Perceptions of Racial Group Size in a Minority-majority Area

Abstract

Americans overestimate the size of minority groups and underestimate the size of the majority. Research on perceived racial group size, however, has focused on areas where non-Hispanic whites are dominant and has neglected to examine the impact of group conflict attitudes other than perceived threats. This study examines perceptions of group size in Harris County, Texas, in 2007, which became a minority-majority area by 2006. It also examines whether perceived racial discrimination influences perceived group size. Analyses demonstrate that people underestimate the size of the largest group, which in this case is the local Hispanic population. The sizes of the black and Asian populations, by contrast, are overestimated while estimates of the white population are accurate. Race and perceived discrimination play important roles in shaping perceptions of group size. Perceptions of group size are also based on the objective size of these groups, but are unrelated to recent changes in group size.

Keywords

group size race Houston prejudice racial and ethnic minorities

Introduction

The United States is in the midst of a demographic transformation through which it is becoming increasingly diverse. In 1970, non-Hispanic whites comprised 83.5 percent of the population (U.S. Census Bureau 2015). Although they were still the majority at 63.7 percent in 2010 (U.S. Census Bureau 2015), the Census Bureau estimates that the United States will become a minority-majority country by 2042 (Roberts 2008). Much of this change is due to the increase in the Hispanic population, which grew from 4.5 to 16.3 percent from 1970 to 2010 (U.S. Census Bureau 2015). This demographic transition, however, is happening faster in some places than in others. In Houston, for example, non-Hispanic whites ceased to be the majority before 2000, and Hispanics became the largest group by 2006.

The ethnic and racial composition of America’s communities is important because it shapes people’s opinions and behaviors, such as their treatment of outgroups, where they prefer to live, and their support of public policies on crime and poverty. Traditional prejudice and opposition to race-targeted policies, for example, are stronger among whites living in areas with larger African American populations (Quillian 1996; Taylor 1998). Whites report that neighborhoods with more blacks and Hispanics are less desirable (Lewis, Emerson, and Klineberg 2011). Those living in areas with larger Hispanic populations are more likely to view Hispanics as threatening and to prefer the more stringent treatment of Hispanics within the criminal justice system (Stewart et al. 2015). People are also less likely to blame the poor for their poverty in areas where the poor are mostly white (Hopkins 2009).

It is often an untested assumption, however, that people accurately perceive the size of ethnic and racial groups. Evidence from multiple surveys suggests that people’s perceptions of group size are often wildly inaccurate (Alba, Rumbaut, and Marotz 2005; Kaiser Family Foundation 2001; Nadeau, Niemi, and Levine 1993; Sigelman and Niemi 2001; Teixeira et al. 2013). Moreover, perceived group size is an important predictor of various outcomes on its own—for example, perceptions of neighborhood disorder and fear of crime (Chiricos, McEntire, and Gertz 2001; Wickes et al. 2013), racial prejudice (Dixon 2006), and immigration-related policy preferences, beliefs about affirmative action, and racial stereotyping (Alba et al. 2005).

A group of studies now exists that examines the sources of perceived racial group size (Alba et al. 2005; Gallagher 2003; Highton and Wolfinger 1992; Nadeau and Niemi 1995; Nadeau et al. 1993; Sigelman and Niemi 2001; Wong 2007).¹ Due to data limitations, however, scholars have focused on perceived racial group size at the national level. Even when they have examined perceived racial group size at the local level, the local areas available in the data are typically majority-majority areas. Also, although these studies of perceived group size are framed by theories of racial threat, only a few control for group conflict attitudes (Alba et al. 2005; Nadeau and Niemi 1995; Nadeau et al. 1993). Moreover, these studies focus on only one type of group conflict attitude (i.e., perceived threats), neglecting others such as perceived racial discrimination.

This article addresses two questions: How accurate are people’s perceptions of local racial group size, and to what extent do perceived racial discrimination and direct observation from the local racial context influence perceived racial group size? It contributes to the literature on perceived racial group size in several ways. First, it examines the perceived size of the local Hispanic, white, black, and Asian populations in Harris County, Texas, in 2007. Harris County, the location of Houston, became a minority-majority area by 2006. Second, this article is the first to examine the role of perceived racial discrimination in shaping perceptions of group size.

Harris County, Texas: The Local Context

Harris County is located in southeast Texas and contains both the city of Houston and surrounding suburbs. Harris County is racially diverse with Hispanics, whites, and blacks comprising roughly 39, 36, and 18 percent of the population in 2007, respectively (see Table 1). It has only recently, however, become a minority-majority area. In 1960, 73.9 percent of Harris County was white, and 19.8 percent was black; Hispanics comprised only 6.0 percent of the population at this time (Klineberg 2014). Changes in federal immigration policy in 1965 dramatically transformed the ethnic and racial distribution of Houston and other cities throughout the United States. Hispanics became the second largest group in Harris County by 1990. They comprised just under one third of the population in 2000, but nearly 40 percent by 2009. Whites, in contrast, comprised 42.1 percent of the population in 2000, but only 35.1 percent by 2009. Census data indicate that Hispanics became the largest group sometime between 2005 and 2006. The relative size of all other groups has remained consistent from 2000 to 2009.²

Table 1.

Trends in the Racial Composition of Harris County, Texas.

Race	1980	1990	2000	2005	2006	2007	2008	2009
Hispanic, any race	15.3	22.9	32.9	37.8	38.2	38.6	39.3	39.8
White, alone	62.6	54.2	42.1	37.9	36.7	36.4	35.9	35.1
Black, alone	19.5	18.7	18.2	17.6	18.4	18.3	17.9	17.9
Asian, alone	1.9	3.8	5.1	5.5	5.4	5.5	5.6	5.7
American Indian/Alaskan, alone	0.2	0.2	0.2	0.2	0.2	0.2	0.2	0.2
Hawaiian/Pacific Islander, alone	—	—	0.0	0.1	0.1	0.1	0.1	0.1
Other, alone	0.4	0.2	0.1	0.2	0.3	0.2	0.2	0.3
Two or more, alone	—	—	1.3	0.8	0.8	0.7	0.8	0.9
Total population	2,409,547	2,818,199	3,400,578	3,647,656	3,886,207	3,935,855	3,984,349	4,070,989

Source. U.S. Census Bureau. 1980 and 1990: Analyzed by the Social Science Data Analysis Network. Accessed via http://www.censusscope.org/; 2000: (PL002) Hispanic or Latino, and Not Hispanic or Latino by Race (http://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=DEC_00_PL_PL002&prodType=table); 2005: (DP01) General Demographic Characteristics (http://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?src=bkmk); 2006–2009: (DP05) ACS Demographic and Housing Estimates (http://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?src=bkmk), Accessed October 17, 2016 via American FactFinder (Tables can be found by entering the table name in the Advanced Search menu at http://factfinder.census.gov/faces/nav/jsf/pages/searchresults.xhtml?refresh=t).

Note. ACS = American Community Survey.

Data from the U.S. Census and American Community Survey (ACS) as reported in the Houston Area Survey (HAS) data indicate that changes in the size of these populations have not been distributed evenly within Harris County (not shown in Table 1). While the average increase in the size of the Hispanic population from 2000 to 2005–2009 was 6.7 percent for Harris County as a whole, changes in the size of the Hispanic population at the Census tract-level ranged from −30.9 to 30.2 percent. The middle 50 percent of Census tracts experienced an increase of between 2.5 and 11.4 percent. The average decrease in the size of the white population for Harris County was 6.7 percent. Changes ranged from −34.7 to 31.0 percent, with the middle 50 percent of Census tracts experiencing a decrease of between 11.6 and 1.0 percent. Differences across Census tracts in the changes of the black and Asian populations are, by contrast, smaller.

Perceptions of Racial Group Size

Previous research on perceived racial group size suggests several general conclusions. First, regardless of their own ethnic and racial identity, people overestimate the size of minority groups, such as blacks, Hispanics, Asians, and Jews (Kaiser Family Foundation 2001). Second, regardless of their own ethnic and racial identity, people underestimate the size of the majority group—that is, non-Hispanic whites (Alba et al. 2005; Wong 2007). Third, misperceptions of group size are entangled with innumeracy, which is “the [in]ability to translate a perception into numerical terms” (Alba et al. 2005:902).

Explaining Misperceptions of Racial Group Size

The approaches used to explain misperceptions of group size differ by race. Some studies focus on white’s perceptions of minority group size (Gallagher 2003; Nadeau et al. 1993). These typically utilize competition-based arguments and are grounded in the racial threat perspectives of scholars, such as Hubert M. Blalock (1967) and Herbert Blumer (1958). Blumer (1958), for example, argues that minorities pose a threat to the dominant position of the majority and that negative attitudes toward minorities are a defensive reaction to these threats. From this perspective, ethnic and racial minorities challenge the dominant position of non-Hispanic whites, leading them to perceive minority groups as a threat and to overestimate their size (Alba et al. 2005; Gallagher 2003). Similarly, some argue that the negative portrayal of African Americans in local television news and the residential concentration of blacks lead whites to perceive them as a “menace” (Gallagher 2003; Nadeau et al. 1993).

Although competition-based approaches focus on majority perceptions of minority groups, competition might lead to the overestimation of any outgroup by any ingroup. This could explain why minorities tend to overestimate the size of other minority groups. Interestingly, however, minorities also tend to overestimate the size of their own group and underestimate the size of the white population (Alba et al. 2005; Wong 2007). Some have speculated that ingroups overestimate the size of their own group out of the need for self-esteem (Wong 2007). Competition-based explanations, though, have difficulty explaining why minorities overestimate the size of their own group and underestimate the size of the majority group.

To better explain why minorities tend to overestimate the size of their own group, scholars have focused on racial context and lived experience (Alba et al. 2005; Sigelman and Niemi 2001; Wong 2007). People’s perceptions of racial group size may be grounded in the extent of residential segregation and intergroup contact. Minorities may overestimate the size of their own group because they tend to live in segregated neighborhoods and most frequently come into contact with other people from their own group. They take this information from their daily experience into account and generalize to other areas, such as the U.S. population as a whole. Research suggests that changes in racial and ethnic composition may play a more important role in shaping perceptions than absolute size itself (Newman and Velez 2014).

Testing Predictors of Perceived Racial Group Size

Previous research supports competition-based explanations and indicates that perceived threats lead to the overestimation of racial group size. Richard Nadeau et al. (1993), for example, demonstrate that people are more likely to overestimate the size of the Hispanic population when they fear that the growing number of Hispanics will reduce jobs, increase taxes and crime, and threaten the English language. People overestimate the size of the black population when they believe the civil rights movement has moved too quickly (Nadeau et al. 1993). Similarly, whites who believe that Hispanics increase crime rates have less accurate perceptions of the size of the Hispanic population (Nadeau and Niemi 1995). Richard Alba et al. (2005) show that perceived group size is associated with perceived threats—for example, that immigrants make it harder to keep the country united, that whites are hurt by affirmative action, and that blacks and Hispanics tend to be violent. Perceived threats are also relevant in studies focusing on the perceived size of the foreign-born population (Herda 2010).

As mentioned above, scholars have also focused on racial context and lived experiences to explain perceptions of group size. The most common measure of racial context and lived experience is objective group size. Previous research demonstrates that this variable is related to perceived group size. The actual size of the local Hispanic population, for example, is positively related to the perceived size of the national Hispanic population (Wong 2007). The actual sizes of the local white, black, Hispanic, Asian, and American Indian populations increase the perceived local sizes of these groups (Wong 2007). Only one study investigates the relationship between contact and perceived racial group size. Increased contact with blacks increases the perceived size of the black population for both whites and blacks (Sigelman and Niemi 2001). Similarly, Daniel Herda (2010) demonstrates that neighborhood contact plays a role in the overestimation of the size of the foreign-born population.

Considerations for Studies of Perceived Group Size

In this section, some important considerations for this and future studies of perceived group size are discussed. These include: examining additional group conflict attitudes, such as perceived racial discrimination, and the importance of adjusting measures of perceived group for innumeracy.

To date, research on perceived group size has focused on perceived threats while other group conflict attitudes have largely been overlooked. Lawrence Bobo (1988:95) defines group conflict motives as “attitudes directly concerned with the competitive aspects of group relations and attempts to alter those relations. They concern the distribution of scarce values and resources . . . [and] attempts to affect the process of and pattern of their distribution.” Although the group conflict motives concept includes perceived threat, it is larger and also includes other attitudes, such as perceptions of incompatible group interests and relative group standing. Bobo (1988) argues that a sense of incompatible group interests and relative group standing are related to the structure of group differences, while perceived threats are related to those attempting to change existing structures.

Although there is clearly evidence that perceived threats are related to perceived group size, other group conflict-related attitudes should be considered. One such possibility is perceived racial discrimination. Racial discrimination is often defined as the unequal treatment of a person or group on the basis of their race (Harnois 2014; Quillian 2006). While “threat” implies the possibility of a future negative outcome, “discrimination” may be more backward-looking. It focuses on the treatment and/or structural conditions that have already led to an unequal outcome. Perceived discrimination is a group conflict attitude because it involves perceptions of a successful attempt to alter existing structures and group grievances based on these unequal outcomes. It may be particularly relevant for understanding perceived group size because of this backward focus on existing outcomes. People may attribute negative outcomes to power imbalances between groups. Feelings of relative powerlessness may lead to a diminished sense of ingroup size and the sense that an outgroup is larger than it really is. These ideas are consistent with the notion of reactive identity, which suggests one’s identity is influenced by intergroup conflict (Portes and Rumbaut 2001).

Finally, one of the challenges faced in previous research is disentangling misperceptions of racial group size from innumeracy. It is possible, for example, to understand the relative size of different racial groups while failing to express this accurately in a survey. Innumeracy is well documented in previous research—for example, people’s estimates should sum to 100 percent, but are routinely quite a bit less or quite a bit more than 100 (Alba et al. 2005). The lack of a fixed maximum value is problematic because it renders perceptions incomparable across respondents. Two responses of 40 for the percent Hispanic, for example, are not equal if one person’s estimates for all groups sum to 80 while those of another sum to 160. When a comprehensive list of ethnic and racial groups is available, it is possible to adjust the estimates such that they sum to 100 percent and are once again relative and comparable (discussed in more detail below).

Research Questions and Hypotheses

Several research questions guide the analysis.

Research Question 1: How accurate are perceptions of local racial group size in a setting that very recently became a minority-majority area?

Research Question 2: To what extent do perceived racial discrimination and direct observation from the local racial context influence perceived racial group size in a new minority-majority area?

The following hypotheses are tested:

Hypothesis 1 (Perceived group discrimination): Those who perceive discrimination of their own group overestimate the size of outgroups and underestimate the size of their ingroup.

It should be noted that it is necessary to include interaction effects between race and perceived discrimination to test Hypothesis 1. This is because the hypothesis predicts that perceived discrimination has a different effect on the perceived size of each racial group depending on the racial group of the respondent. Hypothesis 1 predicts, for example, that perceived discrimination should increase the perceived percent white for those who are Hispanic, but decrease the perceived percent white for those who are white. The direction of the effect of perceived discrimination, in other words, depends on the race of the respondent.

Hypothesis 2a (Direct observation of the racial context): Objective racial group size is positively related to perceptions of racial group size.

Hypothesis 2b (Direct observation of changes in the racial context): Increases in objective racial group size are positively related to perceptions of racial group size.

Data and Method

Sample

Survey data are from the 2007 wave of the HAS (Klineberg 2014). The HAS is a countywide data collection effort in Harris County, Texas, dating back to 1982, which is coordinated by the Kinder Institute at Rice University. The data are a probability sample of adults living in Harris County. Hispanic and black respondents are oversampled such that the data contain approximately equal numbers of Hispanic, black, and white respondents. A weight provided with the data is utilized so that the results are generalizable to Harris County. The data are collected by means of random digit dialing and computer assisted telephone interviewing. Response and cooperation rates are approximately 35 and 50 percent, respectively (Klineberg 2014). The questionnaire is available in English and Spanish, and interviewers are bilingual.

Data from the 2007 HAS provide an interesting opportunity to reexamine perceptions of racial group size for a variety of reasons. First, the focus is on group size at the local level, and the geographic area (Harris County) is defined for the survey respondents, which make it possible to verify the accuracy of each response. Most surveys focus on perceived group size at the national level. When they have included a question targeting the local community, the question wording has made it impossible to verify the accuracy of any single response. Second, the HAS data include respondents from multiple racial and ethnic groups. This makes it possible to compare results across groups to better understand how discrimination and the racial context shape perceived group size. Third, HAS respondents were asked to estimate the size of a comprehensive list of ethnic and racial groups for Harris County. Given this, it is possible to adjust perceptions for innumeracy (Alba et al. 2005). Fourth, the order of the racial groups was randomized. This eliminates the possibility that, in the aggregate, perceived group size is related to question order. Fifth, the data also include perceptions of racial discrimination in Houston as well as measures of objective group size at the zip code, Census tract, and Census block levels.

Dependent Variables

Four items serve as dependent variables and measure perceived group size: “Next, I’d like to ask you about the size of the different ethnic communities in Harris County . . . What percent of the Harris County population would you guess is Anglo/African-American/Hispanic/Asian?” The order of the groups was randomized. Unfortunately, it is not possible to know the order of the groups for each respondent.

Independent Variables

Two variables and their interaction are used to test Hypothesis 1 (perceived group discrimination): race and perceptions of racial discrimination. Race is measured by three dummy variables indicating whether the respondent identifies as white (the reference category), Hispanic, or black. Asians and those in the “Other” category are excluded due to small sample sizes. One item measures people’s perceptions of ingroup racial discrimination: “How often are [R’s ethnicity] discriminated against in Houston?” The response choices are never (the reference), rarely, fairly often, and very often. Fairly often and very often are combined into one category because only a small percentage of white respondents has selected very often. This coding choice does not change the conclusions. The percentage breakdown of perceptions of discrimination for Hispanics, whites, and blacks, respectively, is as follows: never—28.8, 50.9, 27.6; rarely—33.9, 35.5, 32.6; fairly/very often—37.3, 13.5, 39.8. Interactions between race and perceived discrimination are also included to test Hypothesis 1. As mentioned above, the interaction effects make it possible to determine if perceived discrimination has a different effect on the perceived size of each group depending on the race of the respondent. Unfortunately, measures of perceived threat are not available.

Census tract-level variables are included to measure the immediate racial context in 2000 and the five-year average for 2005 to 2009: percent Hispanic, percent white, logged percent black, and logged percent Asian. The change in these percentages from 2000 to 2005–2009 is also included (the change variables are not transformed because they are relatively symmetric). These data are based on estimates from the 2000 U.S. Census and the 2005–2009 ACS and are available within the HAS data file (data are pooled from 2005 to 2009 to provide more reliable estimates of racial group size for this period because the data are from the ACS rather than the decennial Census).

HAS data also include measures of racial composition for zip codes (N = 128, 2000 U.S. Census) and Census blocks (N = 672, 2000 U.S. Census). Scholars often disagree over which is the best contextual unit of analysis. Map-based approaches are an important step forward (Wong et al. 2012), but are not available within the HAS data. Objective group size is measured at the Census tract level because it is only possible to measure changes in racial composition from 2000 to 2005–2009 at the tract level (ACS data for 2005–2009 are not available at the zip code and block levels). Unfortunately, measures of interracial contact are not available in the 2007 HAS data.

Control Variables

Control variables include educational attainment, the number of years living in the Houston area, age, gender, and household income. Educational attainment is measured with a set of dummy variables, including less than high school, high school, some college, college, or a postgraduate degree. Some college is the most common level of educational attainment and serves as the reference category. Years living in the Houston area is centered on the grand mean. The “Houston area” is not defined for survey respondents. This terminology, though, encourages those living in or near the city of Houston to count themselves as living in the Houston area. Age is measured in years and is centered on the grand mean. Sex is measured with a dummy variable where 1 = female and 0 = male. Household income is measured with a set of dummy variables: $15,000 or less, $15,001 to $25,000, $25,001 to $35,000, $35,001 to $50,000, $50,001 to $75,000, and more than $75,000 (the reference). Self-reported knowledge of current events and place of residence (in the city of Houston inside of loop 610, etc.) were included in preliminary models, but neither is related to perceived group size. Other measures of media exposure are not available in the 2007 wave.

Analytic Technique

The data are multilevel (i.e., 977 respondents are nested within 457 Census tracts). Having multilevel data violates the assumption of independence and often leads to heteroskedasticity and correlated errors. Ordinary least squares (OLS) regression with cluster-corrected standard errors is utilized to control for the nesting of respondents in Census tracts (hypotheses are tested using two-tailed t tests with an alpha level of .05). This approach is preferable to multilevel modeling for several reasons. First, multilevel modeling assumes that the level 2 cases are a random sample, which may not be true for the Census tracts represented in the HAS data. Second, OLS regression provides unbiased slope estimates even in the presence of multilevel data. It is the standard errors, not the slopes, that are misestimated (the use of cluster-corrected standard errors addresses this problem). Third, there are only a small number of respondents per Census tract (2.1 on average), and the hypotheses do not require the estimation of Census tract-specific coefficients. See Stephen Raudenbush and Anthony Bryk (2002, chap. 9) for a discussion of the assumptions of multilevel modeling.

Most person-level variables contain only a small amount of missing data due to nonresponse (i.e., 1 percent or less). Missing cases make up 43.0 percent of the data for household income, 6.5 percent for perceptions of group discrimination, 5.5 percent for age, and 4.8 percent for race. Multiple imputation (MI) is used to deal with missing data because it outperforms other approaches (Enders 2010). The number of imputations was set to 40. The chained equations approach is used because many of the person-level variables are categorical. Regression analyses are limited to those cases with complete data on all four dependent variables, with a valid Census tract identifier (ID), and who are Hispanic, white, and black (due to very small sample sizes for Asians and people in the “Other” category). The final sample size for the regression models after these deletions is 977.

Results

How accurate are people’s perceptions of the relative size of racial groups in Harris County, Texas? The descriptive statistics in Table 2 indicate that residents of Harris County believed that whites held a slight numeric edge over other groups in 2007, on average. The mean and median perceived percent white are 42.9 and 40 compared with 36.0 and 30 for Hispanics. The size of the Hispanic population is underestimated by 2.6 percent, although this point estimate is within the survey’s margin of error of 3.5 percent (Klineberg 2014). By contrast, the sizes of the white, Asian, and black populations are overestimated by 6.5, 7.8, and 11.9 percent, respectively. Although not visible in Table 2, answers ending in 0 and, to a lesser extent, 5 dominate the data. The predominance of 0’s and 5’s is not unexpected because this type of response is encouraged by the question wording, “what percent of the Harris County population would you guess is . . .” This pattern is consistent with other research on perceived group size (e.g., Alba et al. 2005).

Table 2.

The Perceived Racial Composition of Harris County, Texas, in 2007: Descriptive Statistics (N = 1,154).

Descriptivestatistics	Perceived percent					Innumeracy adjustedperceived percent^a
Descriptivestatistics	Hispanic	White	Black	Asian	Sum	Hispanic	White	Black	Asian	Sum
Sample mean	36.0	42.9	30.2	13.3	122.4	29.0	36.4	24.6	10.1	100.0
Census estimate	38.6	36.4	18.3	5.5	98.8	38.6	36.4	18.3	5.5	98.8
Average error	−2.6	6.5	11.9	7.8	23.6	−9.6	0.0	6.3	4.6	1.2
Median	30	40	30	10	110	29	35	24	9	100
Minimum	0	0	0	0	0	0	1	4	0	100
Maximum	100	99	100	100	396	88	86	70	51	100
Interquartile range	20	20	15	10	30	14	18	11	8	0
Standard deviation	18.6	15.7	14.2	12.7	38.9	11.0	13.4	8.2	6.9	0.0

Adjusted for innumeracy by dividing each estimate by the sum of all estimates for each respondent.

Although some officially recognized racial groups are not included in the question set, the four groups that are included made up nearly 99 percent of Harris County residents in 2007. We should, therefore, expect the sum of the four answers to be close to 100 percent. On average, the perceived percentages sum to 122, and most people provide answers that are within 39 points of this average. The median value is 110 and the middle 50 percent of people provide answers that sum to between 80 and 140 percent. These measures of central tendency and variability provide evidence of innumeracy.

Alba et al. (2005) argue that the misperception of group size is distinct from innumeracy. To adjust for innumeracy, the estimated percentages are recalculated by dividing them by the sum of the four estimated percentages. The innumeracy adjusted perceived percent Hispanic, for example, is,

y = (\frac{% H i s p a n i c}{% H i s p a n i c + % A n g l o + % B l a c k + % A s i a n}) \times 100 .

Adjusting the estimates in this way may at least partially offset innumeracy because the sum of the four estimates must now be 100 for all respondents. Without this standardization of the scale, it is not possible to compare estimates across respondents. The distributions of these adjusted percentages are also displayed in Table 2.

Once adjusted for innumeracy, people’s perceptions of the size of the white population are equal to the true value. The size of the Hispanic population, however, is underestimated by 9.6 percent while the sizes of the Asian and black populations are overestimated by 4.6 and 6.3 percent, respectively. It is notable that the relative size of the group that is largest (i.e., Hispanics) is underestimated in this minority-majority location. The magnitude of the perception errors is also of note. Whether an error is large or small is, of course, subjective. The errors are considerably smaller in size, though, compared with those for the United States as a whole, as presented in other research. In one study, for example, respondents underestimate the size of the white population by about 11 percent and overestimate the size of the Hispanic, black, and Asian populations by about 11, 18, and 13 percent, respectively (Alba et al. 2005). This may suggest that people have more accurate perceptions of the size of ethnic and racial groups in their local community.

The regression results are presented in Table 3. All four dependent variables are perception errors—that is, the 2007 Census estimate subtracted from the innumeracy adjusted values for each racial group. A score of 0 on a dependent variable represents a correct perception, a negative score indicates underestimation of group size (e.g., a score of −10 indicates underestimation by 10 percentage points), and a positive score indicates overestimation of group size. There are two models for each dependent variable. Model 1 controls for objective group size in the Census tract in 2005 to 2009. Model 2 controls for objective group size in the Census tract in 2000 and the change from 2000 to 2005–2009. Results for perceived racial group size that are unadjusted for innumeracy are available upon request.³

Table 3.

Perception Errors for the Racial Composition of Harris County, Texas, in 2007: Regression Models (N = 977).

Variables	Perception error % Hispanic				Perception error % white				Perception error % black				Perception error % Asian
	Model 1		Model 2		Model 1		Model 2		Model 1		Model 2		Model 1		Model 2
	b	SE	b	SE	b	SE	b	SE	b	SE	b	SE	b	SE	b	SE
Intercept	−14.12	1.94	−13.19	1.80	−3.58	2.53	−3.99	2.56	4.17	1.71	4.30	1.77	5.12	1.42	4.90	1.41
Race
White (ref)
Black	−4.57	2.73	−4.22	2.92	0.48	3.66	0.48	3.65	−0.56	1.86	−0.67	1.88	5.69*	2.07	5.61*	2.09
Hispanic	7.62*	2.84	7.76*	3.25	−2.13	4.06	−2.20	4.08	−2.71	2.49	−2.65	2.52	−1.09	2.19	−0.93	2.19
Group discrimination
Never (ref)
Rarely	1.67	1.54	1.44	1.46	−0.87	2.10	−0.85	2.10	0.92	1.52	0.99	1.57	−1.82	1.25	−1.86	1.26
Fairly/very often	5.78*	1.88	4.40*	1.75	−5.14*	2.42	−5.15*	2.42	0.42	1.68	0.40	1.73	−0.85	1.53	−0.88	1.53
Race × group discrimination
Black × rarely	4.32	3.40	4.31	3.37	−0.30	4.49	−0.15	4.49	1.45	2.47	1.38	2.51	−5.08*	2.39	−4.92*	2.40
Black × fairly/very often	3.14	3.12	2.92	3.26	2.61	3.98	2.77	3.99	−0.39	2.22	−0.42	2.29	−5.15*	2.34	−5.05*	2.36
Hispanic × rarely	−6.88*	3.21	−7.83*	3.53	3.36	4.28	3.48	4.30	1.81	2.91	1.62	2.93	1.20	2.32	1.03	2.32
Hispanic × fairly/very often	−11.04*	3.25	−10.84*	3.54	6.58	4.41	6.74	4.44	2.01	2.70	1.84	2.72	1.47	2.45	1.29	2.46
Objective group size
% Hispanic	0.04	0.02	0.05*	0.02
Change % Hispanic			−0.06	0.05
% white					0.09*	0.02	0.09*	0.02
Change % white							0.05	0.06
% black (logged)									2.30*	0.63	2.47*	0.65
Change % black											0.02	0.06
% Asian (logged)													1.74*	0.69	2.16*	0.67
Change % Asian															0.06	0.09
Education
Less than high school	1.05	1.51	0.30	1.55	−0.06	1.85	−0.01	1.85	1.51	1.35	1.44	1.40	−1.95*	0.96	−2.01*	0.96
High school	0.20	1.18	0.41	1.01	0.99	1.17	0.99	1.17	0.49	0.90	0.44	0.89	−1.50*	0.75	−1.48*	0.75
Some college (ref)
College	0.87	1.10	0.43	1.03	2.37	1.27	2.41	1.27	−1.49	0.77	−1.49	0.77	−1.86*	0.68	−1.89*	0.68
Postgraduate	2.41*	1.22	1.69	1.24	2.05	1.55	2.16	1.57	−1.21	0.92	−1.22	0.92	−3.45*	0.71	−3.46*	0.71
Years in Houston	0.02	0.03	0.00	0.02	−0.02	0.03	−0.02	0.03	0.04	0.02	0.04	0.02	−0.03	0.02	−0.03	0.02
Age	0.03	0.03	0.03	0.03	0.02	0.04	0.02	0.04	−0.06*	0.03	−0.06*	0.03	0.01	0.02	0.01	0.02
Gender (female = 1)	−0.17	0.85	0.22	0.72	2.46*	0.97	2.47*	0.98	−0.91	0.61	−0.94	0.61	−1.54*	0.56	−1.53*	0.55
Household income
<$15,001	0.08	2.06	0.66	1.81	−5.00	2.80	−4.89	2.80	2.12	1.75	1.94	1.72	3.58*	1.43	3.64*	1.45
$15,001–$25,000	−2.35	1.90	−1.01	1.62	2.17	2.32	2.27	2.31	−1.11	1.40	−1.33	1.38	2.29	1.23	2.28	1.23
$25,001–$35,000	−1.98	2.01	−1.77	1.63	−0.39	2.02	−0.35	2.02	1.65	1.34	1.57	1.34	1.46	0.92	1.47	0.91
$35,001–$50,000	0.37	1.83	0.66	1.38	−1.26	1.73	−1.25	1.73	0.14	1.05	0.02	1.04	1.17	1.03	1.16	1.02
$50,001–$75,000	−0.15	1.51	0.12	1.21	−0.16	1.57	−0.14	1.58	0.29	1.01	0.20	1.01	0.39	0.79	0.41	0.79
>$75,000 (ref)
Average RVI	0.17		0.15		0.14		0.13		0.11		0.11		0.12		0.12
Largest FMI	0.44		0.38		0.39		0.38		0.25		0.27		0.32		0.33
Complete DF	455		455		455		455		455		455		455		455
DF average	344		348		361		365		393		394		376		378
F	1.96		1.62		4.42		4.25		2.01		1.99		4.99		4.88
Prob > F	0.01		0.04		0.00		0.00		0.01		0.01		0.00		0.00

Note. The dependent variable is a perception error—that is, it is equal to the perceived percentage minus the true percentage for each group. A score of 0 is a correct perception, a negative score indicates underestimation of group size, and a positive score indicates overestimation of group size. RVI = relative increase in variance; FMI = fraction of missing information; DF = degrees of freedom.

p < .05 (two-tailed).

The results in Table 3 provide some support for Hypothesis 1. Results indicate that whites who perceive the racial discrimination of whites “fairly” or “very often” tend to have higher estimates of the size of the Hispanic population and lower estimates of the size of the white population. On average and net of controls, their estimate of the size of the Hispanic population is 5.78 percentage points higher compared with whites who have “never” perceived the discrimination of whites. Their estimate of the size of the white population is 5.14 percentage points lower compared with whites who have “never” perceived the discrimination of whites. Discrimination is not related to the perceived size of the black and Asian populations for whites. These patterns are depicted in Panel A of Figure 1. It is interesting to note that whites, regardless of the level of perceived discrimination, always underestimate the size of the Hispanic population. The gap between their perceptions of whites and Hispanics, however, is much smaller for those who perceive discrimination “fairly” or “very often” because discrimination leads to smaller estimates of the size of the white population and larger estimates of the size of the Hispanic population.

Figure 1.

Interaction effects: Race and discrimination.

Results also suggest that, in some cases, perceptions of racial discrimination interact with race to shape perceptions of group size. Hispanics perceiving the racial discrimination of Hispanics tend to have lower estimates of the size of the Hispanic population (see Panel B of Figure 1). On average and net of controls, Hispanics who “rarely” perceive the discrimination of Hispanics provide an estimate of the Hispanic population that is 5.21 percentage points lower compared with Hispanics who have “never” perceived the discrimination of Hispanics (the effect of “rarely” is equal to the main effect for whites plus the interaction effect for Hispanics: 1.67 ± 6.88 = −5.21). Also, Hispanics who perceive the discrimination of Hispanics “fairly” or “very often” provide an estimate of the Hispanic population that is 5.26 percentage points lower compared with Hispanics who have “never” perceived the discrimination of Hispanics (5.78 ± 11.04 = −5.26). Although the effect of discrimination on the perceived size of the white population is not different for Hispanics as compared with whites, separate analyses limited to Hispanics (available upon request) indicate that discrimination is not related to the perceived size of the white population (this can also be seen in the relatively flat slope of the “perception error percent white” line in Panel B of Figure 1). Finally, discrimination is not related to the perceived size of the black and Asian populations for Hispanics. As Figure 1 illustrates, Hispanics have accurate perceptions of the size of the white population, but underestimate the size of the Hispanic population. That said, the gap is much wider among those who believe that Hispanics are “rarely” and “fairly/very often” discriminated against in Houston (as compared with “never”) because discrimination leads to smaller estimates of the size of the Hispanic population.

The nonsignificant interaction effects for blacks in the first set of models indicate that discrimination has the same effect on the perceived size of the Hispanic population for blacks as compared with whites. This finding is confirmed by separate analyses that are limited to black respondents (available upon request; both discrimination coefficients are positive and significant in these models). As perceived ingroup discrimination increases, so does the perceived size of the Hispanic population (see Panel C of Figure 1). Although the effect of discrimination on the perceived size of the white population is not different for blacks as compared with whites, separate analyses limited to blacks (available upon request) indicate that discrimination is not related to the perceived size of the white population. Discrimination is also unrelated to perceptions of the size of the black population for blacks. Finally, blacks provide an estimate of the size of the Asian population that is 5.69 percentage points higher than whites. This racial gap in perceptions between blacks and whites, however, depends on the level of perceived ingroup discrimination (because the interactions between the black and group discrimination dummy variables are significant). In opposition to Hypothesis 1, blacks who perceive that their ingroup experiences discrimination “rarely” and “fairly/very often” have lower estimates of the size of the Asian population (as compared with blacks who respond “never”).

In sum, the results for race and perceived discrimination are complex. Several coefficients support Hypothesis 1. As perceived discrimination increases, Hispanics provide smaller estimates for the size of the Hispanic population and larger estimates for the size of the white population. Also, as perceived discrimination increases, whites and blacks provide larger estimates for the size of the Hispanic population. This particular pattern may reflect underlying conflicts between both whites and blacks with Hispanics in Houston as Hispanics recently became the largest group. In opposition to Hypothesis 1, blacks provide smaller estimates of the size of the Asian population as perceived discrimination increases. This pattern may indicate that blacks view Hispanics rather than Asians as the source of discrimination in Houston. Unfortunately, the data do not provide any information on the source of perceived discrimination.

Hypothesis 2a (Direct observation of the racial context) states that perceptions of group size are based at least partly in objective group size. People take cues from their surroundings and use these to make judgments about the objective conditions around them. There is support for Hypothesis 2a for all four racial groups. Those living in Census tracts with a higher percentage of Hispanics in 2000 tend to believe that the Hispanic population of Harris County is larger (see Model 2; p = .068 for the % Hispanic coefficient in Model 1). This same pattern exists in both models for perceptions of the size of the white, black, and Asian populations. It should also be noted that these analyses have been replicated using measures of group size from 2000 at the level of zip codes and Census blocks (available upon request). The conclusions using data at these different units of analysis are the same—that is, objective group size is positively related to perceived group size.

Given increases in the size of the Hispanic population and decreases in the size of the white population in Houston since 2000, this analysis also examines whether changes in group size are related to perceived group size (these results are shown in Model 2 for each outcome). The group size measures in the second model for each outcome are objective group size in 2000 and the change in objective group size from 2000 to 2005–2009. Results indicate that the change in racial group size is not related to perceived group size. None of the slopes for the change measures is statistically significant. There is, therefore, no support for Hypothesis 2b.

Excluding race, perceptions of discrimination, and objective group size, few of the other independent variables are related to people’s perceptions of group size. Only one education dummy variable is significant when predicting the perceived size of the Hispanic population (in Model 1, but not in Model 2). Although education is related to people’s perceptions of the size of the Asian population, those who have completed some college have the least accurate perceptions (because the intercept is positive and the slopes for all four education dummies are negative). Education is unrelated to people’s perceptions of the size of the white and black populations. Older individuals have more accurate perceptions of the size of the black population. The intercept is 4.17, which is the predicted percent for those of average age (because age is centered). Each one year increase in age reduces this prediction error by 0.06. Women have more accurate perceptions of the size of the white and Asian populations compared with men. The perception errors for percent white are −3.58 for men and −1.12 for women (−3.58 + 2.46 = −1.12). The perception errors for percent Asian are 5.12 for men and 3.58 for women (5.12 − 1.54 = 3.58). There is little evidence that household income is related to perceived group size in Houston. Only those in the lowest income category have less accurate perceptions of the size of the Asian population. Finally, length of time living in Houston is not significant in any model.

Discussion and Conclusion

Data from the 2007 Houston Area Survey are used to examine perceived racial group size in a minority-majority area. Results indicate that survey respondents underestimated the size of the local Hispanic population and overestimated the size of the local black and Asian populations. Their perceptions of the size of the white population, by contrast, are accurate (based on the innumeracy adjusted data). In the years immediately preceding the survey, Hispanics became the largest group in Harris County. People, however, continued to believe that whites were the largest group despite media reports of the demographic shifts occurring in Harris County and the United States as a whole (e.g., Pinkerton 2007; Roberts 2007).

What accounts for the variability in people’s perceptions of racial group size? Similar to previous research, there is evidence from Houston that people base answers to factual questions about racial group size on their immediate racial context. Regardless of their race, those living in Census tracts with larger Hispanic, white, black, and Asian populations tend to have higher estimates for the size of these groups in Harris County. Additional analyses at the Census block and zip code levels support the idea that objective group size is positively related to local perceived group size. Despite this, there is no indication that recent changes in group size are related to perceptions of group size in the Houston area.

Previous research suggests that perceived threats lead to the overestimation of minority populations for the United States as a whole. Researchers, however, have not considered the role of other group conflict attitudes. This study focuses on perceived racial discrimination. Results indicate that discrimination shapes perceptions of racial group size in Houston and that these relationships depend upon race. Hispanics who perceive discrimination toward their group tend to provide higher estimates of the size of the white population and lower estimates of the size of the Hispanic population. Similarly, whites who believe that their group has experienced discrimination tend to provide lower estimates of the size of the white population. Finally, blacks who believe that their group has experienced discrimination tend to provide higher estimates of the size of the Hispanic population and lower estimates of the size of the Asian population. This complex pattern indicates that the effect of perceived discrimination on perceived racial group size differs by race and depends on group-specific conflicts in Houston. A different pattern might emerge in other places based on the history of race relations and the current racial context in that area.

This study is among the first to examine whether another type of group conflict attitude (i.e., perceived racial discrimination) is related to perceived group size. Unfortunately, it is not possible to measure perceived threat or other group conflict attitudes beyond perceived discrimination with the HAS data from 2007. Past research suggests that perceived threat would play a significant role in shaping perceptions of group size in Houston. Given the demographic changes that have occurred there, the perceived threat of Hispanics would probably be more important than the perceived threat of other groups. It is also possible that there might be a different conclusion about the effect of perceived discrimination if other group conflict motives were controlled. Perhaps perceived discrimination plays a more important role in shaping the perceived size of the white population among historically disadvantaged groups, while perceived threats are more important in shaping white’s perceptions about historically disadvantaged groups. Additional research is needed to compare and contrast the effects of various group conflict attitudes on perceived group size as well as to test the mechanisms through which these influence people’s perceptions.

Finally, it is noteworthy that people underestimate the Hispanic population in Houston. Results indicate that the size of the Hispanic population is underestimated by 9.6 percent once the estimates are adjusted for innumeracy. In all previous research (in majority-majority settings), it is the white population that is underestimated. What accounts for this curious pattern? It could be a reflection of the relative status and power of Hispanics in Houston. Only 16 percent of U.S.-born Hispanics in Houston, for example, have achieved at least a college degree compared with 25 and 46 percent U.S.-born blacks and whites, respectively, and 51 percent of Hispanic immigrants lack a high school diploma (Klineberg 2014). Similarly, third generation Latinos in Houston are only marginally more likely than first generation Latinos to have incomes greater than $25,000, to be employed in higher status occupations, and to have health insurance (Klineberg 2014). Houston also holds the dubious distinction as never having elected a Latino to Congress despite its large Latino population (Schleifer 2014).

Unfortunately, additional variables that may help us get to the bottom of this puzzle, such as measures of intergroup contact, media exposure, numeracy, and question order for each respondent, are not available in the HAS data. Even with these variables, however, the unexplained variation may remain high given that people are providing their best guess. These questions may contain a large amount of measurement error and random noise. Before conducting additional survey research on the topic, scholars should consider using an experimental design to further examine the generation of relative group size estimates. Such an approach would allow researchers to examine differences in relative group size answers under various conditions, which may lead to improvements in the reliability of these measures. Some possibilities might include varying the order and number of groups, the types of groups (e.g., racial and ethnic groups, foreign- and native-born populations, sex and gender-related groups), and the ability of respondents to revise guesses.

This study shares some limitations with previous research on perceived racial group size. Although it is possible to verify the accuracy of people’s beliefs about racial group size in Harris County, the county boundaries and racial composition of it may not be particularly relevant or known for most people. Also, Census tracts (and Census blocks and zip codes) are probably not the most appropriate areas for informing people’s views on racial group size. Their personal definitions of “community” probably do not line up with these administrative divisions (Wong et al. 2012). People may also utilize information from nonresidential contexts, such as their places of work and worship. Finally, with cross-sectional data, it is difficult to establish the causal direction of the relationship between perceived group size and perceived discrimination or other competition-related variables (see Lawrence and Sides 2014). It is possible that competition and conflict alter perceptions of group size and also that misperceptions of group size shape people’s views on intergroup competition and conflict. These may also all be a function of a more general negative orientation toward outgroups.

Despite these challenges, researchers should continue to explore the sources and consequences of perceived racial group size—both in the United States as a whole and more locally in communities around the country. The United States is in the midst of a demographic transformation through which it will become a minority-majority country within the next 25 years. Some communities have already completed this transition and resemble these Census projections. Although it is somewhat reassuring that people’s perceptions in Houston are based partly on the size of racial groups in their area, it is disconcerting that they are also shaped reactively by perceptions of racial discrimination. Additional research is needed in other communities to determine how common such patterns may be. A comparative approach that takes into account the similarities and differences of these communities and the patterns that emerge may provide guidance to those working to improve race relations as the United States and communities around the country approach minority-majority status.

Footnotes

Acknowledgements

The author wishes to thank the editors and reviewers as well as Michael O. Emerson and Jie Wu of the Kinder Institute for Urban Research at Rice University for their assistance with the Houston Area Survey Data.

Declaration of Conflicting Interests

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

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Robert M. Kunovich is an associate professor of sociology. He conducts research in the areas of race, ethnicity, and nation; political sociology; and the sociology of risk.

References

Alba

Richard

Rumbaut

Ruben G.

Marotz

Karen

. 2005. “A Distorted Nation: Perceptions of Racial/Ethnic Group Sizes and Attitudes toward Immigrants and Other Minorities.” Social Forces 84(2):901–19.

Blalock

Hubert M.

1967. Toward a Theory of Minority-group Relations. New York: John Wiley and Sons.

Blumer

Herbert

. 1958. “Race Prejudice as a Sense of Group Position.” Pacific Sociological Review 1:3–7.

Bobo

Lawrence

. 1988. “Group Conflict, Prejudice, and the Paradox of Contemporary Racial Attitudes.” Pp. 85–109 in Eliminating Racism: Profiles in Controversy, edited by Katz

Phyllis A.

Taylor

Dalmas A.

New York: Plenum Press.

Chiricos

Ted

McEntire

Ranee

Gertz

Mark

. 2001. “Perceived Racial and Ethnic Composition of Neighborhood and Perceived Risk of Crime.” Social Problems 48(3):322–40.

Dixon

Jeffrey C.

2006. “The Ties that Bind and Those that Don’t: Toward Reconciling Group Threat and Contact Theories of Prejudice.” Social Forces 84:2179–204.

Enders

Craig K.

2010. Applied Missing Data Analysis. New York: Guilford Press.

Gallagher

Charles A.

2003. “Miscounting Race: Explaining Whites’ Misperceptions of Racial Group Size.” Sociological Perspectives 46(3):381–96.

Harnois

Catherine E.

2014. “Are Perceptions of Discrimination Unidimensional, Oppositional, or Intersectional? Examining the Relationship among Perceived Racial–Ethnic-, Gender-, and Age-based Discrimination.” Sociological Perspectives 57(4):470–87.

10.

Herda

Daniel

. 2010. “How Many Immigrants? Foreign Born Population Innumeracy in Europe.” Public Opinion Quarterly 74:674–95.

11.

Herda

Daniel

. 2013. “Too Many Immigrants? Examining Alternative Forms of Immigrant Population Innumeracy.” Sociological Perspectives 56(2):213–40.

12.

Herda

Daniel

. 2015. “Beyond Innumeracy: Heuristic Decision-making and Qualitative Misperceptions about Immigrants in Finland.” Ethnic and Racial Studies 38(9):1627–45.

13.

Highton

Benjamin

Wolfinger

Raymond E.

1992. Estimating the Size of Minority Groups. ANES Pilot Study Report, No. nes002291. Retrieved October 7, 2016 (http://www.electionstudies.org/Library/papers/pilotrpt.htm).

14.

Hooghe

Marc

de Vroome

Thomas

. 2015. “The Perception of Ethnic Diversity and Anti-immigrant Sentiments: A Multilevel Analysis of Local Communities in Belgium.” Ethnic and Racial Studies 38(1):38–56.

15.

Hopkins

Daniel J.

2009. “Partisan Reinforcement and the Poor: The Impact of Context on Explanations for Poverty.” Social Science Quarterly 90(3):744–64.

16.

Kaiser Family Foundation. 2001. “Race and Ethnicity in 2001: Attitudes, Perceptions and Experiences.” Washington Post/Kaiser Family Foundation/Harvard University. Retrieved October 7, 2016 (http://kff.org/other/poll-finding/race-and-ethnicity-in-2001-attitudes-perceptions/).

17.

Klineberg

Stephen L.

2014. Kinder Houston Area Survey, 1982–2013: Successive Representative Samples of Harris County Residents (ICPSR20428-v3). Ann Arbor, MI: Inter-university Consortium for Political and Social Research (ICPSR) [distributor]. doi:10.3886/ICPSR20428.v3.

18.

Lawrence

Eric D.

Sides

John

. 2014. “The Consequences of Political Innumeracy.” Research & Politics 1(2):1–8.

19.

Lewis

Valerie A.

Emerson

Michael O.

Klineberg

Stephen L.

2011. “Who We’ll Live with: Neighborhood Racial Composition Preferences of Whites, Blacks and Latinos.” Social Forces 89:1385–408.

20.

Martinez

Michael D.

Wald

Kenneth D.

Craig

Stephen C.

2008. “Homophobic Innumeracy? Estimating the Size of the Gay and Lesbian Population.” Public Opinion Quarterly 72(4):753–67.

21.

Nadeau

Richard

Niemi

Richard G.

1995. “Educated Guesses: The Process of Answering Factual Knowledge Questions in Surveys.” The Public Opinion Quarterly 59(3):323–46.

22.

Nadeau

Richard

Niemi

Richard G.

Levine

Jeffrey

. 1993. “Innumeracy about Minority Populations.” Public Opinion Quarterly 57:332–47.

23.

Newman

Benjamin J.

Velez

Yamil

. 2014. “Group Size versus Change? Assessing Americans’ Perception of Local Immigration.” Political Research Quarterly 67(2):293–303.

24.

Pinkerton

James

. 2007. “Hispanics Exceed Anglos in Harris County: Census Shows Area Leads the Nation in Growth of Minorities.” The Houston Chronicle, August 9. Retrieved October 17, 2016 (http://www.chron.com/news/houston-texas/article/Hispanics-exceed-Anglos-in-Harris-County-1841891.php).

25.

Portes

Alejandro

Rumbaut

Rubeén G.

2001. Legacies: The Story of the Immigrant Second Generation. Berkeley, CA: University of California Press.

26.

Quillian

Lincoln

. 1996. “Group Threat and Regional Change in Attitudes toward African-Americans.” American Journal of Sociology 102(3):816–60.

27.

Quillian

Lincoln

. 2006. “New Approaches to Understanding Racial Prejudice and Discrimination.” Annual Review of Sociology 32:299–328.

28.

Raudenbush

Stephen

Bryk

Anthony

. 2002. Hierarchical Linear Models: Applications and Data Analysis Methods. 2nd ed. Thousand Oaks, CA: Sage Publications.

29.

Roberts

Sam

. 2007. “Minorities Now Form Majority in One-third of Most-populous Counties.” New York Times, August 9. Retrieved October 17, 2016 (http://www.nytimes.com/2007/08/09/us/09census.html?_r=0).

30.

Roberts

Sam

. 2008. “A Generation Away, Minorities May Become the Majority in U.S.” New York Times, August 14. Retrieved October 17, 2016 (http://www.nytimes.com/2008/08/14/washington/14census.html).

31.

Schleifer

Theodore

. 2014. “Houston Most Hispanic Part of Country without Hispanic in Congress.” Houston Chronicle, September 6. Retrieved October 17, 2016 (http://www.houstonchronicle.com/news/politics/texas/article/Houston-most-Hispanic-part-of-country-without-5738987.php).

32.

Semyonov

Moshe

Raijman

Rebeca

Gorodzeisky

Anastasia

. 2008. “Foreigners’ Impact on European Societies: Public Views and Perceptions in a Cross-national Comparative Perspective.” International Journal of Comparative Sociology 49:5–29.

33.

Sigelman

Lee

Niemi

Richard G.

2001. “Innumeracy about Minority Populations: African Americans and Whites Compared.” Public Opinion Quarterly 65:86–94.

34.

Stewart

Eric A.

Martinez

Ramiro

Jr. Baumer

Eric P.

Gertz

Marc

. 2015. “The Social Context of Latino Threat and Punitive Latino Sentiment.” Social Problems 62:68–92.

35.

Taylor

Marylee C.

1998. “How White Attitudes Vary with the Racial Composition of Local Populations: Numbers Count.” American Sociological Review 63:512–35.

36.

Teixeira

Ruy

Halpin

John

with Matt Barreto Pantoja

Adrian

. 2013. “Building an All-in Nation: A View from the American Public.” Center for American Progress and PolicyLink in partnership with the Rockefeller Foundation. Retrieved October 7, 2016 (https://www.americanprogress.org/issues/race/report/2013/10/22/77665/building-an-all-in-nation/).

37.

U.S. Census Bureau. 2007. Table S0501, American Community Survey data for Harris County, Texas in 2007. Retrieved October 7, 2016 (http://factfinder.census.gov/).

38.

U.S. Census Bureau. 2015. “A Look at the 1940 Census.” Retrieved October 7, 2016 (https://www.census.gov/newsroom/cspan/1940census/CSPAN_1940slides.pdf).

39.

Wickes

Rebecca

Hipp

John R.

Zahnow

Renee

Mazerolle

Lorraine

. 2013. “‘Seeing’ Minorities and Perceptions of Disorder: Explicating the Mediating and Moderating Mechanisms of Social Cohesion.” Criminology 51(3):519–60.

40.

Wong

Cara J.

2007. “‘Little’ and ‘Big’ Pictures in Our Heads: Race, Local Context, and Innumeracy about Racial Groups in the United States.” Public Opinion Quarterly 71(3):392–412.

41.

Wong

Cara J.

Bowers

Jake

Williams

Tarah

Simmons

Katherine Drake

. 2012. “Bringing the Person Back in: Boundaries, Perceptions, and the Measurement of Racial Context.” The Journal of Politics 74(4):1153–70.