Need within 4 hours (Deadline 8:45 EST or better)

American Behavioral Scientist 56(11) 1462 –1480

© 2012 SAGE Publications Reprints and permission: http://www. sagepub.com/journalsPermissions.nav

DOI: 10.1177/0002764212458275 http://abs.sagepub.com

ABS458275ABS561110.1177/0002764212458275American Behavioral ScientistHunt and Ray © 2012 SAGE Publications

Reprints and permission: http://www. sagepub.com/journalsPermissions.nav

1Northeastern University, Boston, MA, USA 2University of Maryland, College Park, MD, USA

Corresponding Author: Matthew O. Hunt, Sociology Department, Northeastern University, Boston, MA 02115, USA E-mail: [email protected]

Social Class Identification Among Black Americans: Trends and Determinants, 1974–2010

Matthew O. Hunt1 and Rashawn Ray2

Abstract

Although much research documents the growth of a “professional middle class” among African Americans over the past several decades, we know comparatively little about how Blacks see themselves in social class terms, and whether this has changed over time. In the current study, we use data from the 1974 to 2010 General Social Surveys to analyze trends in, and the determinants of, Blacks’ social class identifications (SCI) over the past four decades. Our results show that Blacks’ tendency to identify as “middle class” has increased in concert with Blacks’ socioeconomic status (SES) gains since the 1970s. Regarding the determinants of SCI, education and household income appear more consequential than occupational prestige and self-employment in shaping Blacks’ self-reports of their own class positions. Finally, we see little evidence of change over time in the relationship between various SES characteristics and SCI, with one exception: Self-employment has become a more potent predictor of Blacks’ SCI over the past several decades. Our results provide an important update to our knowledge of the dynamics of SCI among Black Americans. They also raise important questions for future research on the relationship between, and relative impact of, “race” and “class” in shaping Blacks’ identities and their orientations toward American society.

Keywords

stratification ideology, African Americans, middle class

Article

Hunt and Ray 1463

Among the most important changes in the American social landscape over the past several decades are the well-documented socioeconomic gains of Blacks, and the con- sequent growth of a Black middle class (Hout, 1984; Landry, 1987; Marsh, Darity, Cohen, Casper, & Salters, 2007; Wilson, 1978). Research on social mobility, occupa- tional status, educational attainments, income inequality, residential attainments, and numerous other indicators clearly document the growth of a “professional middle class” among African Americans (see Landry & Marsh, 2011). What is less often examined, however, is how these changes in the “objective status” of Black Americans have impacted Blacks’ subjective sense of their own social class standing. That is, in the face of much work on Blacks’ actual social class positions, we know relatively little about how Blacks see themselves in social class terms. To wit, in their recent review of the literature on the Black middle class, Landry and Marsh (2011) note that “we found little research on how middle class Blacks see themselves,” which these authors suggest represents, “an important area for future research on the Black middle class” (p. 33). We address such issues in the current study. Specifically, drawing on data from the 1974 to 2010 General Social Surveys (GSS), we examine trends in, and key determinants of, Blacks’ social class identification (SCI) across the 1974 to 2010 time period.

Background Social Class—Objective and Subjective Dimensions

As the preceding paragraph suggests, social class has both “objective” and “subjec- tive” referents (Hout, 2008; Jackman & Jackman, 1983; Vanneman & Cannon, 1987). The study of the former generally centers on how people make a living (i.e., their occupations and employment relations), their academic credentials (i.e., education levels), and how well they are remunerated within labor markets (i.e., how much money they make) as the indicators of where people reside in the social class hierar- chy. Subjective assessments of social class position generally focus on peoples’ own reports of where they reside in the class hierarchy and are typically studied via survey questions offering a set of class labels/levels (e.g., lower, working, middle, or upper class) from which respondents choose an identification.1

The study of SCI is traceable to the work of Centers (1949), who documented a clear association between White males’ occupations and their self-placement into dif- ferent social classes. For example, 70% of Centers’s respondents with professional occupations identified as middle-class, whereas 70% of his manual laborer respon- dents identified as working-class.2 Subsequent studies have documented comparable associations between objective class placements and Whites’ SCI using more repre- sentative samples than Centers employed in his seminal work (e.g., Jackman & Jackman, 1983; Vanneman & Cannon, 1987). Research on differences in SCI by race and/or the determinants of SCI among non-White minority populations is scarce— particularly for recent decades (though Hout, 2008, offers some important insights,

1464 American Behavioral Scientist 56(11)

reviewed below). However, using data from a 1975 national survey, Jackman and Jackman (1983) concluded that “socioeconomic status is completely unsuccessful in accounting for the class identification of blacks” as (for middle- and working-class Blacks) “their class identity is decidedly secondary to their racial identity” (p. 83). And, using data from 1952 to 1974, Vanneman and Cannon (1987) observe a similarly weak socioeconomic status SES/SCI relationship among Blacks, though they also document a notable increase in the tendency of Blacks to identify as middle-class from 1952 to 1974, which they attribute to the “upgrading of the Black class and status structure” during that time period (p. 248).3

Building on this important work, we explore trends in, and the determinants of, Blacks’ SCI since the 1970s. Among our central aims is determining whether and how traditional objective class indicators (e.g., education, occupation, income, self- employment) have come to shape SCI among Blacks, and whether this relationship has shifted over time. Answers to such questions stand to provide an important update to past research using data from 1970s (and earlier) and will shed light on long- standing debates over the operation of social class among African Americans, and how best to conceptualize the “Black middle class.”

Defining the Black “Middle Class” In surveying existing definitions of the “Black middle class,” Landry (1987) could locate no scholarly or popular consensus on the topic, though three factors—education, occupation, and income—were most prevalent in prevailing definitions (variously weighted by different observers). This focus on education, income, and occupation is, of course, consistent with conventional sociological treatments of socioeconomic standing (Davis & Robinson, 1998; Hout, 2008; Grusky, 2000) and is traceable to the earliest work on the subject of the Black middle class. For instance, Du Bois (1899/1996) argued that Blacks’ class status was a function of education, income, and occupation but also included sociocultural factors such as membership in “respect- able” families. Subsequent scholars have built on Du Bois’s conception, adding vari- ous nuances such as an emphasis on political power and influence (Drake & Cayton, 1945); values, behavior, and character (Frazier, 1957/1997); marital status and spouse’s work status (Willie, 1979); the class-status of neighborhoods (Pattillo, 2005); and, most recently, income-driven status distinctions within the Black middle class (Lacy, 2007) and even household structure given the growing group of “single and living alone” (SALA) Black middle-class women (Marsh et al., 2007).

Despite such variation in emphases, one strain of thought—traceable to the work of Du Bois (1899/1996, 1903)—highlights education as the most important determinant of middle-class status among African Americans.4 Along these lines, several decades ago Kronus (1971) argued that, with respect to the Black middle class, “education is more frequently emphasized than any other status element, followed by occupation and income respectively” (p. 5). And, more recently, Morris (1994) has postulated that—given Blacks’ unique history of striving for educational equality—education

Hunt and Ray 1465

arguably matters more to Blacks’ class identification than occupational prestige, income, and self-employment. As such, an important question for the current study is whether education demonstrates a distinctive role (vis-à-vis other SES components) in shaping how Blacks’ view themselves in class terms, and whether this has changed over time.

Studying Social Class Among African Americans The lack of consensus on which objective indicators—alone or in combination—are most central to Black middle-class status (and/or identities) is reflected in empirical research, which has a mixed record with respect to the examination of education, occupation, and income. Most often, these class indicators have been studied in isola- tion, with occupational prestige perhaps the most frequently examined factor (Gatewood, 2000; Hwang, Fitzpatrick, & Helms, 1998; Landry, 1987), alongside some studies examining two key SES characteristics (Durant & Sparrow, 1997; Hochschild, 1995; Willie & Reddick, 2003). Rarely, however, are education, income, and occupation examined simultaneously in attempts to operationalize Blacks’ actual class positions and/or to examine the correlates of Blacks’ SCI.

One important recent exception is Hout’s (2008) analysis of SCI using GSS data from 1973 to 2004, which includes an examination of the relationship between SCI and income, education, and occupation.5 Whereas Hout’s primary focus is on Americans generally, he does give selected attention to African Americans’ SCI via (a) the use of race as a predictor (race is a weaker predictor than the examined SES factors; and Blacks are less likely than Whites to identify as middle-class, though this difference narrows over time) and (b) analysis of the relative predictive power of SES on SCI among various societal subgroups, including African Americans. Here, we learn that SES has significant effects among Blacks, but they are weaker than among other subgroups—a finding seen in earlier studies (Jackman, 1979; Jackman & Jackman, 1983; Kluegel, Singleton, & Starnes, 1977).

While offering numerous important insights, Hout’s analysis does not examine over-time change in the nature and determinants of Blacks’ SCI (perhaps by necessity, given his chosen foci and the scope of the study).6 Thus, we take as our primary focus trends in, and the determinants of, SCI, including possible over-time changes in the association between SES and SCI. In addition, we analyze self-employment as a deter- minant of SES alongside income, education, and occupation because of (a) the central- ity of small businesses ownership (i.e., Marx’s “petit bourgeoisie”) in important sociological class-analyses schemes (Wright, 1996) and (b) the growth of the small business/entrepreneurial sector among African Americans in recent decades (Bates, Jackson, & Johnson, 2007). Therefore, our analyses stand to provide important new insights into which African Americans identify with various social classes, why, and how this may be changing. Finally, it is important to point out that, whereas our pri- mary focus is on socioeconomic determinants, we examine numerous sociocultural

1466 American Behavioral Scientist 56(11)

and lifestyle factors that may influence Blacks’ SCI, given the emphases on these fac- tors from classic work on the Black middle class (Du Bois, 1899/1996; Frazier, 1957/1997).

Research Questions and Expectations We pursue three primary research questions in this study. First, how, and to what extent, has Blacks’ SES increased over the past several decades? Our task here is to document—for each of the four SES characteristics we examine: education, house- hold income, occupational prestige, and self-employment—the precise nature and extent of socioeconomic gains by Blacks since the 1970s. Given well-documented changes in the standing of Blacks, we expect to see increases over time in the mea- sures of SES we employ; however, specifying these patterns across four separate indicators, by decade, promises to add new knowledge to our understanding.

Second, we ask whether Blacks’ SES gains have been accompanied by increases in SCI. Our aim here is to demonstrate the extent to which increases in objective status over the past several decades have been accompanied by corresponding changes in how Blacks see themselves in social class terms. Given past evidence of such shifts (Vanneman & Cannon, 1987), alongside recent research indicating growth of the Black middle class (Lacy, 2007; Marsh et al., 2007), it seems reasonable to expect corresponding change in SCI (i.e., a larger percentage of Blacks should identify as middle-class, vis-à-vis other classes, over time). However, given that previous work has found only weak to nonexistent connections between SES and SCI among Blacks (Jackman & Jackman, 1973, 1983), the nature and extent of any over-time shifts dur- ing the past several decades are open questions.

Third and finally, what is the precise relationship between various components of SES and Blacks’ SCI, and has this changed over time? We explore the first part of this question (“3a”) via multivariate models allowing us to specify which SES characteris- tics are paramount in shaping SCI, and the degree to which each operates indepen- dently. This is important because much past work on Blacks’ social class standing, and identifications, has either failed to simultaneously examine multiple SES predictors, or has found them to be decidedly weak determinants of Blacks’ SCI (Jackman & Jackman 1983). This undertaking also allows us to speak to journalistic and scholarly debates concerning the “true” locus of class identity among Blacks, as well as the nature of Black middle-class status more generally.

Subsequently, we examine whether there have been any statistically significant changes over time in the associations between key SES characteristics and SCI among Blacks (“3b”). Given Blacks’ gains in various arenas (e.g., education, labor market, business ownership), it is important to ascertain whether any shifts have occurred in the relationship between various SES characteristics and Blacks’ tendencies to identify in social class terms. Using GSS data from 1973 to 2004, Hout (2008) observed that—among all GSS respondents—the predictive power of income has

Hunt and Ray 1467

increased over time, relative to the effects of education and occupation (whose effects weakened). Whether this generalization holds among Blacks is among the important questions we consider herein.

Data, Measures, and Methods Data

To address our research questions, we draw on data from the 1974 to 2010 GSS. The GSS has been conducted by the National Opinion Research Center (NORC) annually between 1972 and 1994, after which it began being administered biannually. The GSS is designed to yield a nationally representative sample of English-speaking individu- als 18 years and older living in noninstitutional settings within the United States. By retaining the exact wording of survey questions, the GSS lends itself to the analysis of over-time trends in attitudes and opinions. Given our focus on the dynamics of Blacks’ SCI, we limit our analysis to GSS respondents identified as “Black” in the variable RACE. We outline the key variables used in this study below. Our initial analyses (Tables 1 and 2) utilize all available (Black) respondents across the 1974 to 2010 period (see tables for detailed Ns). Subsequent analyses (i.e., our regression models) were run using a list-wise deletion of cases based on our “full model” (Model 6 in Table 3), resulting in an analytic sample of 5,293 for the models presented in Table 3 (and Table 4, though Ns are listed separately by decade there). We also ran each of the individual models comprising Table 3 with the maximum number of avail- able cases for each model (Ns vary from 5,717 to 6,469) and observed results similar to those presented.

Table 1. Socioeconomic Characteristics by Decade for Blacks (means)

1970s 1980s 1990s 2000s

Mean N Mean N Mean N Mean N

Education 12.03a,b,c 789 12.50d,e 2,074 12.83f 1,707 13.09 2,186 Occupational prestige 31.23a,b,c 710 33.78d,e 1,902 38.62f 1,559 40.36 2,028 Household income 20.49 706 19.05d,e 1,868 21.46 1,479 21.53 1,897 Self-employed 0.06 789 0.04d,e 2,074 0.06 1,707 0.06 2,186

Note: Education = years; occupational prestige range = 12–86. household income metric = $1,000 (1986 dollars); self-employed = proportion. a1970s and 1980s differ significantly (p < .05, two-tailed test). b1970s and 1990s differ significantly (p < .05, two-tailed test). c1970s and 2000s differ significantly (p < .05, two-tailed test). d1980s and 1990s differ significantly (p < .05, two-tailed test). e1980s and 2000s differ significantly (p < .05, two-tailed test). f1990s and 2000s differ significantly (p < .05, two-tailed test).

1468 American Behavioral Scientist 56(11)

Dependent Variable

The following question has been asked every year since 1974: “If you were asked to use one of four names for your social class, which would you say you belong in: the lower class, the working class, the middle class, or the upper class?” Individuals were

Table 2. Blacks’ Class Identification Over Time (in percentages)

1970s 1980s 1990s 2000s

Lower 12.29 10.90 10.49 12.21 Working 58.17 56.51 55.42 54.16 Middle 25.35b,c 28.69 31.52 30.33 Upper 04.18b 03.91d 02.58 03.29 N 789 2,074 1,707 2,186

b1970s and 1990s differ significantly (p < .05, two-tailed test). c1970s and 2000s differ significantly (p < .05, two-tailed test). d1980s and 1990s differ significantly (p < .05, two-tailed test).

Table 3. Logistic Regressions of Class Identification on Socioeconomic Characteristics for Blacks, 1974–2010 (Middle and Upper Class = 1).

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6

1980s 0.220* 0.145 0.182 0.224* 0.222* 0.174 (0.106) (0.107) (0.106) (0.107) (0.106) (0.108) 1990s 0.193 0.078 0.094 0.145 0.192 0.069 (0.111) (0.112) (0.112) (0.112) (0.111) (0.113) 2000s 0.246* 0.086 0.118 0.186 0.246* 0.079 (0.107) (0.109) (0.109) (0.109) (0.107) (0.111) Education 0.153*** 0.094*** (0.016) (0.020) Occupational prestige 0.015*** 0.002 (0.002) (0.003) Respondent’s income 0.020*** 0.016*** (0.002) (0.002) Self-employed 0.172 0.146 (0.125) (0.127) Constant –1.555 –3.369 –1.965 –1.674 –1.559 –2.811 Log Likelihood –3,193.003 –3,148.425 –3,172.802 –3,121.034 –3,192.065 –3,103.831 Chi-square 168.066 257.222 208.467 312.003 169.941 346.410 Pseudo R-squared .026 .039 .032 .047 .026 .053 Observations 5,293 5,293 5,293 5,293 5,293 5,293

Note: These models control for gender, age, marital status, parents’ education, work status, school attendance, region, city size, religiosity, Protestant affiliation, church attendance, and political party identification. Standard errors are in parentheses. *p < .05. ***p < .001 (two-tailed tests).

Hunt and Ray 1469

only allowed to make one selection. We refer to this variable as Social Class Identification (or SCI). Our initial descriptive analysis examines all four categories, but because the number of respondents identifying as lower-class and upper-class is small, we collapse “lower” and “working” into a single category, and do likewise with the “middle and “upper” class categories, in subsequent regression analyses. In these analyses, respondents who reported a lower- or working-class identification are assigned a value of 0, whereas those who identified as middle- and upper-class are assigned a value of 1.7 This dichotomous measurement strategy is consistent with previous research on SCI (Davis & Robinson, 1988, 1998; Hout, 2008).

Independent Variables Year of survey. As a key focus of the current study is examining over-time trends, we

disaggregate the sample by decade to compare determinants of social class identifica- tion in four time periods: the 1970s, 1980s, 1990s, and 2000s. We also include dummy variable measures of each decade in some of the presented regression analyses, using the 1970s as the excluded category.

Table 4. Logistic Regressions of Class Identification on Socioeconomic Characteristics for Blacks, 1974–2010 (Middle and Upper Class = 1)

1970s 1980s 1990s 2000s

Education 0.173* 0.053 0.155*** 0.082* (0.077) (0.037) (0.041) (0.032) Occupational prestige 0.008 –0.000 0.006 0.001 (0.009) (0.005) (0.006) (0.005) Respondent’s income 0.003a 0.021*** 0.017*** 0.014*** (0.006) (0.004) (0.004) (0.003) Self-employed –0.779b,c –0.304e 0.405 0.547** (0.477) (0.279) (0.245) (0.208) Constant –4.748 –2.168 –3.383 –2.415 Log likelihood –341.080 –974.374 –723.628 –1,021.600 Chi-squared 61.644 115.284 143.880 106.444 Pseudo R-squared .083 .056 .090 .050 Observations 635 1,658 1,288 1,712

Note: These models control for gender, age, marital status, parents’ education, work status, school attendance, region, city size, religiosity, Protestant affiliation, church attendance, and political party identification. Standard errors are in parentheses. a1970s and 1980s differ significantly (p < .05, two-tailed test). b1970s and 1990s differ significantly (p < .05, two-tailed test). c1970s and 2000s differ significantly (p < .05, two-tailed test). e1980s and 2000s differ significantly (p < .05, two-tailed test). *p < .05. **p < .01. ***p < .001 (two-tailed tests).

1470 American Behavioral Scientist 56(11)

SES. In line with the existing literature on social class identification, occupational mobility, and the Black middle class, we analyze four main socioeconomic character- istics: education, occupational prestige, income, and self-employment. Education is measured in years as a continuous variable ranging from 11 (less than high school degree) to 20 (doctorate). Occupational Prestige is measured as a continuous variable ranging from 12 to 86, and is based on the social standing of occupations as estimated by samples of Americans. The GSS includes two occupational prestige scales (1970 and 1980). The 1970 scale was originally based on the 1960 U.S. Census and then adapted for the 1970 U.S. Census, whereas the 1980 scale is based on the 1980 Cen- sus. To create a single occupational prestige variable, missing values on the 1970 scale are replaced with prestige scores from the 1980 scale.

Respondent’s Household Income is measured as a continuous variable that adjusts for inflation from the 1970s to 2000s. Specifically, the GSS variable REALINC includes respondents from all waves of the survey and transforms respondents’ house- hold incomes to a standard measure (metric = $1,000) adjusted to 1986 dollars. Ligon (1989) and Hout (2004) note in GSS Methodological Reports #64 and #101 that this measure of income is suitable for analyzing the effect of earnings over time because it adjusts for inflation, follows procedures important for not overestimating or underes- timating the higher and lower categories of income, and takes into account the Consumer Price Index for Urban Consumers (CPI-U) as published by The Bureau of Labor Statistics. Finally, Self-Employment is included as a proxy for business owner- ship, and is measured as a dummy variable coded: 0 = not self-employed, 1 = self-employed.

Sociodemographics. In addition to the four main SES variables, we include a series of sociodemographic variables known to be associated with SCI and/or Blacks’ iden- tification with the middle class. Gender is measured as a dummy variable coded 0 = female, 1 = male. Age is measured in years, and ranges from 18 to 88. Marital status is measured as a series of dummy variables (i.e., married, never married, formerly married). Formerly Married is coded 1 if respondents are divorced, widowed, or sepa- rated; 0 otherwise. Full-Time Employment and School Attendance are separate dummy variables coded 1 if respondents are employed full-time or attending school, 0 otherwise. Home Ownership, a proxy for wealth and potential gateway to middle class status, is measured as a dummy variable.8 We also, in line with the status attainment and race literatures, include measures of both mother’s and father’s education. Father’s Education is coded 0 if the father does not have a college degree and 1 if he had at least a college degree (or 16 years of education). Mother’s Education is coded analogously.9

We also include a series of religion measures, given their mention in the Black middle-class literature. Religiosity measures strength of religious beliefs (0 = not reli- gious, 1 = not very strong religious beliefs, 2 = strong religious beliefs, 3 = very strong religious beliefs). Church Attendance measures how often respondents attend church (0 = never, 1 = yearly, 2 = monthly, 3 = weekly). And we include a dummy variable for being Protestant. In addition, we include a series of dummy indicators for region

Hunt and Ray 1471

(i.e., South, East, Midwest, and West, following U.S. Census definitions). To capture the effect of neighborhood context, urbanicity is included to measure population den- sity (0 = population of 100,000 or less; 1 = greater than 100,000). Finally, Political Party Identification is measured as a continuous variable ranging from strong Democrat to strong Republican on a scale from 0 to 6.

Findings Question 1: How, and to what extent, has Blacks’ SES changed from the 1970s

through the 2000s?

Table 1 presents our four indicators of SES, by decade. All six “decade vs. decade” comparisons (e.g., 1970s vs. 80s; 1970s vs. 90s, etc.) are significant for education and occupational prestige, whereas only the 1980s vs. 1990s and 2000s comparisons are significant for household income and self-employment. Education increases mono- tonically from a mean of 12.03 years in the 1970s to 13.09 years in the 2000s. Occupational prestige likewise increases monotonically from a mean of 31.23 in the 1970s to 40.36 in the 2000s. Household income increases from 20.49 in the 1970s to 21.53 in the 2000s, though with a slight dip in the 1980s (at 19.05).10 Only self- employment remains more or less static over time at 6% (though with a slight dip, to 4%, in the 1980s). Thus, without controlling for other ways in which respondents differ, Blacks’ SES (with a few minor exceptions) does appear to have increased over time in the GSS data, consistent with expectations.

Question 2: Have Blacks’ SES gains been accompanied by increases in SCI?

Table 2 reports Blacks’ SCI by decade. Here, results document net gains in middle- class identification, alongside declines in each of the other three class identifications over time (though changes in “lower-class” identification may be better described as curvilinear; more on this below). Specifically, whereas just over 25% of Blacks iden- tified as middle-class in the 1970s, this increased approximately 5 percentage points (to just over 30%) in the first decade of the 21st century (though middle-class identi- fication actually peaked, at 31.5%, in the 1990s). The difference in middle-class identification between the 1970s and 2000s is statistically significant (as is the increase from the 1970s to the 1990s).

Lower-class identification declined from 12.3% in the 1970s to a low point of 10.5% in the 1990s, followed by a modest uptick—to 12.2%—in the most recent decade.11 None of the over-time differences in lower-class identification are statisti- cally significant. Working-class identification declined monotonically from the 1970s (58.2%) to the 2000s (54.2%), though these shifts are also not statistically significant. Finally, upper-class identification was highest, at 4.2%, in the 1970s, and then declined over time to a low of 2.6% in the 1990s; this was followed by a slight uptick to 3.3%

1472 American Behavioral Scientist 56(11)

in the 2000s. The declines in upper-class identification from the 1970s to 1990s, and from the 1980s to the 1990s, are statistically significant.

Thus, our initial descriptive analyses suggest that Blacks’ propensity to identify as middle-class increased between the 1970s and 2000s (with a peak in the 1990s), along- side the previously observed increases in SES. That said, the precise relationship between SES and SCI cannot be gleaned from our Table 1 and 2 results. Therefore, we turn next to a series of logistic regression analyses designed to better specify the asso- ciation between SES and SCI, and whether this has changed over time.

Question 3a: Which determinants of SCI are paramount in a multivariate analysis?

Table 3 presents results of the logistic regression of SCI (middle and upper = 1, lower and working = 0) on decade and SES predictors, along with a host of other controls (whose effects are not shown; see Methods section for a complete listing). We estimated six models. Model 1 reports only decade terms. The subsequent four models each contain one of the four SES characteristics (in addition to the decade terms): Model 2 includes education, Model 3 includes occupational prestige, Model 4 includes household income, and Model 5 includes self-employment. Model 6 includes all of these predictors simultaneously.

These analyses allow us to make several conclusions regarding the determinants of Blacks’ SCI. First, Model 1 reproduces our Table 2 finding of a statistically significant increase in middle-class identification over time, as seen in the positive and significant coefficients for the 1980s and 2000s terms (the 1990s term, at p = .08, barely misses statistical significance at the conventional “.05” level). Models 2 through 4 demon- strate that, when entered separately, education, occupational prestige, and household income all positively and significantly predict SCI among Blacks.12 However, Model 5 shows that self-employment does not. Furthermore, regarding the implications of various SES controls for the observed over-time changes in SCI, education and occu- pational prestige move the formerly significant effects of the decade terms to nonsig- nificance, whereas income and self employment do not (in the “income” model [#4], the 1980s term remains significant, and in the “self-employment” model [#5], both the 1980s and 2000s terms remain significant). These findings suggest that variation in Blacks’ levels of education and occupational prestige may be more central to under- standing over-time change in SCI than self-employment and (to a lesser extent) income. Finally, Model 6 shows that only education and income retain their indepen- dent effects on SCI when all SES predictors are modeled simultaneously, suggesting that the formerly significant impact of occupational prestige is accounted for by these factors.13 In addition, the decade effects are again attenuated in Model 6, suggesting that education trumps occupational prestige in accounting for over-time changes in SCI.

Question 3b: Have there been any significant changes over time in the determi- nants of Blacks’ SCI?

Hunt and Ray 1473

Table 4 focuses on over-time changes in determinants of Blacks’ SCI by presenting results for our full model (minus the decade terms, of course) stratified by decade. Looking within columns (decades), we see that, in the 1970s, only education is sig- nificantly associated with SCI; in the 1980s, only income is; in the 1990s, education and income register significant effects, whereas in the most recent decade, education, income, and self-employment do. Thus, occupational prestige does not significantly impact SCI in any of the decades examined, whereas self-employment does so only during the most recent decade. Education and income appear most consistent in reg- istering significant effects on SCI, as each of these factors does so for three of the four decades we examine (i.e., the 1970s, 1990s, and 2000s for education; and the 1980s, 1990s, and 2000s for income).

That said, there is more to determining the significance of over-time trends in the relationship between SES and SCI than a simple noting of whether selected coeffi- cients are statistically significant or not within given decades. As such, we took the additional step of assessing the statistical significance of any over-time changes in the impact of each SES predictor by running a series of fully saturated “decade * covari- ate” interaction models (not shown, but noted in Table 4). These results demonstrate that only income and self-employment show statistically significant differences over time in their relationships with SCI. Specifically, the effect of household income was significantly stronger in the 1980s (b = .022, p < .001) than the 1970s (b = .003, n.s.). And self-employment shows significant differences in its effect on SCI between the 1970s and 1990s, the 1970s and 2000s, and the 1980s and 2000s, in moving from an inverse association during the 1970s (b = –.779, n.s.) to a positive one in the 2000s (b = .695, p < .01). Thus, across four decades, self-employment appears to have become consistently more central to how Blacks see themselves in social class terms, perhaps owing to the growth in entrepreneurship, business ownership, and the atten- tion these phenomena have received in recent decades.14

Discussion and Conclusions Using data on African Americans from the 1974 to 2010 GSS, we document that Blacks’ SES characteristics, predictably, have increased over-time, as has Blacks’ tendency to identify as “middle class.” Multivariate models show that the over-time increase in SCI holds net of the effects of a wide range of sociodemographic factors, but is accounted for, once key SES variables are included. Regarding the impact of key indicators of SES, education, occupational prestige, and household income all have independent and significant positive effects on SCI net of decade and sociode- mographic controls, whereas self-employment does not. Importantly, however, when entered simultaneously, only education and income remain significant, and account for the formerly significant impact of occupational prestige (and supplemental analy- ses reported in Note 13 suggest that education is the more consequential factor here).

Our finding that SES characteristics significantly shape Blacks’ SCI echoes Hout’s (2008) recent analysis of 2000 to 2004 GSS data, but extends this generalization to a

1474 American Behavioral Scientist 56(11)

considerably wider time frame and adds important descriptive details regarding the specifics of these effects. In so doing, our results shed further light on current debates over definitions of the “Black middle class” by demonstrating that income and educa- tion (perhaps especially the latter) appear to be most central to how Blacks see them- selves in social class terms. Finally, multivariate analyses run by decade (as well as with “decade * covariate” interactions) reveal few noteworthy patterns regarding any over-time changes in the association between SES characteristics and SCI. The one exception is for self-employment, which has clearly become a more potent predictor of SCI in recent decades. These findings provide a potentially important qualification to recent findings on SCI among the wider U.S. population, suggesting that income has become more central in determining how Americans think about themselves in terms of social class (Hout, 2008).

Taken together, our results raise some important questions about how the relation- ship between Blacks’ objective and subjective statuses may have changed over the past four decades. On the one hand, we find significant associations between SES and SCI where earlier work did not (Jackman & Jackman, 1983; Vanneman & Cannon, 1987).15 However, our analyses also suggest that—with the exception of increases in the potency of self-employment as a predictor—relatively little has changed regarding the SES/SCI relationship among Blacks since the 1970s. Our results thus raise ques- tions about the extent to which a sense of “linked collective fate” (Dawson, 1994) or “group identification” (Hunt, 1996) continues to structure how Black Americans of varying socioeconomic positions come to understand their social class locations. Put simply, does race still trump class, as Jackman and Jackman (1983) asserted? And if not, how should we characterize the relative potency of these two dimensions of iden- tity among contemporary African Americans?

Although a definitive answer to this “race versus class” question lies beyond the scope of the current investigation (as such an analysis requires data from more than one race group), future work on race and SCI should seek to address this, and other issues, by adopting a framework that is more explicitly comparative by race/ethnicity. Whereas one recent study involving limited White/Black comparisons has taken some initial steps in this direction (Hout, 2008), future research should increasingly seek to include other major U.S. ethno-racial groups such as Hispanics and Asian Americans given the rapid growth of these populations in an increasingly multiethnic United States. Doing so would add SCI to a growing list of social psychological and attitude- related topics examined within a multiethnic framework (Bobo & Hutchings, 1996; Bobo & Zubrinsky, 1996; Hunt, 2007; Oliver & Wong, 2003).

Future work on race/ethnicity and SCI should also explore a wider variety of pre- dictors. Though recent studies (e.g., Hout, 2008, and the current investigation) take important steps in this direction given their simultaneous estimation of multiple SES determinants alongside the ability to control for a wide range of other important poten- tial confounders, research should seek to move beyond the heavy reliance on sociode- mographic predictors to include more measures of aspects of respondents’ interpersonal and subjective worlds. Specifically, following the lead of “social structure and

Hunt and Ray 1475

personality” research (House, 1981; McLeod & Lively, 2004), future work should seek to explain why and how persons’ locations in larger social structures such as class and race come to impact individuals’ self-concepts by specifying relevant and more “proximate” patterns of relationships through which larger structures exert their effects. In the context of research on SCI, this means better-specifying the “meso- level” channels (e.g., class and racial composition of neighborhoods, social networks, workplaces, churches) and key social psychological mechanisms such as social com- parison processes (Rosenberg, 1981) that underlie how persons locations in social structure ultimately impinge on the person.

Finally, future research should move beyond the analysis of race/ethnic group dif- ferences in SCI generally, to explore in detail how the determinants of SCI may differ across major race/ethnic lines in the United States. Following a growing body of stud- ies documenting race/ethnic differences in the determinants of important beliefs and attitudes (Hughes & Tuch, 1999; Hunt, 2007; Hunt, Powell, Steelman, & Jackson, 2000; Schnittker, Freese, & Powell, 2000), sociologists are increasingly challenging the assumption that the social psychological processes underlying attitude formation are the same across major social structural lines such as those represented by race/ ethnic inequality in the United States. Given the historical oppression of African Americans, the distinct immigration-related experiences of many Hispanic and Asian Americans, and the often-unrecognized nature of “Whiteness” and its associated privi- leges by many Americans of European descent, it seems reasonable to posit that the mechanisms by which identities around “social class” are shaped may differ meaning- fully for the various ethno-racial groups constituting the United States.

Acknowledgments

We thank Brian Powell, Larry Hunt, and Kris Marsh for helpful comments on a prior draft of this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Notes

1. As Hout (2008) notes, over 99% of GSS respondents provide an answer to this question, suggesting that these terms carry meaning for the vast majority of adult Americans.

2. Centers (1949) also found associations between SCI and both political party affiliation and political behavior, suggesting that SCI represents a reasonable proxy for “class consciousness.”

1476 American Behavioral Scientist 56(11)

3. It is important to note that, whereas Vanneman and Cannon (1987) concur with Jackman and Jackman (1983) regarding the relationship between “socioeconomic attainments” (as indexed by incremental differences in income, education, occupational prestige) and SCI among Blacks, they differ on their overall characterization of Blacks’ class identities. Specifically, rather than seeing Blacks’ class identities as overwhelmed by racial con- siderations, Vanneman and Cannon conclude that “class identity is quite strong among Blacks,” (p. 233) but only when class is conceptualized along the categorical “mental- manual” division emphasized in their research.

4. Du Bois’s (1903) “Talented Tenth” notion rested on the premise that college-educated Blacks were positioned to be the leaders of the Black race and create social change. This argument later found support in the form of college-educated Blacks playing central roles at the forefront of civil rights leadership and activism in the 1950s and 60s (Battle & Wright, 2002; Morris, 1994; Parks, 2008).

5. Hout’s (2008) occupation scheme takes into account self-employment, a factor we analyze separately as explained below.

6. Hout’s (2008) other primary foci are exploring the meaning of class (money vs. lifestyle vs. work) and selected consequences of class position (e.g., lifestyles, attitudes, well-being).

7. Analyses excluding “upper” class from the dependent variable yielded similar results to those presented herein.

8. The home ownership question was not asked until the 1985 GSS; consequently, our analy- sis of this variable is restricted to 1985 to 2010. Our analyses reveal that an initial positive and statistically significant effect of home ownership is moved to nonsignificance when we introduce controls for SES and other sociodemographics.

9. Analyses using continuous measures of parent’s education, analogous to our measure of respondent’s education, produced similar results.

10. As noted above, we use an inflation-adjusted measure of household income adjusted to 1986 dollars. Therefore, making a meaningful interpretation in terms of today’s dollars is difficult.

11. Though beyond the scope of this research, the recent increase in lower-class identification could stem from the so-called “Great Recession” of the late 2000s (and beyond), which is known to have hit African Americans especially hard. That same recessionary period could also account for the slight decline in middle class identification from the 1990s to the 2000s. Future research should examine these possibilities.

12. Use of Hout’s (2008) categorical approach to measuring income ($0–less than $20,000; $20,000–$29,999; $30,000–$49,999; $50,000–$74,999; $75,000–$99,999; $100,000 and more) produced results similar to those presented here.

13. And other analyses (not shown) reveal that education is more consequential than income in accounting for the effect of occupational prestige. Specifically, in a model including only decade, occupational prestige, and household income as predictors, occupational prestige retains its significant association with SCI; however, in a model including only decade, occupational prestige, and education as predictors, occupational prestige has a nonsignifi- cant association with SCI. These same basic patterns are replicated in two variations of the

Hunt and Ray 1477

“full model” from Table 3 (where we first excluded only the education term, and then only the household income term).

14. Furthermore, whereas the self-employed still represent a fairly small segment of the Black population (6% overall, and 9% of the self-identified middle class), this group may be central to the upward mobility of Blacks more generally, given the concentration of Black- owned businesses in predominately Black neighborhoods, and because of the key role such businesses play in job creation and economic growth in such areas—a point not lost on some of the earliest observers of the Black middle class (Du Bois, 1899/1996; Frazier, 1957/1997).

15. However, absent direct empirical comparison with Whites, we are reluctant to characterize the relative strength of these associations as have past researchers.

References

Bates, T., Jackson, W. E., & Johnson, J. J. (2007). Introduction: Advancing research on minor- ity entrepreneurship. Annals of the American Academy of Political and Social Science, 613, 10–17.

Battle, J., & Wright, E. (2002). W.E.B. DuBois’s talented tenth: A quantitative assessment. Journal of Black Studies, 32, 654–672.

Bobo, L., & Hutchings, V. L. (1996). Perceptions of racial group competition: Extending Blumer’s theory of group position to a multiracial social context. American Sociological Review, 61, 951–972.

Bobo, L., & Zubrinsky, C. L. (1996). Attitudes on residential integration: Perceived status dif- ferences, mere in-group preference, or racial prejudice? Social Forces, 74, 883–909.

Centers, R. (1949). The psychology of social classes: A study of class consciousness. London: Taylor & Francis.

Davis, N. J., & Robinson, R. V. (1988). Class identification of men and women in the 1970s and 1980s. American Sociological Review, 53, 103–112.

Davis, N. J., & Robinson, R. V. (1998). Do wives matter? Class identities of wives and hus- bands in the United States, 1974-1994. Social Forces, 76, 1063–1086.

Dawson, M. C. (1994). Behind the mule: Race and class in African-American politics. Princeton, NJ: Princeton University Press.

Du Bois, W. E. B. (1899/1996). The Philadelphia Negro. Philadelphia, PA: University of Pennsylvania Press.

Du Bois, W. E. B. (1903). Talented tenth. In The Negro problem: A series of articles by repre- sentative Negroes of to-day. New York, NY: James Pott and Company.

Drake, St. C., & Cayton, H. R. (1945). Black metropolis: A study of Negro life in a northern city. New York, NY: Harper & Rowman.

Durant, T. J., & Sparrow, K. H. (1997). Race and class consciousness among lower and middle class Blacks. Journal of Black Studies, 27, 334–351.

Frazier, E. F. (1957/1997). Black bourgeoisie: The rise of a new middle class. New York, NY: Free Press.

Gatewood, W. B. (2000). Aristocrats of color: The Black elite, 1880-1920. Fayetteville: Uni- versity of Arkansas Press.

1478 American Behavioral Scientist 56(11)

Grusky, D. (2000). Social stratification: Class, race, and gender in sociological perspective. Boulder, CO: Westview.

Hochschild, J. (1995). Facing up to the American dream: Race, class, and the soul of the nation. Princeton, NJ: Princeton University Press.

House, J. S. (1981). Social structure and personality. In M. Rosenberg & R. H. Turner (Eds.), Social psychology: Sociological perspectives (pp. 525–561). New York, NY: Basic Books.

Hout, M. (1984). Occupational mobility of Black men: 1962-1973. American Sociological Review, 49, 308–322.

Hout, M. (2004). Getting the most out of the GSS income measure. GSS Methodological Report No. 101. Chicago, IL: NORC.

Hout, M. (2008). How class works: Objective and subjective aspects of class since the 1970s. In A. Lareau & D. Conley (Eds.), Social class: How does it work? (pp. 25–64). New York, NY: Russell Sage Foundation.

Hughes, M., & Tuch, S. A. (1999). How beliefs about poverty influence racial policy attitudes: A study of Whites, African Americans, Hispanics, and Asians in the United States. In Sears, D. O Sidanius, J & Bobo, L. (pp. 165–190). Chicago, IL: University of Chicago Press.

Hunt, M. O. (1996). The individual, society, or both? A comparison of Black, Latino, and White beliefs about the causes of poverty. Social Forces, 75, 293–332.

Hunt, M. O. (2007). African American, Hispanic, and White beliefs about Black/White inequality, 1977–2004. American Sociological Review, 72, 390–415.

Hunt, M. O., Powell, B., Steelman, L. C., & Jackson, P. B. (2000). Color-blind: The treatment of race and ethnicity in social psychology. Social Psychology Quarterly, 63, 352–364.

Hwang, S. S., Fitzpatrick, K. M., & Helms, D. (1998). Class differences in racial attitudes: A divided Black America? Sociological Perspectives, 41, 367–380.

Jackman, M. R. (1979). The subjective meaning of social class identification in the United States. Public Opinion Quarterly, 43, 443–462.

Jackman, M. R., & Jackman, R. W. (1973). An interpretation of the relation between objective and subjective social status. American Sociological Review, 38, 569–582.

Jackman, M. R., & Jackman, R. W. (1983). Class awareness in the United States. Berkeley: University of California Press.

Kluegel, J., Singleton, R., & Starnes, C. E. (1977). Subjective class identification: A multiple indicator approach. American Sociological Review, 42, 599–612.

Kronus, S. (1971). The Black middle class. Columbus, OH: Merrill. Lacy, K. R. (2007). Blue-chip Black: Race, class, and status in the new Black middle class.

Berkeley: University of California Press. Landry, B. (1987). The new Black middle class. Berkeley: University of California Press. Landry, B., & Marsh, K. (2011). The evolution of the new Black middle class. American Review

of Sociology, 37, 373–394. Ligon, E. (1989). The development and use of a consistent income measure for the general

social survey. GSS Methodological Report No. 64. Chicago, IL: NORC. McLeod, J. D., & Lively, K. J. (2003). Social structure and personality. In J. Delamater (Ed.),

Handbook of social psychology (pp. 77–102). New York, NY: Kluwer Academic/Plenum.

Hunt and Ray 1479

Marsh, K., Darity, W. A., Cohen, P. N., Casper, L. M., & Salters, D. (2007). The emerging Black middle class: Single and living alone. Social Forces, 86, 735–762.

Morris, A. D. (1994). The origins of the civil rights movement: Black communities organizing for change. New York, NY: Free Press.

Oliver, J. E., & Wong, J. (2003). Intergroup prejudice in multiethnic settings. American Journal of Political Science, 47, 567–582.

Parks, G. S. (2008). Black Greek letter organizations in the 21st century: Our fight has just begun. Lexington: University Press of Kentucky.

Pattillo, M. (2005). Black middle-class neighborhoods. Annual Review of Sociology, 31, 305–329. Rosenberg, M. (1981). The self-concept: Social product and social force. In M. Rosenberg.,

& R. H. Turner (Eds.), Social psychology: Sociological perspectives (pp. 593–624). New York, NY: Basic Books.

Schnittker, J., Freese, J., & Powell, B. (2000). Nature, nurture, neither, nor: Black-White differ- ences in beliefs about the causes and appropriate treatment of mental Illness. Social Forces, 78, 1101–1132.

Vanneman, R., & Cannon, L. W. (1987). The American perception of class. Philadelphia, PA: Temple University Press.

Willie, C. V. (1979). The caste and class controversy. Bayside, NY: General Hall. Willie, C. V., & Reddick, R. J. (2003). A new look at Black families (5th ed.). Walnut Creek,

CA: Altamira. Wilson, W. J. (1978). The declining significance of race. Chicago, IL: University of Chicago

Press. Wright, E. O. (1996). The enduring importance of class analysis. Theory and Society, 25,

693–716.

Bios

Matthew O. Hunt is an Associate Professor of Sociology at Northeastern University and a Non-Resident Fellow at the W.E.B. Du Bois Institute for African and African American Research at Harvard University. His primary research interests involve intersections of race/ ethnicity, social psychology, and inequality in the United States. His work has appeared in the American Sociological Review, Social Forces, Social Psychology Quarterly, Social Science Quarterly, Du Bois Review and other publications. He also recently co-edited a Volume (March 2011) of the Annals of the American Academy of Political and Social Science titled “Race, Racial Attitudes, and Stratification Beliefs: Evolving Directions for Research and Policy.”

Rashawn Ray is an Assistant Professor of Sociology at the University of Maryland, College Park. He received a Ph.D. in Sociology from Indiana University in 2010. Ray’s research inter- ests are social psychology, race and ethnic relations, and race-class-gender. His work addresses three key areas: the determinants and consequences of social class identification, men’s treat- ment of women, and how racial stratification structures social life.

Ray is the editor of Race and Ethnic Relations in the 21st Century: History, Theory, Institutions, and Policy. His work has appeared in Ethnic and Racial Studies, Journal of Contemporary

1480 American Behavioral Scientist 56(11)

Ethnography, Journal of Higher Education, and Journal of African American Studies. Ray has been awarded funding from the National Science Foundation, National Institute of Mental Health, American Sociological Association Minority Fellowship Program, Society for the Study of Social Problems, and the Ford Foundation. From 2010-2012 he was a Robert Wood Johnson Health Policy Research Scholar at the University of California, Berkeley/UCSF.