Social Capital and Diversity

Social Capital and Racial Inequality in America
By Rodney E. Hero

“Social capital” is said to be strongly associated with better substantive outcomes as well as with civic and economic dimensions of equality in the United States. Robert Putnam argues in Bowling Alone that “the American states with the highest levels of social capital are precisely the states most characterized by economic and civic equality.” He adds: “[B]oth across space [i.e., states] and across time, equality and fraternity [social capital] are strongly positively correlated. . . . [T]he empirical evidence on recent trends is unambiguous. . . . Community and equality are mutually reinforcing, not mutually incompatible.”1 Yet in seeming contradiction to this “social capital thesis,” social capital and related factors such as “political culture” do not necessarily bridge across racial or ethnic groups; nor do they always engender relative civic and economic equality between blacks and whites. Even more problematic, higher aggregate social capital is sometimes associated with relatively worse outcomes for racial minorities.2 These results underscore the importance of racial/ethnic dimensions of equality in the United States and reveal major limitations of the social-capital thesis. Most analyses of social capital do not adequately confront conditions associated with race; as a result, they come to conclusions more benign than a fuller assessment warrants. Social-capital analyses implicitly emphasize absolute or aggregate indicators of equality and overlook or understate relative outcomes for minorities. They thus obscure important dimensions of the continuing inequality by race and mislead us in our thinking about equality and democracy in American politics. The appropriate assessment of social capital’s impact on American civil society and politics shows that it depends on what dimensions of public life we consider, how we define “better off,” whether one is black or white, and whether one lives in a more or less racially heterogeneous community.3

Arguments and Evidence
In Bowling Alone, Putnam created a “social capital index” (for 1990), which serves as his major explanatory (independent) variable concerning the contemporary importance of social capital. The index has five components and 14 specific variables, measuring community organizational life, engagement in public affairs, community volunteerism, informal sociability, and social trust.4 This index, along with the numerous dependent variables examined in the study, does not disaggregate along racial or ethnic lines, although it includes race as a separate independent variable. How does our understanding change if we inspect the social-capital thesis more directly through the lens of racial diversity, in various arenas of politics and policy? To begin, I examined the 1990 social-capital index for the American states (see Putnam’s Web site, and its relationship to a measure of state “minority diversity” (i.e., percent black, Latino, and Asian, in 1990).5 I found a negative and substantially strong relationship: r = −0.585 (adjusted R2 = 0.33 to 0.43). (The strength of this relationship depends on whether you use a “squared term” to capture an upward swing in the regression line.6) More minority diversity is related to less aggregate social capital; less diversity is related to higher social capital. Indeed, a third or more of social capital appears related to racial/ethnic diversity.7 In short, states with high social capital tend to be relatively homogeneous in racial or ethnic terms.8 For example, the Dakotas and Minnesota have high social-capital scores. Putnam’s index of civic equality is “based on class differences in rates of political participation, as measured in Roper Social and Political Trends surveys, aggregated over 1974 to 1994. For each of the twelve forms of political participation—signing petitions, attending public meetings, and so on”—he constructed a “ratio of the logged incidence in the top quintile of the income distribution to the bottom quintile of the income distribution.”9 Putnam finds a strong relationship between social capital and civic equality—the more social capital, the more civic equality.10 But the civic-equality index, like the one for social capital, is not disaggregated according to race. Nearly all the measures used in the civic-equality index come from surveys; and given the size and distribution of racial/ethnic populations in the United States, appropriate survey evidence is likely not available or even feasible across the 50 states. Other 113

Rodney E. Hero is the Packey J. Dee Professor of American Democracy and chair of the Department of Political Science at the University of Notre Dame ( He specializes in U.S. democracy and politics, especially as viewed through the analytical lenses of Latino and ethnic/minority politics, state/urban politics, and federalism. His book Faces of Inequality: Social Diversity in American Politics received the American Political Science Association’s 1999 Woodrow Wilson Foundation Award.

and Henry Brady14 similarly shows that we should not expect to find more racial equality in other political activities.0958 (. 4.0428) .0324) . for one of the three years.6038*** (.6435**** (. 3.0275) .0903** (. and much contradicts it. ****significant at . **significant at .0257) 1994 Model 4 −. two-tailed tests. according to another set of analyses not reported here. The bivariate rela- Economic Equality: Ratios Putnam’s Gini index of economic inequality. Census Bureau—are illuminating since “the voting population is the closest match in racial and ethnic terms to the population as a whole. Models 1. census data on per capita income and poverty levels.298 44 Model 3 −. like his civic-equality measure.15 Social capital is associated with lower. This gap in data constrains a full examination of civic equality relative to race.. 3. Kay Scholzman. Models 2. substantial evidence of racial differentiation appears.S.12 Indeed. virtually no evidence supports the social-capital thesis.8711**** (.8623**** (. 114 March 2003 Vol. are available. or both (see Table 1.) Note: The Census Bureau does not provide data for all 50 states. higher aggregate social capital is associated with larger gaps between black and white registration. ***significant at .e.139) . as shown in Table 1. 1994.1518) . and 5) between social capital and black/white turnout is negative for all three years. In the same two years. such as minority diversity. I will therefore examine racial civic equality using two CPS indicators. the relationships continue to be negative and significant even after accounting for other factors. 1/No.102 43 Model 5 −.05. but might well find even less. Turnout ratios Social capital also does not go along with more equal black/white voter turnout ratios in the states. Within each state. It is also lower in states with high aggregate levels of civic culture.195 28 . I use these data (for 1990) to examine racial/ethnic “economic equality.1125**** (.0038) .0193) . Poverty ratios Is social capital positively (and significantly) related to better black/white poverty ratios? No. 1 .16 28 Model 1 −.0027) . and for two of the years it is significant.1 level.13 The authoritative story of political participation by Sidney Verba. see Models 2.0036) . does not differentiate by racial groups within states. and 6).349 44 . Civic Equality: Ratios Registration ratios Is social capital significantly related to more equal black/white voter registration ratios in the states? No. civic equality (i.” Here too.121** (.0017 (. Nonetheless. R2 N . higher levels of social capital are associated with larger gaps in voting turnout between blacks and whites even when I control for minority diversity and black poverty (in Table 2.3281** (. There is also no support for the claim that social capital is positively related to more equal ratios after controlling for other factors that affect voter registration. and 6). indicators of organizational activity also do not distinguish by the racial/ethnic backgrounds of participants.0122 (.0686** (. among the eligible population for 1992. 4.2008) .1147) .0025 (.0066* (. by racial/ethnic groups and within the states. The bivariate relationships between social capital and black/white registration ratios are all negative and statistically significant. and 1996. census data for voter registration and turnout— drawn from the Current Population Survey (CPS) of the U. the ratio of black to white registration and turnout) is lower in states with high aggregate levels of capital. not higher. black poverty.0597) .1822) . relative civic equality regarding race. However. registration and turnout. as the social-capital thesis asserts. and 5. In short.Symposium Social Capital and Diversity tionship (in Table 2. (Standard errors in parentheses. The relationship is negative and weakly significant.01. Data beyond 1996 were available but not employed. the connection should appear in voting.0487) Model 2 −. within States Dependent Variable: Black/White Voter Registration Ratio 1992 Independent Variables Social Capital Minority Diversity Black Poverty Rates Constant Adj. States with sparse populations are more likely to be excluded.6805**** (.”11 So if social capital is indeed strongly linked to civic equality.0689 (. states with higher levels of overall social capital show a somewhat greater disparity between levels of black Table 1 Social Capital with Racial Civic Equality: Voter Registration Ratios.8406**** (. see Models 1.265 43 *Significant at . because of their distance in time from the 1990 social-capital index and social-diversity measures.1432) .001.0296) .5988**** (.0376) 1996 Model 6 −.

0037) . social capital is positively and significantly related (at p = 0.”17 These complexities and ambiguities suggest that the relationship between social class and social capital is more complicated than we have realized up until now.0274) . ****significant at . That is. R2 N . the more diverse the state. as Table 4 shows.1497*** (.16 This is the first piece of evidence consistent with the social-capital thesis.001.10) to per capita income ratios. Turnout rates Overall. rates across states—rather than comparing blacks to whites within a given state. When one controls for the index of minority diversity. On the other hand. (Standard errors in parentheses. I address this in Table 3 and subsequent tables.5623**** (. because the ratios are themselves products of the rates for minorities divided by the rates for whites. because “community” is presumably more plausible within states. in other words.18 Multivariate analysis that includes controls for a state’s black population percentage and black poverty rates shows only a weak (nonsignificant) positive relationship. The bivariate www.1614) .1951) .0042) .apsanet.132) .058) . in fact. federal system that are among the most relevant to equality concerns.0006 (. Both types of measurement are valuable. the within-state (ratio) comparison is arguably the most fitting for examining racial/ethnic equality.028) 1994 Model 4 −. Social-capital analyses are ambiguous on this point. That is.0432) 1996 Model 6 . Blacks do not.7362**** (. from living in a state with high levels of social capital. compared with other members of their race.7862**** (.176 44 Model 3 −. whites benefit.1 level.0692 (. It is not clear whether more equal income leads to more social capital. However.) poverty and white poverty.159 43 *Significant at .007 43 Model 5 −. the income of the average black resident is relatively similar to the income of the average white resident.14 (. On the one hand. For this reason.” some observers . Per capita income ratios Unlike poverty ratios.120 28 Model 1 −.0025 (.6848**** (.org 115 Civic Equality: Rates Ratios of the type I have been exploring up to this point are most appropriate for examining relative patterns of equality within states.0316) −. Registration rates High levels of social capital do not go along with high rates of black voter registration across states.0034) .0496) . two-tailed tests. and whites compared to whites)—that is.0493 (. However. Additionally. In the bivariate case.S. argue for examining the racial-group patterns (blacks compared to blacks.0027 (.0708* (.0373) . within States Dependent Variable: Black/White Voter Turnout Ratio 1992 Independent Variables Social Capital Minority Diversity Black Poverty Rates Constant Adj. and because there are many ways to think about the concept of “equality.01. ***significant at .19 In this arena. states exercise the major police power and domestic policy responsibilities in the U. saying only that the patterns “go together. the bivariate and multivariate relationships for all three years are strongly positive and significant (data not shown). or the reverse.8575**** (. But social capital is significantly related to white voter registration rates.1255** (. A pattern of more equitable per capita income ratios but not more equitable black/white poverty ratios associated with higher social capital may suggest black economic polarization. states’ level of minority diversity itself has a negative and weakly significant relationship (p = −0.0892*** (. then. group comparisons across states are fruitful since they recast debates about race and (in)equality into a more directly nation-centered rather than state-centered perspective on American society.2116) . all three relationships are negative and one is significant. to the question of whether overall levels of social capital are associated with different outcomes for each racial group from one state to another.05.021) .Table 2 Social Capital with Racial Civic Equality: Voter Turnout Ratios. the relationship remains negative but is no longer statistically significant (data not shown in table form). I move.6691*** (. the greater the distance between black and white average incomes. they may even be harmed (in comparison with other blacks).15 44 .1475) . extreme (very high or very low) rates for either group will affect the overall ratios. a higher proportion of African Americans are poor. **significant at . in states with higher levels of social capital.001) to black/white per capita income ratios. there is a slightly positive relationship between social capital and levels of black turnout.672**** (.0368 (. on average blacks enjoy greater equality with whites in states with high levels of social capital—but in those same states.228 28 .0451) Model 2 −.1647) .

1 . even though the rates for whites are.484**** (6.75**** (1.329 (2. (Standard errors in parentheses.1 level.835) . **significant at .001.01.1 level.318) .2119) 39.14 28 35. 1/No.1764 (.49) . across States Dependent Variable: Black Voter Registration Rate 1992 Independent Variables Social Capital Percent Black Population Black Poverty Rates Constant Adj.311 28 55.017 43 Model 5 −3. ****significant at .2568) 44.686**** (1.2814) .008 43 Model 5 2.634 (5. ***significant at .023 43 *Significant at .845) . Thus. By not disaggregating along racial lines.41) . Thus there is strong reason to wonder if findings that a sense of community is strongly related to civic equality are largely owing to patterns in states with large white populations and small minority populations.066 28 Model 1 −5. when we compare across states within a given race.286 (2.3227 (.2592) 51.396**** (7.) 116 March 2003 Vol. That is.039 (2. we see that black rates of registration or turnout are seldom significantly more equal where there is higher aggregate social capital. the finding that civic equality is associated with social capital appears to be an artifact of white social capital and white civic participation.748) .508) 1996 Model 6 6. R2 N 62.3364 (.1151 (.20 Social capital is more strongly related to white voter turnout rates in the bivariate and multivariate cases. broad claims about the dynamics of equalization while masking what may be a situation of racial inequality. two-tailed tests.365) .2568) .018 44 Model 3 .3485 (.) relationship in 1992 is negative and weakly significant.035**** (6. the relationships for all three years are positive and significant (data not shown).015) .2099) . ***significant at .629**** (2.151**** (7.158* (3. **significant at .2805) .177) . (Standard errors in parentheses.522) 1996 Model 6 .653) 1994 Model 4 2.6792 (3.695**** (1.547) .3001 (.2395) 37.05.668) .484 (3.4938* (.835) −.7914*** (.026) Model 2 1.4037 (.01.037 44 49.4204* (3.001.3299 (.339 (5.041 43 *Significant at .8462 (1.354) . This evidence on registration and turnout ratios and rates for blacks and whites strongly implies that the civic equality identified in studies of social capital is the result of high(er) levels for whites.047 44 Model 3 −3.066**** (5. two-tailed tests.1949) −.303**** (7. R2 N 52.047**** (1.124 44 60.1891) 37.Symposium Social Capital and Diversity Table 3 Social Capital with Black Voter Registration Rates.818) .036 28 Model 1 −5. the relationship is positive but not significant.3171 (2.05. As Table 4 Social Capital with Black Voter Turnout Rates.734) −.874) 1994 Model 4 .0095 (. bivariate or multivariate.68) .883) Model 2 4.56**** (1.55) . ****significant at . across States Dependent Variable: Black Voter Turnout Rate 1992 Independent Variables Social Capital Percent Black Population Black Poverty Rates Constant Adj. The relationship between social capital and the black turnout rate for 1996 is positive and weakly significant after accounting for black population and black poverty.1978 (.243) −.458) .318* (1. in all other models except one. and perhaps of relative racial homogeneity in a number of states. social-capital analyses incorrectly make strong.2037) 43.

black poverty rates are lower (better) in states where there is more aggregate social capital—a finding consistent with the social-capital thesis.” But my findings have complicated that picture. Instead.22 Because of the issue of murky causality. the specific operationalization and measurement of social capital and of community matters more than it should. Furthermore. and (3) a regression analysis that removes states with particularly small minority populations. even after controlling for black population.apsanet. . which speak to social class but not directly to racial (in)equality—may be attributable to white patterns. These indicators address minority or black patterns (as specified in the individual analyses) and white patters. higher in states with high levels of social capital. The pattern generally holds across the several types of regression analyses.23 As I noted at the beginning of this article. Perhaps earlier findings on income distribution— based on Gini coefficients. or higher levels of social capital? Do well-off blacks move into states with active communities. with or without controls for the level of minority diversity and/or black poverty rates within states. . “kids are better off in high social capital states. This evidence starkly contradicts the social-capital thesis. in the bivariate case. are drawn from Department of Education data and from census data. incarceration. much as the civic-equality ratios did. (2) robust regression.05). Which comes first in a state—lower rates of black poverty or slightly higher rates of black income. The information for the ratios.a.29 Minority school suspension ratios The gap between minority and Anglo suspension rates is. equality. economic prosperity.” “schools work better in high social capital states.”25 Do these strong claims (that better substantive outcomes are associated with higher levels of social capital) hold when we examine 117 . Schlozman. I discuss the findings in a general way and present only the OLS regressions in Tables 5 and 6. “the public’s voice is often loud.” and “mortality is lower in high social capital states. as Table 5.Verba. The pattern holds with or without controls for levels of black poverty rates. this weak. I created relative or disaggregated indicators somewhat similar to previous research26 and considered the indicators in relation to the social-capital index.24 At a minimum. but not after controlling for minority www. The average income of whites in a state declines as social capital increases (not shown here).001). there are boosts in child welfare and educational performance. Minority school graduation ratios The gap between graduation rates for minorities and whites is greater in states with high levels of social capital. i. But this inverse relationship between social capital and the black poverty rate does not hold up when the percent black in a state is controlled.. different indicators of community—social capital rather than civic culture—produce somewhat different findings. Social capital is related to black poverty rates. These findings. minority-to-white. Substantive Equality: Ratios On the whole. which is similar in the multivariate case. with the former divided by the latter to create the ratios.01). it is difficult to interpret these findings on the connections between social capital (or civic culture) and poverty or income. the results here resemble those we found earlier regarding voter registration and turnout: higher levels of social capital do not go along with more equality in outcomes between minorities (or blacks) and whites. black per capita income increases slightly in that state in comparison to black incomes in other states. hence. the bivariate relationship is negative and significant for 1990 (p = 0. which might distort findings. or do states with active communities work harder to reduce levels of black poverty (but not white)? We do not know. shows. similarly. and infant mortality. Social capital is not related to white poverty rates (results not shown here).”21 They too point to race (along with income) as a crucial element in this pattern of political inequality. and in states with high levels of social capital. and with or without controls for levels of minority diversity in states. sometimes clear. although surprising from the standpoint of the socialcapital thesis. and Brady put it in their analysis of voluntary activity in American politics. Putnam and others have found clear evidence that “American states with the highest levels of social capital are precisely the states most characterized by economic . but rarely equal. and health and happiness. it remains after controlling for percentage of black population in a state (at p = 0. insignificant relationship also holds after controlling for the percent of the population that is black. in the case of incarceration patterns: see explanation below) on three of four indicators.a. so rather than present all of them here. are largely consistent with previous findings of “social diversity” analyses of political culture. there is a weakly negative. For example. the data indicate worse relative outcomes for minorities (or for blacks alone. we find a positive relationship that is in fact statistically significant (p = 0. Economic Equality: Rates Poverty rates Let us turn now to rates of economic inequality within race and across states. concerning education outcomes. insignificant relationship. The full analysis includes three approaches: (1) OLS regression. the reality is substantially more complicated than studies have recognized.e. when looked at through a racial lens. indicators rather than absolute or aggregate measures? To address this question.28 Table 5. shows the results. Examining Substantive Equality The social-capital thesis asserts that when the United States has more social capital. In other words. safe and productive neighborhoods. If we compare civic culture with black per capita income.27 The substantive results from each type of analysis are essentially similar. Per capita income rates As social capital in a state increases. That seems especially likely since economic equality as measured by Gini coefficients tends to be highest in racially homogeneous states.

R2 N Model 1 OLS −.Symposium Social Capital and Diversity Table 5 Minority Outcomes. this pattern persists even when controlling for levels of minority diversity and/or black poverty.g.0759) −. I focus only on non-Hispanic blacks and non-Hispanic whites. either in the bivariate case or in the multivariate case.5357) .) diversity alone or when diversity and black poverty are both considered. Dependent Variables: Minority/White Graduation and Suspension Ratios Minority/White Graduation Ratios Controlling for Minority Diversity and Black Poverty Model 2 OLS −. **significant at .0211 (. shows black-to-white incarceration ratios.127*** (.0008 (. Black incarceration ratios Table 5.577*** (. **significant at . Ratios.) 118 March 2003 Vol. 1/No. Therefore.181) −. ***significant at . states with high levels of social capital show worse relative outcomes for blacks.4881) .0179) Bivariate Model 1 OLS .042 (.683**** (. Minority infant mortality ratios Unlike for three previous indicators.a. Asian American or Native American.0138) . two-tailed tests.0496*** (.446**** (.135 47 1.014 40 Minority/White Infant Mortality Ratios Controlling for Minority Diversity and Black Poverty Model 2 OLS −.0021) .b.2702 (.1249) −.873) .b.01.941) .1037) *Significant at .1 level.758) .01.5377 (.015 40 Bivariate Independent Variables Social Capital Minority Diversity Black Poverty Rates Constant Adj.813**** (.) Clearly.125 48 . two-tailed tests.001. ****significant at .. ****significant at .5507) .5535 (3.0472* (. data on Hispanics are imprecise because Hispanics are categorized as “other.0126) 1.504** (.1 level. ***significant at .001.058 48 Minority/White Suspension Ratios Controlling for Minority Diversity and Black Poverty Model 2 OLS .094 48 2.30 It also persists when community is measured by levels of civic culture. 1 .0911 (.652* (2.0247) .05. (Standard errors in parentheses.075) .1045) .1352) *Significant at . the relative gap between minorities and whites with regard to infant mortality does not rise 5. (Standard errors in parentheses.5716) . and Social Capital 5.8022**** (.147 47 5. (In this case.1033) −.5804 (.9413) .078 48 Bivariate Independent Variables Social Capital Minority Diversity Black Poverty Rates Constant Adj. R2 N Model 1 OLS 2.0109 (. Civic culture is not significantly related to minorityto-white suspension ratios.008**** (.0705 (.1201 (.05.558*** (.0785) 8.266* (. Dependent Variables: Black/White Incarceration and Minority Infant Mortality Ratios Black/White Incarceration Ratios Controlling for Minority Diversity and Black Poverty Model 2 OLS 2.” and it is not clear how many people in the category of “other” are not Hispanic—e.7111) Bivariate Model 1 OLS −.0151) 1.

581 48 86. and the relationship with the civic-culture indicator is quite similar when controls are introduced. even after accounting for the proportion of minority population and their poverty rates. To the extent that black incarceration rates are higher in states where there is more aggregate social capital. two-tailed tests.692**** (. social capital within a state rises (see Table 5. infant mortality rates among minorities may be lower (better) where there is more social capital.74 48 2.31 Nor is there when civic culture is substituted for social capital as the independent variable.4936) . minority school suspension rates are significantly higher (much worse) in states where there is more aggregate social capital.337 (5. unlike in public schooling or incarceration. these contrasting findings on graduation and suspension might indicate economic polarization within the minority population.701**** (. higher social capital in a state is not associated with better relative substantive outcomes for racial minorities. Infants.291**** (1.1129) 68. The measure of civic culture also shows higher rates of suspension for minorities in states with a greater civic culture. WIC (Women.). to rates of inequality within a race across states.). There is no statistically significant relationship here.b. and others— play a substantial role in lessening disparities in this dimension of policy.652) −. (Standard errors in parentheses. or after accounting for both (see Table 6. These outcomes contrast substantially with those of the minority populations.0064 (. Perhaps federal government programs—such as Medicare.107) . the data tell an ambiguous story.a. this is true in the bivariate as well as the multivariate cases. Minority school graduation rates Graduation rates for minorities clearly improve as aggregate social capital increases.). nor does it fall.b. Minority school suspension rates In strong contrast to the findings for minority graduation rates. Rates. However. a finding in keeping with the earlier discussion of poverty and ratios of per capita income. Dependent Variable: Minority Graduation and Suspension Rates Minority Graduation Rates Controlling for Minority Diversity and Black Poverty Model 4 OLS 7. Minority infant mortality rates The relationship between social capital and minority infant mortality rates is not statistically significant in the bivariate case.05. and Children).). which control for levels of diversity and black poverty rates (see Table 6.) www. This is so in the bivariate analysis. after accounting for the size of either minority populations or black poverty rates. Earlier findings of better overall outcomes in high social capital states result from a focus on absolute or aggregate measures.a.b.6106**** (.). I turn now from a focus on ratios of racial minorities compared to whites within a state.427 47 Minority Suspension Rates Controlling for Minority Diversity and Black Poverty Model 4 OLS .404 47 Bivariate Independent Variables Social Capital Minority Diversity Black Poverty Rates Constant Adj. Table 6 Minority Outcomes.0902 (. However. including the outcomes for the large nonminority populations in relatively homogeneous states.918**** (.apsanet. ****significant at .1601) −.6708) .35) −1.1 level. if you account for levels of minority diversity either alone or together with black poverty rates (see Table 6. In sum. **significant at .0907) . R2 N Model 1 OLS 10.32 Black incarceration rates Black incarceration rates are no less or more disparate where there is higher aggregate social capital. This impressively supports the social-capital thesis and is reaffirmed when civic culture is substituted for the social-capital measure.832**** (.967**** (1.0138) 1. and Social Capital 6. the essential findings follow in Table 6.1165) *Significant at . but on this measure. ***significant at . Together.001.01. in the bivariate case or in the multivariate case (see Table 6. they are higher (worse) when you control for levels of minority diversity alone and combined with black poverty rates in the robust regression analysis (data not shown).266) Bivariate Model 1 OLS . these outcomes are sometimes worse in states with high(er) levels of social capital.748**** (4. which are mainly affected by states’ policy choices.9787) . The civic-culture measure produces Substantive Equality: Rates As with the discussion of civic and economic equality.a. the data are not consistent with the social-capital thesis’ claims about equality.07**** (.org 119 .

605 (4.614 (1. two-tailed tests. **significant at . The minority graduation rates. community may be as much a problem as it is a solution for a host of social and political concerns in American civil society. to the extent that more community in the aggregate is associated with more economic equality.) roughly similar findings when controls are added. emphasis on its creation may inadvertently shift the balance away from bridging social capital and toward bonding social capital. because the identification and nature of “problems” and potential “solutions” are likely to be defined and addressed differently.Symposium Social Capital and Diversity 6.9371) Bivariate Model 1 OLS −1. however.727) . the evidence on civic equality and substantive equality ratios suggests no support for—and often directly contradicts—the implied claims of the social-capital thesis. The outcomes with regard to rates are sometimes. .221) 8. 1 is associated with more aggregate social capital. Putnam has noted a “.” or “biased bonding. Here. (Standard errors in parentheses. Configurations of social capital are clearly complicated and need to be better understood before we accept them as having generally benevolent consequences.b. .458) . The pattern in high social-capital settings that are relatively racially homogeneous may be better characterized as “bounded bridging. However. Dependent Variable: Black Incarceration and Minority Infant Mortality Rates Black Incarceration Rates Controlling for Minority Diversity and Black Poverty Model 4 OLS 1. or between having a community of people who are likely to express very divergent viewpoints and having a community of people who are likely to participate effectively in shared democratic decision making.113) . social capital is associated with positive outcomes.” Similarly. Social capital and civic culture are negatively and substantially related to racial and ethnic diversity in the states.1 level. The findings in this article also raise normative and theoretical questions about whether we should think about racial-group equality within or across states. it matters whether the federal polity is considered from more nation-centered or state-centered perspectives. race—as well as ethnicity and class—is fundamental in defining American “community” and in understanding American civil society. In some ways. that association occurs only or mostly in the white population and/or around social-class issues.0102 (.624* (. It is primarily for economic equality across the races that some support for the social-capital thesis is indicated.033 40 Minority Infant Mortality Rates Controlling for Minority Diversity and Black Poverty Model 4 OLS −1. Summary and Conclusions The social-capital thesis has deeply influenced thinking about community. Nevertheless. Since social capital is inevitably easier to foster within homogeneous communities. since different understandings of equality are implied by each choice. risk that emphasizing community exacerbates division and exclusion.01.1018) 12.05.8849) −5. thus. But social-capital studies understate the importance of race and ethnicity in their conceptualization and analyses. Rates of infant mortality among minorities.5614) .4*** (3. Minority graduation rates are higher where there is higher social capital. And how should African Americans judge a state in which there is greater racial inequality but also a higher floor. R2 N Model 1 OLS . They focus almost entirely on aggregate outcomes and absolute gains.12**** (.0892) 11.337** (3.33 If these arguments are correct. equality.015 47 9. and democracy in the United States and around the world. ***significant at . ****significant at . the “bridging” extolled in states with high social capital is more limited than most analyses have recognized. but minority suspension rates are also higher. but not always. are lower in states with high levels of social capital.344**** (. as compared with a state in which blacks are on average poorer or less politically active but not so dissimilar to whites? .001.002 47 13. they importantly amend the social-capital thesis.049) .424 (5.168 (. in combination with the suspension and incarceration rates. indicate that social capital may have facilitative effects for some people of color and substantial social-control effects for other people of color.706) −.139 40 Bivariate Independent Variables Social Capital Minority Diversity Black Poverty Rates Constant Adj.1231 (. I found little evidence that racial equality on civic or substantive dimensions 120 March 2003 Vol. In assessing American institutions.5231 (. For instance.”34 Perhaps there is a deep tension between diversity and connectedness. consistent with social-capital claims. 1/No.7673) *Significant at .

851–71. more fitting understanding of American community.Figure 1 Scatterplots: Social Capital and Registration and Turnout Ratios (1994. Caroline J. Paper presented at the American Political Science Association annual meeting. Michael. and Caroline J. Behind the Mule: Race and Class in African-American Politics. Rodney E.. Rodney E. Elazar. Daniel J. 29 August–1 September. New York: Crowell. A more complete. References Dawson. Murray. American Federalism: A View from the States. Urbana: University of Illinois Press. we must appreciate the complexity of American equality. 1998.. Hero. 1967. Edelman. and of its presumed democratic and egalitarian aspirations. 1996. Hero. American Journal of Political Science 40:3. 1972. Boston. 1994. is likely to emerge when we theoretically and empirically integrate that complexity into our thinking. 1996) Most generally. Rodney E. The Symbolic Uses of Politics. Tolbert.S. Tolbert.apsanet. www. 1984. Hero. A racial/ethnic diversity interpretation of politics and policy in the states of the U. Princeton: Princeton University Press. 2002. Oxford: Oxford University Press. scholarship. and public-policy discourse. and Ramona McNeal. Faces of Inequality: Social Diversity in American 121 . “Race” and “community” as influences on political participation: Social diversity and social capital considered.

Hero 1998. Class. Evidence (from 1996 and 2000) on other forms of political participation corroborates these findings. 8 Putnam 2000. cf. cf. Kenneth. Princeton: Princeton University Press. 16 The same pattern holds if one compares Rice and Sumberg’s measure of civic culture to income ratios. 297. See his Web site. 15 Hero et al. 28 Cf. 21 Verba et al. 300. chapter 17. The relationship of the social-capital index with the civicculture indicator is r = 0. 1991. Class. Meier et al. and passim. Putnam 2000. Facing Up to the American Dream: Race. 99–114.. Making social science work across space and time. 26 Hero and Tolbert 1996. 1989 and Meier and Stewart 1991. Benjamin. and Hartz: The multiple traditions in America. Kenneth.—I refer to blacks and whites. 91–2. 359–61. 298. essentially the same patterns occur when a “civic culture” measure (developed in Rice and Sumberg 1997)—a similar yet different indicator of community that correlates highly with the “social capital” index—is examined with registration ratios. 18 And one sees essentially the same findings when the civicculture measure is substituted for the social-capital indicator. Voice and Equality: Civic Voluntarism in American Politics. Smith. when the civic-culture measure replaces the social-capital measure of community. 22 Yet with the same controls. Jr. 291. 389–97. Publius: The Journal of Federalism 27:1. also see Rice and Sumberg 1997. Hero 1998. 6 There is also a strong relationship with the Rice and Sumberg 1997 measure of “civic culture” (a concept that is quite similar to social capital): r = -0. 1989. 106–7. 1/No. n. cf. cf. 105–7. 31 Cf. 495. 1993. Putnam 2000. 29 Hero 1998. 328. 99–100. chapters 16–22. etc.) 13 Moreover. 237. Notes 1 Putnam 2000. Kay Lehman Schlozman. largely from surveys and analyses conducted after he completed Bowling Alone. 19 Similar. 17 Putnam 2000. 24 See Putnam 2000. Hero and Tolbert 1996 for similar findings concerning social diversity and “political culture”. 2000. Tarrow. Tarrow 1996. 1995. 12. 23 Cf. Myrdal. turnout. www. and Education: The Politics of Second Generation Discrimination. Bowling Alone: The Collapse and Revival of American Community. 2 Cf. Madison: University of Wisconsin Press. Beyond Tocqueville. Hochschild 1995. cf. Putnam 2000. American Political Science Review 87:3. Rice and Sumberg 1997. Sumberg. when I discuss issues of civic or economic equality—voter registration and turnout. Hochschild. 1984. The Politics of Hispanic Education. which also find detrimental consequences. Sidney. 1995. and Alexander F. and public policy in the American states. and 359. 1996. Robert D. Hero 1998. 30 Cf. 1995. 1 . especially 297–98. 10 Putnam 2000. New York: Simon and Schuster. Smith 1993. chapter 18. Meier. civic culture and white per capita income are positively and significantly related. Brady. Albany: State University of New York Press. (Incarceration is an exception here. 360. 1989 and Meier and Stewart 1991. 5 Hero 1998. but closely parallel to. Race. relationships materialize when you substitute the civic-culture measure (data not shown) for the social-capital indicator. Putnam. 27 The latter two approaches account for outliers or extreme cases. Elazar 1972. and Joseph Stewart. 20 But this relationship does not hold when you substitute the civic-culture measure of community for the socialcapital indicator. and is actually more pronounced. 330. 25 Putnam 2000. Joseph Stewart. Rice. Dawson 1994.a. although somewhat weaker. 49. 7 Cf. Political Research Quarterly 48:4. 400. 4 Putnam 2000. 9–14. 358–63 (emphasis in original) and chapter 22 more generally.560. Tom W. Radcliff and Saiz 1995. When discussing outcomes such as graduation and infant mortality. Cambridge: Harvard University Press. the data are on blacks and whites. Verba.Symposium Social Capital and Diversity 9 Putnam 2000. 122 March 2003 Vol. the social-capital index. and the Soul of the Nation. poverty rates.818. and Robert England.bowlingalone. 1997. 33 Edelman 1967. 14 Verba et al. additionally. Jr.” such as Hero and Tolbert 1996 and Hero 1998. Radcliff. and Martin Saiz. Jennifer L. 1995. 775–94. Putnam 2000. all emphases added. 509. This pattern holds. Rogers M. Race. Meier et al. also see Hero 1998. emphasis added. see Table 5. for new data and Rice and Sumberg 1997. Putnam 2000. 34 Putnam 2000. 1995. 12 In general. 549–66. 1995. Meier. 3 Putnam’s most recent work demonstrates that he has reached the same conclusion. 11 Verba et al. American Political Science Review 90:2. which was created separately from.. Sidney. emphasis added. 2002. see examinations of race and “political culture. 32 Cf. 497. and Henry E. also see chapter 8 more generally. Civic culture and government performance in the American states. I refer to minorities and whites.