Multiple meanings of minority concentration: Incorporating contextual explanations into the analysis of individual-level U.S. black mortality outcomes

Population Research and Policy Review 23: 309 326, 2004. 2004 Kluwer Academic Publishers. Printed in the Netherlands. 309 Multiple meanings of minority concentration: Incorporating contextual explanations into the analysis of individual-level U.S. black mortality outcomes TROY C. BLANCHARD, JERALYNN S. COSSMAN & MARTIN L. LEVIN Mississippi State University Abstract. Prior research on mortality for U.S. blacks focuses on the detrimental effects of minority concentration and residential segregation in metropolitan areas on health outcomes. To date, few studies have examined this relationship outside of large U.S. central cities. In this paper, we extend current research on the minority concentration and mortality relationship to explain the rural advantage in mortality for nonmetropolitan blacks. Using data from the 1986-1994 linked National Health Interview Survey/National Death Index, we examine the rural-urban gap in mortality for U.S. blacks. Our findings indicate that blacks in nonmetropolitan areas experience a lower risk of mortality than metropolitan central city blacks after indicators of socio-economic and health status are controlled. Our findings also point to the importance of accounting for contextual factors. Net of individual level controls, minority concentration exerts differential effects across metropolitan and nonmetropolitan areas, such that nonmetropolitan black residents experience a lower risk of mortality in high minority concentration areas than blacks in metropolitan central city areas. This finding suggests a reconceptualization of the meaning for minority concentration with respect to studies of health outcomes in nonmetropolitan communities. Keywords: Minority concentration, mortality, rural-urban Introduction Social scientists have long sought to explain rural-urban differences in measures of health and socio-economic well-being. Compared to urban communities, rural communities consistently exhibit lower rates of age-adjusted mortality despite lower levels of health care access, higher rates of poverty, and spatial isolation from the job opportunities available in large metropolitan areas (McLaughlin et al. 2001; Clifford & Brannon 1985). Net of economic barriers, such as poverty and inequality, population density, and racial composition, rural residents experience a mortality advantage. Although prior studies have acknowledged the rural-urban mortality gap for the total population, few studies have examined whether the rural health advantage extends to specific racial or ethnic segments of the population.

310 TROY C. BLANCHARD ET AL. Age-adjusted mortality rates for 1989 1991 disaggregated by race indicate that the protective effect of rural residence does not extend to all segments of the population (NCHS 2002). Nationally, blacks in small U.S. counties (nonmetropolitan counties without a city of at least 10,000 persons) experienced 43.4 fewer deaths per 100,000 persons than those in central city counties of a large MSA (NCHS 2002). In contrast, white age-adjusted mortality rates are higher in rural areas. The age-adjusted mortality rate for whites in small U.S. counties is 936.8 deaths per 100,000, while the rate in central counties of a large MSA is 929.4 per 100,000 (NCHS 2002). If the focus is shifted to the southern portion of the U.S., these disparities increase. For whites, the difference in the mortality rate in large and small southern counties is 77 deaths per 100,000 persons greater in nonmetropolitan counties without a city of 10,000 or more (NCHS 2002). In stark contrast, the black mortality rate in small nonmetropolitan counties is 91.4 deaths per 100,000 persons fewer than the rate in large central counties (NCHS 2002). These statistics suggest that the effects of rural residence vary substantially among blacks and whites. This assertion is further supported in studies that control for communitylevel socio-economic conditions. Blacks in persistently poor areas, such as the black belt in Alabama and the Louisiana Delta, experience longer life expectancies than their counterparts in Watts, Harlem, and the South Side of Chicago, and fare at least as well as their counterparts in nonpoor urban areas (Geronimus et al. 2001). These differences persist despite shortages of health care workers, hospital facilities, and medical specialists, and lower rates of health insurance enrollment observed in rural areas (Raibner 1995; Eggebeen & Lichter 1993). Prior research also demonstrates that differences in mortality between urban and rural areas are not merely functions of population composition, individual level socio-economic status, and behavioral traits (Hayward et al. 1997). Thus, current explanations of health outcomes provide little insight into the paradoxical rural-urban mortality disparity for blacks and point to a growing need for contextual explanations of individual level outcomes. In this paper we build on individual level explanations of mortality by developing a contextual level explanation of the rural-urban gap in black mortality. Drawing on theories of minority relations and historical demographic trends, we develop a perspective for understanding the rural mortality advantage enjoyed by black residents and argue that levels of black mortality net of individual level characteristics are conditioned by differential structures of racial discrimination. Using data from the 1986 1994 linked National Health Interview Survey/National Death Index, we test the hypothesis that rural-

MULTIPLE MEANINGS OF MINORITY CONCENTRATION 311 urban disparities in black mortality are due in part to the differential outcomes of minority concentration between rural and urban areas. Minority concentration and black mortality A vast body of research has been devoted to understanding the relationship between minority concentration and the well-being of minority residents in U.S. central cities. Blalock s (1956) visibility discrimination hypothesis argues that minority concentration is directly related to prejudice, interracial conflict, and discrimination. More generally, this perspective is rooted in the notion that as blacks moved into the central cities of the Northeast and Midwest during the Great Migration from 1910 1970, labor market competition between blacks and whites resulted in a discriminatory response by whites that limited the social and economic prospects for blacks (Lieberson 1980). For example, research on the relationship between minority concentration and earnings for blacks finds an inverse relationship between proportion black and earnings for black males in non-southern metropolitan areas (Cassirer 1996). Although the visibility discrimination hypothesis has not been directly applied to mortality, the most commonly cited structural correlate of mortality for blacks, residential segregation, is a central outcome of the visibility discrimination process (Massey & Denton 1993; Wilson 1987). Residential segregation has been touted as one of the key causal mechanisms in models of black mortality (Jackson et al. 2000; Polednak 1996; Massey et al. 1991). Following the visibility discrimination hypothesis, blacks in metropolitan areas with high black concentration experience higher levels of residential segregation and social isolation due to competition for resources. Because whites seek to maintain power in political and labor arenas, they historically restricted the housing opportunities for blacks, thus generating residential segregation (Massey & Denton 1993). This process becomes an important health determinant for blacks because residential segregation increases the concentration of poverty, eroding institutional effectiveness in key arenas, such as health care, education, and political power (Massey et al. 1991). Although prior research finds minority concentration directly related to the responsiveness of local governments to the needs of the black community (Santoro 1995), residential segregation limits the political effectiveness of the black vote by concentrating black residents in a limited number of neighborhoods. Thus, segregation limits the social, economic, and political capacity of a community to provide a protective environment for residents. Segregation also impacts behavioral and psychological well-being. Researchers point to a culture of violence in highly disadvantaged neighborhoods legitimating violence as a means for achieving status (Anderson

312 TROY C. BLANCHARD ET AL. 1999). This assertion is supported by studies finding a strong relationship between the geographic concentration of disadvantage and homicide rates among blacks in central cities (Krivo & Peterson 2000; Peterson et al. 2000). Additionally, segregation limits the capacity for effective social control mechanisms regulating healthy behavior among residents, especially children (Sampson 1987). Finally, disenfranchisement from the political process minimizes the degree of engagement in collective problem-solving activities for residents. Participation in this type of interactive environment optimizes biological functioning and improves the health of participants (Young & Lyson 2001). Combined with the disappearance of many middle-class black business owners from the central city, segregation limits the organizational capacity of communities by reducing economic stakes in community well-being. Local business owners who both live and work in the community have a vested interest in maintaining, among other things, quality health care facilities that ensure a productive labor force. If the black middle-class disappears from these communities, local leadership is diminished and communities are less effective in addressing local problems (Wilson 1987). Researchers empirically testing this perspective have found strong support for the segregation mortality relationship. At the aggregate level, studies find a positive relation between residential segregation and infant mortality/adult mortality in U.S. central cities (Polednak 1996; LaVeist 1992; Polednak 1991; Massey et al. 1991; Laveist 1989). These studies also demonstrate that the effect of segregation is not subsumed by factors often attributed to segregation, such as poverty, family disruption, and education. However, researchers often suggest that aggregate level research is problematic because it does not account for individual level heterogeneity (Waitzman & Smith 1998). To address this issue, more recent analyses have estimated contextual models that account simultaneously for both contextual and individual level factors. For example, one recent study controlled for individual level income while assessing the relationship between minority concentration and mortality. Even after controlling for age and individual level income, blacks living in neighborhoods with high levels of minority concentration experience higher levels of mortality (Jackson et al. 2000). Similar findings were obtained when controlling for educational attainment and other individual level characteristics (Hart et al. 1998; Waitzman & Smith 1998). LeClere et al. (1997) estimated a comprehensive contextual model that accounted for age, race, income, education, and marital status, while also controlling for minority concentration. These authors found that much of the mortality disadvantage experienced by blacks is accounted for by neighborhood minority concentration.

MULTIPLE MEANINGS OF MINORITY CONCENTRATION 313 Minority concentration and mortality in the nonmetropolitan south Although previous studies demonstrate that segregation accounts for a great deal of variation in black mortality, few studies have attempted to examine residential differences in black mortality or apply the minority concentration perspective to explain rural-urban differences. Prior studies have focused almost exclusively on the mortality process at a national level or within the metropolitan context only. Thus, it is unclear how the mortality process for blacks varies across residential contexts and how minority concentration theories apply to nonmetropolitan areas. The central assumption underlying theories of minority concentration is that minority in-migration into a locale results in a majority response that increases the level of inequality directed towards the minority group. In the nonmetropolitan South, black-white race relations did not emerge from a competitive threat by black labor. By the end of the Great Migration, the spatial distribution of blacks in the U.S. had changed dramatically with the vast majority of the black population locating in urban centers across the U.S. and the remaining nonmetropolitan blacks residing in the rural South (Fligstein 1981). Unlike metropolitan areas, the rural South received little inmigration from 1910-1970. Thus, black concentration decreased or remained constant during this era. Additionally, blacks and whites rarely competed for the same jobs in the rural South in either the agricultural or industrial era. Falk and Lyson (1988) note that industries locating to the nonmetropolitan black belt South do so to hire nonunionized cheap labor. In practice, this economic development strategy largely recreates the agricultural system where blacks and whites worked in close proximity, but with well-defined roles and an authority structure that favors whites. Researchers studying the impact of minority concentration in the South pose alternative explanations differing from the competition based visibility discrimination hypothesis. The white-gains hypothesis argues that whites reap economic and social benefits from the subordination of blacks (Glenn 1966, 1964; Dollard 1937). In this case, the lack of competition between whites and blacks in the labor market is beneficial to whites. Fossett and Siebert (1997) provide an extension of this perspective indicating that if the occupational status difference between black and whites is large, white gains are directly related to minority concentration. In contrast to the visibility discrimination hypothesis characterizing minority concentration as detrimental to white well-being, the white-gains hypothesis suggests that minority concentration has positive effects for whites in the nonmetropolitan South. Where the status of whites and blacks is unequal, there is little need to direct extreme institutional discrimination against blacks, especially in housing markets and other arenas where blacks do not stand to gain status advantages. This process

314 TROY C. BLANCHARD ET AL. is best reflected in the finding that southern whites are opposed to residing with blacks, but are more likely to live in racially mixed neighborhoods than whites from other regions (Hurlbert & Bankston 1998). With respect to black mortality, the white-gains hypothesis characterization provides a number of theoretical extensions. First, given that minority concentration does not pose a threat to whites, there is little need for whites to develop institutional barriers to segregate blacks residentially. In urban areas, the residential segregation of blacks restricts access to high quality healthcare located in more affluent sections of the city. Although rural areas are often characterized by poor health care infrastructure and lower rates of health insurance enrollment (Raibner 1995, Eggebeen & Lichter 1993), the lack of segregation in the nonmetropolitan South provides even access for black residents. A second extension of this perspective is that the economic development strategy of the nonmetropolitan South is oriented towards low-skill labor (Falk & Lyson 1988). For many inner city blacks, industrial restructuring during the late 1970s and 1980s relocated manufacturing activity out of the central city eliminating many entry-level low-skill jobs. This process led to a massive increase in the jobless rate for young black males, an elevation in the concentration of poverty, and growth in the number of female headed families (Wilson 1987). Unlike the central cities of the North and Midwest, the nonmetropolitan South experienced growth in low-skill manufacturing employment supplementing existing agricultural production (Falk & Lyson, 1988). Thus, blacks in the nonmetropolitan South are less likely to experience unemployment and more likely to have employed role models. Although low-skill jobs may not necessarily provide high wages, the existence of employment opportunities for low-skill workers may ensure community stability and a healthier environment through social control networks within the community (Shihadeh & Ousey 1998). A third consequence of minority concentration in the nonmetropolitan South is the presence of black business owners and farmers. Fossett and Siebert (1997) argue that a sizable black consumer base may result in the development of minority owned businesses and a semi-separate economy for blacks. The presence of black business owners has been linked to the wellbeing of black residents (Villemez & Beggs 1984). Middle-class business owners and professionals play a central role in social problem solving and protecting community health and well-being (Lyson et al. 2001; Young & Lyson 2001). Though not agreeing on causality, both Wilson (1987) and Massey and Denton (1993) point to a decline in black middle-class presence in central cities coinciding with the disappearance of manufacturing jobs and the creation of an urban underclass. Massey and Denton (1993) also

MULTIPLE MEANINGS OF MINORITY CONCENTRATION 315 argue that residential segregation limited the ability of middle class blacks to escape distressed neighborhoods. For communities in the nonmetropolitan South, black business owners, professionals, and farmers may enhance the well-being of black residents by serving as a link to political leaders to voice concerns (Duncan 1999). To our knowledge, no study has examined the impact of minority concentration on mortality outcomes for nonmetropolitan blacks. Additionally, researchers have not employed minority concentration as an explanatory framework for understanding why nonmetropolitan blacks experience lower rates of mortality than their metropolitan counterparts. We argue that minority concentration in the nonmetropolitan South is related to positive health outcomes because the structure of race relations results in a less residentially segregated environment with access to low-skill employment, in the potential for black capitalism, and in cohesive communities. Combined, these factors provide a healthy environment for residents. Through these mechanisms, we suspect that higher levels of minority concentration will result in lower levels of mortality for black residents in the nonmetropolitan South, thus accounting for the rural-urban gap in mortality for blacks net of compositional differences between urban and rural locales. Data and methods The data for this study come from the 1986 1994 National Health Interview Survey (NHIS) linked to the National Death Index (NDI). The NHIS is a nationally representative annual household survey designed to measure disease incidence and prevalence. Roughly 49,000 households are sampled each year, resulting in individual level data for approximately 125,000 civilian noninstitutionalized persons. NCHS has linked 1986 1994 NHIS respondents over the age of 18 to the National Death Index through 1997 to provide an indicator of vital status. Survey respondents are matched to NDI records through a probabilistic scheme that assigns weights to social security number, name, age, race, sex, marital status, nativity, and residence at time of death to evaluate the quality of the match (NCHS 2000). Through the matching process, approximately 2% of NHIS respondents are identified as ineligible and dropped from the analysis because these cases do not contain appropriate information for matching to NDI records. An evaluation of the NCHS probabilistic matching technique indicates that this method is highly accurate (Patterson & Bilgrade 1986). To examine metropolitan and nonmetropolitan differences in mortality, U.S.-born black respondents over the age of 18 were selected for the study. The foreign born are excluded from the analysis because of the divergent

316 TROY C. BLANCHARD ET AL. mortality patterns between foreign and U.S. born blacks (Rogers et al. 2000). We limit our analysis to the nonmetropolitan South and metropolitan areas to account for the uneven distribution of blacks in the U.S. nonmetropolitan areas. Following the Great Migration the vast majority of the nonmetropolitan black population was located in the South (Fligstein 1981). Additionally, the 1986 1994 NHIS contains only 353 black respondents living in nonmetropolitan areas outside of the South. We conducted alternative analyses including both non-southern and southern nonmetropolitan blacks and obtained similar findings. Because our analysis pooled respondents from all U.S. metropolitan areas with those residing in the nonmetropolitan South, we tested for regional variability in mortality patterns by estimating models separately for metropolitan residents. Our findings indicated that there was no significant regional variability in the probability of mortality for metropolitan blacks residing outside of a central city. For blacks residing in metropolitan central city areas, residents of the West and Midwest experienced a lower probability of mortality than blacks in the South and Northeast. Thus, by pooling metropolitan central city respondents from the West and Midwest with the South and Northeast, our models may understate the mortality gap between blacks in metropolitan central city areas and the nonmetropolitan South. We also tested for differences in mortality among blacks residing in metropolitan central city, metropolitan not central city areas, and nonmetropolitan areas within the South. Our models for the South yielded results similar to those presented in the text indicating that mortality processes in the nonmetropolitan South operate differently from those in other metropolitan areas. Our dependent variable in this study is the vital status of the respondent. We limit our analyses to non-external causes of death. The dependent variable is defined by causes of death coded from 004-780 in the International Classification of Diseases (U.S. Department of Health and Human Services 1990). The resulting dataset consists of 658,680 person years with 7,579 deaths. Pooling nine years of NHIS survey respondents provides information on a large number of blacks allowing for a metropolitan and nonmetropolitan comparison not available in most individual level mortality data sources. We selected only non-external causes to generate conservative estimates of mortality differences between blacks in metropolitan central cities and those in metropolitan non-central city areas and the nonmetropolitan South. The 507 deaths occurring from external cases account for a small share (6.3%) of all deaths occurring to blacks in the 1986 1994 NHIS. Blacks in the nonmetropolitan South experienced substantially lower rates of death due to external causes (homicide, suicide, and accidents) than blacks in metropolitan central cities. Although blacks in the nonmetropolitan South do experience higher

MULTIPLE MEANINGS OF MINORITY CONCENTRATION 317 rates of mortality due to accidents than their metropolitan counterparts, this difference is offset by the higher rates of suicide and homicide in metropolitan central city areas. Thus, the inclusion of external causes of death may inflate the observed mortality gap between blacks in metropolitan central cities and the nonmetropolitan South, suggesting a greater health advantage for black residents of the nonmetropolitan South. In addition to the models presented here, we also estimated additional models that pooled internal and external causes of death. These models yielded results nearly identical to those derived from the internal causes of death only models. We also estimated models examining external causes of death only. Although the direction and magnitude of the coefficients in the model are similar to those in our internal causes of death only models, the coefficients were not significant. The results from the external cause of death only models suggest that the complex processes underlying homicide, suicide, and accidental death may be outside of the scope of our conceptual framework. Thus, we focus our analyses on internal causes of death only. Our independent variable of interest is the proportion black of the population, which captures the level of minority concentration in a given area. Public use versions of NHIS data contain no sub-regional geographic information for respondents, such as state, Metropolitan Statistical Area, county, or tract identifiers. Because the NHIS does contain Primary Sampling Unit (PSU) identifiers, we calculate the proportion black using NHIS respondent information tabulated to the PSU level. PSUs consist of a single county or a county group depending on county population size. In the most urbanized areas, a PSU may consist of a single county, while the most rural PSUs may be comprised of multiple counties. The proportion black in the average PSU in our analysis is 0.242, with a range of 0.003 to 0.65. A second key variable in our analysis is residential type. Nonmetropolitan residence is coded as a binary variable where nonmetropolitan residents are contrasted with metropolitan residents. We also include a series of control variables in our model to adjust probabilities of mortality for demographic, socio-economic, and health status. Age, sex, and marital status are included as demographic controls. Age is entered as a continuous variable ranging from 18 to 108 years. Sex is coded as a binary variable with females as the contrast group. Marital status is categorized into never married, widowed, divorced/separated, or married, with married as the reference group. Socioeconomic indicators include income, education, and labor force participation. Income is measured using an income equivalence scale because the NHIS only reports the family income for respondents (Rogers et al. 2000). Income equivalence scales express family income relative to family size to capture the purchasing power of a family. Income equivalence is expressed in units

318 TROY C. BLANCHARD ET AL. of $10,000. Education is categorized as less than 12 years of education, high school graduate, or 13 or more years of education, with high school graduate as the contrast group. We categorize labor force participation as unemployed, not in the labor force, or employed (reference category). We also include self-rated health status in our model. The NHIS contains five categories of health rated health status that identifies the respondent s perception of his or her health as excellent, very good, good, fair, or poor. We include excellent health as the reference category. A second health status variable in our model is the reported number of bed days due to illness. We enter this variable as a dichotomous measure where 1 = less than 30 days and 0 = 31 or more days. We conduct our analyses using a discrete time hazard model (Allison 1984). Model estimation requires NHIS person record files to be converted to person year files that include one record for each year the individual is alive, including the year of death. As interviews occur throughout the year, each person is weighted at 0.5 for the year they were surveyed and given a weight of 1 for the following years. All variables in the model, with the exception of age, are time invariant. Because discrete time hazard models are estimated using standard logistic regression procedures, model coefficients are reported as odds ratios. Allison (1984) demonstrates that discrete time hazard and proportional hazard modeling techniques yield similar results. The complex sampling design of the NHIS requires the use of statistical software that weights cases according to the NHIS multistage stratified design. Our models are estimated using SUDAAN to produce correct standard errors for regression coefficients from complex samples (Shah et al. 1997). Results Descriptive statistics for all U.S. born noninstitutionalized black respondents in U.S. metropolitan central cities, metropolitan non-central city areas, and the nonmetropolitan South are reported in Table 1. Statistics are presented for the full sample and disaggregated by residence. With respect to demographic variables, the key difference between black respondents in each of the three residential types is that blacks in metropolitan central cities are less likely to be married. Measures of socio-economic status indicate that blacks in the nonmetropolitan South have lower levels of income and education. Blacks in the nonmetropolitan South also experience a health status disadvantage. Blacks in metropolitan central city and non-central city areas are more likely to report excellent or very good health status. These findings suggest that across socio-economic and health status dimensions blacks in the nonmetropolitan South are more disadvantaged on key predictors of mortality status than blacks in other residential types. Descriptive statistics for the contextual

MULTIPLE MEANINGS OF MINORITY CONCENTRATION 319 factor in our model, proportion black, indicate that nonmetropolitan South PSUs have the largest concentrations of black residents. When disaggregated by vital status, black residents of metropolitan central city PSUs with an above average proportion black experience a greater risk of mortality compared to blacks in other residential types. Odds ratios from models predicting mortality for blacks are presented in Table 2. Differences in the likelihood of dying by residential type net of individual level covariates of mortality are examined in Models 1 6. Results observed from Model 1 indicate that net of demographic controls, residents of metropolitan not central city locales have the lowest risk of mortality from nonexternal causes, though central city and nonmetropolitan South areas experience similar risks. Measures of socio-economic status are included in Model 2. The effects of income equivalence, labor force status, and education on mortality risk for blacks follow patterns observed for the total population in prior research (Rogers et al. 2000). The inclusion of these measures account, in part, for the observed difference in mortality between central and non-central city metropolitan areas. Odds ratios for the hazard of mortality for blacks, controlling for demographic variables and indicators of health status, are reported in Model 3. The findings indicate that, after controlling for health status of black residents, residence in the nonmetropolitan South is related to a lower risk of mortality. Relative to other residential types, blacks in the nonmetropolitan South have lower levels of self-rated health status, especially within the excellent and very good categories. Descriptive statistics from Table 1 also demonstrate that excellent and very good health status categories account for a larger share of mortality in metropolitan areas than in the nonmetropolitan South. These findings suggest that poorer levels of health status in the nonmetropolitan South do not translate directly to increased risk of mortality. In Models 4 6 we simultaneously control for socio-economic and health status indicators adding the contextual effect, proportion black, in Models 5-6. After controlling for socio-economic and health status, the difference in the risk of mortality for blacks in the nonmetropolitan South and metropolitan central city areas increases. In Model 5, the proportion black in the respondent s PSU is added to the model. The results indicate that minority concentration has little effect on black mortality across all residential types. Because we hypothesize differential effects of minority concentration in metropolitan and nonmetropolitan areas, we estimate interaction terms in Model 6. The results from the interaction model (Model 6) indicate substantial differences in the effect of minority concentration across residential types.

320 TROY C. BLANCHARD ET AL. Table 1. Descriptive statistics of demographic variables, socio-economic status indicators, health status measures, and minority concentration by vital status and residence for Blacks, 1986 1997 Metro, Central City Not Central City Nonmetro South Survived Died Survived Died Survived Died Demographic variables Sex Male 60.7 51.0 58.7 47.8 59.5 51.3 Female 39.3 49.0 41.3 52.2 40.5 48.7 Age (mean) 46.0 68.2 43.8 66.8 46.5 69.9 Marital status Never married 32.0 11.7 28.0 10.9 29.5 11.1 Widowed 9.3 29.4 6.5 26.9 10.7 33.1 Divorced/separated 18.1 17.7 14.8 14.6 13.6 11.6 Married 40.6 41.2 50.8 47.6 46.3 44.2 Socio-economic status Income equivalence Mean in $10,000s 1.3 0.9 1.7 1.1 0.9 0.6 Education Less than 12 years 33.1 59.7 24.7 60.0 46.7 76.9 12 years 39.7 25.5 39.9 24.0 39.1 16.8 Greater than 12 years 27.2 14.8 35.4 16.1 14.3 6.3 Labor force status Unemployed 8.0 3.2 7.0 3.8 6.7 1.9 Not in labor force 37.5 75.5 27.0 68.2 38.2 76.9 Employed 54.5 21.4 66.0 28.0 55.2 21.1 Health status Self-rated health status Excellent 23.9 9.4 28.8 11.8 17.0 5.8 Very good 24.4 14.0 26.5 14.3 20.7 10.8 Good 31.0 26.0 29.1 27.7 34.6 23.1 Fair 15.1 28.5 11.5 26.3 19.2 29.8 Poor 5.7 22.2 4.1 19.9 8.5 30.5 Bedsickdays 31 or more 4.8 14.0 3.8 12.9 4.3 14.4 0 30 95.3 86.0 96.2 87.1 95.7 85.62 PSU Level measure of minority concentration Percent at or above the mean proportion black 45.4 47.1 53.5 53.3 55.6 55.5 (Mean Proportion Black) (0.25) (0.23) (0.32) Person years and deaths 393,876 4,652 141,091 1,207 88,225 1,086

MULTIPLE MEANINGS OF MINORITY CONCENTRATION 321 Table 2. Odds ratios for hazards of mortality for Blacks by demographic variables, socio-economic status indicators, health status measures, and minority concentration, 1986 1997 Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Demographic variables Sex Male 1.82 1.89 1.89 1.92 1.92 1.92 Female ref ref ref ref ref ref Age (mean) 1.08 1.07 1.07 1.06 1.06 1.06 Marital status Never married 1.46 1.23 1.45 1.30 1.30 1.30 Widowed 1.20 1.08 1.18 1.11 1.11 1.11 Divorced/separated 1.22 1.15 1.16 1.13 1.13 1.13 Married ref ref ref ref ref ref Socio-economic status Income Equivalence (mean) 0.89 0.93 0.93 C0.93 Education Less than 12 years 1.09 1.02 1.03 1.02 12 years ref ref ref ref Greater than 12 years 0.93 + 0.96 0.96 0.96 Labor force status Unemployed 1.26 1.16 + 1.16 + 1.16 + Not in labor force 1.83 1.48 1.48 1.48 Employed ref ref ref ref Health status Self-rated health status Excellent ref ref ref ref Very good 1.07 1.04 1.04 1.04 Good 1.41 1.36 1.36 1.36 Fair 2.02 1.83 1.83 1.83 Poor 2.81 2.39 2.40 2.40 Bed sick days 0-30 0.66 0.69 0.69 0.69 31 or more ref ref ref ref Minority concentration Proportion black (Mean) 0.96 1.24 Residence Metro Central City ref ref ref ref ref ref Metro, Not Central City 0.91 0.97 0.92 0.96 0.96 0.69 Nonmetro South 0.99 0.92 + 0.88 0.85 0.86 1.02 Residence Interactions Metro, not Central City times Minority concentration 0.72 Nonmetro South times Minority concentration 0.55 2 Log Likelihood 66,349 62,591 63,227 60,074 60,074 60,068 + p<0.10, p<0.05.

322 TROY C. BLANCHARD ET AL. Figure 1. Predicted probabilities by proportion Black. Only the slope for nonmetropolitan South is significant (p <0.05, two tailed). We summarize these findings from Model 6 in Figure 1. For metropolitan central city black residents, there is a positive association between minority concentration and the risk of death. For example, a 0.25 increase in minority concentration in metropolitan central city PSUs yields a 4.5% increase in the risk of mortality. Black residents of metropolitan not central city and nonmetropolitan South PSUs experience a lower risk of mortality in PSUs with higher minority concentration. This finding largely conforms to our theoretical expectations. Metropolitan not central city PSUs demonstrate a divergent pattern from central city areas. Substantively, the effect of minority concentration in noncentral city metropolitan PSUs is negligible, with the risk of death decreasing by 1% for every 0.10 increase in the proportion black. This finding is unique given prior research linking minority concentration to residential segregation in suburban locations (Stearns & Logan 1986). Thus, even though increased minority concentration may imply greater residential segregation, it has little effect on mortality risk. In the nonmetropolitan South, a 0.10 increase in the proportion black reduces the risk of death by 4%. Thus, blacks in the nonmetropolitan South

MULTIPLE MEANINGS OF MINORITY CONCENTRATION 323 enjoy a lower risk of mortality in areas with higher minority concentration. Although the risk of mortality in nonmetropolitan South and metropolitan central city PSUs is similar when the proportion black is less than 0.10, the gap in the risk of mortality grows substantially with increases in the proportion black. If the proportion black equals 0.35, blacks in metropolitan central city PSUs are 21% more likely to experience mortality than blacks in the nonmetropolitan South. 0 This large difference in mortality risk suggests that minority concentration yields health benefits for blacks in the nonmetropolitan South. Conclusions In this paper we have assessed the relationship between minority concentration and mortality for U.S. native-born blacks across different types of residence: metropolitan central city, metropolitan not central city, and the nonmetropolitan South. Prior research has suggested that minority concentration yields higher rates of mortality for black residents. Our findings suggest that minority concentration indeed has marked effects on mortality risks for blacks, but these effects vary across residential types. For metropolitan central city areas, the impact of minority concentration is largely consistent with theoretical expectations originating with the Blalock hypothesis (Blalock 1956). Outside of the central city, minority concentration has a weak effect on mortality for metropolitan non-central city blacks and a substantial effect in the nonmetropolitan South. Our analyses demonstrate a strong negative relationship between minority concentration and the risk of mortality in the nonmetropolitan South after controlling for demographic, socio-economic, and health status. Although further research is required to clarify this relationship, prior theoretical work on race relations in the South suggests that minority concentration may not lead to the same detrimental impacts on black well-being as in other regions (Fossett & Siebert 1997; Glenn 1966, 1964; Dollard 1937). One central difference between the South, especially the nonmetropolitan South, and other regions is the link between minority concentration and residential segregation. The in-migration of blacks into northern central cities that led to harsh residential inequalities for blacks was also mirrored by a simultaneous out- migration of blacks from the rural South. Historical economic conditions during the antebellum era and massive out-migration limited the level of competition between blacks and whites, decreasing the need for residential segregation. A second key difference is that unlike central cities outside of the South, nonmetropolitan blacks did not suffer the deleterious effects of

324 TROY C. BLANCHARD ET AL. de-industrialization which left many urban blacks racially and economically segregated from the remainder of society (Massey & Denton 1993). Because of these differences in the effect of minority concentration, social support structures in nonmetropolitan areas may differ considerably from those for blacks in central cities. For rural residents social networks are more likely to be rooted in kinship ties and neighborhood solidarity, rather than friendship (Beggs et al. 1996). In nonmetropolitan communities with substantial minority concentration, we suspect that these ties yield higher levels of social support and network resources leading to more positive health outcomes. Given the similar level of poverty and socio-economic deprivation experienced by blacks in the nonmetropolitan South, it seems unlikely that nonmetropolitan blacks would enjoy a substantial health advantage over central city blacks. Our findings suggest the contrary, pointing to a key difference between these two residential types, namely the manner in which minority concentration translates into housing market inequalities that limit the life prospects for black residents. Prior research also suggests that black population concentration, largely occurring in black belt areas of the rural South, may reduce the overall level of mortality through the presence of a black middle class and the greater sense of community generated by large racial enclaves (McLaughlin & Stokes 2002). Our findings provide stronger support for the validity of community explanations of mortality. If minority concentration does not generate residential segregation in the nonmetropolitan South, we suspect that a central dimension of the nonmetropolitan mortality advantage for blacks is rooted in greater community cohesion associated with lower levels of residential segregation. Acknowledgements Partial funding for this paper was provided by grant number 4D1ARH00005-01-01 from the Office of Rural Health Policy of the Department of Health and Human Services through the Rural Health, Safety, and Security Institute, Social Science Research Center, Mississippi State University. References Allison, P.D. (1984). Event history analysis: Regression for longitudinal event data. New York: Sage University Press. Anderson, E. (1999). Code of the street: Decency, violence, and the moral life of the inner city. New York: W.W. Norton.

MULTIPLE MEANINGS OF MINORITY CONCENTRATION 325 Beggs, J.J., Haines, V. & Hurlbert, J. (1996). Revisiting the rural-urban contrast: Personal networks in nonmetropolitan and metropolitan settings, Rural Sociology 61(2): 306 326. Blalock, H.M. (1956). Economic discrimination and Negro increase, American Sociological Review 21: 584 588. Cassirer, N. (1996). Race composition and earnings: Effects by race, region, and gender, Social Science Research 25(4): 375 399. Clifford, W.B. & Brannon, Y.S. (1985). Rural-urban differentialsinmortality, Rural Sociology 50: 210 224. Dollard, J. (1937). Caste and class in a Southern town. New Haven, CT: Yale University Press. Duncan, C. (1999). World s apart: Why poverty persists in rural America. New Haven, CT: Yale University Press. Eggebeen, D.J. & Lichter, D. (1993). Health and well-being among rural Americans: Variations across the life course, Journal of Rural Health 7: 153 169. Falk, W.W. & Lyson, T.A. (1988). High tech, low tech, no tech: Recent industrial and occupational change in the South. Albany, NY: State University of New York Press. Fligstein, N. (1981). Going North: Migration of blacks and whites from the South 1900 1950. New York: Academic Press. Fossett, M.A. & Seibert, M.T. (1997). Long time coming: Racial inequality in the nonmetropolitan South, 1940 1990. Boulder, CO: Westview Press. Geronimus, A.T., Bound, J., Waidman, T.A., Colen, C. & Steffick, D. (2001). Inequality in life expectancy, functional status, and active life expectancy across selected black and white populations in the United States, Demography 38(2): 227 251. Glenn, N.D. (1964). The relative size of the Negro population and Negro Glenn, N.D. (1966). White gains from Negro subordination, Social Problems 14: 159 178. Hart, K.D, Kunitz,S. & Sell, R. (1998). Metropolitan governance, residential segregation, and mortality among African Americans, American Journal of Public Health 88: 434 438. Hayward, M.D., Pienta, A. & McLaughlin, D. (1997). Inequality in men s mortality: The socioeconomic status gradient and geographic context, Journal of Health and Social Behavior 38(December): 313 330. Hurlbert, J.S. & Bankston, W. (1998). Cultural distinctiveness in the face of structural transformation: The new old South, pp. 168 188 in D. Hart (ed.), The Rural South Since World War II. Baton Rouge: Louisiana State University Press. Jackson, S.A., Anderson, R., Johnson, N. & Sorlie, P. (2000). The relation of residential segregation to all-cause mortality: A study in black and white, American Journal of Public Health 90(4): 615 617. Krivo, L.J. & Peterson, R. (2000). The structural context of homicide: Accounting for differences in process, American Sociological Review 65(4): 547 559. LaVeist, T.A. (1989). Linking residential segregation to the infant mortality race disparity in U.S. cities, Social Science Research 73: 90 94. LaVeist, T.A. (1992). The political empowerment and health status of African Americans: Mapping a new territory, American Journal of Sociology 97(4): 1080 1095. LeClere, F.B., Rogers, R. & Peters, K. (1997). Ethnicity and mortality in the United States: Individual and community correlates, Social Forces 76(1): 169 198. Lieberson, S. (1980). A piece of the pie: Blacks and American immigrants since 1980. Berkeley, CA: University of California Press. Lyson, T.A., Torres, R. & Welsh, R. (2001). Scale of agricultural production, civic engagement, and community welfare, Social Forces 80(1): 311 328. Massey, D.S. & Denton, N. (1993). American apartheid: Segregation and the making of the underclass. Cambridge, MA: Harvard University Press.

326 TROY C. BLANCHARD ET AL. Massey, D.S., Gross, A. & Eggers, M. (1991). Segregation, the concentration of poverty, and the life chances of individuals, Social Science Research 20: 397 420. McLaughlin, D.K. & Stokes, C.S. (2002). Income inequality and mortality in U.S. counties: Does minority concentration matter? American Journal of Public Health 92(1): 99 104. McLaughlin, D.K., Stokes, C.S. & Nonoyama, A. (2001). Residence and income inequality: Effects on mortality among U.S. counties, Rural Sociology 66(4): 579 598. National Center for Health Statistics (2000). National health interview survey: Multiple cause of death public use data file: 1986-1997 Survey years. Hyattsville, MD: National Center for Health Statistics. National Center for Health Statistics (2002). Health, United States 2002 with chartbook on trends in the health of Americans. Hyattsville, MD: National Center for Health Statistics. Patterson, B.H. & Bilgrade, R. (1986). Use of the national death index in cancer studies, Journal of the National Cancer Institute 77: 877 881. Peterson, R.D., Krivo, L. & Harris, M. (2000). Disadvantage and neighborhood violent crime: Do local institutions matter?journal of Research in Crime and Delinquency 37(1): 31 63. Poldenak, A.P. (1991). Black-white differences in infant mortality in 38 standard metropolitan statistical areas, American Journal of Public Health 81(11): 1480 1482. Poldenak, A.P. (1996). Trends in U.S. urban black infant mortality by degree of residential segregation, American Journal of Public Health 86(5): 723 726. Raibner, D.J. (1995). Patterns and predictors of noninstitutional health care utilization by older adults in rural and urban America, Journal of Rural Health 11: 259 273. Rogers, R. G., Hummer, R. & Nam, C. (2000). Living and dying in the USA: Behavioral, health, and social differentials of adult mortality. New York: Academic Press. Sampson, R.J. (1987). Urban black violence: The effect of male joblessness and family disruption, American Journal of Sociology 93: 348 382. Santoro, W.A. (1995). Black politics and employment policies: The determinants of local government affirmative action, Social Science Quarterly 76(4): 794 806. Shah, B., Barnwell, B.G. & Bieler, G. (1997). SUDAAN user s manual, release 8. Research Triangle Park, NC: Research Triangle Institute. Shihadeh, E.S. & Ousey, G. (1998). Industrial restructuring and violence: The link between entry-level jobs, economic deprivation, and black and white homicide, Social Forces 77(1): 185 201. Stearns, L.B. & Logan, J. (1986). The racial structuring of the housing market and segregation in suburban areas, Social Forces 65(1): 28 42. U.S. Department of Health & Human Services. (1990). International classification of diseases, 3rd edition, 9th revision, volume 1. Washington, D.C.: USGPO. Villemez, W.J. & Beggs, J. (1984). Black capitalism and black inequality: Some sociological considerations, Social Forces 63(1): 117 144. Waitzman, N.J. & Smith, K. (1998). Separate but lethal: The effects of economic segregation on mortality in metropolitan America, Milbank Quarterly 76(3): 341 374. Wilson, W.J. (1987). The truly disadvantaged: The inner-city, underclass, and public policy. Chicago: University of Chicago Press. Young, F.W. & Lyson, T. (2001). Structural pluralism and all-cause mortality, American Journal of Public Health 91(1): 136 138. Address for correspondence: Troy C. Blanchard, P.O. Box C, Mississippi State, MS 39762, U.S.A. Phone: (662) 325-7886; Fax: (662) 325-4564; E-mail: tcb44@ra.msstate.edu