Introduction
During the 2016 presidential primary season, then-candidate Hillary Clinton appeared on the popular morning radio show, the Breakfast Club, to discuss her candidacy, policy positions, and to appeal to potential Black voters. An interesting turn happened near the latter part of the interview when co-host Angela Yee asked Clinton, “What’s something you always carry with you?” Pondering that question, Clinton answered, “Hot sauce.” This appeared to be a reference to Beyonce’s 2016 record ‘Formation’ thus prompting Charlemagne Tha God to say, “Now listen. I just want you to know people are going to see this and say, ‘Okay, she pandering to Black people,”’ to which Clinton replied, seemingly in jest, “Okay… Is it working?” Clinton quickly clarified that she carried hot sauce due to health reasons, but the interview gained attention, hurting her standing with the Black community. Throughout her campaign, Clinton was plagued with accusations that she was not genuine and instead, only pandering for Black votes. What could she have done differently? How can politicians combat perceptions of pandering?
Here, we seek to better understand if perceptions of political pandering among Black voters can be mitigated by elite endorsements. We draw on the literatures on racial appeals, Black political ideology, and trust to develop expectations that White politicians’ attempts to cater to the Black community may not be successful if they are not paired with an endorsement from a credible source. Without alignment to a credible source like a Black elite, Black voters will likely perceive the positive, explicit appeal as disingenuous pandering rather than genuine commitment.
We test this expectation using an original preregistered experiment with approximately 400 Black Americans. We expose respondents to one of four news stories that cover the supposed candidacy of a White politician for governor. The respondents either read about (1) the candidate’s positive, explicit racial appeal to Black voters, (2) the endorsement the candidate received from a Black elite, (3) both the candidate’s pro-Black appeal and endorsement, or (4) coverage of the candidate without reference to race (the control).
We find, contrary to our expectations, that all conditions increase Black voters’ positive assessment of the politician, relative to the control—and that there is no difference among the treatment conditions themselves. In other words, a positive racial appeal and an endorsement both increase willingness to vote for the candidate and increase positive evaluations of him (i.e., that is he authentic, trustworthy, would support people like them, and that he stands with the Black community). There is no difference, though, between the effect of a positive racial appeal and an endorsement on evaluations of the White politician.
To better understand this lack of difference, we go on to conduct a qualitative analysis of the Black voters’ written responses to the treatment. We find that the White candidate is seen as pandering to the Black community in all conditions, including the control. This sheds light on the null effects between the treatment conditions by clarifying that it was not only the positive, explicit racial appeal that was seen as pandering—it was everything. A White candidate attending a church in the context of running for office is inherently seen as pandering. A positive racial appeal is not necessary to amplify that sentiment, and an endorsement does not successfully mitigate it.
When it comes to heterogeneity within the Black electorate, we do find that Black voters respond to positive, explicit appeals and endorsements differently based on their political ideology and level of linked fate. Black conservatives support the White politician in all conditions where a positive explicit racial appeal is present, but they withdraw support in the endorsement-only condition. Somewhat similarly, Black respondents with low levels of linked fate are most likely to trust the candidate when he makes a positive, explicit racial appeal and least likely to trust him in the endorsement-only condition.
These findings add to our understanding of the ways that campaign strategies are received by Black voters. While positive, explicit racial appeals and endorsements can both function to increase support for a candidate, relative to an event that employs neither (the control), they still may be seen as pandering. Below, we outline our broad theoretical framework that shapes our expectations, describe the preregistered survey experiment, present the findings, and conclude with directions for future research.
Racial Appeals
Racial appeals are a longstanding feature of American political campaigns. Racial appeals are references to race that are intended to bring racial considerations to mind when evaluating candidates or policies. These can be explicit, which are characterized by the use of racial nouns (like “White,” “Black,” or “race”) or implicit, which are meant to be deniable, like coded language (e.g., “inner city” to cue “Black Americans”) (Mendelberg Reference Mendelberg2001).
Politicians engage both in negative racial appeals, disparaging a minoritized group to appeal to hostile racial attitudes of (often, White) voters, and positive racial appeals, which cater to a particular racial group with the intention of garnering support from that community (McIlwain and Caliendo, Reference McIlwain and Caliendo2011). While much of the literature has focused on the way that White racism is mobilized through negative racial appeals (Christiani Reference Christiani2023; Hutchings and Jardina, Reference Hutchings and Jardina2009; Hutchings et al., Reference Hutchings, Walton and Benjamin2010; Mendelberg Reference Mendelberg2001, Reference Mendelberg2008; Nteta et al., Reference Nteta, Lisi and Tarsi2016; Reny et al., Reference Reny, Valenzuela and Collingwood2020; Valentino et al., Reference Valentino, Hutchings and White2002), some have also considered the way that politicians cater to minoritized communities through positive racial appeals as well (Burge et al., Reference Burge, Wamble and Laird2020; Holman et al., Reference Holman, Schneider and Pondel2015; McIlwain and Caliendo, Reference McIlwain and Caliendo2011; Stout Reference Stout2020; White Reference White2007; Zárate Reference Zárate2023; Zárate et al., Reference Zárate, Quezada-Llanes and Armenta2024). Indeed, Christopher T. Stout (Reference Stout2020) argues, in the increasingly racially diverse United States, politicians will need to make positive, explicit racial appeals to garner votes from racially minoritized communities and White voters with more egalitarian racial attitudes. The relevance of this question, then, amidst a diversifying United States and increasingly polarizing racial attitudes (Engelhardt Reference Engelhardt2023) is paramount.
Here, we focus this study exclusively on positive, explicit racial appeals directed at Black voters. Thus, we center our study on statements from politicians that use racial nouns (making them explicit), but in a complimentary, positive manner. For example, politicians are using positive, explicit racial appeals when they say that they “stand with the Black community” or that “Black Lives Matter.” This is explicit because the politician is clearly identifying race (by saying “Black”) and it is positive because they are characterizing the group favorably.
Positive explicit racial appeals can be tricky, though, as they may be perceived as disingenuous pandering for votes. Such a perception could lead to backlash among the electorate. For the purpose of this study, backlash is defined as Black voters being less likely to support a White candidate and/or seeing them as inauthentic. When the racial appeal does not garner the expected positive effect, the backlash effect occurs, resulting in a net negative for the candidate (Hersh and Schaffner, Reference Hersh and Schaffner2013). This kind of reaction is common. The anecdote opening this manuscript illustrated an instance where Clinton’s attempt to appeal to the Black community was seen as inauthentic, ultimately hurting her popularity.
Positive Explicit Racial Appeals or Pandering?
Recent work has explored when positive racial appeals are received as pandering. Camille D. Burge and colleagues (Reference Burge, Wamble and Laird2020) argue that “identity-based targeting that relies on stereotypes to address and connect with Black voters will be viewed negatively as Blacks will perceive these advertisements to be pandering to the Black community” (p. 426). They test this proposition with a survey experiment, and they find that Black respondents reacted most negatively to the campaign advertisement that targeted the Black community with rap music. Using rap music to reach out to the Black community resulted in perceptions of pandering and inauthenticity, and a decline in support (Burge et al., Reference Burge, Wamble and Laird2020).
Burge and colleagues (Reference Burge, Wamble and Laird2020) point to the invocation of stereotypes as a primary mechanism that results in perceptions of pandering. This makes sense, as stereotypes homogenize an entire group and assign particular characteristics to that group (Bouchard Reference Bouchard2022; Hainmueller and Hopkins, Reference Hainmueller and Hopkins2014; Peffley and Hurwitz, Reference Peffley, Hurwitz, Borgida, Federico and Sullivan2009; Pérez Reference Pérez2010; Stangor et al., Reference Stangor, Jhangiani and Terry2014). Unfortunately, White candidates have a long history of deploying racial stereotypes. Often, they have evoked these stereotypes to mobilize White resentment (Hutchings and Jardina, Reference Hutchings and Jardina2009; Mendelberg Reference Mendelberg2001; Valentino Reference Valentino1999; Valentino et al., Reference Valentino, Hutchings and White2002). However, even when playing into those racial stereotypes is meant to appeal to the group, like with the rap music experiment (Burge et al., Reference Burge, Wamble and Laird2020), they can often fall flat, with the politician’s caricatures making them appear disingenuous. Examples of these miscalculations abound. In 2020, then-presidential candidate Joe Biden collaborated with battle rappers, DNA and Charlie Clips, to develop a Get Out the Vote battle rap PSA clearly targeted at Black Americans in urban areas. Trying to look “cool,” engaging with rappers to mobilize Black voters, as an elderly White man, fell flat.
However, it is not only the use of stereotyping that can transform a positive racial appeal into pandering. Other factors matter as well. Marques G. Zárate (Reference Zárate2023) argues that the expectations for the group that the politician is a part of matter immensely for whether their actions are seen as pandering. He argues that perceptions of pandering are tied to whether a politician’s actions are surpassing or falling short of expectations that voters hold for their group. When a politician is seen as making a significant investment into the community to which they want to appeal, they can be seen as authentic rather than pandering (Zárate et al., Reference Zárate, Quezada-Llanes and Armenta2024).
Finally, shared racial identity (or a lack thereof) is often a strong cue for how authentic a politician is perceived when they say that they share concerns of the group of voters to which they are trying to appeal (Bobo and Gilliam, Reference Bobo and Gilliam1990; Colburn and Adler, Reference Colburn and Adler2001; Gay Reference Gay2001; Philpot and Walton, Reference Philpot and Walton2007; Swain Reference Swain1993). Michael C. Dawson’s (Reference Dawson1994) canonical work demonstrates that Black voters, in particular, have a sense of linked fate—that what happens in their individual life is strongly tied to what happens to their racial group. Black candidates, then, as opposed to White candidates, are likely to be seen as more authentic in their appeals to the Black community. In particular, White candidates have a history of reaching out to the Black community for votes, but then abandoning the group’s interest once in office (Smith Reference Smith1996). Thus, White candidates are likely to be seen as pandering when reaching out to the Black community on the campaign trail—even, and perhaps especially, when professing to support the Black community (McGraw et al., Reference McGraw, Lodge and Jones2002).
Thus, recent work demonstrates that there are at least three key factors driving perceptions of pandering: (1) the use of stereotypes, (2) pre-conceived expectations of the politicians’ identity group, and (3) racial identity. Racial appeals that rely on stereotyping, confirm pre-conceived biases about the candidate’s identity group, and/or come from a politician of a racial out-group are more likely to be seen as pandering. While racial appeals are highly contextual and there are multiple factors that matter, these three characteristics have been identified as particularly meaningful in contemporary empirical work.
Defining Political Pandering
For this study, we draw on Zárate (Reference Zárate2023) and Burge et al. (Reference Burge, Wamble and Laird2020) to define political pandering as a form of insincere or strategic communication in which a candidate or politician makes superficial explicit appeals to a particular group to gain short term favor and electoral support, without making a deeper commitment to the values or concerns of the targeted group. Pandering may include one or more of the characteristics identified above: (1) use of stereotypes, (2) failing to meet or exceed expectations, and (3) out-group racial identity. Pandering differs from authentic appeals, which may be seen as rooted in concrete policy proposals (Canes-Wrone et al., Reference Canes-Wrone, Herron and Shotts2001; Jacobs and Shapiro, Reference Jacobs and Shapiro2000) and/or coming from a politician that has a longstanding, trusted relationship with the group.
Endorsements
One way that White politicians often try to increase their credibility among Black voters is through the solicitation of endorsements from Black elites. Endorsements are when a person or other source (e.g., civic organization, newspaper) indicates their support for something or someone. Here, we are focused on endorsements of politicians by other elites (e.g., community, religious, or civic leaders).
Endorsements are effective at garnering votes (Benjamin Reference Benjamin2017a,Reference Benjaminb; Benjamin and Miller, Reference Benjamin and Miller2019; Boudreau et al., Reference Boudreau, Elmendorf and MacKenzie2019; Lucero and Robles, Reference Lucero and Robles2024) and can help voters make informed decisions when they do not have full information about policy particulars (Grossman and Helpman, Reference Grossman and Helpman1999) or candidates. This is especially true in lower-information environments, such as state and local elections or for ballot initiatives (Benjamin Reference Benjamin2017b; Gerber and Phillips, Reference Gerber and Phillips2003; Krebs Reference Krebs1998; Lupia Reference Lupia1994). Ethnic group endorsements at the local level can actually increase candidate support above and beyond ideological congruence (Boudreau et al., Reference Boudreau, Elmendorf and MacKenzie2019) and lead to support for policies that would benefit ethnic out-groups (Lucero and Robles, Reference Lucero and Robles2024). This indicates the power that elite endorsements can have, especially at the local level.
Endorsements are powerful because they act as a “signal or cue… that a particular candidate will best represent their interests” (Benjamin and Miller, Reference Benjamin and Miller2019, p. 644). Benjamin (Reference Benjamin2017b) finds, through extensive experimental analyses, that Black Americans, in particular, rely on endorsements when selecting candidates—to a greater degree than Latinx voters, for example. Such endorsements can be particularly important for achieving cross-racial coalition building, as they serve as a signal that a candidate, even if they do not share a racial identity with the voter, is going to advance the voter’s group’s interests while in office (Benjamin Reference Benjamin2017b). While there can be backlash from voters when Black elites are seen as too closely associated with the White community (i.e., seen as “selling out” or as “Uncle Toms”), these accusations are largely remedied if the White candidate being endorsed is a Democrat as opposed to a Republican (White and Laird, Reference White and Laird2020).
Endorsements are so powerful for Black voters in part because Black Americans have historically been a fairly united political force, with high levels of group consciousness (Dawson Reference Dawson1994; Harris-Perry Reference Harris-Perry2004; Tate Reference Tate2004; White and Laird, Reference White and Laird2020). This grows out of Black Americans’ reaction to racist exclusion and oppression by White Americans throughout U.S. history. They have pushed back against this exclusion by building social movements and schools of thought that oppose racial oppression (Browning et al., Reference Browning, Marshall and Tabb1984; Dawson Reference Dawson2001; Gause Reference Gause2022; Gillion Reference Gillion2013; McAdam Reference McAdam1999; Toure and Hamilton, Reference Toure and Hamilton1967; White and Laird, Reference White and Laird2020). Indeed, the Black experience of living under racial oppression, particularly one that resulted in racial segregation, led Black Americans to adopt high levels of group unity (White and Laird, Reference White and Laird2020).
While it is clear that this unity in struggle against racial oppression has resulted in high levels of group consciousness, there are competing theories about why such a link has emerged. Dawson’s (Reference Dawson1994) theory of linked fate argues that Black Americans are united because what is best for the (racial) group often translates to what is best for the individual. In other words, Black Americans use the status of their racial group as an efficient way to infer the status of their own individual prospects. Given the way that Black Americans have been oppressed collectively throughout U.S. history, this heuristic functions well to provide information about particular political choices, even when information is fairly low (Dawson Reference Dawson1994). Other work has pointed to the role that social norms rooted in Black liberation politics have played to create a unified (pro-Democratic) voting bloc (White and Laird, Reference White and Laird2020). In Ismail K. White and Chryl N. Laird’s (Reference White and Laird2020) account, it is not an individual-level utility heuristic that produces this, but instead a group-level “social constraint” (p. 15). In other words, group norms that are rooted in movements against racial oppression have led to high levels of unity and enforcement of such unity. While the precise way in which Black group unity has emerged is beyond the scope of this manuscript, both of these processes get us to a place in which Black political unity leads to increased levels of support for Black candidates and for the valuation placed on Black elite endorsements.
High levels of group cohesion has meant higher levels of trust among Black Americans for Black politicians than for White politicians. Shayla C. Nunnally (Reference Nunnally2012) finds that Black voters even trust Black Republican politicians more than White Democratic politicians—despite the fact that Black voters, on average, tend to identify as Democrats (White and Laird, Reference White and Laird2020). She suggests that despite Black Republicans’ outlying views about Black group interest (Harris-Perry Reference Harris-Perry2004), they still have a sense of racial identification that White Democrats cannot relate to (Nunnally Reference Nunnally2012; Toure and Hamilton, Reference Toure and Hamilton1967; Walton Reference Walton1985). Thus, even when Black voters do not share the same partisanship with the Black politician, shared racial identity lends more confidence that the politician will look out for the Black community. As a result, an endorsement from a Black elite may be able to increase the credibility of a White politician’s positive, explicit racial appeal (Benjamin Reference Benjamin2017a). It may signal to Black voters that this politician’s appeal to their community has more credibility, as it is aligned with an elite in whom they likely have greater levels of trust.
Expectations
Bringing together the literatures on racial appeals, pandering, and endorsements, we propose that when White politicians attempt to cater to the Black community, their positive explicit racial appeal will not be successful unless they align themselves with a credible source. Without alignment to that credible source, Black voters will likely perceive the appeal as disingenuous pandering rather than genuine commitment (Dawson Reference Dawson1994; Nunnally Reference Nunnally2012). Black voters, through their racial socialization experiences, learn to distrust White Americans the most, as they are the group that has historically perpetrated the most discrimination against them (Nunnally Reference Nunnally2012; Smith Reference Smith1996). Thus, we expect that endorsements from a Black elite will be more successful in garnering support than positive explicit appeals—but that the combination of a positive explicit appeal and an endorsement may lead to the greatest gains in support. This leads to the following preregistered hypothesesFootnote 1 as it relates to a White politician:
-
1. A pro-Black explicit appeal combined with an endorsement from a Black elite will lead to the most support among Black voters
-
2. A pro-Black appeal without an endorsement from a Black elite will lead to either no effect or a backlash effect (decline in support) among Black voters
-
3. An endorsement from a Black elite without a pro-Black appeal will lead to an increase in support (but the support will still be lower than if it were combined with a pro-Black appeal) among Black voters
Despite the fact that a political legacy of uniting to resist racial oppression has led to high levels of group cohesion when it comes to politics, Black Americans are not a monolith. There are significant differences in Black Americans’ politics, including their self-identified political ideology (Jefferson Reference Jefferson2020, Reference Jefferson2023; Philpot Reference Philpot2017) and the interaction of multiple overlapping identities (Cohen Reference Cohen2009; Combahee River Collective 1977; Crenshaw Reference Crenshaw1989; Harris-Perry Reference Harris-Perry2011). Thus, it is unlikely that all Black voters will respond to positive racial appeals or endorsements in the same manner. As a result, we included a section in our preregistration where we noted that we planned to examine heterogeneous treatment effects by individual-level characteristics. Below, we examine not only the overall treatment effects, but also those moderated by linked fate, religiosity, partisanship, ideology, age, and gender.
Data and Methods
To investigate these questions, we turned to an original, preregistered survey experiment of Black Americans (N = 408). We fielded the survey on Lucid, an online survey aggregator whose samples have been demonstrated to track well with U.S. national benchmarks and suitable for experimental research (Coppock and McClellan, Reference Coppock and McClellan2019).
Respondents first answered a series of pre-treatment items, composed of demographic and ideological variables. In particular, respondents indicated their party identification, political ideology, religiosity, ethnicity, and level of linked fate. We include these items pre-treatment as we expect that they may condition the relationship between the treatment and outcome variables. After respondents answered the pre-treatment items, they participated in an attention check. Respondents were not dropped from the dataset due to failing the attention check, as these checks have been demonstrated to track with politically relevant characteristics (Berinsky et al., Reference Berinsky, Margolis and Sances2014).
Once respondents finished the attention checks, they were asked to “Please read the following excerpt from a news article and let us know what you think.” The excerpt was fictional, which they were debriefed about at the end of the survey. Respondents were randomized into one of four conditions: (1) appeal only, (2) endorsement only, (3) appeal + endorsement, or (4) control. The full text of the appeal and endorsement conditions are reprinted in Figure 1. The combined appeal + endorsement is reprinted in Figure 2a and the control is in Figure 2b. The text that was varied between the conditions is underlined in these figures, but was not underlined in the survey for respondents. In each condition, a White politician is appealing to a group of voters. In every condition except for the control, these voters are clearly identified as Black members of a historically Black church. When the politician is making the appeal directly, he speaks to the audience and vows to fight for the Black community now and consistently into the future. When he is being endorsed, he is a guest of the Reverend Floyd Brown Sr. In the endorsement condition, the White politician does not speak at all. Then, in the appeal + endorsement condition, he is a guest of the Reverend who also makes a direct, positive, explicit appeal to the Black community. We chose to use the context of the Black church as it has long been a bedrock institution for political mobilization and socialization within the Black community (Harris-Perry Reference Harris-Perry2004; Walton Reference Walton1985).
Figure 1. The Appeal and Endorsement Treatment Conditions
Note: The text above was not underlined in the experiment. They are only underlined here to indicate the differences between the two treatment conditions.
Figure 2. The Appeal + Endorsement and Control Conditions
Note: The text above was not underlined in the experiment. They are only underlined here to indicate the differences between the two treatment conditions.
Note that this study design does not distinguish between an authentic appeal and a pandering appeal. Instead, we make the assumption that the positive, explicit, racial appeal from the White politician will likely be seen as pandering (absent an endorsement). The design here is to test whether an endorsement can mitigate perceptions of pandering. Based on the literature we reviewed in this article, the racial appeal here fits three characteristics that tend to result in perceptions of pandering. First, the politician is part of a racial out-group, as he is White and the community is Black. Second, the politician is feeding into perceptions that many hold about White politicians—that they are visiting and making promises in exchange for votes. There is no mention in the article that he has a positive track record with the Black community. Third, he is visiting a historically Black church, which may be seen as stereotypical. In the qualitative analyses we conduct, we are able to test the assumption that this appeal is seen as pandering by respondents—and we confirm that it is. Still, while this appeal contains attributes of pandering, it was also designed such that it was not so clearly disingenuous that there would be no possible redemption. Because we want to test whether an endorsement can save a racial appeal from being seen as pandering, we had to construct a racial appeal that was not too strongly pandering to allow for this possibility.
Further, while the appeal + endorsement condition may seem heavy-handed, note that it is not uncommon for politicians to use appeals in combination with endorsements when campaigning. For example, in 2020, then-candidate Joe Biden coupled his extensive policy agenda for Black America (Eligon and Burch, Reference Eligon and Audra2020) with an endorsement from Congressman Jim Clyburn (Strauss Reference Strauss2020). With his positive appeals and elite endorsement, he was able to revitalize his flailing campaign, win the South Carolina Democratic primary and ultimately, the nomination. At the state-level, take former Illinois governor, Bruce Rauner, as an example. He relied on appeals to the Black community, like noting his role at Morehouse College (Geiger and Garcia, Reference Geiger and Garcia2014) and endorsements from Black church leaders (KSDK 2014) in the Chicago area to win the governorship in 2014. White politicians looking for support among the Black electorate often rely on both racial appeals and endorsements.
As a final design note, we did not choose to identify partisanship of the candidate in the experimental conditions. This is never an easy decision, as identifying partisanship can carry assumptions but at the same time, not identifying partisanship can lead to respondents guessing the partisan affiliation of the candidate (though note that our qualitative evidence revealed a lack of consensus about the candidate’s presumed partisanship—we address this further in the conclusion). However, our key aim was to test the effects of racial appeals and race-based endorsements. As we did not have a large enough sample size to vary partisan affiliation and retain statistical power for the key tests we were interested in, we chose to omit references to partisanship. We return to this discussion in the conclusion of this article where we draw out the potential implications of omitting partisan cues.
After respondents read the excerpt from the news article, they were asked: “What is your first impression of the candidate for governor, Mark Smith? Write 2–5 sentences describing what you think about him.” Providing a space for open-ended reflection after a treatment can deepen treatment effects (Condon and Wichowsky, Reference Condon and Wichowsky2020) and the qualitative data can provide an insight into how respondents received the treatment. Then, respondents were exposed to several dependent variables. These variables ask that respondents evaluate the candidate for governor on a variety of dimensions. Respondents were asked:
-
• Based on what you read, if he was running in your state, would you vote for Mark Smith? (5-point scale)
-
• How authentic does the candidate for governor, Mark Smith, seem to be? (5-point scale)
-
• How trustworthy does the candidate for governor, Mark Smith, seem to be? (5-point scale)
-
• Do you think that the candidate for governor, Mark Smith, would support people like you if he got into political office? (5-point scale)
-
• Based on what you read, we want to get your impression of the candidate for governor, Mark Smith using a feeling thermometer. (100-point scale)
-
• Do you think that Smith stands with the black community? (4-point scale)
All of these variables are coded such that higher values indicate a more positive evaluation. Using these dependent variables, we are able to estimate the way that the White candidate’s interaction with the Black community affected respondents’ attitudes about him. Not only do we measure the extent to which the respondents would support the candidate electorally, but we also try to tap elements of pandering by asking about the extent to which the candidate seems genuine, authentic, and trustworthy. We also measure whether they believe that this candidate would support people like them, and the broader Black community, if elected to office. These varied measures allow us to capture a broad picture of the respondents’ evaluation of this candidate. After respondents answered these items, they were debriefed about the true purpose of the survey.
Analyses and Findings
With these measures, we turn to evaluating whether the treatment affected respondents’ perception of the White candidate. Note that a manipulation check confirms that participants understood and retained information regarding the treatments—and balance tests indicate that randomization was successful. See the Appendix for details of these tests.
First, we evaluate the effect of each treatment on the dependent variables by specifying a series of OLS models that use an indicator for treatment condition as the independent variable. These results are presented in Table 1. By and large, the treatment conditions tended to increase the respondents’ evaluation of the candidate, relative to the control condition. Respondents in any of the three treatment conditions were more likely to say that they would vote for the candidate, that he was authentic, that he supports people like them, and that he stands with the Black community. Respondents in the endorsement condition were also more likely to say that he seems trustworthy—and respondents in the appeal condition were more likely to rate him highly on the thermometer. These findings demonstrate that making explicit appeals to the Black community and receiving endorsements do increase the politician’s ratings relative to the control condition, but they do not adjudicate between the tactics.
Table 1. Average Treatment Effects for the Full Sample
***p < 0.01; **p < 0.05; *p < 0.1
To evaluate which strategies are most effective at increasing ratings for the politician, we respecify the OLS models, omitting the control condition. Now, the appeal condition is the baseline and the other two treatments are compared to that condition. These models are reported in Table 2.
Table 2. Average Treatment Effects for the Full Sample; Excluding the Control Condition
***p < 0.01; **p < 0.05; *p < 0.1
Here, there are no statistically significant effects across the conditions. That is, there are no differences in candidate ratings based on treatment condition when the treatment conditions are compared to one another, rather than to the control. The appeal, endorsement, and appeal + endorsement conditions all increased ratings for the candidate, but no condition was more or less effective at this. This amounts to a lack of evidence for hypothesis one, as we do not see a pro-Black appeal combined with an endorsement leading to more support than either an appeal alone or endorsement alone. It also amounts to a lack of evidence for the second hypothesis, as we do not see that a pro-Black appeal without an endorsement elicits backlash. Finally, we do not have evidence for the third hypothesis either, as the endorsement does not increase support to a lesser extent than when combined with an appeal. On the whole, we do not find evidence that these strategies have differential effects on the population, at least when it comes to the full sample.
To investigate this relationship further, we respecified OLS models with interaction effects for variables that we believed may moderate the effect of the treatments on the dependent variables. There were no statistically significant interaction effects between the treatment conditions and age, gender, party identification, identification as a Christian, or level of Christianity (identification as a Christian, plus the amount of church attendance). There was not enough variation to explore whether ethnic identity influenced the treatment (e.g., Black, African, African American, AfroLatino, etc.). However, there were interaction effects for the two moderators: ideology and linked fate. Ideology interacted with the treatment across four of the six models, while linked fate only interacted with the treatment in one of the six models. Below, we present both sets of interactions and their associated plots. First, Table 3 reports from the interaction between treatment condition and linked fate.
Table 3. Linked Fate Moderates the Effects of the Treatment, for the Full Sample
***p < 0.01; **p < 0.05; *p < 0.1
There is only one statistically significant interaction across the six models, and thus, we should take this finding with a grain of salt. Nevertheless, the third model demonstrates a positive and statistically significant interaction between the endorsement + appeal condition and linked fate. To better visualize this relationship, we plot the predicted values of the dependent variable, “trust,” in Figure 3. This demonstrates that for Black respondents with high levels of linked fate, there is no discernible difference between the treatment conditions. However, for respondents low in linked fate, the appeal alone condition leads to the highest trust for the candidate, while the endorsement + appeal condition leads to the lowest levels of trust. Perhaps Black respondents with low levels of linked fate are less persuaded by endorsements as they do not feel as strongly tied to their racial group. However, this would not explain why the endorsement alone condition fares better than the endorsement + appeal condition. Nevertheless, the fact that the appeal alone condition works best for Black respondents with low levels of linked fate is unexpected, as we anticipated that the appeal alone condition may be seen as pandering and thus elicit backlash.
Figure 3. Linked Fate Moderates the Effect of the Treatment Conditions on Evaluations of the Candidate
Next, Table 4 present the results from a series of OLS regressions that interact the treatment indicator with a measure of political ideology—with the control condition omitted. Ideology interacts with the endorsement condition to predict the extent to which the respondent will vote for the candidate, believes he is authentic, and sees him as trustworthy. The endorsement + appeal condition interacts with ideology to predict the extent to which the respondent would vote for the candidate and the extent to which they see the candidate as supporting people like them.
Table 4. Ideology Moderates the Effects of the Treatment, for the Full Sample
***p < 0.01; **p < 0.05; *p < 0.1
To visualize the relationship uncovered here, we plot the predicted values for each of these dependent variables (vote, authentic, trust, and support) in Figure 4. As respondents become more conservative, the endorsement condition (solid line) depresses their evaluations of the candidate. The appeal condition (dotted line) does not have much of an effect; and the endorsement + appeal condition (dashed line) has a slight positive effect on evaluations.
Figure 4. Ideology Moderates the Effect of the Treatment Conditions on Evaluations of the Candidate
Note: higher values of ideology indicate greater conservatism.
When we respecify the models with ideology as a factor variable, so that it not constrained to a linear format, it becomes clear that it is strong conservatives who are driving this relationship (these analyses are presented in the Appendix). That is, the treatments have no differential effect on evaluations of the candidate among Black respondents who identify as liberal—all of the treatments increase positive evaluations of the candidate, relative to the control condition. Relative to each other, there is no difference. However, for Black conservatives, the endorsement alone condition significantly decreases evaluations of the candidate relative to the appeal only and appeal + endorsement conditions.
Before diving into the theoretical implications of this, keep in mind that the number of strong conservatives in this sample is small (N = 16). Conservatives (N = 18) and slight conservatives (N = 15) is similarly small. The vast majority of respondents identify as moderate or middle of the road (N = 180). Compared to conservatives, larger numbers identify as slightly liberal (N = 42), liberal (N = 68) or strongly liberal (N = 69). Thus, the conclusions we draw from this analysis should be contextualized with the notion that this is not a large group of respondents who is moving the needle in response to these treatments.
Nevertheless, one possible explanation is that Black conservatives may view the Black pastor as a likely Democrat—and thus worry that his politics may not align with theirs. Another possibility is that Black conservatives may view the Black elite in the endorsement-only condition as “playing the race card,” a practice they find to be counterproductive and detrimental (Dawson Reference Dawson2001). In contrast, the conditions with direct, explicit appeals may signal greater authenticity and balanced messaging—especially because Dawson (Reference Dawson2001) argues that Black conservatives are not as bothered by racial appeals made by White politicians and because conservatives tend to believe that White sensibilities on the issues of race and economics should be considered (Dawson Reference Dawson2001). While these forces are potentially at work, the low sample size urges caution in over-interpreting any of these results.
On the whole, we find that the treatments largely increase evaluations of the candidate, but only relative to the non-racial control condition. When we investigate whether there are differences between the conditions—that is, differences between only making a pro-Black appeal, only receiving an endorsement, or making a pro-Black appeal and receiving an endorsement, we do not find statistically significant differences in the whole sample. While there are some differences by ideology and linked fate in terms of how Black Americans distinguished between the positive, explicit racial appeal and the elite endorsement, these should be taken with caution as the sample size starts to dwindle with these divisions. By and large, Black Americans do not show different levels of support for the candidate based on whether he uses a positive appeal or an endorsement. Instead, they show similar levels of support in all treatment conditions.
It’s All Pandering: Examining Qualitative Responses to the Treatment
To better understand why there was no overall difference in how Black Americans reacted to the positive, explicit racial appeal and the endorsement from the Black elite, we turn to the qualitative responses that they provided to the treatment they saw. After respondents read the news article, they were asked: “Based on what you read… What is your first impression of the candidate for governor, Mark Smith? Write 2–5 sentences describing what you think about him.” Overall, 87.5% of respondents wrote an intelligible response to this question that directly answered it in some way.
All of the responses were hand coded by the researcher.Footnote 2 To better understand how Black Americans saw this politician, responses were coded for whether they identified that the candidate was pandering to them or their community.Footnote 3 If the response said that the candidate was just visiting for votes or that they were not genuine in their visit, then we coded this response as pandering. Importantly, the response needed to indicate some disingenuity in order to be coded as pandering—responses that were neutral or positive in the description of the politician trying to get votes were not included.
We had initially expected that Black Americans would see the positive, explicit racial appeal by a White politician as pandering, but that the presence of an endorsement from a Black elite would mitigate this perception. However, when we examined the qualitative responses, we see that the White politician was perceived as pandering across all experimental conditions. Specifically, nearly 20% of the sample, across each treatment condition, perceived the politician as pandering in a disingenuous manner—this is high given there was no direct prompt to reflect on whether he was pandering, specifically. The mean proportion of respondents who said that the politician was pandering was 0.20 in the control condition, 0.20 in the appeal condition, 0.12 in the endorsement condition, and 0.23 in the endorsement + appeal condition. According to an ANOVA analysis,Footnote 4 these differences between the conditions are not statistically significant. Reading the qualitative responses, it was clear that there were elements of all of these news stories that made it seen as though the politician was pandering—specifically, that he chose to visit a church (which was perceived as stereotypical), that he had no documented history or track record with the Black community (in line with negative expectations about his racial group), and his own racial identity (White).
The respondents’ own words illustrate these elements best. One respondent in the control condition called out the stereotype directly, saying, “Mark Smith appears to be the typical candidate the [sic] thinks showing up for Sunday service will give a broader appeal to voters of color. Most candidates don’t realize that… seeking voters in church is outdated and stereotypical.” Even though this respondent was in the control condition, they saw the image of the politician so they knew he was White—and they picked up on the stereotypical nature of attending Sunday service to appeal to Black voters, specifically. Another respondent in the appeal condition noted that this White politician was living up to the negative expectations that they have for this group, saying “I feel like he’s pandering and it’s making me mad. These White politicians always come in and plead for the sake black votes saying they’re going to xyz just to lie…” Finally, just like the previous quote, there was a lot of mention of the candidate’s race. While one respondent in the endorsement condition noted that it is possible for White politicians to be genuine, it depends on their track record, writing, “All hype and no facts make me question this White candidate. I am not sure what it means to ‘attempt’ to push back racial and other divisions. Although there are many good White elected officials, we need facts and a solid track record. I also want to know why this church is backing him.” Another respondent (in the appeal + endorsement condition) called out the performativity of the visit, saying “I don’t trust this because it’s a tactic almost equivalent to kissing a black baby.”
This qualitative analysis helps explain the lack of difference across the quantitative analyses of the treatment conditions: the politician was perceived as pandering in every single condition. It is not just the appeal condition that is seen as pandering, as was hypothesized, but instead, all conditions. This explains why there are no differences in effects between the treatment conditions.
Discussion and Conclusion
The findings from this survey experiment indicate that positive, explicit racial appeals and endorsements can increase support for a White politician among Black voters—contrary to our expectations that racial appeals would only increase support when paired with an endorsement. However, the qualitative findings provide a more illuminating picture. Across all conditions, including the control, respondents in the sample perceived the candidate to be engaged in pandering.
It was not only the positive, explicit racial appeal that was seen as pandering—it was everything. Thus, because the politician was seen as pandering in all conditions, the presence or absence of pandering did not significantly influence Black respondents’ evaluations of the candidate. Instead, the treatment conditions still managed to push some respondents toward greater support for the candidate, demonstrating that perceived insincerity by the candidate failed to entirely undermine the positive effects of the political strategies of making appeals and receiving endorsements.
Interestingly, despite the breadth of research demonstrating the importance of co-ethnic endorsements (Benjamin Reference Benjamin2017a,Reference Benjaminb; Benjamin and Miller, Reference Benjamin and Miller2019; Boudreau et al., Reference Boudreau, Elmendorf and MacKenzie2019; Lucero and Robles, Reference Lucero and Robles2024), the endorsement here did not perform better than the positive racial appeal alone. This suggests that, in the eyes of respondents, the endorsement from the Black elite did little to substantiate the candidate’s claims or intentions beyond what the appeal itself accomplished. We think that a couple of factors could be driving this.
First, it could be because the elite was fictional and thus, is not a known, trusted figure for respondents in this sample. An elite cue is more powerful when the cue-giver is known and trusted (Boudreau Reference Boudreau, Suhay, Grofman and Trechsel2020; Lupia and McCubbins, Reference Lupia and McCubbins1998). If the elite does not have credibility with respondents, then they may be seen as someone focused on their own personal gain or status, thus devaluing the endorsement (Dawson Reference Dawson2001; Toure and Hamilton, Reference Toure and Hamilton1967). The lack of an established identity for the pastor may have blunted the effect of the endorsement treatment. In this experiment, we did not choose to use a known, influential figure, because we did not want to complicate the design with respondents’ preconceived attitudes toward that individual. However, this may have contributed to the null effect for the endorsement condition. Future work should build on this to examine whether the endorsement would function to boost support, relative to the racial appeal, when coming from a known and trusted individual.
Second, it could be that a pastor is not an elite who is best positioned to shape attitudes in contemporary Black politics. The open-ended responses to the treatments indicated that respondents saw the reliance on the Black church as stereotypical and, at times, outdated. In our survey, 42.33% of respondents reported that they attend church infrequently (never, less than once a year, once a year). This aligns with research suggesting that the Black church’s influence on political mobilization has diminished due to the growing availability of alternative avenues for social and political engagement among Black Americans (Calhoun-Brown Reference Calhoun-Brown1996) and that the United States is secularizing (Campbell et al., Reference Campbell, Layman and Wayde2024; Navarro-Rivera Reference Navarro-Rivera and Pinn2018; Voas and Chaves, Reference Voas and Chaves2016). Additionally, even frequent churchgoers may not seek to merge their faith with their political behavior if they do not attend a political church (Calhoun-Brown Reference Calhoun-Brown1996; Harris-Perry Reference Harris-Perry2004). Thus, it may be that a pastor—or any religious figure—is not well situated to sway behavior or attitudes today.
Indeed, in contemporary politics, Democrats and Republicans attempt to win endorsements from a broader range of institutions and elites, such as celebrities, community organizers, influencers, and content creators (Goodwin et al., Reference Goodwin, Joseff, Riedl, Lukito and Woolley2023; Jackson et al., Reference Jackson, Nownes and Norton2024; Knoll and Matthes, Reference Knoll and Matthes2017; Nownes Reference Nownes2017; Pease and Brewer, Reference Pease and Brewer2008). While the Black church has historically been a stalwart institution in the Black community, the decline in religiosity, especially among younger groups, is motivating a shift in strategy. From this, we see an important direction for future research in exploring how endorsements from different types of Black elites (e.g., religious leaders vis-à-vis secular community leaders, celebrities, influencers, or activists) affect candidate support. These different endorsements may also function differentially across the Black population—by age, religiosity, ideology, and more. Understanding the impact of these various elites could offer valuable insights into how political candidates should approach endorsements in an increasingly diverse Black community.
One aspect of contemporary politics that this article did not examine is the potential influence of candidate partisanship. As mentioned in the data and methods section, we chose not to identify candidate partisanship in order to preserve statistical power to test the effects of a racial appeal, endorsement, and the combination. However, given the importance of partisanship, especially in something like a statewide gubernatorial race, future work may consider how it interacts with both racial appeals and endorsements to shape support among the Black community. In the qualitative analysis, very few respondents organically mentioned partisanship in their response to the treatment. But of those who did, they were split on whether the candidate was a Democrat or Republican (i.e., one person said he was a Democrat, one said maybe a Republican, two said he was a Republican, and one compared him to President Biden). Thus, there was no consistent perception on the part of the respondents as to his likely partisanship.
Nevertheless, given the broad support for the Democratic Party that exists in the Black electorate, as well as the social norms governing such allegiance (White and Laird, Reference White and Laird2020), varying partisanship would likely have strong effects on support. For example, it may be the case that White Republican candidate would not have the same boost in support in response to appeals or endorsements as White Democrats—at least among the majority of the Black population. However, it is also possible that Black voters could overlook ideological congruence in favor of endorsements when in a lower-information or local election (Boudreau Reference Boudreau, Suhay, Grofman and Trechsel2020). Future work may examine the effects of partisanship at all levels of the process to better understand the conditions under which appeals and endorsements can boost support among the Black electorate.
This study provides evidence that positive, explicit racial appeals and endorsements can increase support for a White politician among Black voters, even when such appeals are perceived as pandering. However, there is more work to be done to understand how the identity of the endorser and/or the particular format of the racial appeal shape perceptions of pandering as well as political support. It is likely that when the appeal is not perceived as pandering, it would fare even better. However, it is unclear whether a White politician can make an appeal that is not seen as pandering, given the long history of such tactics in U.S. politics and race relations. Future work may examine the specific nature of the appeal and how it could work to garner or depress support, when attempting to engage in cross-racial mobilization.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S1742058X25000013.
Acknowledgments
We would like to thank Nathan J. Kelly, Jana Morgan, Shayla C. Nunnally, Christopher Ojeda, Justin Rose, participants of the 2022 American Political Science Association’s annual meeting, and the anonymous reviewers for their thoughtful feedback that greatly improved this project. We would also like to thank the APSA Centennial Grant and University of Tennessee for their financial support in funding this project.
Open access
