Beliefs about the Causes of Racial Inequality: The Persisting Impact of Urban and Suburban Locations?

The purpose of this paper is to assess the classical theoretical propositions of Wirth and Stouffer regarding the independent impact of urban and suburban residency on beliefs about racial inequality. This paper further assesses the impacts of these social locations over a three-decade period. While scholars pose that city size positively impacts behavior, there is debate regarding the impact of these locations over time. Using the General Social Survey, findings demonstrate that while urbanism continues to have an independent impact on beliefs, the impact of suburban residency is much weaker and inconsistent. Analyses of over time trends reveal that the gap between urban and rural residents appears to have increased, with rural residents expressing more intolerant beliefs about the causes of racial inequality over time. These findings are further discussed in a broader theoretical context.


Introduction
In their now classic works on city size and subcultures, Louis Wirth [1] and Samuel Stouffer [2] proposed that urbanization creates an environment that encourages more tolerant attitudes among inhabitants.Testing this proposition, subsequent research has shown that individuals in larger communities tend to exhibit more tolerant attitudes toward a host of "different" groups, lifestyles, and characteristics [3][4][5][6][7][8][9][10][11].
In terms of beliefs deemed "unpopular," "different," and/or "deviant," Wilson [12], for example, found greater levels of tolerance among urban residents than among rural residence.Specifically, Wilson [12] found a greater willingness to extend civil liberties to groups like atheists, communists, homosexuals, racists, and militarists among urbanites than among rural residents.Greater tolerance toward homosexuality, extramarital sex, premarital sex, and pornography is also found among urban than nonurban dwellers [9,12,13].Abrahamson and Carter [3] more generally show that nonurban residents' disapproval of issues sometimes used to signal tolerance including civil liberties, euthanasia, and prohibition.
Of immediate concern to the current paper, past research has looked at the impact of urban on racial attitudes in particular [4,10,14].Tuch [10], for example, found more liberal attitudes toward interracial marriage and voting for an African American for US president among urban residents than among rural residents.
More recent research, however, suggests a more complex relationship between urbanism and racial attitudes.In terms of survey items measuring "principles of equality" [8], urbanites exhibit more liberal attitudes than rural residents."Implementation" measures, however, have been shown to actually diminish the impact of urbanism on racial attitudes.Principle survey questions emphasize abstract principles of racial equality, whereas implementation questions tap into peoples' attitudes toward policies that would actually put those professed principles into practice [14].Accordingly, Carter and colleagues [14] suggest that the influence of urbanism as such may not be indicative of deep-seated "tolerant" racial attitudes.Rather, they suggest that the impact of urbanism as such may simply reflect a more shallow measure.
Although past research has shown an independent effect of urbanism on tolerance, relatively few have examined the place of suburban residency in this context.Yet, suburban communities are distinct from urban central cities and rural areas in terms of size, demographic composition, and social structure [15].More specifically, suburban communities differ from rural and metropolitan central cities in terms of family structure, racial composition, and socioeconomic status [16,17].As Lamb [18] notes, there is also substantial variation between urban central cities and suburban communities in terms of wealth, business and commercial development, employment opportunities, educational quality, health care, recreational facilities, among a host of other important variables.Lamb [18], however, notes that perhaps the most visible difference between urban/central cities and suburban areas is differences in racial composition: Black cities and White suburbs.African Americans, other racial and ethnic minorities, and the less affluent generally live in urban cities and/or suburbs surrounding those cities, while Whites tend to live in suburban areas further away from central cities.Yet, scholars have also observed that suburban communities are increasingly diverse [19].In general, central cities and their surrounding suburbs tend to have high population density but a weak tax base and residential instability [20][21][22].
According to Baldassare [15], suburban locations are "understudied in sociology, relative to their size and importance.They are also not well understood, subject to urban biases and cultural myths about suburban living" (Page 475).Moreover, suburbia is now a ubiquitous phenomenon in contemporary America.Since the 1970s, suburbs of metropolitan areas surpassed central cities in population size [15].Indeed, although population growth has decreased slightly from the late 2000s for suburban locations, majority of the population, considered to be located in metro areas, continues to be located in suburban areas [23].
Theoretically, Wirth [1] was much more vague about the impact of suburbia on residents.He was very specific, however, in noting that the impact of social location is directly associated with city size, heterogeneity, and density.Individuals from larger cities that are heterogeneous in terms of cultural and social backgrounds and tend to be densely populated will express greater tolerance than individuals from smaller, homogeneous, and loosely packed locations.By extension, we would expect suburbia to be the point of transition from tolerant urban locations to intolerant nonurban locations.Thus, we propose that measures of tolerance would range from highest among urban residents, intermediate among suburban residents, and lowest among rural residents.
This paper also attempts to assess the change in the impact of these locations over time.Wirth's later ideas [24] specifically suggested that, over time, the impact of urbanism would eventually spread-effectively eliminating the localized and unique impact of rural locales-to more rural areas.Wirth [24] noted that the demise of the urban/rural difference would be a result of improvements in technology, transportation, and communication.We propose a convergence in racial attitudes would not only affect the urban/rural divide but the suburban/rural divide as well, as suggested by Wirth.
Accordingly, the goal of this paper is two-fold.First, along with reassessing the impact of urbanism on beliefs about racial inequality, we also assess the impact of suburban residency on the same beliefs.Second, we assess how attitudes among urbanites, suburbanites, and rural residents have shifted over time.In accomplishing the tasks outlined in the preceding paragraph, the issues of the paper will be taken in the following order.We begin with a background review of the literature on urbanization and levels of tolerance.We then describe our study and report its results.We conclude with theoretical and practical implications of the findings and suggestions for future research.

Background
Wirth [1] was one of the first sociologists to develop a formal theoretical account of the unique impact of urbanism.According to Wirth [1], urbanites are unique in character in that metropolitan areas bring individuals of different social and cultural backgrounds into contact with one another within a relatively confined space.Metropolitan areas thus promote the formation of secondary relationships of an impersonal nature, as oppose to primary relationships more likely to be prominent in rural areas.Secondary relationships, in turn, promote tolerant attitudes of others with differing social and cultural backgrounds [1].
Wirth [1] proposed specifically that the impact of urbanism is a function of three factors: size of the population, density of the population, and heterogeneity of the population.In general, residents of densely populated large areas with heterogeneous groups tend to be more accepting, psychologically sophisticated, and cosmopolitan than individuals living in areas that are small, more dispersed, and homogeneous.
Stouffer [2] similarly proposed that urban areas produce unique qualities in inhabitants that are distinct from characteristics produced by rural residency.According to Stouffer [2], urbanites are exposed to the "strange and the different," which forces them to develop toleration of social and cultural differences.Specifically testing these propositions, Stouffer [2] found urbanites more likely to express tolerant attitudes toward variant ideas, such as communism, than rural residents.More recent research testing the basic propositions of Wirth [1] and Stouffer [2] have found substantial support for the effects of urbanism as such on tolerance.Urban residents are shown to express more tolerant attitudes toward a variety of lifestyles and characteristics viewed "different" [3,4,6,7,[9][10][11]25].
The impact of urbanism on racial attitudes, however, has revealed varying and often contradictory results.Early studies documented a significant liberalizing effect of urban residency [10,26].Using more contemporary data, however, more recent studies have brought into question this liberalizing effect [4,14,27].While finding urban residents expressing more tolerant attitudes when principles of equality measures are employed, Carter and associates [14], for example, also found this effect to diminish when attitudes toward political policy are assessed.
Moreover, while previous research has shown notable difference in attitudes between urban and rural residents, predictions of future attitudinal trends have been inconclusive.Some theorists predict an ephemeral effect of urbanism, where the impact of living in the city diminishes over time.Wirth [1] noted that technological advances in entertainment, transportation, and communication will ultimately eliminate any differences in attitudes between urban and non-urban residents.More recently, Glenn and Hill [26] argued that the urban effect would diminish because of the diffusion of information through the mass media across urban boundaries.In other words, differences in attitudes will converge over time in that non-urban residents will eventually become as tolerant toward differences as their urban counterparts.Others, however, posed that, because of cultural lag, the unique impact of urbanism on residents would never diminish [6].
Nonetheless, only a limited amount of research assessing the impact of urbanism over time exists.Using questions mainly focusing on principles of racial equality, past research has shown a continued impact of urbanism over time [14,28,29].Carter [4] found the impact of urbanism and its effect across time to be much more variable and dependent on how racial tolerance is measured-principle or implementation.Moreover, while previous research has shown an independent effect of urbanism on tolerance, relatively few have examined the place of suburban residency in this context.
To that end, we do not only reassess the impact of urbanism on beliefs about racial inequality, but we also assess the impact of suburbia on racial beliefs as well.Furthermore, we take on the task of assessing the impact of social location on beliefs about racial inequality over time.Scholarship in this area posits that explanations of inequality fall along two lines: structural (tolerant) and individual (intolerant).The structural explanation is a more racially tolerant explanation of inequality and posits that inequality is a consequence of external factors, such as discrimination [30,31].By contrast, the individual explanation is a more intolerant explanation and suggests that inequality results from individual characteristics, such as a lack of work ethic, rather than racial discrimination [31,32].Specifically, we examine whether differences in believes about the causes of racial inequalityexternal structural limitations or individual-exist between urban, suburban, and rural residents.Analyzing earlier data, Carter [27] found the impact of urbanism greatly diminished once different indicators, such as explanation questions of attitudes, are used to assess prejudice.By using more up-todate data and different indicators to assess racial toleration, this research further investigates the distinct impact of place of residency (urban, suburban, and rural).Finally, in light of Wirth's [24] proposition regarding the transient nature of urbanism due to improvements in technology, this paper also assesses the distinct effects of urban, suburban, and rural residency over time.

Data and Measures
3.1.Data.To assess the impact of urban and suburban residency on beliefs about inequality, this analysis uses crosssectional data collected by the National Opinion Research Center's General Social Survey (GSS).Respondents included in this random sample are noninstitutionalized, adults over the age of 17 and are stratified by region and metropolitan residence.The GSS was annually administered between 1972 and 1994 and biannually thereafter [33].Since this project focuses on White racial beliefs, the analysis only includes White respondents.All models were ran using the SPSS data analysis software.
Initially, all years from GSS that asked about beliefs about racial inequality were pooled together, and OLS regression models were calculated.Findings showed the independent impacts of urban and suburban residency (rural was the comparison category) via unstandardized parameter estimates on beliefs about the causes of racial inequality.Along with assessing the general relationship between city size and racial beliefs, the paper also sought to assess changes in these beliefs over an extended time period.To accomplish these goals, this project needed a data set that tracked attitudes and beliefs toward Blacks as well as a data set that asked these questions over several decades.The GSS met each of these criteria.
Analytically, to assess change in beliefs over an extended time period, the data was pooled into three time periods: 1980s (including 1977), 1990s, and 2000s.For example, the years 2000,2002,2004,2006,2008, and 2010, were collapsed together and labeled the 2000s decade.The decades of 1980s and 1990s were completed in the same fashion.Change in beliefs over time was then assessed by running separate OLS regression models for each decade.To assess whether a significant change in the unstandardized parameter estimates occurred from one decade to another, t-tests (see Results for further discussion) were calculated [34].It must be noted that this type of method does have some obvious limitations.For instance, this method does not allow for a within-decade assessment of change.However, it does have strengths as well.This method assesses nonlinear shifts in beliefs over time [35].

Dependent
Variable.This project attempts to assess the impact of city size on beliefs about racial inequality and how those beliefs have changed over time.The GSS includes several questions asking respondents their beliefs about the root causes of racial inequality in the United States.These questions were included in the GSS from 1977 to 2010.Schuman et al. [8] posited that these questions tend to get at different dimensions of racial attitudes and range from more individualistic (intolerant) to structural (tolerant) explanations.Respondents were asked how strongly they feel the unequal distribution of jobs, income and housing is caused by discrimination (Racdiff1), in-born ability to learn (Racdiff2), lack of opportunity for a good education (Racdiff3), and/or lack of motivation (Racdiff4).Scores ranged from 4 to 8, with higher scores representing structural (tolerant) explanations and lower scores representing individualistic (intolerant) explanations.

Independent Variables.
In order to assess the independent impact of urbanism, Gans [36] argued that certain factors associated with urban/rural differences need to be included in any model, including race, social status, and life-cycle stage.Gans [36] posed that if these variables were sufficiently accounted for, the impact of urban residency would be negligible.
To assess the impact of urban and suburban residency, this paper includes several control variables, including marital status, sex, age, education, income level, and level of religious fundamentalism.Marital status was treated as a dummy variable, comparing married (1) and nonmarried (0) respondents.The sex variable was also a dummy variable comparing male (1) and female (0) respondents.The income variable was treated as an ordinal variable with 12 categories (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12), ranging from low income to higher incomes.Assuming a linear effect, the education and age variables were treated as continuous variables.To assess level of religiosity, two variables (level of fundamentalism and church attendance) were included in the models.The religious fundamentalism and religious moderate variables were treated as dummy variables, with the liberal respondents being the reference group for both.The church attendance variable was treated as an ordinal variable ranging from 0 (never attending) to 8 (more than once a week).This research also included a control for region of residency, with the region variable being treated as a dummy variable comparing South (1) to nonsouth (0) locations.
Finally, this paper includes two variables to assess size of residency to directly test our hypotheses.The size of residency variable differentiates between SMSA and non-SMSA residence in the GSS generally.Specifically, this paper included two dummy variables comparing urban and suburban residents to rural residents.While scholars in this area argue that partitioning of residency size into two categories tends to be the more conservative approach [10,12], we attempt to more thoroughly assess the impact of city size by looking at urban and suburban residents.Individuals who live in an SMSA with fifty thousand or more residents, were coded as urban, while individuals who live in suburban or unincorporated areas associated with an SMSA were coded as suburban residents.Individuals who live in a non-SMSA area were coded as rural residents and were treated as the reference group in the regression models.

General Trends. Table 1 provides unadjusted Beliefs
Index mean scores for urban, suburban, and rural respondents overall and for three decades (1980s; 1990s; 2000s).Again, higher Beliefs Index mean scores reflect more tolerant or structural explanations of racial inequality, while lower Beliefs Index means scores reflect less tolerant or individualistic explanations of racial inequality.Beliefs results from Table 1 reveal a few notable findings.First, for the overall scores, and ANOVA test reveals a hierarchy among the groups.Urban residents maintain the highest mean followed by suburban and rural residents, who maintain the lowest mean scores, and these mean score differences did reach significance at the .001probability level.This finding supports the idea that suburban locations produce notates a significant difference at the .001probability level that fall in a hierarchy when cross comparing the locations.Accordingly, with a higher mean score, urban residents maintain more tolerant or structural beliefs about the causes of racial inequality and, with lower scores, rural residents maintain more intolerant or individualistic beliefs about the cause of racial inequality and, with a mean score in the middle, uburban residents fall in between their beliefs.That is to say, suburban residents maintain beliefs about racial inequality that is significantly less tolerant than urban residents, yet more tolerant than their rural counterparts.This holds for the pooled data as well as when the data is split by decade.# The Explanation Index includes the following four questions in reference to the following statement: "on the average (Negroes/Blacks/African-Americans) have worse, jobs, income, and housing than White people." The questions include (1) do you think these differences are mainly due to discrimination (0: No; 1: Yes); (2) because most (Negroes/Blacks/African-Americans) have less in-born ability to learn?(0: Yes; 1: No); (3) because most (Negroes/Blacks/African-Americans) don't have the chance for education that it takes to rise out of poverty?(0: No; 1: Yes); and (4) because most (Negroes/Blacks/African-Americans) just don't have the motivation or will power to pull themselves up out of poverty?(0: Yes; 1: No).
unique individuals that maintain their own perspective when discussing beliefs about racial inequality.Turning to the by decade analysis, findings reveal a similar hierarchical pattern over the three decades (see 1970s for exception where no difference is found between the three groups).In terms of change over time, the Δ terms show notable shifts over time within each group which may actually increase the belief gap (shown in the overall means) that separates urban/suburban and rural residents.First, urban and suburban residents show increasing means across the decades, which implies that beliefs are becoming more tolerant and structural.However, for rural residents, the mean scores are actually declining over time, which implies that attitudes are becoming more intolerant or individualistic.Second, looking more specifically at the change over time, the shift in the means score among the groups appears to take on the same hierarchical form again, with urban residents (Δ = +.22)showing a greater positive shift of the time period followed by suburban (Δ = +.11) and rural residents (Δ = −.11), who actually appear to be becoming less tolerant over time. 2 provides unstandardized parameter estimates and standard errors for two models assessing the impact of urban/suburban on the Beliefs Index, while controlling for pertinent extraneous variables.Model 1 includes the primary independent variables: urban and suburban while, Model 2 includes the other control variables noted in the literature as possibly affecting racial beliefs.

Multivariate Analysis. Table
Looking at Model 1, urban and suburban residents were significantly more likely to maintain more tolerant or structural explanations for racial inequality than rural residents.These relationships were significant at the .001probability level.That is, urban and suburban residents were more likely than rural residents to see discrimination and lack of education as important factors impacting continued racial inequality.When including life cycle and other control variables as proposed by Gans [36] in Model 2, the absolute impact of both urban and suburban residency diminishes greatly.However, the impact of urban residency on beliefs about racial inequality remains significant at the .05probability level, whereas the impact of suburban residency did not.Consequently, and in line with the arguments of Gans [36], the control variables did mediate the impact of urbanism to an extent.The impact of the other control variables was quite consistent with past research [10,27], with non-Southerners, singles, women, younger, more educated, less income, and more religiously liberal more likely to express tolerant or structural explanations of racial inequality.On a final and interesting note, the year variable, which would show any linear shifts over time in beliefs, was found to be insignificant, which suggest that attitudes are not shifting in a positive or negative manner over time.This finding is consistent with the assertion that questions that tap into deeper level racial attitudes or beliefs tend to be quite stable and stagnant [8]..147.148.138* P < .05;* * P < .01;* * * P < .001.# Note that the standard errors in parentheses.Also note that the parameter estimates are calculated net of the same control variables included in Table 2. 3 provides unstandardized parameter estimates and fit statistics for three models assessing the impact of urban/suburban residency on beliefs about racial inequality across decades.It should be noted that all models in Table 3 include the control variables (available upon request) included in Model 2 in Table 2.The purpose of calculating regression models by decade is to assess whether beliefs of urban/suburban and rural residents are "converging" or "diverging" or "staying the same" across an extended time period.The determination of one of these trends is done by looking at the absolute shifts (or lack thereof) in the regression coefficients over the three decades.

Across Time Trends. Table
We further use t-tests for differences (formula for the t-test: ((b1 − b2)/ √ se1 2 + se2 2 )) to compare the regression coefficients across each decade, allowing for a more definitive assessment of change (increase, decrease, or none) over time.More specifically, the t-test compares all possible pairwise comparisons (e.g., 1980s versus 1990s; 1980s versus 2000s; and 1990s versus 2000s) for statistically significant changes in the urban and suburban variables over time.Knoke et al. [34] note that this type of test (t-test) is commonly used to assess whether regression coefficients differ statistically between groups in a given population.
Several interesting trends over time are observed.Taking a cursory look at the coefficients, it is interesting to see that the effects of urban and suburban residency follow similar trends.First, in the 1980s and in the 1990s, the difference between urban/suburban and rural residents is minimal at best.However, by the 2000s, the effects of urban and suburban residency are significant with both indicating more tolerant or structural beliefs about the causes of racial inequality.Second, the trend is for the gap that separates urban/suburban residents from rural residents to be increasing over time.Whereas in the 1980s and 1990s the coefficients are insignificant, there is a clear pattern of moving from negative to positive, ultimately resulting in a significant result in the 2000s.
Using the more definitive t-test to assess statistical significant shifts over time in the regression coefficients reveals similar but more nuanced findings.From the 1980s to the 1990s, shifts in the regression coefficients for urban (t = −1.35,P > .05)and suburban residents (t = −1.26,P > .05)are insignificant.That is to say, the gap that separates urban/suburban and rural residents is quite small and has not shifted significantly over time.Looking at urban residents, the shift in the coefficient from the 1980s to 2000s (t = 2.26, P < .05)and from the 1990s to 2000s (t = −2.84,P < .05)reveals a significant shift.These shifts reflect a growing gap between urban and rural residents over time with, as noted above, urban residents maintaining more tolerant or structural beliefs about racial inequality.Turning to suburban residents, the trend is quite similar with the shift from the 1980s to 2000s (t = −3.16,P < .05)and from the 1990s to 2000s (t = −2.62,P < .05)showing a demonstrable positive shift, resulting in more tolerant or structural beliefs about racial inequality by the 2000s.
A final comment on the observed hierarchical pattern of beliefs among the groups.Using the same t-test noted above to compare across decade patterns, we can examine the difference between urban and suburban groups overall (return to Table 2) and across the three decades (Table 3).Looking back at Table 2, this comparison only focused on urban and suburban in relation to rural residents.When directly comparing urban and suburban, urban residents maintain significantly greater tolerance or structural beliefs about racial inequality than suburban residents (t = 2.13, P < .05).This finding shows suburban residents to hold beliefs quite similar (intolerant) to those expressed by rural residents.But what about the pattern across the three decades (Table 3).In both the 1980s and 1990s, the differences between all groups (urban/suburban/rural) are negligible at best.That is to say, beliefs about racial inequality were quite similar over different city sizes and conditions.However, when comparing urban, suburban and rural residents in the 2000s, a hierarchy appears, with urban residents espousing more tolerant or structural beliefs than suburban residents (t = 1.90, P < .05)and rural residents.

Conclusions
The findings from this research are complicated but quite consistent.With reference to overall findings, the impact of urbanism on beliefs about racial inequality is significant.The theoretical proposition of Wirth [1] posed that this difference can be attributed to the secondary relationships fostered in urban locations.Individuals in larger, heterogeneous populations that are densely packed will express greater toleration than individuals from smaller, homogeneous populations that are loosely packed.However, the overall impact of suburban residency does not necessarily reflect the picture proposed by Wirth [1].Specifically, the impact of suburban locations, in relation to rural residents, is negligible at best.This finding demonstrates quite similar beliefs to those who live in suburban and rural locations.In this light, suburban locations do not seem to create the conditions as outlined by Wirth that produces the associated tolerant viewpoints.Similar to rural residents, suburban locations are more likely than urban residents to be smaller in size, homogeneous, and less densely spaced.Indeed, Baldassare [15] stated that suburbia was initially characterized by one-way movements of White middle class families out of diverse urban locales.With reference to over-time trends, the pattern is clear at least with respect to the urban/rural gap.In particular, although very little difference is found between urban and rural residents in the 1980s and 1990s, by the 2000s the difference reaches statistical significance.Indeed, it is quite clear that the regression coefficient for the urban variable is increasing systematically over the three-decade period.Although to a much lesser degree, the impact of suburban residency in relation to rural residents follows the same pattern.However, by the 2000s, the impact of suburban residents only reaches borderline significance.It is worth noting that regression analyses as such cannot tell us why the gap is increasing (i.e., which group is increasing or decreasing over time).Descriptive statistics can provide some insight.From the descriptive statistics in Table 1, it appears that the cause of the gap increase is likely in growing intolerance or individualistic explanations for racial inequality among rural residents.Although some increase in the mean scores is noted in urban and suburban residents, the most notable shift is the decrease observed for rural residents (movement toward individualistic explanations).
The over-time results of the regression and descriptive statistics, while not definitive, do not support the notion proposed by Wirth [24], who predicted that increases in technology, transportation and communication would eventually erode the independent effect of urbanism.Moreover, note that in this argument Wirth further implied that rural residents would become more tolerant and that increase would eventually eliminate any urban/rural differences.Our findings are inconsistent with this proposition in two ways.First, the gap appears to be increasing and not decreasing over time.Second, the gap seems to be a product of growing intolerance among rural residents.An explanation of these findings may be found in the growing Group Positioning literature [37,38].Group Positioning theory proposes that racial resentment may simply stem from group competition, where one group sense that valued resources are being threatened.In this case, individuals who live in rural areas may be feeling a greater sense of competition from minorities in general and blacks in particular over time.
In conclusion, the findings of this paper support the idea promoted by Wirth [1] that homogenous relationships do not promote a great level of toleration of social and cultural differences.Indeed, it seems apparent that even secondary relationships that are born in diverse locations may have positive effects on residents who inhabit larger urban cities.
* −.11 N = 17267 N = 5984 N = 6769 N = 4514 * Using a Tukey HSD post hoc test associated with a one-way ANOVA, this

Table 3 :
Unstandardized coefficients (main effects) and standard errors for the Beliefs Index by Decade.#