.13  1.00  458,681.00  Male  0.42  --  1.00  1.00  Black  0.43  --  0.00  1.00  White  0.54  --  0.00  1.00  Other  0.03  --  0.00  1.00  Childhood health  0.04  0.19  0.00  1.00  Age 2013  51.94  4.99  42.00  63.00  N. Inc. 1970 Missing  0.27  --  0.00  1.00  N. Inc.2013 Missinga  0.00  --  0.00  1.00    Mean/Proportion  SD  Minimum  Maximum  Functional Limitations  0.29  0.90  0.00  6.00  HH Income 1970  10,269.84  7,492.34  1.00  91,660.00  HH Income 2013  87,680.46  147,396.00  1.00  3,316,000.00  Neighborhood Income 1970  10,468.69  3,620.94  3,664.92  47,720.74  Neighborhood Income 2013  79,373.95  40,936.13  1.00  458,681.00  Male  0.42  --  1.00  1.00  Black  0.43  --  0.00  1.00  White  0.54  --  0.00  1.00  Other  0.03  --  0.00  1.00  Childhood health  0.04  0.19  0.00  1.00  Age 2013  51.94  4.99  42.00  63.00  N. Inc. 1970 Missing  0.27  --  0.00  1.00  N. Inc.2013 Missinga  0.00  --  0.00  1.00  Note: n = 2,344; This is reported based on the 1st of 10 imputed datasets (summary statistics for all 10 available upon request). aThis value is less than 0.01 because of rounding. View Large Multivariable Results Results of negative binomial regression predicting adult disability are shown in Table 2. In model 1, both of the measures of socioeconomic present (2013 household- and neighborhood-level income) are included (net of controls). Both measures are associated with lower disability. In model 2, both of the measures of socioeconomic history (1970 household- and neighborhood-level income) are included. Household income is associated with lower disability in adulthood, but neighborhood-level income is not. In model 3, both measures of individual resources (1970 and 2013 household-level income) are included, and both measures are associated with lower disability. In model 4, both measures of contextual resources (1970 and 2013 neighborhood-level income) income are included. In this model, 2013 neighborhood-level is associated with lower disability in adulthood, but 1970 neighborhood-level income is not. In model 5, all four measures of income are included and, as was the case in the previous models, both 1970 and 2013 household income are associated with lower disability, as is 2013 neighborhood-level income, but 1970 neighborhood-level income is not associated with disability in adulthood. Table 2. Negative Binomial Regression of Functional Limitations   (1)  (2)  (3)  (4)  (5)  HH Income 1970a    −0.396**  −0.265*    −0.289*      (0.120)  (0.107)    (0.114)  HH Income 2013a  −0.317***    −0.367***    −0.310***    (0.061)    (0.066)    (0.059)  Neighborhood Income 1970a    −0.128    −0.016  0.109      (0.289)    (0.278)  (0.277)  Neighborhood Income 2013a  −0.673***      −0.956***  −0.669***    (0.182)      (0.178)  (0.185)  Maleb  −0.689***  −0.647***  −0.683***  −0.652***  −0.661***    (0.138)  (0.139)  (0.138)  (0.141)  (0.138)  Blackc  −0.063  0.087  −0.070  0.118  −0.251    (0.163)  (0.182)  (0.171)  (0.183)  (0.185)  Otherc  0.310  0.166  0.292  0.251  0.191    (0.331)  (0.337)  (0.336)  (0.331)  (0.341)  Childhood health  0.503*  0.577**  0.482*  0.593**  0.521*    (0.218)  (0.217)  (0.221)  (0.218)  (0.222)  Age 2013  0.060***  0.063***  0.060***  0.061***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd    −0.241    −0.167  −0.358*      (0.175)    (0.173)  (0.174)  N. INC. 2013 Missingc  −17.995***      −17.714***  −18.046***    (1.117)      (0.581)  (1.123)  Constant  6.591**  0.334  2.009  6.454*  8.021**    (2.156)  (2.682)  (1.367)  (3.064)  (3.027)    (1)  (2)  (3)  (4)  (5)  HH Income 1970a    −0.396**  −0.265*    −0.289*      (0.120)  (0.107)    (0.114)  HH Income 2013a  −0.317***    −0.367***    −0.310***    (0.061)    (0.066)    (0.059)  Neighborhood Income 1970a    −0.128    −0.016  0.109      (0.289)    (0.278)  (0.277)  Neighborhood Income 2013a  −0.673***      −0.956***  −0.669***    (0.182)      (0.178)  (0.185)  Maleb  −0.689***  −0.647***  −0.683***  −0.652***  −0.661***    (0.138)  (0.139)  (0.138)  (0.141)  (0.138)  Blackc  −0.063  0.087  −0.070  0.118  −0.251    (0.163)  (0.182)  (0.171)  (0.183)  (0.185)  Otherc  0.310  0.166  0.292  0.251  0.191    (0.331)  (0.337)  (0.336)  (0.331)  (0.341)  Childhood health  0.503*  0.577**  0.482*  0.593**  0.521*    (0.218)  (0.217)  (0.221)  (0.218)  (0.222)  Age 2013  0.060***  0.063***  0.060***  0.061***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd    −0.241    −0.167  −0.358*      (0.175)    (0.173)  (0.174)  N. INC. 2013 Missingc  −17.995***      −17.714***  −18.046***    (1.117)      (0.581)  (1.123)  Constant  6.591**  0.334  2.009  6.454*  8.021**    (2.156)  (2.682)  (1.367)  (3.064)  (3.027)  Note: n = 2,344; *p < .05. **p < .01. ***p < .001. Standard errors are shown in parentheses. Pooled results from 10 imputation datasets. aAll income measures are transformed using the natural log. References are: b= Females; c= Whites; and d= Nonmissing. View Large Table 2. Negative Binomial Regression of Functional Limitations   (1)  (2)  (3)  (4)  (5)  HH Income 1970a    −0.396**  −0.265*    −0.289*      (0.120)  (0.107)    (0.114)  HH Income 2013a  −0.317***    −0.367***    −0.310***    (0.061)    (0.066)    (0.059)  Neighborhood Income 1970a    −0.128    −0.016  0.109      (0.289)    (0.278)  (0.277)  Neighborhood Income 2013a  −0.673***      −0.956***  −0.669***    (0.182)      (0.178)  (0.185)  Maleb  −0.689***  −0.647***  −0.683***  −0.652***  −0.661***    (0.138)  (0.139)  (0.138)  (0.141)  (0.138)  Blackc  −0.063  0.087  −0.070  0.118  −0.251    (0.163)  (0.182)  (0.171)  (0.183)  (0.185)  Otherc  0.310  0.166  0.292  0.251  0.191    (0.331)  (0.337)  (0.336)  (0.331)  (0.341)  Childhood health  0.503*  0.577**  0.482*  0.593**  0.521*    (0.218)  (0.217)  (0.221)  (0.218)  (0.222)  Age 2013  0.060***  0.063***  0.060***  0.061***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd    −0.241    −0.167  −0.358*      (0.175)    (0.173)  (0.174)  N. INC. 2013 Missingc  −17.995***      −17.714***  −18.046***    (1.117)      (0.581)  (1.123)  Constant  6.591**  0.334  2.009  6.454*  8.021**    (2.156)  (2.682)  (1.367)  (3.064)  (3.027)    (1)  (2)  (3)  (4)  (5)  HH Income 1970a    −0.396**  −0.265*    −0.289*      (0.120)  (0.107)    (0.114)  HH Income 2013a  −0.317***    −0.367***    −0.310***    (0.061)    (0.066)    (0.059)  Neighborhood Income 1970a    −0.128    −0.016  0.109      (0.289)    (0.278)  (0.277)  Neighborhood Income 2013a  −0.673***      −0.956***  −0.669***    (0.182)      (0.178)  (0.185)  Maleb  −0.689***  −0.647***  −0.683***  −0.652***  −0.661***    (0.138)  (0.139)  (0.138)  (0.141)  (0.138)  Blackc  −0.063  0.087  −0.070  0.118  −0.251    (0.163)  (0.182)  (0.171)  (0.183)  (0.185)  Otherc  0.310  0.166  0.292  0.251  0.191    (0.331)  (0.337)  (0.336)  (0.331)  (0.341)  Childhood health  0.503*  0.577**  0.482*  0.593**  0.521*    (0.218)  (0.217)  (0.221)  (0.218)  (0.222)  Age 2013  0.060***  0.063***  0.060***  0.061***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd    −0.241    −0.167  −0.358*      (0.175)    (0.173)  (0.174)  N. INC. 2013 Missingc  −17.995***      −17.714***  −18.046***    (1.117)      (0.581)  (1.123)  Constant  6.591**  0.334  2.009  6.454*  8.021**    (2.156)  (2.682)  (1.367)  (3.064)  (3.027)  Note: n = 2,344; *p < .05. **p < .01. ***p < .001. Standard errors are shown in parentheses. Pooled results from 10 imputation datasets. aAll income measures are transformed using the natural log. References are: b= Females; c= Whites; and d= Nonmissing. View Large As shown in Table 3, we tested all six possible interaction terms comprised of our four measures of income. Although two of those interactions were nonsignificant (1970 household income × 1970 neighborhood income, and 1970 household income × 2013 neighborhood income), the four remaining interactions were significant. The negative association between each of these interactions and disability suggests either (a) when individuals have/had lower levels of income in one dimension but higher levels of income in another dimension, the deleterious association of low income is partially mitigated by having higher levels of income in another dimension, or (b) having more income in two dimensions combines in such a way that helps mitigate the risks of disability to a greater extent than one would expect by just considering the direct benefit of each of the two dimensions of income. In short, these four significant interaction terms suggests the more dimensions in which one has/had high income, the better it is for their physical health in adulthood. Table 3. Negative Binomial Regression of Functional Limitations   (1)  (2)  (3)  (4)  (5)  (6)  HH Income 1970 X  −0.332             Neighborhood Income 1970  (0.313)            HH Income 1970 X    −0.175**           HH Income 2013    (0.063)          Neighborhood Income 1970 X      −1.290*         Neighborhood Income 2013      (0.518)        HH Income 1970 X        −0.197       Neighborhood Income 2013        (0.213)      HH Income 2013 X          −0.412*     Neighborhood Income 1970          (0.166)    HH Income 2013 X            −0.449**   Neighborhood Income 2013            (0.146)  HH Income 1970a  2.753  1.536*  −0.260*  1.892  −0.268*  −0.276*    (2.869)  (0.676)  (0.110)  (2.345)  (0.112)  (0.111)  HH Income 2013a  −0.307***  1.193*  −0.301***  −0.307***  3.453*  4.646**    (0.059)  (0.523)  (0.058)  (0.059)  (1.501)  (1.587)  Neighborhood Income 1970a  3.105  0.181  14.478*  0.158  4.405*  0.171    (2.815)  (0.276)  (5.781)  (0.283)  (1.755)  (0.275)  Neighborhood Income 2013a  −0.642**  −0.590**  11.158*  1.093  −0.630**  4.097**    (0.184)  (0.187)  (4.737)  (1.908)  (0.184)  (1.524)  Maleb  −0.675***  −0.660***  −0.656***  −0.664***  −0.667***  −0.657***    (0.138)  (0.139)  (0.139)  (0.138)  (0.139)  (0.138)  Blackc  −0.252  −0.259  −0.231  −0.247  −0.262  −0.240    (0.183)  (0.185)  (0.182)  (0.185)  (0.183)  (0.181)  Otherc  0.191  0.192  0.187  0.193  0.196  0.170    (0.341)  (0.336)  (0.345)  (0.342)  (0.343)  (0.336)  Childhood health  0.510*  0.538*  0.525*  0.522*  0.521*  0.512*    (0.221)  (0.222)  (0.220)  (0.222)  (0.223)  (0.221)  Age  0.064***  0.062***  0.062***  0.063***  0.063***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd  −0.415*  −0.335  −0.446*  −0.370*  −0.393*  −0.401*    (0.181)  (0.172)  (0.178)  (0.173)  (0.176)  (0.171)  N. INC. 2013 Missingd  −17.889***  −19.268***  −18.155***  −18.329***  −19.353***  −19.272***    (1.039)  (1.445)  (1.078)  (1.063)  (1.643)  (1.255)  Constant  −19.659  −9.120  −123.937*  −11.914  −31.810*  −45.234**    (26.039)  (6.880)  (52.857)  (21.552)  (16.227)  (17.054)    (1)  (2)  (3)  (4)  (5)  (6)  HH Income 1970 X  −0.332             Neighborhood Income 1970  (0.313)            HH Income 1970 X    −0.175**           HH Income 2013    (0.063)          Neighborhood Income 1970 X      −1.290*         Neighborhood Income 2013      (0.518)        HH Income 1970 X        −0.197       Neighborhood Income 2013        (0.213)      HH Income 2013 X          −0.412*     Neighborhood Income 1970          (0.166)    HH Income 2013 X            −0.449**   Neighborhood Income 2013            (0.146)  HH Income 1970a  2.753  1.536*  −0.260*  1.892  −0.268*  −0.276*    (2.869)  (0.676)  (0.110)  (2.345)  (0.112)  (0.111)  HH Income 2013a  −0.307***  1.193*  −0.301***  −0.307***  3.453*  4.646**    (0.059)  (0.523)  (0.058)  (0.059)  (1.501)  (1.587)  Neighborhood Income 1970a  3.105  0.181  14.478*  0.158  4.405*  0.171    (2.815)  (0.276)  (5.781)  (0.283)  (1.755)  (0.275)  Neighborhood Income 2013a  −0.642**  −0.590**  11.158*  1.093  −0.630**  4.097**    (0.184)  (0.187)  (4.737)  (1.908)  (0.184)  (1.524)  Maleb  −0.675***  −0.660***  −0.656***  −0.664***  −0.667***  −0.657***    (0.138)  (0.139)  (0.139)  (0.138)  (0.139)  (0.138)  Blackc  −0.252  −0.259  −0.231  −0.247  −0.262  −0.240    (0.183)  (0.185)  (0.182)  (0.185)  (0.183)  (0.181)  Otherc  0.191  0.192  0.187  0.193  0.196  0.170    (0.341)  (0.336)  (0.345)  (0.342)  (0.343)  (0.336)  Childhood health  0.510*  0.538*  0.525*  0.522*  0.521*  0.512*    (0.221)  (0.222)  (0.220)  (0.222)  (0.223)  (0.221)  Age  0.064***  0.062***  0.062***  0.063***  0.063***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd  −0.415*  −0.335  −0.446*  −0.370*  −0.393*  −0.401*    (0.181)  (0.172)  (0.178)  (0.173)  (0.176)  (0.171)  N. INC. 2013 Missingd  −17.889***  −19.268***  −18.155***  −18.329***  −19.353***  −19.272***    (1.039)  (1.445)  (1.078)  (1.063)  (1.643)  (1.255)  Constant  −19.659  −9.120  −123.937*  −11.914  −31.810*  −45.234**    (26.039)  (6.880)  (52.857)  (21.552)  (16.227)  (17.054)  Note: n = 2,344; *p < .05. **p < .01. ***p < .001. Standard errors are shown in parentheses. Pooled results from 10 imputation datasets. aAll income measures are transformed using the natural log. References are: b= Females; c= Whites; and d= Nonmissing. View Large Table 3. Negative Binomial Regression of Functional Limitations   (1)  (2)  (3)  (4)  (5)  (6)  HH Income 1970 X  −0.332             Neighborhood Income 1970  (0.313)            HH Income 1970 X    −0.175**           HH Income 2013    (0.063)          Neighborhood Income 1970 X      −1.290*         Neighborhood Income 2013      (0.518)        HH Income 1970 X        −0.197       Neighborhood Income 2013        (0.213)      HH Income 2013 X          −0.412*     Neighborhood Income 1970          (0.166)    HH Income 2013 X            −0.449**   Neighborhood Income 2013            (0.146)  HH Income 1970a  2.753  1.536*  −0.260*  1.892  −0.268*  −0.276*    (2.869)  (0.676)  (0.110)  (2.345)  (0.112)  (0.111)  HH Income 2013a  −0.307***  1.193*  −0.301***  −0.307***  3.453*  4.646**    (0.059)  (0.523)  (0.058)  (0.059)  (1.501)  (1.587)  Neighborhood Income 1970a  3.105  0.181  14.478*  0.158  4.405*  0.171    (2.815)  (0.276)  (5.781)  (0.283)  (1.755)  (0.275)  Neighborhood Income 2013a  −0.642**  −0.590**  11.158*  1.093  −0.630**  4.097**    (0.184)  (0.187)  (4.737)  (1.908)  (0.184)  (1.524)  Maleb  −0.675***  −0.660***  −0.656***  −0.664***  −0.667***  −0.657***    (0.138)  (0.139)  (0.139)  (0.138)  (0.139)  (0.138)  Blackc  −0.252  −0.259  −0.231  −0.247  −0.262  −0.240    (0.183)  (0.185)  (0.182)  (0.185)  (0.183)  (0.181)  Otherc  0.191  0.192  0.187  0.193  0.196  0.170    (0.341)  (0.336)  (0.345)  (0.342)  (0.343)  (0.336)  Childhood health  0.510*  0.538*  0.525*  0.522*  0.521*  0.512*    (0.221)  (0.222)  (0.220)  (0.222)  (0.223)  (0.221)  Age  0.064***  0.062***  0.062***  0.063***  0.063***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd  −0.415*  −0.335  −0.446*  −0.370*  −0.393*  −0.401*    (0.181)  (0.172)  (0.178)  (0.173)  (0.176)  (0.171)  N. INC. 2013 Missingd  −17.889***  −19.268***  −18.155***  −18.329***  −19.353***  −19.272***    (1.039)  (1.445)  (1.078)  (1.063)  (1.643)  (1.255)  Constant  −19.659  −9.120  −123.937*  −11.914  −31.810*  −45.234**    (26.039)  (6.880)  (52.857)  (21.552)  (16.227)  (17.054)    (1)  (2)  (3)  (4)  (5)  (6)  HH Income 1970 X  −0.332             Neighborhood Income 1970  (0.313)            HH Income 1970 X    −0.175**           HH Income 2013    (0.063)          Neighborhood Income 1970 X      −1.290*         Neighborhood Income 2013      (0.518)        HH Income 1970 X        −0.197       Neighborhood Income 2013        (0.213)      HH Income 2013 X          −0.412*     Neighborhood Income 1970          (0.166)    HH Income 2013 X            −0.449**   Neighborhood Income 2013            (0.146)  HH Income 1970a  2.753  1.536*  −0.260*  1.892  −0.268*  −0.276*    (2.869)  (0.676)  (0.110)  (2.345)  (0.112)  (0.111)  HH Income 2013a  −0.307***  1.193*  −0.301***  −0.307***  3.453*  4.646**    (0.059)  (0.523)  (0.058)  (0.059)  (1.501)  (1.587)  Neighborhood Income 1970a  3.105  0.181  14.478*  0.158  4.405*  0.171    (2.815)  (0.276)  (5.781)  (0.283)  (1.755)  (0.275)  Neighborhood Income 2013a  −0.642**  −0.590**  11.158*  1.093  −0.630**  4.097**    (0.184)  (0.187)  (4.737)  (1.908)  (0.184)  (1.524)  Maleb  −0.675***  −0.660***  −0.656***  −0.664***  −0.667***  −0.657***    (0.138)  (0.139)  (0.139)  (0.138)  (0.139)  (0.138)  Blackc  −0.252  −0.259  −0.231  −0.247  −0.262  −0.240    (0.183)  (0.185)  (0.182)  (0.185)  (0.183)  (0.181)  Otherc  0.191  0.192  0.187  0.193  0.196  0.170    (0.341)  (0.336)  (0.345)  (0.342)  (0.343)  (0.336)  Childhood health  0.510*  0.538*  0.525*  0.522*  0.521*  0.512*    (0.221)  (0.222)  (0.220)  (0.222)  (0.223)  (0.221)  Age  0.064***  0.062***  0.062***  0.063***  0.063***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd  −0.415*  −0.335  −0.446*  −0.370*  −0.393*  −0.401*    (0.181)  (0.172)  (0.178)  (0.173)  (0.176)  (0.171)  N. INC. 2013 Missingd  −17.889***  −19.268***  −18.155***  −18.329***  −19.353***  −19.272***    (1.039)  (1.445)  (1.078)  (1.063)  (1.643)  (1.255)  Constant  −19.659  −9.120  −123.937*  −11.914  −31.810*  −45.234**    (26.039)  (6.880)  (52.857)  (21.552)  (16.227)  (17.054)  Note: n = 2,344; *p < .05. **p < .01. ***p < .001. Standard errors are shown in parentheses. Pooled results from 10 imputation datasets. aAll income measures are transformed using the natural log. References are: b= Females; c= Whites; and d= Nonmissing. View Large Discussion In this article, we developed conceptual matrix of socioeconomic status, and tested five hypotheses derived from a set of three assumptions upon which our conceptual matrix rests. Results indicated both 1970 and 2013 household income were associated with lower disability in adulthood, as was 2013 neighborhood-level income, but 1970 neighborhood-level income was not associated with disability in adulthood. Further, four of the six possible interactions between the multiple dimensions of income were associated with significant reductions in disability. These findings provide several important empirical insights, but also help inform a framework for thinking about the multidimensional ways in which SES relates to health more generally. First, the hypothesis that there would be a negative association between childhood income and level of disability in adulthood, net of income in adulthood was partially supported; childhood family income was associated with a reduction in level of disability in adulthood. In contrast, childhood neighborhood income was never associated with a reduction in level of disability. Additionally, the hypothesis that there would be negative association between adult income and level of disability in adulthood, net of income in childhood was supported; both individual- and neighborhood-level income were associated with a reduction in level of disability in adulthood. Taken together, these findings support the first assumption of our conceptual matrix of SES and health, that is: current health is a function of both an individual’s socioeconomic present but also an individual’s socioeconomic history. Second, both the hypothesis that there would be a negative association between household-level income and adult disability, net of neighborhood-level income, and the hypothesis that there would be a negative association between neighborhood-level income and disability, net of household-level, we generally supported. As mentioned above, in all of our models, both 1970 and 2013 household income, as well as 2013 neighborhood income (but not 1970 neighborhood income) were always associated with a reduction in level of disability. Therefore, these findings support the second assumption of our conceptual matrix of SES and health, that is: current health is a function of both individual resources and an individual’s contextual resources. Third, we hypothesized the association between any one dimension of income and disability in adulthood would be conditioned by the other dimensions of income. Four of the six interactions were significant, so this hypothesis was generally supported. These findings support the second assumption of our conceptual matrix of SES and health, that is: The relationship within and between the temporal and spatial dimensions of SES and health are not just additive, but rather, interact with each other in a multiplicative fashion to shape adult health. This research contributes to the literature on SES and health in two primary ways. First, we have provided a conceptual matrix of SES by temporal and spatial dimensions in order to advance the ways in which we think about how various forms of SES may relate to general health. Second, to our knowledge, we are the first to empirically document how child and adult income, measured as both household income and neighborhood income, relate to adult disability. These contributions are limited by a variety of factors. First, our research was guided three theoretical perspectives; however, we are not explicitly testing any of these theories. Second, although we believe our conceptual matrix of SES should be relevant to forms of SES beyond just income and types of health beyond disability, we only tested the association between our four dimensions of income and disability here. Therefore, we cannot make empirical claims beyond those relationships. Third, it is possible that childhood health impacts both adult income and adult disability, but we are unable to address that with the current data. Hopefully, future data collections efforts will provide additional insights. Finally, it is possible census tracts do not adequately capture neighborhoods as residences perceive them (Coulton, Korbin, Chan, & Su, 2001). We acknowledge this limitation, but, unfortunately, this is the linkage we are able to make with the data we have. Conclusion In sum, these findings provide a much more nuanced understanding of and a conceptual model for interpreting the ways in which the temporal and spatial dimensions of socioeconomic status relate to disability. We argue the association between SES and disability is comprised of both socioeconomic circumstances and socioeconomic resources. At the same time, the association between SES and disability is related to both individual resources and collective resources. Moreover, these four components generally function interactively to shape adult health. We hope our conceptual matrix of SES by temporal and spatial dimensions will provide a usual typology to guide future research on the relationship between various forms of SES and wider array of health outcomes, particularly in an era where the ability to link geocoded longitudinal individual data with spatial data is becoming increasingly possible. The degree of disability in adulthood is an important concern for population health because it is highly predictive of future health declines. According to our results, interventions seeking to reduce or delay disability need to take a wide view and consider the temporal and spatial dimensions of SES that contribute to the development of disability. For instance, policies and programs that alleviate childhood poverty are likely to produce lasting effects for health into adulthood. At the same time, policies and programs that help target the deleterious consequences of adult poverty and living in poor neighborhoods are essential for physical health as well. We believe our conceptual matrix of SES by temporal and spatial dimensions provides a useful tool to help guide research as it disentangles the complex web of associations between SES and health. Funding The collection of data used in this study was partly supported by the National Institutes of Health under grant number R01 HD069609 and the National Science Foundation under award number 1157698. Conflict of Interest None reported. Acknowledgments B. L. Kail would like to thank Dawn C. Carr for providing feedback on some of the ideas written about in this manuscript. References Adler, N. E., Boyce, T., Chesney, M. A., Cohen, S., Folkman, S., Kahn, R. L., & Syme, S. L. ( 1994). Socioeconomic status and health. The challenge of the gradient. The American Psychologist , 49, 15– 24. doi:10.1037/0003-066X.49.1.15 Google Scholar CrossRef Search ADS   Aneshensel, C. S., & Sucoff, C. A. ( 1996). The neighborhood context of adolescent mental health. Journal of Health and Social Behavior , 37, 293– 310. doi:10.2307/2137258 Google Scholar CrossRef Search ADS   Antonovsky, A. ( 1967). Social class, life expectancy and overall mortality. The Milbank Memorial Fund Quarterly , 45, 31– 73. doi:10.2307/3348839 Google Scholar CrossRef Search ADS   Balfour, J. L., & Kaplan, G. A. ( 2002). Neighborhood environment and loss of physical function in older adults: Evidence from the Alameda County Study. American Journal of Epidemiology , 155, 507– 515. doi:10.1093/aje/155.6.507 Google Scholar CrossRef Search ADS   Beauchamp, M. K., Bean, J. F., Ward, R. E., Kurlinski, L. A., Latham, N. K., & Jette, A. M. ( 2015). How should disability be measured in older adults? An analysis from the Boston rehabilitative impairment study of the elderly. Journal of the American Geriatrics Society , 63, 1187– 1191. doi: 10.1111/jgs.13453 Google Scholar CrossRef Search ADS   Bloome, D., & Western, B. ( 2011). Cohort change and racial differences in educational and income mobility. Social Forces , 90, 375– 395. doi:10.1093/sf/sor002 Google Scholar CrossRef Search ADS   Brault, M. W. ( 2012). Americans with disabilities: 2010. Current Population Reports , 7, 0– 131. Clarke, P., Morenoff, J., Debbink, M., Golberstein, E., Elliott, M. R., & Lantz, P. M. ( 2014). Cumulative exposure to neighborhood context: Consequences for health transitions over the adult life course. Research on Aging , 36, 115– 142. doi: 10.1177/0164027512470702 Google Scholar CrossRef Search ADS   Cohen, S., Janicki-Deverts, D., Chen, E., & Matthews, K. A. ( 2010). Childhood socioeconomic status and adult health. Annals of the New York Academy of Sciences , 1186, 37– 55. doi: 10.1111/j.1749-6632.2009.05334.x Google Scholar CrossRef Search ADS   Coulton, C. J., Korbin, J., Chan, T., & Su, M. ( 2001). Mapping residents’ perceptions of neighborhood boundaries: a methodological note. American Journal of Community Psychology , 29, 371– 383. doi: 10.1023/A:1010303419034 Google Scholar CrossRef Search ADS   Cutrona, C. E., Wallace, G., & Wesner, K. A. ( 2006). Neighborhood characteristics and depression: An examination of stress processes. Current Directions in Psychological Science , 15, 188– 192. doi: 10.1111/j.1467-8721.2006.00433.x Google Scholar CrossRef Search ADS   Diez Roux, A. V., & Mair, C. ( 2010). Neighborhoods and health. Annals of the New York Academy of Sciences , 1186, 125– 145. doi: 10.1111/j.1749-6632.2009.05333.x Google Scholar CrossRef Search ADS   ElderJr, G. H. ( 1999). Children of the great depression: Social change in life experience . Boulder, CO: Westview Press. Ellen, I. G., Mijanovich, T., & Dillman, K.-N. ( 2001). Neighborhood effects on health: Exploring the links and assessing the evidence. Journal of Urban Affairs , 23, 391– 408. doi:10.1111/0735- 2166.00096 Google Scholar CrossRef Search ADS   Ferraro, K. F. ( 2011). Health and aging: Early origins, persistent inequalities? In A. R. Settersten & L. J. Angel (Eds.), Handbook of sociology of aging  (pp. 465– 475). New York, NY: Springer New York. Google Scholar CrossRef Search ADS   Ferraro, K. F., & Shippee, T. P. ( 2009). Aging and cumulative inequality: How does inequality get under the skin? The Gerontologist , 49, 333– 343. doi: 10.1093/geront/gnp034 Google Scholar CrossRef Search ADS   Ferraro, K., Shippee, T., & Schafer, M. ( 2009). Cumulative inequality theory for research on aging and the life course. In Bengtson, V., Silverstein, M., Norella, P., & Gans, D. (Eds.), Handbook of theories of aging  ( 2nd ed., pp. 413– 433). New York: Springer Publishing. Freedman, V. A., & Martin, L. G. ( 1999). The role of education in explaining and forecasting trends in functional limitations among older Americans. Demography , 36, 461– 473. doi: 10.2307/2648084 Google Scholar CrossRef Search ADS   Galster, G. C. ( 2012). The mechanism (s) of neighbourhood effects: Theory, evidence, and policy implications. In van Ham, M., Manley, D., Bailey, N., Simpson, L., & Maclennan, D. (Eds.), Neighbourhood effects research: New perspectives  (pp. 23– 56). Dordrecht, Netherlands: Springer. doi:10.1007/978-94-007-2309-2 Google Scholar CrossRef Search ADS   Gayman MD, Turner RJ, & Cui M. ( 2008). Physical limitations and depressive symptoms: Exploring the nature of the association. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences , 63, S219– S228. Google Scholar CrossRef Search ADS   GeoLytics. ( 2014). CensusCD Neighborhood Change Database 1970–2010 tract data . Retrieved from http://www.geolytics.com.ezproxy.gsu.edu/USCensus,Neighborhood-Change-Database-1970–2000,Data,Features,Products.asp Haas, S. ( 2008). Trajectories of functional health: The ‘long arm’ of childhood health and socioeconomic factors. Social Science & Medicine , 66, 849– 861. doi: 10.1016/j.socscimed.2007.11.004 Google Scholar CrossRef Search ADS   House, J. S., Kessler, R. C., & Herzog, A. R. ( 1990). Age, socioeconomic status, and health. The Milbank Quarterly , 68, 383– 411. doi: 10.2307/3350111 Google Scholar CrossRef Search ADS   House, J. S., Lepkowski, J. M., Kinney, A. M., Mero, R. P., Kessler, R. C., & Herzog, A. R. ( 1994). The social stratification of aging and health. Journal of Health and Social Behavior , 35, 213– 234. doi:10.2307/2137277 Google Scholar CrossRef Search ADS   Hout, M., & Beller, E. ( 2006). Intergenerational social mobility: The United States in comparative perspective. The Future of Children , 16, 19– 36. Jette, A. M. ( 2006). Toward a common language for function, disability, and health. Physical Therapy , 86, 726– 734. Google Scholar CrossRef Search ADS   Kail, B. L., & Taylor, M. G. ( 2014). Cumulative inequality and racial disparities in health: private insurance coverage and black/white differences in functional limitations. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences , 69, 798– 808. doi: 10.1093/geronb/gbu005 Google Scholar CrossRef Search ADS   Link, B. G., & Phelan, J. ( 1995). Social conditions as fundamental causes of disease. Journal of Health and Social Behavior , Spec No, 80– 94. doi:10.2307/2626958 Google Scholar CrossRef Search ADS   Livermore, G., Stapleton, D. C., & O’Toole, M. ( 2011). Health care costs are a key driver of growth in federal and state assistance to working-age people with disabilities. Health Affairs (Project Hope) , 30, 1664– 1672. doi: 10.1377/hlthaff.2011.0247 Google Scholar CrossRef Search ADS   Marmot, M. ( 2015). The health gap: The challenge of an unequal world . London: Bloomsbury Publishing. Montez, J. K. ( 2013). The socioeconomic origins of physical functioning among older U.S. adults. Advances in Life Course Research , 18, 244– 256. doi: 10.1016/j.alcr.2013.08.001 Google Scholar CrossRef Search ADS   Nagi, S. Z. ( 1976). An epidemiology of disability among adults in the United States. The Milbank Memorial Fund Quarterly. Health and Society , 54, 439– 467. doi:10.2307/3349677 Google Scholar CrossRef Search ADS   Panel Study of Income Dynamics. ( 2013). Restricted use dataset. Institute for social research survey research center . Ann Arbor, MI: University of Michigan. Phelan, J. C., Link, B. G., & Tehranifar, P. ( 2010). Social conditions as fundamental causes of health inequalities: Theory, evidence, and policy implications. Journal of Health and Social Behavior , 51 ( Suppl), S28– S40. doi: 10.1177/0022146510383498 Google Scholar CrossRef Search ADS   Pickett, K. E., & Pearl, M. ( 2001). Multilevel analyses of neighbourhood socioeconomic context and health outcomes: A critical review. Journal of Epidemiology and Community Health , 55, 111– 122. http://dx.doi.org/10.1136/jech.55.2.111 Google Scholar CrossRef Search ADS   Poulton, R., Caspi, A., Milne, B. J., Thomson, W. M., Taylor, A., Sears, M. R., & Moffitt, T. E. ( 2002). Association between children’s experience of socioeconomic disadvantage and adult health: A life-course study. Lancet (London, England) , 360, 1640– 1645. doi: 10.1016/S0140-6736(02)11602-3 Google Scholar CrossRef Search ADS   Reardon, S. F., & Bischoff, K. ( 2011). Income inequality and income segregation. American Journal of Sociology , 116, 1092– 1153. doi: 10.1086/657114 Google Scholar CrossRef Search ADS   Robert, S. A. ( 1998). Community-level socioeconomic status effects on adult health. Journal of Health and Social Behavior , 39, 18– 37. doi: 10.2307/2676387 Google Scholar CrossRef Search ADS   Ross, C. E., & Mirowsky, J. ( 2001). Neighborhood disadvantage, disorder, and health. Journal of Health and Social Behavior , 42, 258– 276. doi:10.2307/3090214 Google Scholar CrossRef Search ADS   Ross, C. E., & Mirowsky, J. ( 2008). Neighborhood socioeconomic status and health: Context or composition? City & Community , 7, 163– 179. doi:10.1111/j.1540-6040.2008.00251.x Google Scholar CrossRef Search ADS   Sampson, R. J., Morenoff, J. D., & Gannon-Rowley, T. ( 2002). Assessing “neighborhood effects”: Social processes and new directions in research. Annual Review of Sociology , 28, 443– 478. Google Scholar CrossRef Search ADS   Schootman, M., Jeffe, D. B., Baker, E. A., & Walker, M. S. ( 2006). Effect of area poverty rate on cancer screening across US communities. Journal of Epidemiology and Community Health , 60, 202– 207. doi: 10.1136/jech.2005.041020 Google Scholar CrossRef Search ADS   Schoeni, R. F., Stafford, F., McGonagle, K. A., & Andreski, P. (2013). Response rates in national panel surveys. The Annals of the American Academy of Political and Social Science, 645, 60–87. Seeman, T. E., Bruce, M. L., & McAvay, G. J. ( 1996). Social network characteristics and onset of ADL disability: MacArthur studies of successful aging. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences , 51, S191– S200. doi:10.1093/geronb/51B.4.S191 Google Scholar CrossRef Search ADS   Segerstrom, S. C., & Miller, G. E. ( 2004). Psychological stress and the human immune system: A meta-analytic study of 30 years of inquiry. Psychological Bulletin , 130, 601– 630. doi: 10.1037/0033-2909.130.4.601 Google Scholar CrossRef Search ADS   Skinner, M. A., Zautra, A. J., & Reich, J. W. ( 2004). Financial stress predictors and the emotional and physical health of chronic pain patients. Cognitive Therapy and Research , 28, 695– 713. doi:10.1023/B:COTR.0000045572.33750.6c Google Scholar CrossRef Search ADS   Stafford, M., Gimeno, D., & Marmot, M. G. ( 2008). Neighbourhood characteristics and trajectories of health functioning: A multilevel prospective analysis. European Journal of Public Health , 18, 604– 610. doi: 10.1093/eurpub/ckn091 Google Scholar CrossRef Search ADS   StataCorp. ( 2015). Stata statistical software: Release 14 . College Station, TX: StataCorp LP. Taylor, M. G. ( 2010). Capturing transitions and trajectories: The role of socioeconomic status in later life disability. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences , 65, 733– 743. doi: 10.1093/geronb/gbq018 Google Scholar CrossRef Search ADS   Taylor, S. E., Repetti, R. L., & Seeman, T. ( 1997). Health psychology: What is an unhealthy environment and how does it get under the skin? Annual Review of Psychology , 48, 411– 447. doi: 10.1146/annurev.psych.48.1.411 Google Scholar CrossRef Search ADS   Tomey, K., Diez Roux, A. V., Clarke, P., & Seeman, T. ( 2013). Associations between neighborhood characteristics and self-rated health: A cross-sectional investigation in the Multi-Ethnic Study of Atherosclerosis (MESA) cohort. Health & Place , 24, 267– 274. doi: 10.1016/j.healthplace.2013.10.001 Google Scholar CrossRef Search ADS   Wadsworth, M., Butterworth, S., Marmot, M., & Wilkinson, R. ( 2006). In Marmot, Michael, & Wilkinson R., (Eds.), Social determinants of health . Oxford:OUP, 2005. Wheaton, B. ( 1994). Sampling the stress universe. Stress and mental health  (pp. 77– 114). Boston, MA: Springer. Google Scholar CrossRef Search ADS   Wheaton, B., & Clarke, P. ( 2003). Space meets time: Integrating temporal and contextual influences on mental health in early adulthood. American Sociological Review , 68, 680– 706. doi: 10.2307/1519758 Google Scholar CrossRef Search ADS   White, I. R., Royston, P., & Wood, A.M. (2011). Multiple imputation using chained equations: issues and guidance for practice. Statistics in medicine, 30, 377–399. Published by Oxford University Press on behalf of The Gerontological Society of America 2018. This work is written by (a) US Government employee(s) and is in the public domain in the US. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The Journals of Gerontology Series B: Psychological Sciences and Social Sciences Oxford University Press

A Conceptual Matrix of the Temporal and Spatial Dimensions of Socioeconomic Status and Their Relationship with Health

Loading next page...
 
/lp/ou_press/a-conceptual-matrix-of-the-temporal-and-spatial-dimensions-of-OGyk6ycU0Y
Publisher
Oxford University Press
Copyright
Published by Oxford University Press on behalf of The Gerontological Society of America 2018. This work is written by (a) US Government employee(s) and is in the public domain in the US.
ISSN
1079-5014
eISSN
1758-5368
D.O.I.
10.1093/geronb/gby025
Publisher site
See Article on Publisher Site

Abstract

Abstract Objectives In this study, we (a) draw on fundamental cause theory, the life course perspective, and neighborhood effects to develop conceptual matrix of socioeconomic status (SES) by temporal and spatial dimensions in order to highlight the multidimensional ways in which SES relates to general health, and then (b) assess the multidimensional ways in which income (as a measure of SES) is related to disability in adulthood. Methods Data from the Panel Study of Income Dynamics were linked with Census data to assess (a) which temporal and spatial dimensions of income were associated with disability in adulthood, and (b) whether the various components of income interact with each other when predicting disability. Results Negative binomial regression results indicated both 1970 and 2013 household income were associated with lower levels of disabilities in adulthood, as was 2013 neighborhood-level income, but 1970 neighborhood-level income was not associated with disability in adulthood. Further, 4 of the 6 possible interactions between the multiple dimensions of income were associated with significant reductions in adult disability. Discussion These findings provide several important empirical insights, but also help inform a framework for thinking about the multidimensional ways in which SES relates to health. Socioeconomic status, Disability, Life course, Neighborhood Background According to fundamental cause theory (Link & Phelan, 1995), SES-related resources are important when they may be accessed to help prevent health risks, and reduce the consequences of poor health (Link & Phelan, 1995). Therefore, those with higher SES are better positioned to avoid health problems through the deployment of knowledge, money, support networks, and psychosocial coping resources (Phelan, Link, & Tehranifar, 2010). For instance, those with less education are at greater risk of earlier onset of disability, and this increased risk is attenuated by variations in income (House et al., 1994). Thus, SES clearly represents a primary social determinant of health (Wadsworth, Butterworth, Marmot, & Wilkinson, 2006). However, SES represents a multidimensional construct involving both time and space, and the relationship between SES and health likely varies across different dimensions of SES. Little is known, however, about the combined way the temporal and spatial dimensions of SES relate to health. If the association between SES and health does indeed vary by time and space, this is important for social scientists and policy makers developing prevention–intervention efforts aimed at improving population health. Therefore, in the following sections, first, we review the literature regarding how the temporal and spatial dimensions of SES relate to adult health, and we articulate a conceptual matrix of SES in order to highlight the multidimensional ways in which SES relates to health. Second, we apply the conceptual matrix of SES to the empirical case of the relationship between household income (as a measure of SES) and disability (as a measure of health). Conceptual Matrix of SES by Temporal and spatial Dimensions In this section, we offer a conceptual matrix of the multidimensional ways in which SES may relate to health. The purpose of this conceptual matrix is to provide a guide to (a) understand the multidimensional ways in which SES relates to health in general, and (b) to specifically situate the present study, in which we are assessing how income (as an indicator of SES) is related to disability (an example of physical health). The goal here is to combine insights from fundamental cause theory, the life course perspective, and neighborhood effects to develop a unified typology for how SES relates to health. Link and Phelan (1995:p 87) argue “the essential feature of fundamental social causes, is that they involve access to resources that can be used to avoid risks or to minimize the consequences of disease once it occurs.” To this, we add a caveat, informed by the life course perspective. Although we agree that one way SES relates to health is through the ability to access current economic resources, we think it is additionally important to acknowledge individuals’ past circumstances relating to SES (i.e., their socioeconomic history). Socioeconomic history is distinguished from current SES because historical resources are no longer accessible, yet they continue to shape health. Indeed, according to life course sociology, current health conditions are a product of both proximal and distal antecedents (Elder, 1999; Ferraro, 2011; Ferraro & Shippee, 2009). One’s current circumstances are certainly important for health (proximal), but current circumstances are also shaped by one’s history of prior circumstances (distal). Furthermore, according to cumulative inequality theory, childhood is part of the past that tends to influence people’s health as adults (Ferraro, 2011; Ferraro & Shippee, 2009; Ferraro, Shippee, & Schafer, 2009). Socioeconomic status during childhood/adolescence provides the foundation for early life course development and circumstances that are related to a wide array of adult health conditions (Haas, 2008; Poulton et al., 2002). As shown in Figure 1, in our temporal dimension of SES, SES may be delineated into both (a) socioeconomic present and (b) socioeconomic history. Socioeconomic present represents SES one currently has at their disposal. Past resources, however, are qualitatively difference than current resources—in large part because they are no longer accessible to use to help reduce health risks or mitigate the consequences of poor health. Thus, we define socioeconomic history, as representative of an individual’s SES history (or part of their SES history). Socioeconomic history is importantly different from an individual’s socioeconomic present, because the conditions of socioeconomic history (a) are no longer accessible, but (b) nevertheless play an important role in adult health because they influence people’s health behaviors and health trajectories. Figure 1. View largeDownload slide Conceptual matrix of SES by temporal and spatial dimensions. NOTE: The examples of SES (i.e., income) in the grey boxes are just that: examples that fit within the typology defined by the matrix. These particular examples were selected because they are the variables used in the present study. We believe this matrix would be useful for studying other forms of SES as well (e.g., parental education and individual education along the temporal dimension, or city income per capita and state income per capita along the spatial level). Figure 1. View largeDownload slide Conceptual matrix of SES by temporal and spatial dimensions. NOTE: The examples of SES (i.e., income) in the grey boxes are just that: examples that fit within the typology defined by the matrix. These particular examples were selected because they are the variables used in the present study. We believe this matrix would be useful for studying other forms of SES as well (e.g., parental education and individual education along the temporal dimension, or city income per capita and state income per capita along the spatial level). Taken together, our model offers the following assumption about the temporal relationship between SES and health: Assumption #1: Current health is a function of both (i) an individual’s socioeconomic present, because these accessible resources may help mitigate health risks and poor health, but also (ii) an individual’s socioeconomic history, because this history shapes people’s past opportunities, behaviors, and decisions in such a way that matter for current health. In addition to the temporal dimension, the relationship between SES and health also involves a spatial dimension. Investigations into neighborhood effects (Sampson, Morenoff, & Gannon-Rowley, 2002), have identified four overarching mechanisms linking neighborhood context to health behaviors and outcomes. These include (a) institutional mechanisms; (b) social interactive mechanisms; (c) geographical mechanisms; and (d) environmental mechanisms (Galster, 2012). Within these overarching mechanisms, Galster (2012) articulates a total of 15 specific mechanisms linking neighborhoods to health. Examples of institutional mechanisms include the extent to which the neighborhood provides access to institutions like good schools and medical centers, and the extent to which there are public stereotypes held about the residents. Examples of social-interactive mechanisms include the degree of social control within the neighborhood, neighborhood-based social networks that individuals can call upon, and the collective norms and values of the neighborhood. An example of a geographical mechanism is access to employment opportunities (e.g., the geographic feasibility of transportation to and from work). Finally, examples of environmental mechanisms include the extent to which a neighborhood exposes residence to violence, deteriorated structures, and environmental toxins (for a full list of the 15 mechanisms, see Galster (2012)). In short, neighborhood socioeconomic conditions may both expose people to health risks but also provide access to health related resources. On one hand, economically-disadvantaged neighborhoods might expose individuals to health risks like high rates of poverty, unemployment, and crime (Aneshensel & Sucoff, 1996; Cutrona, Wallace, & Wesner, 2006; Sampson et al., 2002). They may also expose people to social conditions that negatively impact health, for example, by limiting the opportunities for social integration or limiting access to resource-dense social networks (Taylor, Repetti, & Seeman, 1997). On the other hand, wealthier neighborhoods may provide people with access to important health resources like health care options, quality transportation, and places to exercise (Ellen, Mijanovich, & Dillman, 2001). Wealthier neighborhoods may also promote more health enhancing social norms, better public services because of a larger per capita tax base, and include powerful social actors who are positioned and knowledgeable enough to advocate on behalf of the neighborhood (Galster, 2012). Thus, neighborhood-level SES may have both positive and negative associations with health. As shown in Figure 1, following Ross and Mirowsky (2008), here we delineate between what we are calling individual resources and contextual resources. Individual resources refers to the SES resources and circumstances possessed by an individual (or an individual’s household), while contextual resources represents resources and circumstances shaped by the environment in which a person resides. Although our focus is on neighborhoods, people find themselves embedded in a variety of spatial contexts (e.g., states, counties, or countries) that shape their circumstances and resources, which subsequently impact their health (Marmot, 2015; Ross & Mirowsky, 2008). This spatial distinction is generally buttressed by empirical evidence. For instance, low neighborhood-level SES is associated with poor physical functioning, independent of individual-level SES (Pickett & Pearl, 2001; Tomey, Diez Roux, Clarke, & Seeman, 2013). However, individual-level SES is associated with physical functioning, net of neighborhood SES (Stafford, Gimeno, & Marmot, 2008). Taken together, our model offers the following assumption about the spatial relationship between SES and health: Assumption #2: Current health is a function of both (i) individual resources, because these individual resources are important for mitigating one’s health risks and poor health, but also (ii) an individual’s contextual resources, because the context within which an individual resides provides access to health opportunities and exposes individuals to health risk factors. In addition to direct relationships between socioeconomic history, socioeconomic present, individual resources, and contextual resources with current health, we further suspect that these four domains interact with one another to shape current adult health. For instance, childhood circumstances do not preordain specific later life health. While many adults grow up to have similar class backgrounds to those in which they grew up—this is not case for all adults (Bloome & Western, 2011; Hout & Beller, 2006). Similarly, the correlation between neighborhood SES in childhood and neighborhood SES in adulthood is modest enough to suggest there is variation in the kinds of neighborhoods people live in during childhood and adulthood (Clarke et al., 2014; Wheaton & Clarke, 2003). In this context, adult socioeconomic status may be thought of as a resource that may be mobilized to offset the exposure to poor socioeconomic conditions in childhood (Ferraro & Shippee, 2009). Similarly, although there is a nontrivial amount of income segregation in the United States, there is also a considerable amount of within neighborhood income variation as well (Reardon & Bischoff, 2011). Thus, the health benefits of living in a wealthy neighborhood are likely different for someone with relatively lower personal income, and, conversely, the health consequences of living in a poorer neighborhood are likely different for someone with relatively greater personal income. Moreover, there is evidence the relationship between individual-level SES and neighborhood-level SES on health outcomes are not merely additive, but combine in a multiplicative fashion. For instance, the association between childhood neighborhood disadvantage and mental health varies across levels of parental education, such that effect of neighborhood disadvantage is greatest among adults whose parents had lower levels of education (Wheaton & Clarke, 2003). Thus, our model offers a final assumption about the spatial relationship between SES and health: Assumption #3: The relationship within and between the temporal and spatial dimensions of SES and health are not just additive, but rather, interact with each other in a multiplicative fashion to shape adult health. To evaluate our conceptual matrix of SES, in the following sections, we bring our model to bear on the empirical case of the ways that household and neighborhood-level income in both childhood and adulthood relate to level of disability in adulthood. Assessing the Conceptual Matrix of SES: The Empirical Case of Household Income and Disability A significant share of the U.S. adult population experiences physical disability during their lives. According to Census estimates, 17% of nonelderly adults live with some form of physical disability (Brault, 2012). Moreover, an estimated 12% of the federal budget is spent on nonelderly adults with disabilities (Brault, 2012; Livermore, Stapleton, & O’Toole, 2011). Therefore, both in terms of personal and societal costs, physical disability is an important feature in the lives of many adults in the United States. Disability can be conceptualized as “the gap between a person’s intrinsic capabilities and the demands created by the social and physical environment—a product of the interaction of the individual with the environment” (Jette, 2006). That is, disability represents how a person’s physical condition limits their ability to perform the social role requirements necessary for their daily lives. The typical measurement of disability involves asking respondents how much they are limited in doing a particular task (e.g., using the restroom, preparing a meal, etc.) (Beauchamp et al., 2015). In order to understand and limit the prospects of developing a disability, and to potentially identify interventions to reduce these prospects, it is important to identify the factors that are associated with elevated risk of becoming disabled (Seeman, Bruce, & McAvay, 1996). Although there are numerous risk factors associated with disability, one of the primary risk factors for becoming disabled is low socioeconomic status (SES) (Antonovsky, 1967; Link & Phelan, 1995). In the following sections, we consider how SES may relate to disability across temporal and spatial dimensions according to our conceptual matrix of SES. SES and disability: The temporal dimension According to the conceptual matrix of SES, both (a) socioeconomic history and (b) socioeconomic present should shape the extent to which one experiences disability in adulthood, and this is largely borne out by the empirical evidence. On one hand, early life SES is related to a wide array of adult health conditions, including disability (Haas, 2008; Link & Phelan, 1995; Montez, 2013; Poulton et al., 2002). Moreover, childhood and adolescence are periods where nutrition and physical activity are important for the development of bone mass, and related to subsequent “risk of fracture and osteoporosis” (Cohen, Janicki-Deverts, Chen, & Matthews, 2010)—factors relevant to disability in adulthood. Because the socioeconomic circumstances to which one is exposed in childhood are related to late life disability, we developed our first hypothesis: Hypothesis 1: There will be a negative association between childhood income and level of disability in adulthood, net of income in adulthood. On the other hand, adult SES is clearly related to level of disability in adulthood (Haas, 2008; House et al., 1990; Montez, 2013). Lower levels of SES in general, and income in particular, are associated with greater levels of disability in adulthood (House et al., 1990). Inasmuch as income in adulthood is robustly linked to disability, we developed our second hypothesis: Hypothesis 2: There will be a negative association between adulthood income and level of disability in adulthood, net of income in childhood. SES and disability: The spatial dimension According to the conceptual matrix of SES, both (a) individual resources and (b) contextual resources should shape the extent to which one experiences disability in adulthood. In general, this is supported by the empirical evidence. On one hand, household-level SES can both protect against disability, but the lack of SES resources can be a risk factor for disability. Those with higher SES can use more available resources to manage and overcome life hardships, as well as access higher quality medical treatment and preventative care (Freedman & Martin, 1999; Phelan et al., 2010). Conversely, having lower household-level SES can be a financial stress which, in turn, can have negative consequences for poor physical health (Robert, 1998; Skinner, Zautra, & Reich, 2004; Wheaton, 1994). The link between household SES and disability is well established (Adler et al., 1994; Kail & Taylor, 2014; Nagi, 1976; Taylor, 2010). For example, lower individual-level SES is associated greater disability (Gayman, Turner, & Cui, 2008). Similarly, individual-level SES is associated with greater risk of becoming functionally limited, initial level of functional limitation, and subsequent growth of disability over time (Kail & Taylor, 2014). Clearly, individual-level SES is robustly linked with physical health. As such, based on this empirical evidence and our conceptual matrix of SES, we developed our third hypothesis: Hypothesis 3: There will be a negative association between household-level income and adult disability, net of neighborhood-level income. On the other hand, at the neighborhood-level, socioeconomic position shapes the very conditions in which one lives (Phelan et al., 2010). Indeed, neighborhoods represent the physical and social spaces in which adults spend much of their nonworking lives. Because different neighborhoods provide access to resources and expose residents to various risk and protective factors, neighborhoods represent another important spatial dimension of SES. Indeed, neighborhood-level SES is widely linked to physical health and disability (Diez Roux & Mair, 2010; Robert, 1998; Ross & Mirowsky, 2001). For example, poor neighborhoods are associated with declining physical functioning (Balfour & Kaplan, 2002; Schootman, Jeffe, Baker, & Walker, 2006). Neighborhoods involving high levels of economic strain may have negative health consequences by reducing the availability of health promoting resources and behaviors like physical exercise (Diez Roux & Mair, 2010; Phelan et al., 2010). Living in economically deprived neighborhoods may also produce stress which increases the risk of illness by threatening the immune system and accelerating the aging process (Segerstrom & Miller, 2004). Although the particular linking mechanisms are beyond the scope of the current study, there is evidence neighborhood SES has important implications for disability. Thus, based on our conceptual matrix of SES and existing empirical evidence, we developed our fourth hypothesis: Hypothesis 4: There will be a negative association between neighborhood-level income and disability, net of household-level. Finally, there is limited evidence to suggest that various dimensions of SES interact to shape physical functioning among adults (Montez, 2013). Therefore, based on this evidence, and the assumptions of the interactive nature of SES upon which our matrix of SES is predicated, we derive our final hypothesis, specifically: Hypothesis 5: The association between any one dimension of income and disability in adulthood will be conditioned by the other dimensions of income. In summary, based on insights from fundamental cause theory, the life course perspective, and neighborhood effects we have provided a Conceptual Matrix of SES by Temporal and Spatial Dimensions. We believe (1) the association between SES and health (in this case, the association between income and disability) involves (a) a temporal dimension involving (i) socioeconomic history and (ii) socioeconomic present, and (b) a spatial dimension involving (iii) individual resources and (iv) contextual resources, and (2) these four dimensions interact with one another to help shape adult health. Data And Method This study uses data from the Panel Study of Income Dynamics (2013) (PSID) for the years 1970 and 2013. The PSID is a nationally-representative, longitudinal study of U.S. families beginning in 1968 and has continued to follow sample members and their descendants over time. Our empirical objective is to evaluate the impact of adult and childhood income on disability in adulthood. Thus, for our analysis, we selected respondents who were under age 18 in 1970, and who were also in the sample as adults in 2013. Our final sample includes 2,344 individuals, who ranged in age from 1 to 17 in 1970, and 42 to 63 in 2013. Of the 7,590 children in the 1970 PSID sample, approximately 33% remain in the sample in 2013. The remainder dropped out of the sample, mostly because of refusal/nonresponse, or because of a PSID sample reduction occurring in 1997. The main sources of attrition in our sample are refusal/nonresponse (77% of those who dropped out), a PSID sample reduction occurring in 1997 (22%), death (0.6%), and jail/prison (0.2%). Although the PSID has maintained a reinterview rate between 95% and 98% across virtually all survey waves, even small attrition from wave to wave accumulates over time (Panel Study of Income Dynamics, 2013). Assessments of the representativeness of the PSID find that income and health estimates align closely with other cross-sectional surveys, despite accumulating attrition (Schoeni, Stafford, McGonagle, & Andreski, 2013). We conducted our own investigation of respondents who drop out of our sample. We found nonrespondents have 1970 family incomes that are about 20% lower than respondents who remain in the sample; childhood health among these two groups is roughly equivalent (based on two-tailed t tests comparing 1970 family income and reports of poor health in childhood among respondents and nonrespondents). We further restrict our analysis to respondents designated as household heads or spouses in 2013, because questions on disability were only asked of these household members. This eliminates 91 sample members, all of whom were residing in the home of other relatives or nonrelatives. Most commonly, they were residing the home of their parents. Second most commonly, they were residing in the home of their adult children. Further investigation reveals these sample members have family incomes that were about 18% lower in 1970 and 40% lower in 2013 than those who are retained in our sample (based on two-tailed t tests comparing 1970 and 2013 family incomes among 2013 heads/wives and non-heads/wives). Given the selectivity of our sample, it is important to note that results may not be representative of families with low incomes. Our final sample includes 2,344 individuals, who ranged in age from 1 to 17 in 1970, and 42 to 63 in 2013. Discrepancies between 1970 and 2013 age arise based on the timing of interviews. Ages of individuals are asked and reported in each wave of the study. Because interviews are seldom taken exactly 1 year apart for the same individual wave to wave, age gaps can arise. Also, individuals’ ages or birthdates can be misreported. The PSID conducts internal consistency checks for age discrepancies, but they do no alter ages if it cannot be determined which age is correct. Dependent Variable Our outcome variable is disability in adulthood. Disability is measured in 2013 with six items asking whether respondents had any difficulty doing the following activities by themselves and without special equipment: using the toilet; getting outside; walking; getting in and out of bed; eating; and dressing. Responses were coded as either “yes” or “no,” and then summed into an index of “yes” responses. Thirty-seven respondents were missing values on some or all of the disability items; their values were multiply imputed. Study Variables Socioeconomic status (SES) is measured at two time points—childhood and adult—and at two levels—household and neighborhood. SES is a construct representing an individual’s economic and social position, generally based on income, education, and occupation. Because income, education, and occupation are highly correlated, we rely on a single indicator—income—to represent SES in our study. Household SES is measured in 1970 (childhood) and in 2013 (adulthood) as a continuous measure representing the total income of all family members in the previous year including all taxable, transfer, and social security income. Income is self-reported by the household head for all family members. In 1970, negative values (i.e., net losses) were bottom coded at $1. To be consistent, we also recode negative values in 2013 to $1. All results are reported in real 1970 and 2013 dollars, and are not adjusted for inflation. Data exploration revealed a significant right skew produced by a few families with exceedingly high incomes. To adjust for non-normality and improve the fit of our predictive models, we use the natural log of childhood and adult family income in our regression analyses. One respondent is missing the value for 1970 family income, and was multiply imputed. Neighborhood SES was captured with a continuous measure representing the average income in the neighborhood surrounding the respondent in childhood and adulthood. To identify neighborhoods, we use the PSID’s supplemental Geospatial Match Files to link addresses of PSID respondents in 1970 and 2013 to corresponding codes for census tracts. Tract data come from the 1970 U.S. decennial census and the 2013 American Community Survey 5-Year Estimates, with tract boundaries normalized to 2010 (GeoLytics, 2014). Tract average family income was calculated by dividing the total aggregate income in the previous year of all families in the tract by the total number of families in the tract in 1970 and 2013. Neighborhood family income is also non-normally distributed and exhibits a right skew. We therefore used the natural log of neighborhood family income in our regression models. Missing values for neighborhood SES required a special strategy. In 1970, many less-populated areas of the United States had not yet been divided into census tracts. Consequently, some tracts have missing data in 1970, including 635 respondents in our sample. For respondents with missing 1970 tract values, we substitute the overall tract mean and incorporate a dichotomous variable indicating this substitution into all models. We are precluded from employing a more sophisticated imputation strategy for these tracts because all 1970 tract variables are missing (not only income), leaving us with minimal information to impute from. However, we felt it was important to retain individuals with missing 1970 tract values in the analysis to preserve the representativeness of our sample. In sensitivity analyses, we used a measure of tract level poverty rate instead of tract level income. Those results were substantively similar to those presented below and are available upon request. Independent Variables Several time-invariant covariates of disability are also included in the analyses. They include the respondent’s sex (male = 1, female = 0) and race/ethnicity (dummy coded as black, white, and other). Childhood health is included to account for the possibility respondents with poor health in childhood are more likely to develop disability in adulthood. Childhood self-reported health is based on a 2013 survey question asking respondents to retrospectively recall their health before they were 17 years old. Responses were recorded as “excellent,” “very good,” “good,” “fair,” or “poor.” In our analyses, we dichotomize the responses so a value of “0” represents excellent/very good/good, and a value of “1” represents fair/poor self-reported childhood health. Finally, age is included as a covariate in adulthood, since older individuals may be more likely to experience disability. In our sample, 10 respondents are missing values for race/ethnicity and 55 respondents are missing values for childhood health. These values are multiply imputed. Following White et al., (2011), all of our model’s covariates and the outcome variable (i.e., functional limitations) are included in the multiple imputation models predicting race/ethnicity and childhood health. According to the imputed data, respondents who were missing values for race are slightly less likely to be identified as white and slightly more likely to be identified as black or other, compared to respondents without missing values. There is no substantial difference in childhood health in our imputed and nonimputed samples. Analysis Plan Our empirical goal is to assess how the temporal and spatial dimensions of income impact disability in adulthood. Thus, we rely on a negative binomial regression equation to predict the number of disability items with which an individual expediencies difficultly. The negative binomial model is specifically designed to predict overdispersed count variables like our dependent variable. It is more appropriate than standard OLS for count variables (which are based on integers and cannot be negative) rather than continuous values. Indeed, model fit statistics reveal a negative binomial model is preferred to OLS in our data estimates. All models utilize Huber-White clustered standard errors to account for the clustering of siblings who grew up in the same 1970 household. All analyses were conducted in STATA/MP 14.2 (StataCorp, 2015). Because we were concerned about the extent selection into 2013 neighborhood might be biasing our results, we ran additional robustness checks. Those are available as an online supplement to this manuscript, but in short, findings were comparable, and so we present the more parsimonious results below. Results Univariate Results As shown in Table 1, the average person in this sample experienced 0.29 of a disability in adulthood (and 13% had one or more disability). Average household income in 1970 was $10,270 and $87,680 in 2013, and average neighborhood income was $10,469 in 1970 and $79,374 in 2013. Table 1. Summary Statistics   Mean/Proportion  SD  Minimum  Maximum  Functional Limitations  0.29  0.90  0.00  6.00  HH Income 1970  10,269.84  7,492.34  1.00  91,660.00  HH Income 2013  87,680.46  147,396.00  1.00  3,316,000.00  Neighborhood Income 1970  10,468.69  3,620.94  3,664.92  47,720.74  Neighborhood Income 2013  79,373.95  40,936.13  1.00  458,681.00  Male  0.42  --  1.00  1.00  Black  0.43  --  0.00  1.00  White  0.54  --  0.00  1.00  Other  0.03  --  0.00  1.00  Childhood health  0.04  0.19  0.00  1.00  Age 2013  51.94  4.99  42.00  63.00  N. Inc. 1970 Missing  0.27  --  0.00  1.00  N. Inc.2013 Missinga  0.00  --  0.00  1.00    Mean/Proportion  SD  Minimum  Maximum  Functional Limitations  0.29  0.90  0.00  6.00  HH Income 1970  10,269.84  7,492.34  1.00  91,660.00  HH Income 2013  87,680.46  147,396.00  1.00  3,316,000.00  Neighborhood Income 1970  10,468.69  3,620.94  3,664.92  47,720.74  Neighborhood Income 2013  79,373.95  40,936.13  1.00  458,681.00  Male  0.42  --  1.00  1.00  Black  0.43  --  0.00  1.00  White  0.54  --  0.00  1.00  Other  0.03  --  0.00  1.00  Childhood health  0.04  0.19  0.00  1.00  Age 2013  51.94  4.99  42.00  63.00  N. Inc. 1970 Missing  0.27  --  0.00  1.00  N. Inc.2013 Missinga  0.00  --  0.00  1.00  Note: n = 2,344; This is reported based on the 1st of 10 imputed datasets (summary statistics for all 10 available upon request). aThis value is less than 0.01 because of rounding. View Large Table 1. Summary Statistics   Mean/Proportion  SD  Minimum  Maximum  Functional Limitations  0.29  0.90  0.00  6.00  HH Income 1970  10,269.84  7,492.34  1.00  91,660.00  HH Income 2013  87,680.46  147,396.00  1.00  3,316,000.00  Neighborhood Income 1970  10,468.69  3,620.94  3,664.92  47,720.74  Neighborhood Income 2013  79,373.95  40,936.13  1.00  458,681.00  Male  0.42  --  1.00  1.00  Black  0.43  --  0.00  1.00  White  0.54  --  0.00  1.00  Other  0.03  --  0.00  1.00  Childhood health  0.04  0.19  0.00  1.00  Age 2013  51.94  4.99  42.00  63.00  N. Inc. 1970 Missing  0.27  --  0.00  1.00  N. Inc.2013 Missinga  0.00  --  0.00  1.00    Mean/Proportion  SD  Minimum  Maximum  Functional Limitations  0.29  0.90  0.00  6.00  HH Income 1970  10,269.84  7,492.34  1.00  91,660.00  HH Income 2013  87,680.46  147,396.00  1.00  3,316,000.00  Neighborhood Income 1970  10,468.69  3,620.94  3,664.92  47,720.74  Neighborhood Income 2013  79,373.95  40,936.13  1.00  458,681.00  Male  0.42  --  1.00  1.00  Black  0.43  --  0.00  1.00  White  0.54  --  0.00  1.00  Other  0.03  --  0.00  1.00  Childhood health  0.04  0.19  0.00  1.00  Age 2013  51.94  4.99  42.00  63.00  N. Inc. 1970 Missing  0.27  --  0.00  1.00  N. Inc.2013 Missinga  0.00  --  0.00  1.00  Note: n = 2,344; This is reported based on the 1st of 10 imputed datasets (summary statistics for all 10 available upon request). aThis value is less than 0.01 because of rounding. View Large Multivariable Results Results of negative binomial regression predicting adult disability are shown in Table 2. In model 1, both of the measures of socioeconomic present (2013 household- and neighborhood-level income) are included (net of controls). Both measures are associated with lower disability. In model 2, both of the measures of socioeconomic history (1970 household- and neighborhood-level income) are included. Household income is associated with lower disability in adulthood, but neighborhood-level income is not. In model 3, both measures of individual resources (1970 and 2013 household-level income) are included, and both measures are associated with lower disability. In model 4, both measures of contextual resources (1970 and 2013 neighborhood-level income) income are included. In this model, 2013 neighborhood-level is associated with lower disability in adulthood, but 1970 neighborhood-level income is not. In model 5, all four measures of income are included and, as was the case in the previous models, both 1970 and 2013 household income are associated with lower disability, as is 2013 neighborhood-level income, but 1970 neighborhood-level income is not associated with disability in adulthood. Table 2. Negative Binomial Regression of Functional Limitations   (1)  (2)  (3)  (4)  (5)  HH Income 1970a    −0.396**  −0.265*    −0.289*      (0.120)  (0.107)    (0.114)  HH Income 2013a  −0.317***    −0.367***    −0.310***    (0.061)    (0.066)    (0.059)  Neighborhood Income 1970a    −0.128    −0.016  0.109      (0.289)    (0.278)  (0.277)  Neighborhood Income 2013a  −0.673***      −0.956***  −0.669***    (0.182)      (0.178)  (0.185)  Maleb  −0.689***  −0.647***  −0.683***  −0.652***  −0.661***    (0.138)  (0.139)  (0.138)  (0.141)  (0.138)  Blackc  −0.063  0.087  −0.070  0.118  −0.251    (0.163)  (0.182)  (0.171)  (0.183)  (0.185)  Otherc  0.310  0.166  0.292  0.251  0.191    (0.331)  (0.337)  (0.336)  (0.331)  (0.341)  Childhood health  0.503*  0.577**  0.482*  0.593**  0.521*    (0.218)  (0.217)  (0.221)  (0.218)  (0.222)  Age 2013  0.060***  0.063***  0.060***  0.061***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd    −0.241    −0.167  −0.358*      (0.175)    (0.173)  (0.174)  N. INC. 2013 Missingc  −17.995***      −17.714***  −18.046***    (1.117)      (0.581)  (1.123)  Constant  6.591**  0.334  2.009  6.454*  8.021**    (2.156)  (2.682)  (1.367)  (3.064)  (3.027)    (1)  (2)  (3)  (4)  (5)  HH Income 1970a    −0.396**  −0.265*    −0.289*      (0.120)  (0.107)    (0.114)  HH Income 2013a  −0.317***    −0.367***    −0.310***    (0.061)    (0.066)    (0.059)  Neighborhood Income 1970a    −0.128    −0.016  0.109      (0.289)    (0.278)  (0.277)  Neighborhood Income 2013a  −0.673***      −0.956***  −0.669***    (0.182)      (0.178)  (0.185)  Maleb  −0.689***  −0.647***  −0.683***  −0.652***  −0.661***    (0.138)  (0.139)  (0.138)  (0.141)  (0.138)  Blackc  −0.063  0.087  −0.070  0.118  −0.251    (0.163)  (0.182)  (0.171)  (0.183)  (0.185)  Otherc  0.310  0.166  0.292  0.251  0.191    (0.331)  (0.337)  (0.336)  (0.331)  (0.341)  Childhood health  0.503*  0.577**  0.482*  0.593**  0.521*    (0.218)  (0.217)  (0.221)  (0.218)  (0.222)  Age 2013  0.060***  0.063***  0.060***  0.061***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd    −0.241    −0.167  −0.358*      (0.175)    (0.173)  (0.174)  N. INC. 2013 Missingc  −17.995***      −17.714***  −18.046***    (1.117)      (0.581)  (1.123)  Constant  6.591**  0.334  2.009  6.454*  8.021**    (2.156)  (2.682)  (1.367)  (3.064)  (3.027)  Note: n = 2,344; *p < .05. **p < .01. ***p < .001. Standard errors are shown in parentheses. Pooled results from 10 imputation datasets. aAll income measures are transformed using the natural log. References are: b= Females; c= Whites; and d= Nonmissing. View Large Table 2. Negative Binomial Regression of Functional Limitations   (1)  (2)  (3)  (4)  (5)  HH Income 1970a    −0.396**  −0.265*    −0.289*      (0.120)  (0.107)    (0.114)  HH Income 2013a  −0.317***    −0.367***    −0.310***    (0.061)    (0.066)    (0.059)  Neighborhood Income 1970a    −0.128    −0.016  0.109      (0.289)    (0.278)  (0.277)  Neighborhood Income 2013a  −0.673***      −0.956***  −0.669***    (0.182)      (0.178)  (0.185)  Maleb  −0.689***  −0.647***  −0.683***  −0.652***  −0.661***    (0.138)  (0.139)  (0.138)  (0.141)  (0.138)  Blackc  −0.063  0.087  −0.070  0.118  −0.251    (0.163)  (0.182)  (0.171)  (0.183)  (0.185)  Otherc  0.310  0.166  0.292  0.251  0.191    (0.331)  (0.337)  (0.336)  (0.331)  (0.341)  Childhood health  0.503*  0.577**  0.482*  0.593**  0.521*    (0.218)  (0.217)  (0.221)  (0.218)  (0.222)  Age 2013  0.060***  0.063***  0.060***  0.061***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd    −0.241    −0.167  −0.358*      (0.175)    (0.173)  (0.174)  N. INC. 2013 Missingc  −17.995***      −17.714***  −18.046***    (1.117)      (0.581)  (1.123)  Constant  6.591**  0.334  2.009  6.454*  8.021**    (2.156)  (2.682)  (1.367)  (3.064)  (3.027)    (1)  (2)  (3)  (4)  (5)  HH Income 1970a    −0.396**  −0.265*    −0.289*      (0.120)  (0.107)    (0.114)  HH Income 2013a  −0.317***    −0.367***    −0.310***    (0.061)    (0.066)    (0.059)  Neighborhood Income 1970a    −0.128    −0.016  0.109      (0.289)    (0.278)  (0.277)  Neighborhood Income 2013a  −0.673***      −0.956***  −0.669***    (0.182)      (0.178)  (0.185)  Maleb  −0.689***  −0.647***  −0.683***  −0.652***  −0.661***    (0.138)  (0.139)  (0.138)  (0.141)  (0.138)  Blackc  −0.063  0.087  −0.070  0.118  −0.251    (0.163)  (0.182)  (0.171)  (0.183)  (0.185)  Otherc  0.310  0.166  0.292  0.251  0.191    (0.331)  (0.337)  (0.336)  (0.331)  (0.341)  Childhood health  0.503*  0.577**  0.482*  0.593**  0.521*    (0.218)  (0.217)  (0.221)  (0.218)  (0.222)  Age 2013  0.060***  0.063***  0.060***  0.061***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd    −0.241    −0.167  −0.358*      (0.175)    (0.173)  (0.174)  N. INC. 2013 Missingc  −17.995***      −17.714***  −18.046***    (1.117)      (0.581)  (1.123)  Constant  6.591**  0.334  2.009  6.454*  8.021**    (2.156)  (2.682)  (1.367)  (3.064)  (3.027)  Note: n = 2,344; *p < .05. **p < .01. ***p < .001. Standard errors are shown in parentheses. Pooled results from 10 imputation datasets. aAll income measures are transformed using the natural log. References are: b= Females; c= Whites; and d= Nonmissing. View Large As shown in Table 3, we tested all six possible interaction terms comprised of our four measures of income. Although two of those interactions were nonsignificant (1970 household income × 1970 neighborhood income, and 1970 household income × 2013 neighborhood income), the four remaining interactions were significant. The negative association between each of these interactions and disability suggests either (a) when individuals have/had lower levels of income in one dimension but higher levels of income in another dimension, the deleterious association of low income is partially mitigated by having higher levels of income in another dimension, or (b) having more income in two dimensions combines in such a way that helps mitigate the risks of disability to a greater extent than one would expect by just considering the direct benefit of each of the two dimensions of income. In short, these four significant interaction terms suggests the more dimensions in which one has/had high income, the better it is for their physical health in adulthood. Table 3. Negative Binomial Regression of Functional Limitations   (1)  (2)  (3)  (4)  (5)  (6)  HH Income 1970 X  −0.332             Neighborhood Income 1970  (0.313)            HH Income 1970 X    −0.175**           HH Income 2013    (0.063)          Neighborhood Income 1970 X      −1.290*         Neighborhood Income 2013      (0.518)        HH Income 1970 X        −0.197       Neighborhood Income 2013        (0.213)      HH Income 2013 X          −0.412*     Neighborhood Income 1970          (0.166)    HH Income 2013 X            −0.449**   Neighborhood Income 2013            (0.146)  HH Income 1970a  2.753  1.536*  −0.260*  1.892  −0.268*  −0.276*    (2.869)  (0.676)  (0.110)  (2.345)  (0.112)  (0.111)  HH Income 2013a  −0.307***  1.193*  −0.301***  −0.307***  3.453*  4.646**    (0.059)  (0.523)  (0.058)  (0.059)  (1.501)  (1.587)  Neighborhood Income 1970a  3.105  0.181  14.478*  0.158  4.405*  0.171    (2.815)  (0.276)  (5.781)  (0.283)  (1.755)  (0.275)  Neighborhood Income 2013a  −0.642**  −0.590**  11.158*  1.093  −0.630**  4.097**    (0.184)  (0.187)  (4.737)  (1.908)  (0.184)  (1.524)  Maleb  −0.675***  −0.660***  −0.656***  −0.664***  −0.667***  −0.657***    (0.138)  (0.139)  (0.139)  (0.138)  (0.139)  (0.138)  Blackc  −0.252  −0.259  −0.231  −0.247  −0.262  −0.240    (0.183)  (0.185)  (0.182)  (0.185)  (0.183)  (0.181)  Otherc  0.191  0.192  0.187  0.193  0.196  0.170    (0.341)  (0.336)  (0.345)  (0.342)  (0.343)  (0.336)  Childhood health  0.510*  0.538*  0.525*  0.522*  0.521*  0.512*    (0.221)  (0.222)  (0.220)  (0.222)  (0.223)  (0.221)  Age  0.064***  0.062***  0.062***  0.063***  0.063***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd  −0.415*  −0.335  −0.446*  −0.370*  −0.393*  −0.401*    (0.181)  (0.172)  (0.178)  (0.173)  (0.176)  (0.171)  N. INC. 2013 Missingd  −17.889***  −19.268***  −18.155***  −18.329***  −19.353***  −19.272***    (1.039)  (1.445)  (1.078)  (1.063)  (1.643)  (1.255)  Constant  −19.659  −9.120  −123.937*  −11.914  −31.810*  −45.234**    (26.039)  (6.880)  (52.857)  (21.552)  (16.227)  (17.054)    (1)  (2)  (3)  (4)  (5)  (6)  HH Income 1970 X  −0.332             Neighborhood Income 1970  (0.313)            HH Income 1970 X    −0.175**           HH Income 2013    (0.063)          Neighborhood Income 1970 X      −1.290*         Neighborhood Income 2013      (0.518)        HH Income 1970 X        −0.197       Neighborhood Income 2013        (0.213)      HH Income 2013 X          −0.412*     Neighborhood Income 1970          (0.166)    HH Income 2013 X            −0.449**   Neighborhood Income 2013            (0.146)  HH Income 1970a  2.753  1.536*  −0.260*  1.892  −0.268*  −0.276*    (2.869)  (0.676)  (0.110)  (2.345)  (0.112)  (0.111)  HH Income 2013a  −0.307***  1.193*  −0.301***  −0.307***  3.453*  4.646**    (0.059)  (0.523)  (0.058)  (0.059)  (1.501)  (1.587)  Neighborhood Income 1970a  3.105  0.181  14.478*  0.158  4.405*  0.171    (2.815)  (0.276)  (5.781)  (0.283)  (1.755)  (0.275)  Neighborhood Income 2013a  −0.642**  −0.590**  11.158*  1.093  −0.630**  4.097**    (0.184)  (0.187)  (4.737)  (1.908)  (0.184)  (1.524)  Maleb  −0.675***  −0.660***  −0.656***  −0.664***  −0.667***  −0.657***    (0.138)  (0.139)  (0.139)  (0.138)  (0.139)  (0.138)  Blackc  −0.252  −0.259  −0.231  −0.247  −0.262  −0.240    (0.183)  (0.185)  (0.182)  (0.185)  (0.183)  (0.181)  Otherc  0.191  0.192  0.187  0.193  0.196  0.170    (0.341)  (0.336)  (0.345)  (0.342)  (0.343)  (0.336)  Childhood health  0.510*  0.538*  0.525*  0.522*  0.521*  0.512*    (0.221)  (0.222)  (0.220)  (0.222)  (0.223)  (0.221)  Age  0.064***  0.062***  0.062***  0.063***  0.063***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd  −0.415*  −0.335  −0.446*  −0.370*  −0.393*  −0.401*    (0.181)  (0.172)  (0.178)  (0.173)  (0.176)  (0.171)  N. INC. 2013 Missingd  −17.889***  −19.268***  −18.155***  −18.329***  −19.353***  −19.272***    (1.039)  (1.445)  (1.078)  (1.063)  (1.643)  (1.255)  Constant  −19.659  −9.120  −123.937*  −11.914  −31.810*  −45.234**    (26.039)  (6.880)  (52.857)  (21.552)  (16.227)  (17.054)  Note: n = 2,344; *p < .05. **p < .01. ***p < .001. Standard errors are shown in parentheses. Pooled results from 10 imputation datasets. aAll income measures are transformed using the natural log. References are: b= Females; c= Whites; and d= Nonmissing. View Large Table 3. Negative Binomial Regression of Functional Limitations   (1)  (2)  (3)  (4)  (5)  (6)  HH Income 1970 X  −0.332             Neighborhood Income 1970  (0.313)            HH Income 1970 X    −0.175**           HH Income 2013    (0.063)          Neighborhood Income 1970 X      −1.290*         Neighborhood Income 2013      (0.518)        HH Income 1970 X        −0.197       Neighborhood Income 2013        (0.213)      HH Income 2013 X          −0.412*     Neighborhood Income 1970          (0.166)    HH Income 2013 X            −0.449**   Neighborhood Income 2013            (0.146)  HH Income 1970a  2.753  1.536*  −0.260*  1.892  −0.268*  −0.276*    (2.869)  (0.676)  (0.110)  (2.345)  (0.112)  (0.111)  HH Income 2013a  −0.307***  1.193*  −0.301***  −0.307***  3.453*  4.646**    (0.059)  (0.523)  (0.058)  (0.059)  (1.501)  (1.587)  Neighborhood Income 1970a  3.105  0.181  14.478*  0.158  4.405*  0.171    (2.815)  (0.276)  (5.781)  (0.283)  (1.755)  (0.275)  Neighborhood Income 2013a  −0.642**  −0.590**  11.158*  1.093  −0.630**  4.097**    (0.184)  (0.187)  (4.737)  (1.908)  (0.184)  (1.524)  Maleb  −0.675***  −0.660***  −0.656***  −0.664***  −0.667***  −0.657***    (0.138)  (0.139)  (0.139)  (0.138)  (0.139)  (0.138)  Blackc  −0.252  −0.259  −0.231  −0.247  −0.262  −0.240    (0.183)  (0.185)  (0.182)  (0.185)  (0.183)  (0.181)  Otherc  0.191  0.192  0.187  0.193  0.196  0.170    (0.341)  (0.336)  (0.345)  (0.342)  (0.343)  (0.336)  Childhood health  0.510*  0.538*  0.525*  0.522*  0.521*  0.512*    (0.221)  (0.222)  (0.220)  (0.222)  (0.223)  (0.221)  Age  0.064***  0.062***  0.062***  0.063***  0.063***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd  −0.415*  −0.335  −0.446*  −0.370*  −0.393*  −0.401*    (0.181)  (0.172)  (0.178)  (0.173)  (0.176)  (0.171)  N. INC. 2013 Missingd  −17.889***  −19.268***  −18.155***  −18.329***  −19.353***  −19.272***    (1.039)  (1.445)  (1.078)  (1.063)  (1.643)  (1.255)  Constant  −19.659  −9.120  −123.937*  −11.914  −31.810*  −45.234**    (26.039)  (6.880)  (52.857)  (21.552)  (16.227)  (17.054)    (1)  (2)  (3)  (4)  (5)  (6)  HH Income 1970 X  −0.332             Neighborhood Income 1970  (0.313)            HH Income 1970 X    −0.175**           HH Income 2013    (0.063)          Neighborhood Income 1970 X      −1.290*         Neighborhood Income 2013      (0.518)        HH Income 1970 X        −0.197       Neighborhood Income 2013        (0.213)      HH Income 2013 X          −0.412*     Neighborhood Income 1970          (0.166)    HH Income 2013 X            −0.449**   Neighborhood Income 2013            (0.146)  HH Income 1970a  2.753  1.536*  −0.260*  1.892  −0.268*  −0.276*    (2.869)  (0.676)  (0.110)  (2.345)  (0.112)  (0.111)  HH Income 2013a  −0.307***  1.193*  −0.301***  −0.307***  3.453*  4.646**    (0.059)  (0.523)  (0.058)  (0.059)  (1.501)  (1.587)  Neighborhood Income 1970a  3.105  0.181  14.478*  0.158  4.405*  0.171    (2.815)  (0.276)  (5.781)  (0.283)  (1.755)  (0.275)  Neighborhood Income 2013a  −0.642**  −0.590**  11.158*  1.093  −0.630**  4.097**    (0.184)  (0.187)  (4.737)  (1.908)  (0.184)  (1.524)  Maleb  −0.675***  −0.660***  −0.656***  −0.664***  −0.667***  −0.657***    (0.138)  (0.139)  (0.139)  (0.138)  (0.139)  (0.138)  Blackc  −0.252  −0.259  −0.231  −0.247  −0.262  −0.240    (0.183)  (0.185)  (0.182)  (0.185)  (0.183)  (0.181)  Otherc  0.191  0.192  0.187  0.193  0.196  0.170    (0.341)  (0.336)  (0.345)  (0.342)  (0.343)  (0.336)  Childhood health  0.510*  0.538*  0.525*  0.522*  0.521*  0.512*    (0.221)  (0.222)  (0.220)  (0.222)  (0.223)  (0.221)  Age  0.064***  0.062***  0.062***  0.063***  0.063***  0.064***    (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  (0.014)  N. INC. 1970 Missingd  −0.415*  −0.335  −0.446*  −0.370*  −0.393*  −0.401*    (0.181)  (0.172)  (0.178)  (0.173)  (0.176)  (0.171)  N. INC. 2013 Missingd  −17.889***  −19.268***  −18.155***  −18.329***  −19.353***  −19.272***    (1.039)  (1.445)  (1.078)  (1.063)  (1.643)  (1.255)  Constant  −19.659  −9.120  −123.937*  −11.914  −31.810*  −45.234**    (26.039)  (6.880)  (52.857)  (21.552)  (16.227)  (17.054)  Note: n = 2,344; *p < .05. **p < .01. ***p < .001. Standard errors are shown in parentheses. Pooled results from 10 imputation datasets. aAll income measures are transformed using the natural log. References are: b= Females; c= Whites; and d= Nonmissing. View Large Discussion In this article, we developed conceptual matrix of socioeconomic status, and tested five hypotheses derived from a set of three assumptions upon which our conceptual matrix rests. Results indicated both 1970 and 2013 household income were associated with lower disability in adulthood, as was 2013 neighborhood-level income, but 1970 neighborhood-level income was not associated with disability in adulthood. Further, four of the six possible interactions between the multiple dimensions of income were associated with significant reductions in disability. These findings provide several important empirical insights, but also help inform a framework for thinking about the multidimensional ways in which SES relates to health more generally. First, the hypothesis that there would be a negative association between childhood income and level of disability in adulthood, net of income in adulthood was partially supported; childhood family income was associated with a reduction in level of disability in adulthood. In contrast, childhood neighborhood income was never associated with a reduction in level of disability. Additionally, the hypothesis that there would be negative association between adult income and level of disability in adulthood, net of income in childhood was supported; both individual- and neighborhood-level income were associated with a reduction in level of disability in adulthood. Taken together, these findings support the first assumption of our conceptual matrix of SES and health, that is: current health is a function of both an individual’s socioeconomic present but also an individual’s socioeconomic history. Second, both the hypothesis that there would be a negative association between household-level income and adult disability, net of neighborhood-level income, and the hypothesis that there would be a negative association between neighborhood-level income and disability, net of household-level, we generally supported. As mentioned above, in all of our models, both 1970 and 2013 household income, as well as 2013 neighborhood income (but not 1970 neighborhood income) were always associated with a reduction in level of disability. Therefore, these findings support the second assumption of our conceptual matrix of SES and health, that is: current health is a function of both individual resources and an individual’s contextual resources. Third, we hypothesized the association between any one dimension of income and disability in adulthood would be conditioned by the other dimensions of income. Four of the six interactions were significant, so this hypothesis was generally supported. These findings support the second assumption of our conceptual matrix of SES and health, that is: The relationship within and between the temporal and spatial dimensions of SES and health are not just additive, but rather, interact with each other in a multiplicative fashion to shape adult health. This research contributes to the literature on SES and health in two primary ways. First, we have provided a conceptual matrix of SES by temporal and spatial dimensions in order to advance the ways in which we think about how various forms of SES may relate to general health. Second, to our knowledge, we are the first to empirically document how child and adult income, measured as both household income and neighborhood income, relate to adult disability. These contributions are limited by a variety of factors. First, our research was guided three theoretical perspectives; however, we are not explicitly testing any of these theories. Second, although we believe our conceptual matrix of SES should be relevant to forms of SES beyond just income and types of health beyond disability, we only tested the association between our four dimensions of income and disability here. Therefore, we cannot make empirical claims beyond those relationships. Third, it is possible that childhood health impacts both adult income and adult disability, but we are unable to address that with the current data. Hopefully, future data collections efforts will provide additional insights. Finally, it is possible census tracts do not adequately capture neighborhoods as residences perceive them (Coulton, Korbin, Chan, & Su, 2001). We acknowledge this limitation, but, unfortunately, this is the linkage we are able to make with the data we have. Conclusion In sum, these findings provide a much more nuanced understanding of and a conceptual model for interpreting the ways in which the temporal and spatial dimensions of socioeconomic status relate to disability. We argue the association between SES and disability is comprised of both socioeconomic circumstances and socioeconomic resources. At the same time, the association between SES and disability is related to both individual resources and collective resources. Moreover, these four components generally function interactively to shape adult health. We hope our conceptual matrix of SES by temporal and spatial dimensions will provide a usual typology to guide future research on the relationship between various forms of SES and wider array of health outcomes, particularly in an era where the ability to link geocoded longitudinal individual data with spatial data is becoming increasingly possible. The degree of disability in adulthood is an important concern for population health because it is highly predictive of future health declines. According to our results, interventions seeking to reduce or delay disability need to take a wide view and consider the temporal and spatial dimensions of SES that contribute to the development of disability. For instance, policies and programs that alleviate childhood poverty are likely to produce lasting effects for health into adulthood. At the same time, policies and programs that help target the deleterious consequences of adult poverty and living in poor neighborhoods are essential for physical health as well. We believe our conceptual matrix of SES by temporal and spatial dimensions provides a useful tool to help guide research as it disentangles the complex web of associations between SES and health. Funding The collection of data used in this study was partly supported by the National Institutes of Health under grant number R01 HD069609 and the National Science Foundation under award number 1157698. Conflict of Interest None reported. Acknowledgments B. L. Kail would like to thank Dawn C. Carr for providing feedback on some of the ideas written about in this manuscript. References Adler, N. E., Boyce, T., Chesney, M. A., Cohen, S., Folkman, S., Kahn, R. L., & Syme, S. L. ( 1994). Socioeconomic status and health. The challenge of the gradient. The American Psychologist , 49, 15– 24. doi:10.1037/0003-066X.49.1.15 Google Scholar CrossRef Search ADS   Aneshensel, C. S., & Sucoff, C. A. ( 1996). The neighborhood context of adolescent mental health. Journal of Health and Social Behavior , 37, 293– 310. doi:10.2307/2137258 Google Scholar CrossRef Search ADS   Antonovsky, A. ( 1967). Social class, life expectancy and overall mortality. The Milbank Memorial Fund Quarterly , 45, 31– 73. doi:10.2307/3348839 Google Scholar CrossRef Search ADS   Balfour, J. L., & Kaplan, G. A. ( 2002). Neighborhood environment and loss of physical function in older adults: Evidence from the Alameda County Study. American Journal of Epidemiology , 155, 507– 515. doi:10.1093/aje/155.6.507 Google Scholar CrossRef Search ADS   Beauchamp, M. K., Bean, J. F., Ward, R. E., Kurlinski, L. A., Latham, N. K., & Jette, A. M. ( 2015). How should disability be measured in older adults? An analysis from the Boston rehabilitative impairment study of the elderly. Journal of the American Geriatrics Society , 63, 1187– 1191. doi: 10.1111/jgs.13453 Google Scholar CrossRef Search ADS   Bloome, D., & Western, B. ( 2011). Cohort change and racial differences in educational and income mobility. Social Forces , 90, 375– 395. doi:10.1093/sf/sor002 Google Scholar CrossRef Search ADS   Brault, M. W. ( 2012). Americans with disabilities: 2010. Current Population Reports , 7, 0– 131. Clarke, P., Morenoff, J., Debbink, M., Golberstein, E., Elliott, M. R., & Lantz, P. M. ( 2014). Cumulative exposure to neighborhood context: Consequences for health transitions over the adult life course. Research on Aging , 36, 115– 142. doi: 10.1177/0164027512470702 Google Scholar CrossRef Search ADS   Cohen, S., Janicki-Deverts, D., Chen, E., & Matthews, K. A. ( 2010). Childhood socioeconomic status and adult health. Annals of the New York Academy of Sciences , 1186, 37– 55. doi: 10.1111/j.1749-6632.2009.05334.x Google Scholar CrossRef Search ADS   Coulton, C. J., Korbin, J., Chan, T., & Su, M. ( 2001). Mapping residents’ perceptions of neighborhood boundaries: a methodological note. American Journal of Community Psychology , 29, 371– 383. doi: 10.1023/A:1010303419034 Google Scholar CrossRef Search ADS   Cutrona, C. E., Wallace, G., & Wesner, K. A. ( 2006). Neighborhood characteristics and depression: An examination of stress processes. Current Directions in Psychological Science , 15, 188– 192. doi: 10.1111/j.1467-8721.2006.00433.x Google Scholar CrossRef Search ADS   Diez Roux, A. V., & Mair, C. ( 2010). Neighborhoods and health. Annals of the New York Academy of Sciences , 1186, 125– 145. doi: 10.1111/j.1749-6632.2009.05333.x Google Scholar CrossRef Search ADS   ElderJr, G. H. ( 1999). Children of the great depression: Social change in life experience . Boulder, CO: Westview Press. Ellen, I. G., Mijanovich, T., & Dillman, K.-N. ( 2001). Neighborhood effects on health: Exploring the links and assessing the evidence. Journal of Urban Affairs , 23, 391– 408. doi:10.1111/0735- 2166.00096 Google Scholar CrossRef Search ADS   Ferraro, K. F. ( 2011). Health and aging: Early origins, persistent inequalities? In A. R. Settersten & L. J. Angel (Eds.), Handbook of sociology of aging  (pp. 465– 475). New York, NY: Springer New York. Google Scholar CrossRef Search ADS   Ferraro, K. F., & Shippee, T. P. ( 2009). Aging and cumulative inequality: How does inequality get under the skin? The Gerontologist , 49, 333– 343. doi: 10.1093/geront/gnp034 Google Scholar CrossRef Search ADS   Ferraro, K., Shippee, T., & Schafer, M. ( 2009). Cumulative inequality theory for research on aging and the life course. In Bengtson, V., Silverstein, M., Norella, P., & Gans, D. (Eds.), Handbook of theories of aging  ( 2nd ed., pp. 413– 433). New York: Springer Publishing. Freedman, V. A., & Martin, L. G. ( 1999). The role of education in explaining and forecasting trends in functional limitations among older Americans. Demography , 36, 461– 473. doi: 10.2307/2648084 Google Scholar CrossRef Search ADS   Galster, G. C. ( 2012). The mechanism (s) of neighbourhood effects: Theory, evidence, and policy implications. In van Ham, M., Manley, D., Bailey, N., Simpson, L., & Maclennan, D. (Eds.), Neighbourhood effects research: New perspectives  (pp. 23– 56). Dordrecht, Netherlands: Springer. doi:10.1007/978-94-007-2309-2 Google Scholar CrossRef Search ADS   Gayman MD, Turner RJ, & Cui M. ( 2008). Physical limitations and depressive symptoms: Exploring the nature of the association. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences , 63, S219– S228. Google Scholar CrossRef Search ADS   GeoLytics. ( 2014). CensusCD Neighborhood Change Database 1970–2010 tract data . Retrieved from http://www.geolytics.com.ezproxy.gsu.edu/USCensus,Neighborhood-Change-Database-1970–2000,Data,Features,Products.asp Haas, S. ( 2008). Trajectories of functional health: The ‘long arm’ of childhood health and socioeconomic factors. Social Science & Medicine , 66, 849– 861. doi: 10.1016/j.socscimed.2007.11.004 Google Scholar CrossRef Search ADS   House, J. S., Kessler, R. C., & Herzog, A. R. ( 1990). Age, socioeconomic status, and health. The Milbank Quarterly , 68, 383– 411. doi: 10.2307/3350111 Google Scholar CrossRef Search ADS   House, J. S., Lepkowski, J. M., Kinney, A. M., Mero, R. P., Kessler, R. C., & Herzog, A. R. ( 1994). The social stratification of aging and health. Journal of Health and Social Behavior , 35, 213– 234. doi:10.2307/2137277 Google Scholar CrossRef Search ADS   Hout, M., & Beller, E. ( 2006). Intergenerational social mobility: The United States in comparative perspective. The Future of Children , 16, 19– 36. Jette, A. M. ( 2006). Toward a common language for function, disability, and health. Physical Therapy , 86, 726– 734. Google Scholar CrossRef Search ADS   Kail, B. L., & Taylor, M. G. ( 2014). Cumulative inequality and racial disparities in health: private insurance coverage and black/white differences in functional limitations. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences , 69, 798– 808. doi: 10.1093/geronb/gbu005 Google Scholar CrossRef Search ADS   Link, B. G., & Phelan, J. ( 1995). Social conditions as fundamental causes of disease. Journal of Health and Social Behavior , Spec No, 80– 94. doi:10.2307/2626958 Google Scholar CrossRef Search ADS   Livermore, G., Stapleton, D. C., & O’Toole, M. ( 2011). Health care costs are a key driver of growth in federal and state assistance to working-age people with disabilities. Health Affairs (Project Hope) , 30, 1664– 1672. doi: 10.1377/hlthaff.2011.0247 Google Scholar CrossRef Search ADS   Marmot, M. ( 2015). The health gap: The challenge of an unequal world . London: Bloomsbury Publishing. Montez, J. K. ( 2013). The socioeconomic origins of physical functioning among older U.S. adults. Advances in Life Course Research , 18, 244– 256. doi: 10.1016/j.alcr.2013.08.001 Google Scholar CrossRef Search ADS   Nagi, S. Z. ( 1976). An epidemiology of disability among adults in the United States. The Milbank Memorial Fund Quarterly. Health and Society , 54, 439– 467. doi:10.2307/3349677 Google Scholar CrossRef Search ADS   Panel Study of Income Dynamics. ( 2013). Restricted use dataset. Institute for social research survey research center . Ann Arbor, MI: University of Michigan. Phelan, J. C., Link, B. G., & Tehranifar, P. ( 2010). Social conditions as fundamental causes of health inequalities: Theory, evidence, and policy implications. Journal of Health and Social Behavior , 51 ( Suppl), S28– S40. doi: 10.1177/0022146510383498 Google Scholar CrossRef Search ADS   Pickett, K. E., & Pearl, M. ( 2001). Multilevel analyses of neighbourhood socioeconomic context and health outcomes: A critical review. Journal of Epidemiology and Community Health , 55, 111– 122. http://dx.doi.org/10.1136/jech.55.2.111 Google Scholar CrossRef Search ADS   Poulton, R., Caspi, A., Milne, B. J., Thomson, W. M., Taylor, A., Sears, M. R., & Moffitt, T. E. ( 2002). Association between children’s experience of socioeconomic disadvantage and adult health: A life-course study. Lancet (London, England) , 360, 1640– 1645. doi: 10.1016/S0140-6736(02)11602-3 Google Scholar CrossRef Search ADS   Reardon, S. F., & Bischoff, K. ( 2011). Income inequality and income segregation. American Journal of Sociology , 116, 1092– 1153. doi: 10.1086/657114 Google Scholar CrossRef Search ADS   Robert, S. A. ( 1998). Community-level socioeconomic status effects on adult health. Journal of Health and Social Behavior , 39, 18– 37. doi: 10.2307/2676387 Google Scholar CrossRef Search ADS   Ross, C. E., & Mirowsky, J. ( 2001). Neighborhood disadvantage, disorder, and health. Journal of Health and Social Behavior , 42, 258– 276. doi:10.2307/3090214 Google Scholar CrossRef Search ADS   Ross, C. E., & Mirowsky, J. ( 2008). Neighborhood socioeconomic status and health: Context or composition? City & Community , 7, 163– 179. doi:10.1111/j.1540-6040.2008.00251.x Google Scholar CrossRef Search ADS   Sampson, R. J., Morenoff, J. D., & Gannon-Rowley, T. ( 2002). Assessing “neighborhood effects”: Social processes and new directions in research. Annual Review of Sociology , 28, 443– 478. Google Scholar CrossRef Search ADS   Schootman, M., Jeffe, D. B., Baker, E. A., & Walker, M. S. ( 2006). Effect of area poverty rate on cancer screening across US communities. Journal of Epidemiology and Community Health , 60, 202– 207. doi: 10.1136/jech.2005.041020 Google Scholar CrossRef Search ADS   Schoeni, R. F., Stafford, F., McGonagle, K. A., & Andreski, P. (2013). Response rates in national panel surveys. The Annals of the American Academy of Political and Social Science, 645, 60–87. Seeman, T. E., Bruce, M. L., & McAvay, G. J. ( 1996). Social network characteristics and onset of ADL disability: MacArthur studies of successful aging. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences , 51, S191– S200. doi:10.1093/geronb/51B.4.S191 Google Scholar CrossRef Search ADS   Segerstrom, S. C., & Miller, G. E. ( 2004). Psychological stress and the human immune system: A meta-analytic study of 30 years of inquiry. Psychological Bulletin , 130, 601– 630. doi: 10.1037/0033-2909.130.4.601 Google Scholar CrossRef Search ADS   Skinner, M. A., Zautra, A. J., & Reich, J. W. ( 2004). Financial stress predictors and the emotional and physical health of chronic pain patients. Cognitive Therapy and Research , 28, 695– 713. doi:10.1023/B:COTR.0000045572.33750.6c Google Scholar CrossRef Search ADS   Stafford, M., Gimeno, D., & Marmot, M. G. ( 2008). Neighbourhood characteristics and trajectories of health functioning: A multilevel prospective analysis. European Journal of Public Health , 18, 604– 610. doi: 10.1093/eurpub/ckn091 Google Scholar CrossRef Search ADS   StataCorp. ( 2015). Stata statistical software: Release 14 . College Station, TX: StataCorp LP. Taylor, M. G. ( 2010). Capturing transitions and trajectories: The role of socioeconomic status in later life disability. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences , 65, 733– 743. doi: 10.1093/geronb/gbq018 Google Scholar CrossRef Search ADS   Taylor, S. E., Repetti, R. L., & Seeman, T. ( 1997). Health psychology: What is an unhealthy environment and how does it get under the skin? Annual Review of Psychology , 48, 411– 447. doi: 10.1146/annurev.psych.48.1.411 Google Scholar CrossRef Search ADS   Tomey, K., Diez Roux, A. V., Clarke, P., & Seeman, T. ( 2013). Associations between neighborhood characteristics and self-rated health: A cross-sectional investigation in the Multi-Ethnic Study of Atherosclerosis (MESA) cohort. Health & Place , 24, 267– 274. doi: 10.1016/j.healthplace.2013.10.001 Google Scholar CrossRef Search ADS   Wadsworth, M., Butterworth, S., Marmot, M., & Wilkinson, R. ( 2006). In Marmot, Michael, & Wilkinson R., (Eds.), Social determinants of health . Oxford:OUP, 2005. Wheaton, B. ( 1994). Sampling the stress universe. Stress and mental health  (pp. 77– 114). Boston, MA: Springer. Google Scholar CrossRef Search ADS   Wheaton, B., & Clarke, P. ( 2003). Space meets time: Integrating temporal and contextual influences on mental health in early adulthood. American Sociological Review , 68, 680– 706. doi: 10.2307/1519758 Google Scholar CrossRef Search ADS   White, I. R., Royston, P., & Wood, A.M. (2011). Multiple imputation using chained equations: issues and guidance for practice. Statistics in medicine, 30, 377–399. Published by Oxford University Press on behalf of The Gerontological Society of America 2018. This work is written by (a) US Government employee(s) and is in the public domain in the US.

Journal

The Journals of Gerontology Series B: Psychological Sciences and Social SciencesOxford University Press

Published: Mar 5, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off