Legal Issues in Minority Aggregation

The Fifteenth Amendment bars the denial or abridgment of the right to vote on account of race. It also grants the United States Congress the power to enforce the amendment “by appropriate legislation.” The VRA, passed in 1965, is such a piece of legislation. It prohibits “first-generation” forms of voting discrimination such as literacy tests. It also prevents states and localities from adopting election rules that weaken the effectiveness or strength of minority voters, a strategy referred to as “vote dilution.” Under certain conditions, for example, a municipality with a large African American population may be prohibited from using an at-large election system if it prevents African Americans from electing a preferred candidate to the city council.

Between 1965 and 2013, changes to election rules in several states and counties, mostly located in the Deep South, had to be precleared by the Department of Justice or a federal court. If election rules were found to have a retrogressive effect on minority voter strength, they could be stopped. But the Supreme Court’s decision in Shelby v. Holder in 2013 ended this preclearance protection, and those seeking to prevent minority voter dilution have had to rely on litigation.

Members of a protected class of citizens can establish a violation through litigation if, “based on the totality of circumstances, it is shown that the political processes leading to nomination or election in the State or political subdivision are not equally open to participation.” In Thornburg v. Gingles in 1986, Justice Brennan further articulated three conditions that must be present when establishing a violation. They are:

1. 1. The protected class of citizens must be sufficiently large and geographically compact to constitute a majority in a single-member district;

2. 2. The protected class of citizens must be “politically cohesive”;

3. 3. The non-protected majority must vote sufficiently as a bloc to enable it to defeat the protected class of citizens’ preferred candidate.

A protected class of citizens may include groups such as African Americans or Latinos. But do African Americans and Latinos together constitute a class of citizens? Can minority populations be aggregated when presenting claims of vote dilution to help meet the first condition laid out in Gingles?

The Supreme Court has never ruled directly on this question, and the federal circuit courts have reached diverging conclusions. The Sixth Circuit has explicitly rejected the idea of minority aggregation, the Fifth Circuit endorsed it before reversing itself in 2024, and others have ruled on the issue only implicitly.

Legal debate centers on a few key issues, such as congressional intent and whether minorities behave with enough political cohesiveness to be appropriately considered a class of citizens. In this paper, I take up the latter issue. Although plaintiffs would need to establish cohesion between African Americans and Latinos in a specific jurisdiction to meet the second condition in Gingles, overall patterns of cohesion throughout the country are nonetheless informative for demonstrating that minorities, generally speaking, should be thought of as forming a class of citizens.

Social Science Research on Minority Aggregation

What do we know about political cohesion between African Americans and Latinos? Very little, it turns out. Most scholars who have studied the political relationship between these two groups focus on intergroup attitudes. Gay (Reference Gay2006), for example, found that African Americans were more likely to see their political interests as being compatible with those of Latinos if Latinos were economically disadvantaged.

Kauffman’s (Reference Kaufmann2003a) work focused on Latinos’ feelings of commonality with African Americans. Such feelings, Kauffman argued, are a necessary precursor to coalition building. She found that pan-Latino affinity, or feelings connected to all Latinos regardless of region or national origin, promotes solidarity between Latinos and African Americans. This argument was echoed by Wilkinson (Reference Wilkinson2014), who also showed that Latinos with a strong sense of in-group solidarity felt close to African Americans. McClain et al. (Reference McClain, Carter, DeFrancesco Soto, Lyle, Grynaviski, Nunnally, Scotto, Kendrick, Lackey and Cotton2006) found that a similar attitude, linked fate, reduced the negativity toward African Americans among Latino immigrants in the South. Thus, the factors that structure Latino-linked fate, such as economic marginalization (Sanchez and Masuoka Reference Sanchez and Masuoka2010), may also promote political cohesion between Latino and African Americans. Interracial social contact increases feelings of interminority commonality as well (Wilkinson Reference Wilkinson2014), so the coexistence of Latinos and African Americans in the same geographic space facilitates coalition building.

Hero and Preuhs (Reference Hero and Preuhs2013) moved this literature away from an exclusive focus on preexisting attitudes by highlighting the role of elites and considering how federalism interacts with shared interests and ideology to structure cooperation, conflict, and political independence between African Americans and Latinos. They found little evidence of interminority conflict at the national level, which they attribute to the higher degrees of shared interest and ideology on matters that are the purview of the federal government and to the organizational actions of elites and advocacy groups.

A few authors have directly tackled the issue of cohesion in voting and elections. Relying on both survey experiments and exit poll data, Benjamin (Reference Benjamin2017) identified a new predictor of racial cohesion in local elections: elite cues. Co-ethnic elites can form and solidify biracial coalitions through endorsements. Her finding links with Hero and Prehus’s emphasis on elites and other factors beyond mass attitudes.

Hajnal and Trounstine (Reference Hajnal and Trounstine2014) drew on exit poll responses from 56,000 voters from five cities (Chicago, Detroit, Houston, Los Angeles, and New York) between 1985 and 2000 and found that support for winning candidates among African Americans and Latinos differed by over 24 percentage points. This difference grew to 33 percentage points when studying mayoral elections across the nation’s 25 largest cities.

In his analysis of roughly 1,500 school board elections, Rocha (Reference Rocha2007) found little evidence of political cohesion between African Americans and Latinos although with some important caveats. First, his conclusion applied only to nonpartisan elections and majority-White voting districts. When focusing on majority–minority jurisdictions, Rocha (Reference Rocha2007) found that Latinos were more likely to support African American candidates. He also found evidence of cohesion across 200 school districts that relied on partisan elections. Despite this ancillary evidence, Rocha generally portrayed African American and Latino voters as lacking cohesion although his text was inattentive to moderating factors that are important in the voting rights law. What is missing from the literature is a straightforward test of how the presence of Latinos affects the electoral fortunes of African Americans, and vice versa, specifically in voting jurisdictions that are majority–minority.

It is also worth revisiting Rocha’s (Reference Rocha2007) study because of the amount of time that has passed. His conclusions are based on data from 2001, and social scientists and legal scholars are concerned with whether old patterns of racialized voting behavior persist. In Shelby v. Holder, Chief Justice Roberts expressed weariness about the continued applicability of decisions that rely on decades-old data. He reasoned that we must account for the fact that history did not end when Congress passed the VRA; new patterns can emerge. One question in this paper, therefore, is what patterns of racial cohesion look like now as opposed to two decades ago.

The issue of partisanship, which was the focus of Rocha’s (Reference Rocha2007) analysis, requires deep consideration. Do African Americans and Latinos support similar candidates solely because they are more likely on average to be Democrats? Hero and Prehus (Reference Hero and Preuhs2013) suggested that strong partisanship was part of the reason they did not observe conflict between African Americans and Latinos in Congress. But there is some evidence to suggest that minority groups support each other’s candidates at high rates even after accounting for shared partisanship. Kaufmann (Reference Kaufmann2003b), in particular, found that African Americans supported local Latino candidates more often than what would be expected given their partisan affiliation.

The answer to this overarching question has significant legal ramifications. The courts have drawn a distinction between election rules used to dilute voting strength on the basis of partisanship and on the basis of race. Dilution on account of partisanship is permissible; dilution on account of race is not. The courts view redistricting and other election procedures as the “traditional domain of state legislative authority”; therefore, challenges, particularly to redistricting, are often nonjusticiable. One exception occurs when states engage in racial rather than partisan gerrymandering, which violates the protection enshrined in the 14th Amendment and, therefore, invites strict scrutiny from the courts. (See Justice Altio’s majority opinion in Alexander v. South Carolina State Conference of the NAACP.)

But it is often difficult to differentiate between the racial and partisan considerations because the two are so highly correlated. Opponents of the VRA sometimes claim plans that dilute minority voting strength are in fact legally acceptable attempts at gaining partisan advantage. Those challenging district lines must somehow “disentangle” racial from partisan consideration.

Fortunately, studying local elections allows a unique way to disentangle the two. Most cities in the United States rely on nonpartisan elections, but a sizable minority, roughly 25%–30%, use partisanship in a way that is similar to state or federal elections (Hajnal and Lewis Reference Hajnal and Lewis2003). This allows me to study how minority voter cohesion differs in partisan and nonpartisan cities. If minority cohesion occurs simply because both groups are generally Democratic, then cohesion should be highly present in cities that use partisan elections, and cohesion should be low in cities in which shared partisanship is not a tie that can be seen on the ballot. Below, I specify how this and my other expectations should manifest empirically.

The presence of a rainbow coalition implies that as the size of the African American population within an area grows, Latino representation would likewise increase and vice versa. A lack of cohesion manifests as a null result or even a negative relationship between the size of one minority group in a jurisdiction and the level of representation held by the other minority group (see Meier and Stewart Reference Meier and Stewart1991). Thus, I test for evidence of minority cohesion by examining the following hypotheses:

• African American population size is positively associated with Latino representation in elected office.

• Latino population size is positively associated with African American representation in elected office.

To examine whether minority cohesion differs in majority-minority and majority-white contexts, I consider whether the effect of African American/Latino population size on Latino/African American representation is moderated by the demographics of the city. I test this by inserting a dummy variable for majority-minority cities and interacting it with the African American and Latino population measures.

Last, to address the issue of partisanship, I split the sample based on whether each city holds partisan or nonpartisan elections. (Rocha used a similar approach.) If minority cohesion is simply the result of shared partisanship, I would expect positive association between African American/Latino population size on Latino/African American representation to be present only in cities that held partisan elections. If the positive association between African American/Latino population size on Latino/African American representation is present in cities that hold nonpartisan elections, there is evidence that minority cohesion exists independently of partisanship.

Data and Methods

To address these issues, I ask whether the ability of African Americans to win elections to public office is affected by the size of the Latino population in a jurisdiction. I also ask whether the ability of Latinos to win elections is affected by the size of the African American population. This is similar to the research design employed by Rocha (Reference Rocha2007) as well as older studies such as Meier and Stewart (Reference Meier and Stewart1991).

My data come from the ICMA’s 2018 Municipal FOG Survey and the US Census Bureau’s 2018 American Community Survey (ACS). The FOG survey was mailed to city clerks in each of the 12,761 municipalities found in the ICMA’s database. Clerks had the option to respond using a prepaid return envelope or by linking to an online survey. The response rate was 32.2%. (For more details, see https://icma.org/2018-municipal-fog-survey.) The ICMA data are especially useful for comparing outcomes across cities and have been used regularly by studies of local elections (see, for example, Altema McNeely Reference Altema McNeely2019; Bae and Feiock Reference Bae and Feiock2013; DeSantis and Renner Reference DeSantis and Renner2002; Feiock, Krause, and Hawkins Reference Feiock, Krause and Hawkins2017; Kinney Reference Kinney2008; León-Moreta Reference León-Moreta2021; Lubell et al. Reference Lubell, Feiock and De La Cruz2009; Trounstine Reference Trounstine2013; Trounstine and Valdini Reference Trounstine and Valdini2008; Wei, Butler and Jennings Reference Wei, Butler and Jennings2019). Trounstine (Reference Trounstine2013) referred to the ICMA FOG survey as “the best data available for studying large numbers of cities” and noted that it was “fairly representative of the national urban population and provides relatively accurate measures of local structure and conditions.” Trounstine and Valdini (Reference Trounstine and Valdini2008) offered a deeper validation of the FOG survey, noting that the proportion of city councilors elected from single-member district and at-large elections are nearly identical in the ICMA FOG dataset and the U.S. Census Bureau’s Census of Governments.

The survey contains questions about the racial background of city council members and the total number of positions. With this, I am able to develop a measure of the percentage of seats held by African Americans and Latinos.

I pair this information with census data from the ACS 2018 5-year estimates, which contain measures of the size of the African American and Latino citizen voting age population (CVAP). This allows me to predict the percentage of seats held by a group with that size of the African American and Latino electorate. The ACS data also allow me to control for socioeconomic conditions within communities. Specifically, I account for educational attainment among minorities, measured as the percentage of African Americans or Latinos over 25 who have received a bachelor’s degree. Previous studies have noted that minority electoral success is more likely to be observed in areas with a high number of poor Whites (see Rocha Reference Rocha2007), and, therefore, I account for the percentage of Whites living below the poverty line.

I also control whether the jurisdiction was formerly covered by Section 5 of the VRA. Jurisdictions covered by Section 5 have long histories of discrimination against minority voters, especially African Americans. In such communities, race and ethnicity are more salient when voters are choosing candidates. The saliency of race, Benjamin (Reference Benjamin2017) showed, increases cohesion among minority voters.

I include a dichotomous measure to account for whether the jurisdiction is majority–minority. For the purpose of this study, this measure only accounts for whether African Americans and Latinos together constitute 50% or more of the CVAP.

Last, the ICMA data contain information about the electoral structures used in each city. City clerks were asked to report on the system used to elect members to the council. I account for two rules, the percentage of seats elected from single-member districts and whether candidates use partisan labels. Minorities are more likely to be elected when single-member districts are used if their population is geographically compact (Abott and Magazinnik Reference Abott and Magazinnik2020). African Americans and Latinos are more likely to hold office in partisan systems. Furthermore, partisan elections make it more likely that African Americans and Latinos will support similar candidates.

Below I offer a fuller analysis of the extent to which cohesion exists in cities using partisan systems and in those using nonpartisan systems. If cohesion between African Americans and Latinos can be seen in cities that use nonpartisan systems, there is reason to believe that minority cohesion exists for reasons beyond shared partisanship. Contingent on an investigation into the particular circumstances in a locality, we may appropriately consider African Americans and Latinos to be a single class of citizens that is politically cohesive.

Findings

Descriptive plots for all the variables used in the analysis are available in the appendix. Table 1 offers the first test of minority cohesion. There I examine whether the size of the Latino electorate in a city is positively or negatively associated with the ability of African Americans to be elected. At this stage, I do not account for the use of partisan elections, but I do examine how this association differs in a majority–minority context. My goal is to determine whether the relationship between the presence of Latinos and the election of African Americans is systematically different in majority–minority areas. If it is, then it would be inappropriate to use previous research whose inferences about cohesion were drawn from mostly majority-White contexts to argue against the drawing of minority aggregation districts.

Table 1.

Dependent variable = % African American council members

* p < .05.

** p < .01.

(standard errors in parentheses).

The models in Table 1 are OLS estimates. The dependent variable in each model is the percentage of city council seats held by African Americans. My key independent variables include the size of the African American and Latino CVAP and dichotomous variables indicating whether the area is majority–minority. I also include an interaction term between the two to determine whether the marginal effect of CVAP on minority officeholding is different in majority-White and majority–minority contexts.

The results validate my concerns about the usefulness of past work when thinking about minority aggregation. In majority-White contexts, the size of the Latino electorate is positively associated with the election of African American candidates. This finding contradicts Rocha’s (Reference Rocha2007) earlier work and suggests some level of cohesion even in majority-White areas. There is, however, no statistically significant relationship between the size of the African American electorate and the election of Latino candidates in majority-White contexts.

When examining majority–minority contexts a different pattern emerges both in terms of the size of the substantive effects and, at some points, statistical significance. On average, African Americans are more likely to hold office in cities with a large Latino electorate. The marginal effect, displayed in Figure 1, is .292 and shows that if the Latino electorate grew by 10 percentage points, the number of African American officeholders is estimated to grow by roughly 3 percentage points.

Figure 1.

Racial cohesion in majority-white and majority-minority jurisdictions.

Is this evidence of minority cohesion confined to partisan systems? No. Contrary to previous work (Rocha Reference Rocha2007), the findings presented in Table 1 replicate in both partisan and nonpartisan systems. When examining cities that use partisan elections, the marginal effect of Latino voting strength on African American officeholding in a majority-minority contexts is .397 (p = .00). In nonpartisan systems, it is .291 (p = .00).

Overall, the results also offer some evidence for the assertion that African Americans are less likely to hold council seats in cities that were formerly covered by Section 5 of the VRA. In cities that use partisan systems and where preclearance protection has ended, African Americans hold 2.5% fewer seats after controlling for the size of the electorate and the other control variables used in the analysis. The VRA promotes the use of single-member districts; attempts to move away from single-member districts to at-large elections are often viewed as retrogressive under Section 5. Table 1 confirms that African Americans are more likely to hold council seats if a city relies on single-member districts. Together, these findings affirm the importance of the protection the VRA has afforded African Americans.

In Table 2, the dependent variable in each model is the percentage of city council seats held by Latinos. Examining all cities, Figure 1 shows that if the African American electorate increased by 10 percentage points, the number of Latino officeholders would go up by roughly 2 percentage points. The marginal effect of African American voting strength on Latino officeholding in majority–minority contexts is .339 (p = .00) in partisan systems and .146 (p = .00) in nonpartisan systems.

Table 2.

Dependent variable = % Latino council members

* p < .05.

** p < .01.

(standard errors in parentheses).

In summary, I observe evidence of minority voter cohesion even when party labels do not appear on the ballot. Partisanship does strengthen cohesion, but the evidence does not support the idea that partisanship rather than race is what leads minority groups to support each other electorally. Socioeconomic conditions also affect Latino representation. Latinos are more likely to hold council seats in cities in which a larger share of the Latino population has graduated from college and where Whites have a higher rate of poverty.

As a check on these results, I also approached the analysis using propensity score matching.Footnote ¹ The results, which are available in the appendix, continue to show evidence of racial cohesion among African Americans and Latinos in majority–minority contexts. African American population size is associated with an increase in the number of Latino city council members, and this effect is larger in the majority–minority contexts. Latino population size is associated with an increase in the number of African American city council members although this effect does not appear to differ in majority–minority or majority-White cities. Importantly, there is no evidence of interminority conflict, a finding which again contradicts Rocha (Reference Rocha2007).