Abstract
Voters often use candidate ethnicity as a shortcut in low-information elections, but partisan cues might shift the weight of ethnic signals. This article examines how partisan information moderates the relationship between co-ethnicity and vote choice in 103 multiethnic, nonpartisan mayoral elections in California (2010–2021). Although party labels are absent from ballots, candidates sometimes publicly signal a partisan affiliation. Using precinct-level election returns and voter registration data, I find that Latino candidates receive greater support as the share of Latino registrants in a precinct increases—but this relationship changes when party information is revealed. When facing a known non-Latino Democrat, Latino candidates receive less support than their nonpartisan counterparts; when facing a known non-Latino Republican, they gain support. These effects are strongest in precincts with more Latino Democrats. The findings highlight how party cues reshape ethnic voting, even without formal partisan labels.
Introduction
A large body of research on vote choice in local elections has established that voters tend to cast ballots in favor of candidates that match their racial or ethnic identities when those candidates are on the ballot. Scholars originally conceived this phenomenon as “ethnic” voting and studied it mostly in the context of non-Protestant white immigrant groups under the influence of political machines in 19th and twentieth century urban settings (Wolfinger 1965). Although political machines’ power has waned, many still observe an ethnic divide in voter behavior at the local level, especially among Black and Latino voters (Barreto 2007; Benjamin 2017; Hero 1992; Leighley 2001; Manzano and Sanchez 2010; Matson and Fine 2006; Sanchez 2006).
One common explanation used to account for the prevalence of ethnic voting in local politics is that voters use race or ethnicity as a heuristic to judge candidates in low-information electoral environments (Crowder-Meyer, Gadarian and Trounstine 2020; Matson and Fine 2006; McDermott 1998, 1998; Schaffner and Streb 2002). Because local elections often lack key informational signals helpful to voters in national politics, especially candidate partisanship and widespread media coverage, candidate characteristics may be among the few available cues to guide vote choice. When voters know little else about candidates than their race or ethnicity, they may resort to relying on stereotypes associated with those groups to infer their positions on policy and vote accordingly.
If ethnicity serves as a heuristic, however, its relevance may depend on the broader informational context in which voters make decisions. When clearer or more politically informative cues—such as candidate partisanship (Rahn 1993)—are available, voters may find those cues more useful than ethnicity for inferring candidate ideology, policy preferences, or likely behavior in office, particularly when party affiliation and ethnic identity diverge (Michelson 2005). Depending on the amount of information readily available, the salience of candidate ethnicity in guiding vote choice may recede as voters shift toward more direct signals of policy alignment.
At the same time, a large body of research emphasizes that ethnic voting may be an outgrowth of a unique aspect of the local political environment. Rather than being dominated by ideological and partisan competition, local politics is often concerned with the distribution of scarce public resources (Anzia 2021). Because of this, voters may view city politics as a contest between groups over access to government services, attention, and representation. Under this conceptualization, voters prioritize electing co-ethnic candidates not only because ethnicity provides a best guess at ideology or partisanship, but to ensure members of their group win influence and the attention of local government (Barreto 2007; Collet 2005; Kaufmann 2004; Liu and Vanderleeuw 2007; Trounstine 2016). As such, ethnic voting may persist even in the presence of clearer or more traditional ideological cues like party labels. These perspectives are not necessarily contradictory; rather, they highlight different mechanisms through which ethnicity may shape vote choice, depending on what voters know and the salience of group conflict in their cities.
A recent series of survey experiments support an informational explanation. When groups of mostly white respondents are presented with minimal information to differentiate candidates (Crowder-Meyer, Gadarian and Trounstine 2020) or are otherwise cognitively taxed (Crowder-Meyer et al. 2020), hypothetical candidates of color receive lower levels of support. As more information is introduced, the rate of cross-ethnic voting increases. While these experimental designs isolate the individual effects of specific informational cues, there is reason to question how well the survey experiment environment reflects how voters behave in real-world low-information elections.
This study extends the question of whether ethnic voting is still as prevalent in different informational contexts by comparing rates of ethnic voting in contests when candidate partisanship is not publicly known to those where it is, even in officially nonpartisan elections. I analyze precinct-level election returns from 103 multiethnic mayoral contests in California between 2010 and 2021. While party labels are not printed on the ballot, candidates sometimes publicly affiliate with a political party prior to or as they run for office. I leverage this variation in candidate public affiliation to compare how Latino and non-Latino candidates perform in precincts with different Latino voter population concentrations, examining whether co-ethnic support diminishes or intensifies when partisan cues are available.
I find a consistent positive association between the share of Latino registered voters in a precinct and the share of votes Latino candidates receive in contests where no candidate partisan information is publicly available. However, this relationship is moderated by an out-group candidate's partisanship when it is revealed. In the face of a non-Latino publicly affiliated Democrat competitor, Latino candidate support in heavily Latino precincts slackens. Conversely, when up against a non-Latino Republican, Latinos earn even more votes in precincts with large Latino voter populations. This is especially true in precincts with more Latino Democratic voters than Latino Republicans. I further find that in cities where ethnic group-based political conflict at the local level may be more likely, the negative penalties Latino Republican candidates face in heavily Latino precincts disappear. These relationships suggest that the use of the co-ethnic cue may be dependent on context—Latino voters seem to prioritize partisan over ethnic considerations when party affiliation is visible, except in instances when factors that may precede group-based conflict are present.
These findings contribute to our understanding of how voters respond to informational cues and the role ethnicity plays in structuring local politics. Previous research efforts that have established the trend of ethnic voting usually rely on analyses of one or a few elections at a time (Barreto 2007; Collet 2005; Hero 1989; Kaufmann 2004; Michelson 2005). This article can better speak to these trends because it employs a much larger sample of 103 mayoral elections. Moreover, it provides observational evidence that complements findings from survey experiments (Crowder-Meyer, Gadarian and Trounstine 2020; Crowder-Meyer et al. 2020; Kirkland and Coppock 2018) which show that voters behave differently under different informational environments. Unlike survey experiments, which evaluate responses to hypothetical candidates in carefully structured conditions, this article examines how voters respond to partisan and ethnic cues as they naturally arise in real elections. The results presented here show that similar patterns emerge in actual electoral behavior, where candidate information provision is messier, less controlled, and arguably more meaningful to voters.
Understanding Vote Choice in Low-Information Environments
Democratic governance requires voters to frequently make political decisions through elections. At their most visible, national elections dominate media circles and public conversation such that individuals enter voting booths with a host of information about the pertinent candidates, parties, and issues. Yet American voters are also expected to cast ballots in a wide range of down-ballot offices, some of which cover more obscure policy domains and receive far less media scrutiny. This uneven distribution of attention means that voters often confront information deficits in local elections.
A spatial theory of voting predicts that individuals make vote decisions with regard to which candidate would move policy closest to their ideal preference (Downs 1957). However, this framework assumes that voters operate under the required amount of information to make such a rational decision. Low-information elections, such as those for local school board, city council, and mayoral positions often offer little to no information whereby voters can judge differences between candidates’ personalities or policy preferences. In these low-information environments, few voters will be able to properly discern any changes in policy that could occur with one candidate in office over another.
In many low-information elections the only pieces of information many voters likely receive prior to casting their vote are the candidates’ names as they are listed on the ballot. The names of candidates may give voters a noisy but somewhat reliable estimate of their racial and ethnic identities (Gaddis 2017). Lacking more precise information, voters may rely on stereotypes associated with racial or ethnic groups to create potentially noisy inferences of candidate policy preferences, and vote accordingly.
Additionally, some scholars of local politics argue that racial/ethnic group conflict undergirds political behavior at the local level (Hajnal and Trounstine 2014), and that a sense of competition between racial and ethnic groups over limited public resources steers that conflict. Under this group conflict perspective, the prevalence of ethnic voting may have little to do with the amount of information available, and more to do with a desire for increased co-ethnic representation, political incorporation, and distribution of public resources to neighborhoods where co-ethnics reside. Below, I review the informational and group conflict perspectives on ethnic voting. These perspectives are not necessarily competing explanations. Instead, they highlight different mechanisms that may shape how voters use ethnicity in decision-making, with these mechanisms sometimes operating simultaneously but becoming more or less salient depending on the informational context. I then develop a framework for understanding when and how these dynamics matter for vote choice.
Ethnicity as Heuristic in Low-Informational Settings
In an environment where candidates’ policy positions are perfectly known, voters would have no problem determining which candidate's victory would maximize their utility by moving policy closest to their ideal. This state of perfect information is more of a lofty ideal than a realistic expectation. It would require that voters both know and understand each candidate's policy proposals in a given election and that voters themselves have meaningful preferences concerning those issues.
Even when studying voting behavior in highly publicized, national elections, political scientists have concluded that most voters do not have very meaningful policy preferences and often lack the information necessary to meaningfully judge between candidates and parties (Converse 2006; Campbell et al. 1960; Delli Carpini and Keeter 1996). Instead of voting strictly in accordance with their preferences, the uninformed electorate commonly operates under a strong attachment to a political party, referred to as “partisan identification” or “party ID” (Campbell et al. 1960). To the voter with inconsistent preferences, party ID provides a shortcut whereby she can judge candidates and associated policies. Voters evaluate candidates with appeals to the stereotypes they attribute to either of the political parties or the reputations that the parties’ “brands” carry (Aldrich 2011; Rahn 1993). As such, a candidate's affiliation with one of the two major political parties sends a signal whereby voters can reasonably infer policy preferences. Thus, voters need not seek out swaths of information on candidates’ proposed policy platforms to cast their vote, but instead simply cast a vote in favor of the candidate associated with the political party they identify with.
The assortment of nonpartisan, primary, judicial retention, and otherwise low-visibility elections complicate these prevailing theories behind vote choice. In these elections, candidates run without or share a party label. As such, the ability to use a candidate's party affiliation as a shortcut is dampened. The original intent of local nonpartisan elections was to push voters to seek their own information rather than rely on party cues in an era dominated by party machines. But in effect, the additional cognitive effort required by nonpartisan elections represses overall voter turnout (Schaffner, Streb, and Wright 2001). In surveys, less educated and less politically active respondents are less likely to express a vote preference in elections without party labels (Schaffner and Streb 2002). However, in high-visibility nonpartisan elections, perceptive voters may still find cues that allow them to vote along partisan lines (Rock and Baum 2010).
Those studying nonpartisan elections have turned to possible information shortcuts beyond party whereby voters may translate the limited information made available to them into a vote choice. Some avenues whereby voters may obtain information that sends signals on the policy positions of those running in nonpartisan elections include newspaper endorsements (Boudreau, Elmendorf and MacKenzie 2015), publicized incumbency performance (Berry and Howell 2007), and the preferences of leading industry elites (Lupia 1994). In every case, voters employ these informational shortcuts to save themselves from the stresses of obtaining encyclopedic levels of knowledge about every political decision. And in effect, all of these informational shortcuts provide voters a noisy method whereby they can suppose the partisanship and by extension policy preferences of candidates that lack party labels.
Beyond informational shortcuts from relevant sources, research suggests that voters also use candidate characteristics in crafting vote choices in nonpartisan low-information elections. Among these characteristics are candidate gender, race/ethnicity, and prior occupation. In low-information environments absent public knowledge of partisan affiliations, incumbent performance, candidate policy preference, or even the policy domain of the office in question, these candidate characteristics may be the only information whereby a voter can judge between candidates. Indeed, it is likely that in many cases the only information available to voters when casting their ballots are the names of the candidates as they are listed on the ballot—which can prompt voters to consider candidate gender and ethnicity (Gaddis 2017; Jacobson 1987).
When voters know little else about candidates than descriptive characteristics, they may resort to relying on stereotypes associated with the candidates’ affiliated groups to evaluate their fit for office and infer their positions on policy. Existing scholarship provides evidence that voters use candidate race and gender as a proxy for ideology and issue ownership (McDermott 1997, 1998). A strong historical association between minority racial and ethnic groups and the Democratic Party (Uhlaner and Garcia 2005) could similarly prompt individuals to infer Democratic affiliation among candidates belonging to minority groups. The most straightforward application of such stereotypes occurs when a voter assumes that a candidate who appears demographically similar will also share her policy preferences. In the absense of other cues, voters may thus favor candidates who share their demographic characteristics, believing this alignment implies shared values and policy priorities.
Both observational and experimental evidence on low-information electoral environments suggests that voters do tend to rely on candidate characteristics to make their vote choice absent other relevant information. In Miami-Dade County, Florida's first community advising and zoning board elections—an environment absent information on incumbent performance and media coverage—non-Hispanic voters overwhelmingly supported non-Hispanic candidates over their Hispanic opponents (Matson and Fine 2006). In experimental settings, groups of mostly white respondents prefer hypothetical white candidates over candidates of color when the only information given to them are the candidates’ names (Crowder-Meyer, Gadarian and Trounstine 2020).
Thus, if we assume that voters in low-information political environments are using the limited information available to them to suppose candidate policy preferences, ethnicity provides a simple, albeit noisy shortcut whereby voters may approximate those preferences. Most commonly, it seems this shortcut simply points voters to support in-group candidates and disfavor those from out-groups when little other information is readily available. However, reliance on ethnicity as a heuristic need not exhaust the explanation of ethnic voting.
Ethnic Voting as an Outgrowth of Local Conflict
In addition to the informational account, some scholars argue that the policy domains covered by local government are not ideological or partisan by nature, and thus ideological or partisan concerns are not top of mind for voters participating in local elections. Instead, what local politics is largely about is the provision of public goods (Anzia 2021). Because ethnic minority groups like Latinos are heavily concentrated in urban areas in the United States, local government may be the most viable arena where they may exercise political power, expect political incorporation, and the point the distribution of public resources to the geographic areas where members of their ethnic group live (Hero 1992). A legacy of structural racism at the local level has left neighborhoods segregated and local government service provision unequal, even long after the removal of explicitly racist ordinances (Trounstine 2018). More segregated cities tend to be more politically polarized (Trounstine 2016). Some go as far as arguing that in some local contexts, a racial divide better explains voter behavior than partisan, class, or ideological divides (Hajnal and Trounstine 2014). And when voters perceive that group competition for resources is high, so too is the salience of race and ethnicity in shaping vote choices (Kaufmann 2004).
Ethnic considerations are most prevalent when a minority candidate is on the ballot (Wolfinger 1965). Minority candidates tend to emerge and succeed in electoral environments where their group makes up a significant portion of the electorate (Cuéllar 2014; Ocampo 2017) and act as mobilizers and coalition builders for co-ethnic voters (Barreto 2007; Benjamin 2017). In nonpartisan local elections, minority candidates and campaigns have the potential to attract the support of in-group voters in an effort to secure more government resources and attention to the needs of the group. For many Latino voters, Latino ethnicity itself is a highly salient issue shaping vote choice (Sanchez 2006). Those with a strong commitment to their ethnic identity tend to prefer co-ethnic candidates, even when an out-group candidate appears more qualified (Manzano and Sanchez 2010).
Recent pushes to reform local political institutions emphasize the ostensibly important role race and ethnicity may play in defining local politics. The California Voting Rights Act (CVRA) of 2001 softened the legal requirements by which minority group plaintiffs can challenge the presence of minority vote dilution in at-large city council and school board electoral systems—under the supposition that a transition to single-member district elections could enhance minority representation and incorporation. Scholarship in the aftermath of the CVRA finds that school boards that did transition to district based elections did see a boost in the number of Latino school board members, but only in highly segregated school districts (Abott and Magazinnik 2020). Further, research suggesting that off-cycle local elections diminish the representativeness of local electorates (Hajnal, Kogan and Markarian 2022) bolsters an effort to reform local elections to be conducted concurrently with state and federal elections.
These group conflict-based explanations highlight why ethnicity may remain salient in local elections regardless of partisan or ideological considerations if the perceived level of competition between groups is high. From the information-based perspective, however, the use of ethnic cues seems to be conditional on information—when more precise signals like party affiliation are available, voters may rely on those instead. The differing implications of these two perspectives raise an important empirical question: when partisan information is available, how does it alter the weight that voters place on ethnicity in their decisions? Rather than being mutually exclusive, the informational and group conflict perspectives suggest that ethnicity may matter in different ways depending on the informational and political context.
Partisanship in Nonpartisan Elections?
Taken together, the heuristic and group conflict perspectives suggest that the salience of ethnicity in vote choice is not fixed, but conditional on context. At times, voters may lean on ethnicity as a heuristic in the absence of other cues; in other moments, ethnic voting may be better understood as an expression of group-based competition over representation and resources. I therefore treat the heuristic and group conflict perspectives as context dependent, with the salience of ethnicity in vote choice conditioned by both the availability of partisan cues and the depth of pre-existing ethnic political divides in the city.
Although local elections in California are officially nonpartisan, candidates sometimes publicly signal their party affiliation through prior runs for partisan offices, campaign materials, or in media interviews. For instance, in the 2016 Sacramento mayoral election, Darrell Steinberg ran as a nonpartisan, yet his affiliation with the Democratic Party was widely known due to his previous service in the California State Senate. Similarly, in Long Beach's 2014 mayoral election, Latino Democrat Robert Garcia ran against Damon Dunn, a former Republican candidate for California Secretary of State—making both candidates’ partisan affiliations publicly known despite the race's nonpartisan format. Candidates that have previously ran for or served in a partisan office have no choice but to have their partisanship revealed to the public. Other candidates may strategically reveal their partisanship through the media or in campaign materials (see Appendix Section A.2).
A candidate's public affiliation with a political party may provide voters with a stronger and more direct signal of their likely policy preferences than ethnicity alone. For partisan voters, knowing that a candidate is a Democrat or Republican may offer more actionable information than knowing the candidate's racial or ethnic identity. For example, a Latino Democratic voter might prefer a non-Latino candidate they know to be a Democrat over a Latino candidate with no party affiliation. Indeed, prior research has shown that in Congressional elections, shared partisanship tends to outweigh shared ethnicity for Latino Democratic voters when the two cues conflict (Michelson 2005). Voters may favor candidates whose policy positions they can more reliably infer, and party affiliation often facilitates those inferences more effectively than ethnicity.
Conversely, if ethnic voting primarily stems from group-based political considerations at the local level, the presence of partisan information may do little to diminish the role of ethnicity in shaping vote choice. Barreto (2007) finds that a selection of Democratic, Republican, and Green Party Latino mayoral candidates mobilize and earn similar levels of support from Latino voters because they tap into feelings of the ethnic group's political salience in their cities. If ethnicity's salience plays a foundational role in how voters perceive local politics, then ethnic voting may persist even when voters are more informed of candidates’ partisan affiliations.
Table 1 summarizes the mechanisms and expectations derived from the informational and group conflict perspectives, highlighting how their influence varies across contexts. The empirical analyses that follow test the predictions and observable implications of both perspectives. Specifically, I examine the relationship between the proportion of Latinos in an election precinct and votes cast for Latino candidates, and whether this relationship is moderated by partisan cues provided by candidates. I also investigate whether these patterns differ across cities where local politics are more likely to be shaped by group-based or distributional conflicts.
Summary of Theoretical Frameworks and Observable Implications.
Data and Measurement
To test whether the rate of ethnic voting changes under different informational and political environments, I gather precinct-level registration data and election results from 103 mayoral contests in which at least on Latino candidate ran against at least one non-Latino candidate in California between 2010 and 2021. I identify these elections using the American Local Government Elections Database (De Benedictis-Kessner et al. 2023), which compiles information on local contests and candidates in jurisdictions with populations over 50,000 across three decades. The database records candidates’ race and ethnicity 1 using a combination of administrative data and machine-learning predictions trained on national voter file information. I draw precinct-level election results from statements of votes publicized in official county and city election archives.
The American Local Government Elections Database also flags candidates who have officially signaled a partisan affiliation. I independently validated these classifications by searching through candidate biographies and local news pieces published at the time of the elections in which they ran. I summarize how voters had access to information on the candidates’ partisanship at the time of their elections in Section A.2 of the Appendix. Common forms of partisan signaling include disclosure in media interviews, previous campaigns for or service in partisan offices, and public involvement in local party organizations.
Figure 1 visualizes the ethnic and partisan breakdown of the 431 candidates that ran in these elections. Most candidates, Latino and non-Latino alike, are “true” nonpartisan candidates in the sense that no public information at the time of the election explicitly indicated a partisan affiliation. However, 87 candidates, roughly one-fifth of the total, did publicly disclose their partisanship. Known partisanship was more common among non-Latino candidates than among Latino candidates, with Latino Republicans being especially rare.
Despite being officially nonpartisan, some candidates publicly signal their partisan affiliation.
Figure 2 and Table 2 show how these candidates were distributed across the 103 contests over time. A slight majority of contests (57/103) were fully nonpartisan, meaning that no candidate in the race publicly affiliated with a political party. The share of Latino residents in the cities in which these contests took place range from 9.8% (Irvine, 2010) to 81% (Madera, 2020), with the sample's average being 48.5% Latino. Of the 103 contests, 48 occurred in majority Latino municipalities.
Mayoral contests in sample over time and by candidate partisan status.
Mayoral Contests in Sample by Candidate(s) Partisanship and Ethnicity.
A candidate's decision to reveal their partisanship is, of course, not random. Candidates may differ systematically in their incentives or opportunities to signal their party affiliation. To explore this, I examine observable correlates of partisanship revealing behavior in Appendix Table A2. The analysis indicates that candidates are somewhat more likely to reveal their partisanship in larger cities, in places with fewer Latino residents, and when they are incumbents. Contests in which at least one candidate reveals also tend to be decided by larger margins. These patterns are consistent with the idea that revealing may be more common in higher-profile or less competitive settings. Importantly, the analyses presented in the Results section include city-year fixed effects, which absorb much of this contest-level variation, helping to mitigate concerns about confounding from these contextual differences.
I draw precinct-level registration data from California's Statewide Redistricting Database. This database contains the number of individuals registered to vote by election precinct for each even-year primary and general election. For elections occurring in odd years, I used the most recent even-year registration data available. The data also breaks down the number of registered voters in each precinct by party using public registration information and by ethnicity by matching each registered voter's surname with the U.S. Census Bureau's Passel-Word list of Spanish surnames.
The primary measured implication of ethnic voting I am interested in is a positive relationship between the share of registered voters that are Latino and share of votes cast toward Latino candidates in a precinct. Because my unit of analysis is a voting precinct rather than an individual voter, I am limited in the types of inferences I can make—a problem of ecological inference. If precincts with larger concentrations of Latino voters cast far more votes for Latino candidates, I cannot conclude with certainty that individual Latinos are more likely to vote for co-ethnic candidates, just that heavily Latino precincts demonstrate high rates of support for Latino candidates. In addition, unobserved heterogeneity among Latino registrants—such as differences in age, socioeconomic status, or political engagement—could influence voting behavior in ways not captured at the precinct level. However, precinct-based analyses can still offer valuable insights on the ethnic voting relationship because the precinct is the smallest unit at which we have access to the actual vote counts for each candidate. This granular data allows me to detect patterns of electoral support that align with hyper-local demographic characteristics, even if it does not perfectly map onto individual-level voting behavior.
Figure 3 visualizes the distribution of Latino voter registration across the precincts in the data as well as the distribution of Latino voters by political party. The top panel shows that the distribution of Latinos across precincts is right-skewed, indicating that in most precincts, the share of Latino voters is relatively low. On average, about 25% of a precinct's registered voters are Latino. Despite the right-skew, there is still a decent number of precincts in which Latinos make up the plurality or majority of registered voters. The bottom left panel shows the distribution of registered Latinos that affiliate with the Democratic Party across precincts. This distribution is more heavily right-skewed, with most precincts containing small proportions of Latino Democrats. The bottom right panel shows the distribution of registered Latinos that affiliate with the Republican Party, demonstrating that the share of Republican Latinos is minimal across most precincts.
Distribution and partisanship of registered Latinos within precincts.
The data, although extensive, do have certain limitations. The most notable is that the universe of elections analyzed comes exclusively from one state—California—and is limited to municipalities with populations over 50,000. However, California's diverse population and the presence of numerous multiethnic electoral contests makes it an ideal setting for studying ethnic voting behavior. Furthermore, findings observed in larger municipalities can serve as indicators of broader trends in urban areas where ethnic diversity is often most pronounced. It is in these types of cities where multiethnic contests most frequently occur. Although the context is somewhat narrow, the results still deepen our understanding of ethnic voting dynamics in similarly diverse urban settings and lay a foundation for future research. Previous studies examining the trend of ethnic voting in local politics have typically focused on a small number of carefully selected highly-urban cities (Michelson, 2005; Barreto, 2007; Collet, 2005; Hero, 1989). By contrast, this study examines data from 103 contests—every mayoral contest in California cities over 50,000 population with at least one Latino candidate during an eleven-year period—enabling a more comprehensive analysis of ethnic voting trends across a wider range of contexts than previously employed.
Results
A Baseline Trend of Ethnic Voting
I begin by presenting graphical evidence of the voting patterns of the precincts across all contests. Figure 4 presents a scatter plot of the relationship between the share of registered voters that are Latino and the share of votes Latino candidates receive in a precinct. The vertical axis maps the share of votes given to all Latino candidates in an election. Each plotted point represents an individual precinct in an election. Darker points indicate that multiple precincts share similar demographics and shares of votes. A dashed diagonal line indicates where points would be if a perfect one-to-one relationship between the two variables were observed. The solid line and surrounding shaded region visualize the slope of a bi-variate regression between the two variables and its 95% confidence interval.
A general pattern of ethnic voting across all contests; precincts with higher concentrations of Latino voters award more votes to Latino candidates.
The plot indicates that the baseline relationship between the share of registered Latino voters in a precinct and Latino candidate vote share is positive and significant. Latino candidates in a precinct make up of entirely Latino voters are predicted to receive a vote share 57.2 percentage points higher than they would receive in a precinct with no Latino voters.
These results suggest that a general trend of ethnic voting exists across all 103 contests. I now move to testing how revealed partisan information moderates this relationship.
The Moderating Role of Revealed Partisanship
To test whether the strength of the relationship between voters’ ethnicity and Latino vote share changes under different informational contexts, I run a series of regression models that account for whether candidates publicly affiliate with one of the major political parties. These regression models interact the share of Latino voters in a precinct with four indicator variables that designate the presence of a known partisan candidate in an election: two indicators for if the partisanship of a Latino candidate is publicly known, and two indicators for if the partisanship of a non-Latino candidate is known. 2 A positive coefficient on any of the interactions suggests that the partisan information increases the rate at which ethnic voting occurs (precincts with more Latino voters are giving even more votes to the Latino candidate), while a negative coefficient suggests the information diminishes the rate of ethnic voting. Each model includes city-by-year fixed effects. These fixed effects aim to capture any unique conditions in each city across time such as the local political climate, incumbency status of certain candidates, economic conditions, and any other city specific electoral dynamic that changes over time but not directly measured by the independent variables. Table 3 presents the results.
Revealed Candidate Partisan (Particularly Out-Group Candidate) Information Alters the Rate of Ethnic Voting.
Note: Coefficients for “Precinct Share Latino Reg.” indicate the ethnic voting relationship in contests without candidate party information. Interaction coefficients capture how this relationship changes when candidate partisan affiliation is revealed. A positive interaction coefficient suggests increased support for Latino candidates in precincts with more Latino voters; a negative interaction coefficient suggests the opposite. The model does not separately estimate coefficients for the candidate partisanship indicator variables due to fixed effects—these variables do not vary within contests. Standard errors are clustered at the contest level.
* p < 0.05, ** p < 0.01, *** p < 0.001.
Column 1 of Table 3 presents the estimated ethnic voting patterns across the precincts within all 103 contests in the sample. The dependent variable is the share of votes received by all Latino candidates in a precinct. In contests with more than one Latino candidate, the numerator sums all votes received by Latino candidates, while the denominator is the total number of votes cast in the precinct.
The model reveals three key associations. First, after accounting for city-by-year fixed effects, Latino candidates perform better in precincts with larger Latino populations at baseline, indicating a solid pattern of ethnic voting absent party information. Second, the presence of publicly affiliated non-Latino opponents moderates this relationship. When an opposing candidate is a non-Latino known Democrat, Latino support declines in more heavily Latino precincts. By contrast, Latino candidates receive significantly greater support in increasingly Latino precincts when facing a non-Latino known Republican. Third, when Latino candidates publicly affiliate with either of the political parties they do not experience significantly different levels of support as precinct voter demographics shift. Together, these results suggest that while ethnic voting seems to occur in contests absent party information, out-group candidate partisanship, when it is revealed, meaningfully conditions this relationship. For ease of interpretation, Figure 5 visualizes these interaction effects graphically.
Revealed partisanship of non-Latino candidates moderates the ethnic voting relationship.
Column 2 presents the results of a model that restricts the sample to contests with only one potentially viable Latino candidate. These are mayoral contests where only one Latino candidate received more than 15% of the overall votes in the contest. Here, the dependent variable is the share of the vote the viable Latino candidate received in a precinct. This specification enables a cleaner assessment of how precinct-level Latino demographics and public partisan affiliation shape support for a single co-ethnic candidate when there is one seemingly clear option. It also ensures that vote shares are not inflated in contests featuring multiple Latino candidates.
This model's results validate the findings of the first while providing a notable distinction. Here too, we observe a significant ethnic voting relationship in contests absent party information, with that relationship being conditioned by public partisan affiliation of out-group candidates. However, unlike in the full sample, viable Latino candidates who are publicly affiliated with the Republican Party receive significantly less support in more heavily Latino precincts. This suggests that even when a single potentially competitive Latino candidate is on the ballot, Republican affiliation may undermine co-ethnic support in heavily Latino precincts.
Column 3 further narrows the analysis to cases where only two candidates ran for mayor: one Latino, one non-Latino. This specification isolates the clearest head-to-head electoral contexts and limits voters’ choices to one ethnic alternative over another. The dependent variable is the share of votes the Latino candidate received in a precinct.
The results continue to affirm the core ethnic voting pattern. In contests absent party information, a significant positive association between the share of Latino voters in a precinct and the share of votes the Latino candidate receive exists. And, as in prior models, this relationship is moderated by candidate partisanship when it is revealed. In these cases, voters in heavily Latino precincts tend to reject Latino candidates if they are known Republicans and lend increasing support to Latinos when facing against a non-Latino known Republican. These findings suggest that even in the most straightforward electoral match-ups, partisan affiliation—when it is revealed—continues to shape to extent to which the ethnic voting relationship exists.
Together, the three models provide consistent evidence of ethnic voting in contests without partisan cues, with Latino candidates generally receiving greater support in more heavily Latino precincts. However, across all specifications, the availability of candidate partisan affiliation significantly shapes the strength of this relationship. These results underscore that while candidate ethnicity remains a powerful heuristic, its observable effect depends on the informational context of a local contest. 3
Accounting for Candidate Valence
Because public knowledge of a candidate's partisanship may be correlated with other candidate attributes that influence electoral success—such as name recognition, prior political experience, and financial resources—the effects of partisan information estimated in previous models could partially reflect broader candidate valence. To address this, I estimate an additional fixed-effects regression that interacts the share of registered Latino voters in a precinct with indicators for whether a candidate's partisanship is known due to prior candidacy for higher partisan office or because the candidate is the incumbent mayor. 4 These interactions allow for a supplemental test of whether the moderating effects of known candidate partisanship are attributed to the partisan information itself or to candidate valence. Table 4 presents the results. As before, a positive interaction coefficient suggests that the relationship between the share of registered Latinos in a precinct and the share of votes Latino candidates receive is stronger, while negative interactions suggest the opposite.
Highly Latino Precincts Respond to Partisan Information Similarly Regardless of Candidate Valence.
Note: Coefficients for “Precinct Share Latino Reg.” indicate the ethnic voting relationship in contests without candidate party information. Interaction coefficients capture how this relationship changes when candidate partisan affiliation is revealed and when candidate valence varies. A positive interaction coefficient suggests increased support for Latino candidates in precincts with more Latino voters; a negative interaction coefficient suggests the opposite. The model does not estimate an interaction coefficient for Latino Republican incumbents because no Latino Republican was an incumbent in any of the contests in the sample.
* p < 0.05, ** p < 0.01, *** p < 0.001.
The interaction coefficients show that candidate valence most strongly moderates the ethnic voting relationship when candidates are well-known Republicans. The coefficient for the share of Latino registrants in a precinct is positive and significant, indicating that in contests where candidate partisanship is obscured, precincts with more Latino voters award more votes to Latino candidates. This relationship does not change significantly when a Latino candidate is a known Democrat, even if that Democrat previously ran for higher partisan office or is the sitting incumbent. Similarly, the presence of a low-valence Latino Republican candidate does not significantly alter the ethnic voting relationship.
However, when a Latino Republican with prior candidacy for higher partisan office is on the ballot, the ethnic voting relationship reverses direction: the significant negative interaction coefficient (−0.353) means that, in these contests, Latino candidates receive fewer votes in precincts with higher concentrations of Latino registered voters. In other words, heavily Latino precincts shift their support to non-Latino competitors when facing a high-valence Latino Republican.
Valence also moderates the effect of partisan information for non-Latino Republican candidates. The negative and significant interaction for non-Latino known Democrats suggests that increasingly Latino precincts prefer a known Democratic candidate over a non-partisan Latino. This effect does not differ substantially when the non-Latino Democrat is high-valence, either through prior candidacy or incumbency. Conversely, the presence of a known non-Latino Republican candidate amplifies the ethnic voting relationship, with Latino precincts giving even more support to Latino candidates. This effect is attenuated when the non-Latino Republican is a prior candidate, nearly halving the magnitude of the effect.
Importantly, the inclusion of candidate valence indicators—prior candidacy and incumbency—does not substantially alter the interpretation of the main results presented earlier. The effects of partisan information on the ethnic voting relationship remain significant and directionally consistent, independent of candidate valence. In other words, while candidate recognition and experience can moderate the strength of the relationship in specific contexts, the overall pattern persists: public knowledge of candidate partisanship continues to shape the share of votes Latino candidates receive in increasingly Latino precincts in predictable manners, regardless of whether the candidate is more or less well-known. Precincts with more Latino voters tend to support Latino candidates more in the absence of partisan information, even more when the out-group competitor is a known Republican, and less when the out-group competitor is a known Democrat.
Precinct Partisan Composition Shapes Support for Latino Candidates and Publicly Affiliated Out-Group Opponents
The results thus far suggest that Latino voters’ general alignment with the Democratic Party may shape how partisan information conditions ethnic voting—evidenced by the rejection of Republican candidates and greater support for non-Latino Democrats. To test the extent to which these results reflect Latino Democrats seeking out and voting for Democrats while rejecting Republicans, I estimate an additional model that examines whether the ethnic and partisan composition of a precinct's voters predicts Latino candidate support in differing informational contexts.
Rather than focusing on the overall share of Latino voters, this model uses the net Democratic Latino registration rate—defined as the difference between the share of a precinct's registrants that are Latino and registered Democrat and the share that are Latino Republicans—as the key independent variable. This allows for a more direct test of whether support for co-ethnic candidates varies depending on partisan alignment. The model also controls for the partisan balance of non-Latino registrants in each precinct, accounting for the possibility that precincts with more Latino Democrats may also have larger populations of non-Latino Democrats. Table 5 presents the results.
Precincts with More Latino Democrats Respond to Candidate Partisan Information, Giving More Support to Latino Candidates in the Face of a Republican Out-Group Competitor and Preferring Out-Group Democrats to Nonpartisan Latinos.
Note: The coefficient for “Net Democratic Registration (Latino)” captures the support Latino candidates receive with increasing Latino Democratic Registration in contests where candidate partisan-ship is not revealed. Interaction coefficients capture how this relationship changes when candidate partisan affiliation is revealed. A positive interaction coefficient suggests increased support for Latino candidates in precincts with more Latino Democrats than Republicans; a negative interaction coefficient suggests the opposite. The model does not separately estimate coefficients for the candidate partisanship indicator variables due to fixed effects—these variables do not vary within contests. Standard errors are clustered at the contest level.
* p < 0.05, ** p < 0.01, *** p < 0.001.
The results reinforce and extend the earlier findings in two key ways. First, at base-line, Latino candidates receive greater support in precincts where Latino Democrats increasingly outnumber Latino Republicans—consistent with the idea that co-ethnic voting among Latino Democrats may reflect an assumption that even nonpartisan Latino candidates lean toward the Democratic Party. Second, this relationship is significantly moderated by the partisan affiliation of out-group candidates: Latino candidates perform worse in Democratic-leaning Latino precincts when facing a non-Latino known Democrat, but receive increased support in those same precincts when facing a non-Latino known Republican.
Here too, the interaction terms involving a Latino candidate's own partisan affiliation are statistically insignificant. This may indicate that the presumed baseline alignment between Latino Democrats and Latino candidates hold regardless of whether the Latino candidate is publicly affiliated with a party. Alternatively, voters in precincts with high concentrations of Latino Democrats splitting their votes between multiple Latino candidates in crowded contests, even when one reveals their partisanship, may be diluting any detectable partisan effects. Instead, support is more responsive to the partisan identity of an out-group opponent—a pattern that indicates that, on aggregate, Latino Democrats project shared partisan identity onto nonpartisan co-ethnic candidates but adjust their support downward when a known partisan-aligned out-group alternative is available.
Notably, the coefficient for net Democratic registration among non-Latinos, while positive, is small and statistically insignificant. The divergence between the effects of Latino and non-Latino Democratic registration suggests an interesting asymmetry in how the partisan composition of precincts correlates with support for nonpartisan Latino candidates. Precincts with higher concentrations of Latino Democrats tend to offer stronger support to Latino candidates, even in the absence of explicit partisan cues, whereas precincts with more non-Latino Democrats do not exhibit a similar pattern. This may reflect a tendency for Latino Democrats, at least at the aggregate level, to infer partisan alignment from co-ethnic candidates in nonpartisan contests—an inference that does not appear to operate similarly among non-Latino Democrats. These results support the interpretation that assumptions of shared partisanship in the absence of explicit cues are stronger among co-ethnics, and that Latino candidate ethnicity does not signal Democratic affiliation as clearly to non-Latino voters.
How Revealed Partisanship May Interact with Group-Based Conflict
Up to this point, the analyses have tested the observable implications of the theory that suggests ethnic voting is a consequence of a low-information environment. I now turn to testing the implications of the theoretical framework that argues ethnic voting is a product of heightened inter-group conflict for distributional resources in local politics. Specifically, I probe whether partisan information plays less of a role on vote choice in contests where group conflict is likely to be more salient.
As the presence of inter-group conflict is difficult to observe in an aggregated study of numerous elections, I use two measures that aim to capture conditions where ethnic group-based competition for government resources will be likely. The first is a measure of Latino–Non-Latino residential segregation, operationalized via the Index of Dissimilarity (Logan 2013). Residential segregation can serve as a likely proxy for the level of ethnic group-based conflict in city politics because patterns of segregation often reflect a legacy of exclusionary policies that have historically structured political and economic inequality along ethnic lines (Trounstine 2018). And, in contexts where public goods and resources are distributed spatially, increased residential segregation heightens the potential for distributive conflict between groups (Trounstine 2016; Almy 1973; Enos and Celaya 2018). The second measure is an index of socioeconomic disparities between non-Hispanic white and Latino residents in a city. In cities where non-Hispanic whites enjoy disproportionately better socioeconomic outcomes than Latinos, the alignment of economic inequality with ethnic boundaries may heighten perceptions of group threat and foster a politics defined by competition over resources and representation.
I calculate the Index of Dissimilarity for each city-year in the sample using population data from the American Community Survey. This Index represents the proportion of a city's residents who would need to relocate to a different Census tract for each tract in a city to mirror the city's overall demographic composition. In this sample of city-years, the Index of Dissimilarity ranges from 0.07 to 0.56, with a mean of 0.27 and a standard deviation of 0.11.
I construct a socioeconomic disparity index for each city-year using ACS data on median income, poverty rate, educational attainment, homeownership rate, and receipt of public assistance (SNAP) by ethnicity. For each measure, I calculate the gap between outcomes for Latino and non-Hispanic white residents. I then standardize each of these five gap measures and take their average to form a single composite index. The index is directional, with positive values indicating higher than average socioeconomic disadvantage among Latino residents relative to whites, and negative values indicating the reverse.
To assess whether group conflict conditions the influence of partisan information on ethnic voting, I estimate two additional fixed-effects regression models that include triple inter-actions among precinct demographics, indicators for the presence of partisan information, and continuous measures of group conflict (see Table 6). Although the Index of Dissimilarity and the standardized socioeconomic disparity index do not vary within city-years, their interactions with precinct-level demographics and partisan information do. The triple interaction coefficients capture how the relationship between revealed partisan information and precinct demographics changes across levels of group conflict. A positive coefficient indicates that higher segregation or greater socioeconomic disparities amplify the effect of partisan information in more heavily Latino precincts, while a negative coefficient indicates the opposite.
Additional Contextual Measures Aimed at Captured Inter-Group Conflict Further Condition Ethnic Voting Relationship.
* p < 0.05, ** p < 0.01, *** p < 0.001.
The results suggest that residential segregation moderates, rather than overturns, Latino voters’ partisan preferences. In less segregated cities, high Latino precincts respond by showing pronounced resistance to Latino Republicans. As segregation increases, the penalty attached to Latino Republican candidates is reduced—potentially a consequence of segregation making ethnic group divides more salient than partisanship or ideology. This reduction in penalty is not enough, however, to make Latino candidates more appealing than non-Latino Democrats. In other words, segregation softens but does not erase partisan divides: it makes ethnicity somewhat more salient, yet out-group Democratic partisanship still remains a strong cue.
Turning to the socioeconomic disparities model, a similar moderating effect on the role of Latino Republican information emerges. In cities with lower socioeconomic disparities between non-Hispanic whites and Latinos, as precincts become more homogeneously Latino support for known Republican Latino candidates decreases. This relationship reverses, however, in contexts where socioeconomic disparities are higher. Also notable is how Latino candidates fare better in precincts with more Latino voters when competing against a non-Latino known Republican candidate, regardless of the level of socioeconomic disparity.
For ease of interpretability, Figure 6 visualizes how residential segregation and socioeconomic disparities condition the ethnic voting relationship in contests where a Latino candidate is a known Republican. For both measures, the negative relationship between Latino candidate vote share and concentration of Latino voters in a precinct reverses as the conditions that might foster group-based political conflict increase.
Known Latino republican candidates are predicted to perform better in more Latino precincts in contexts where group conflict Is likely higher.
These findings provide suggestive evidence that, in contexts where conditions that are likely to predicate inter-group distributional conflict are more present, Latino voters may prioritize co-ethnic solidarity over information gleaned from partisan cues. This is especially the case when a Latino candidate has revealed their Republican partisanship. At the same time, out-group candidate partisan information still meaningfully moderates the ethnic voting relationship in these contexts—with heavily Latino precincts more strongly favoring Latino candidates in the face of a Republican out-group candidate regardless of the level of socioeconomic disparity, and with heavily Latino precincts still more strongly favoring non-Latino known Democrats over nonpartisan Latinos in more segregated cities.
Discussion
While a substantial body of research documents the role that co-ethnicity plays in shaping vote choice in local, low-information elections (Crowder-Meyer, Gadarian and Trounstine 2020; Hajnal and Trounstine 2014; Barreto 2007; Matson and Fine 2006; McDermott 1998), this study advances the literature by looking for observational evidence of when co-ethnic cues are most influential—particularly in the presence or absence of partisan information and environments prone to more or less distributional conflict between ethnic groups. Two theoretical frameworks simultaneously account for the prominence of ethnic voting at the local level. One argues that because local elections operate within low-information environments that lack party cues, voters rely on other easily available information shortcuts to approximate candidate policy preferences and ideology. A prominent shortcut voters seem to employ connects candidate co-ethnicity to shared preferences. At the same time, others have pointed to the increased salience of group competition at the local level, where policy battles typically cover the allocation of scarce public resources rather than ideological divides, to contend that ethnic voting can be a manifestation of a unique role ethnicity plays in defining local politics.
This article first tests a core implication of the heuristic perspective: that voters should rely less on ethnic cues when more informative signals like partisan affiliation are available. I leverage the fact that, despite the official nonpartisan nature of California local mayoral elections, some candidates publicly affiliate themselves with a political party. Drawing from precinct-level election results in 103 multiethnic mayoral elections from an 11-year period, I find a strong positive relationship between the concentration of Latino voters and the share of votes Latino candidates receive in a precinct. However, the presence of information on candidate partisanship in a contest greatly conditions this relationship.
When running against a non-Latino candidate publicly affiliated with the Democratic Party, the share of votes nonpartisan Latino candidates receive in increasingly Latino precincts declines. Conversely, increasingly Latino precincts award even more votes to nonpartisan Latino candidates in the face of a non-Latino Republican competitor. Latino candidates publicly affiliated with the Republican Party also see declines in support in increasingly Latino precincts when those candidates are the only viable Latino candidate in the contest.
These effects appear especially in precincts with higher concentrations of Latino Democrats. In precincts with more Latino Democrats, nonpartisan Latino candidates earn consistently higher levels of support. Support for these candidates in these precincts is even greater when running against a known Republican non-Latino candidate, yet declines in response to a known Democrat non-Latino. Interestingly, there is no significant relationship between the share of non-Latino Democratic voters in a precinct and the share of votes Latino candidates receive—suggesting that while Latino Democrats, at aggregate, appear to project shared partisanship onto Latino candidates in the absence of party cues, non-Latino voters do not respond similarly.
The article further tests an implication of the group-based conflict perspective: that in contexts where local politics is more likely to be defined by ethnic, distributional conflict, the salience of a need for co-ethnic solidarity will be the primary driver of vote choice. I find that in contexts where conditions likely to preclude group-based conflict are more prevalent, the relationship between the concentration of Latino voters in a precinct and Latino candidate vote share in contests where a known Latino Republican is on the ballot reverses from positive to negative.
What might the implications of these findings be for candidate strategy? The clearest answer might be that, except for in cases where group-based distributional conflict is highly salient, Latino Republican candidates would be better off keeping their partisanship obscured if they hope to win the votes of co-ethnic voters. Latino Democrats tend to do no better in high Latino precincts than Latino nonpartisan candidates, so their partisan disclosure is not much of a strategic advantage. On the other hand, non-Latino candidates may see gains among Latino voters by publicly signaling Democratic partisanship.
Taken together, these results suggest that the availability of partisan cues in nonpartisan elections plays a major role in shaping the relationship between ethnicity and vote choice. That being said, the study does not come without its limitations. Because the data relies on contests from cities with over 50,000 residents in California, I am unable to definitively say whether these results would hold in other environments. At the same time, the landmark studies that identify the trend of ethnic voting have historically limited their analyses to small selections of either one or a few handpicked highly urban environments at a time (Barreto 2007; Collet 2005; Hero 1989; Michelson 2005). This study incorporates the results of a substantially large number of contests (103) over time, the largest set used to study this phenomenon to my knowledge. This broader scope, although limited to one state, provides a stronger foundation for drawing conclusions about ethnic voting trends by offering insights that are less likely to be confined to the peculiarities of specific urban environments.
It is also important to note that these elections took place during a period when the California Voting Rights Act and subsequent debates about structural reforms to municipal governance brought issues of minority representation to the forefront of local political discourse. This heightened attention to descriptive representation likely increased the salience of ethnicity in local elections, potentially dampening the relative influence of partisan information. As a result, any evidence that partisan cues conditioned ethnic voting relationships in these contests may represent a conservative estimate of their broader effect—partisan information could play an even stronger role in contexts where ethnic representation is less central to the political conversation.
Another potential limitation relates to the endogeneity of public partisan disclosure. A candidate's decision to publicly affiliate with a political party in an officially nonpartisan race is not random and may reflect strategic considerations by candidates. Indeed, candidates more often reveal their partisanship in larger cities with fewer Latino residents and when they are incumbents (see Appendix Table A2). While this indicates that the availability of partisan cues is not entirely exogenous, the use of city-by-year fixed effects helps mitigate concerns that such contextual factors bias the main estimates. Future research could more directly study the strategic decision to reveal partisanship and its implications for voter behavior.
As with all observational studies, caution is warranted when interpreting these results as causal. That said, the goal of this analysis is theory-driven: to test how the availability of partisan information moderates the role of ethnicity in vote choice. While the study cannot fully randomize partisan signaling, the design mitigates key sources of confounding. City-by-year fixed effects absorb contest- and context-level variation. This feature provides partial leverage by allowing for theory-testing while acknowledging that unobserved factors could still influence outcomes. The strength of this approach lies in its ability to detect systematic variation in ethnic voting across multiple informational contexts and electoral contests—an empirical question poorly suited to experimental manipulation but well suited to a comparative, context-sensitive observational design such as this.
Although local elections attract less attention from the media, voters, and often political scientists (Trounstine 2009), they constitute a large share of the arenas where democratic decision-making in the United States takes place. And while being closest to voters in terms of geographic proximity and everyday governance, these contests often take place in highly informationally constrained environments. The findings here suggest that in the absence of party labels, voters may rely on ethnicity as a heuristic—one that, while often meaningful, is imperfect and potentially distorting. Over-reliance on such cues could lead voters to support candidates who do not best align with their preferences or to overlook viable representatives from different ethnic backgrounds. In terms of studying local political behavior, this study highlights the need to consider not just candidate supply and voter demand, but the informational conditions that shape how voters assess those candidates.
Supplemental Material
sj-docx-1-uar-10.1177_10780874251414062 - Supplemental material for When Cues Collide: Partisan Signals and the Dynamics of Ethnic Voting in Nonpartisan Local Elections
Supplemental material, sj-docx-1-uar-10.1177_10780874251414062 for When Cues Collide: Partisan Signals and the Dynamics of Ethnic Voting in Nonpartisan Local Elections by E. Grant Baldwin in Urban Affairs Review
Footnotes
Ethical Approval and Informed Consent Statements
This study uses publicly available aggregate election data that contain no individual-level or identifiable information. As such, ethnical approval and informed consent were not required.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental material for this article is available online.