Starting as early as the 1980s, political scientists demonstrated that blacks participated more in politics when there was a black candidate on the ballot (Abney and Hutcheson Reference Abney and Hutcheson1981; Griffin and Keane Reference Griffin and Keane2006; Preston Reference Preston1983; Tate Reference Tate1991; Washington Reference Washington2006). For example, several studies showed that blacks were more likely to vote in an election with a black mayoral, U.S. Senate, gubernatorial, or presidential candidate (Abney and Hutcheson Reference Abney and Hutcheson1981; Preston Reference Preston1983; Tate Reference Tate1991; Washington Reference Washington2006). While descriptive politicians’ influence on black political activity differed at various levels of government,Footnote 1 a number of studies demonstrated that blacks were more participatory in races with co-racial candidates. The question that naturally arises from these studies is why does descriptive representationFootnote 2 increase black political activity?
The earliest work addressing this question suggested that the relationship between descriptive representation and higher levels of black political participation could be explained through improvements in feelings of external efficacy when a co-racial politician holds elected office (Bobo and Gilliam Reference Bobo and Gilliam1990). Additionally, one recent study showed that black candidates spur greater levels of black political activity through these candidates’ mobilization efforts among already supportive black voters (Philpot, Shaw, and McGowen Reference Philpot, Shaw and McGowen2009). Most recent studies on the topic argue that the presence of black candidates is not really the cause of higher levels of turnout at all; instead black candidates tend to campaign in districts where blacks are already empowered because they make up the majority of the electorate (Fraga Reference Fraga2016).
While this burgeoning literature improves our understanding of the mediating factors between descriptive representation and black political participation, much of this literature assumes that under the right context, the race of the candidate is inconsequential in altering black political activity. As a result, previous research overlooks the importance of the personal characteristics of the candidate. In addition to the aforementioned factors, it is possible that black candidates increase minority political activity because they are perceived as being more capable, moral, and empathetic than comparable white elected officials. This perception of black candidates as being better on a host of traits may incentivize blacks to participate in politics to improve the standing of co-racial candidates.
In this study, I argue that a significant portion of the relationship between descriptive representation and higher levels of black political activity can be attributed to perceptions among blacks that co-racial candidates are more capable leaders, more empathetic, more moral, and more knowledgeable and intellectually capable than comparable white candidates. Moreover, I test whether perceptions of all of these traits combined can improve our understanding of the link between descriptive representation and participation for blacks. To assess these hypotheses, I begin by exploring how perceptions of candidate traits may influence levels of political participation. I then discuss how perceived candidate traits may play a mediating role between the presence of an African American candidate on the ballot and higher levels of black political activity.
To test the mediating effect of perceptions of candidate traits on descriptive representation and participation for blacks, I explore differences in political attitudes and behaviors in two elections (2008 and 2012) with black candidate Barack Obama and two elections (1992 and 1996) with white candidate Bill Clinton. Using the American National Election Study (ANES) from these years and the mediation analysis approach designed by Imai, Keele, and Tingley (Reference Imai, Keele and Tingley2010) and Imai et al. (Reference Imai, Keele, Tingley and Yamamoto2011), I find that a significant portion of increased black political participation during the Obama years is mediated by perceptions of Obama as holding more positive qualities than Clinton.
I further test my hypothesis using a survey experiment with a hypothetical black and white candidate who are presented as seeking the 2020 Democratic presidential nomination. Similar to the Obama–Clinton analysis, I find that blacks viewed the hypothetical co-racial candidate as being better across a host of traits and that these more positive assessments explain a significant and substantial amount of why blacks are more enthusiastic about voting when the hypothetical black candidate is on the ballot in 2020. The consistency in results across a real-world pair of candidates and among hypothetical candidates in a controlled setting provide substantial support for the claim that the candidate matters in explaining why descriptive representation increases black political participation.
CANDIDATE'S TRAITS AND TURNOUT
Research has long documented the influence of perceptions of candidates’ characteristics on political behavior. Most of this research demonstrates that voters make decisions about whether to support a candidate based on the perceived strengths of the politician such as their leadership abilities, how intelligent they are, and their levels of empathy (Funk Reference Funk1999; Hayes Reference Hayes2010; Kinder Reference Kinder, Lau and Sears1986; Popkin Reference Popkin1994; Rapoport, Metcalf, and Hartman Reference Rapoport, Metcalf and Hartman1989). Given the importance of candidate traits in attracting support, it is not surprising to find that candidates work to cultivate positive images of themselves (Fenno Reference Fenno1978; Fridkin and Kenney Reference Fridkin and Kenney2011). However, pre-existing expectations about candidate traits make this task easier or more difficult depending on the appearance and partisanship of the candidate. Conover and Feldman (Reference Conover and Feldman1989) show that citizens quickly infer information about a political candidate's issue positions and traits with relatively little information about the candidate. In particular, respondents often infer certain qualities about candidates based almost exclusively on candidates’ phenotypical characteristics such as their race (Petrow Reference Petrow2010).
There are good reasons to suspect that individuals may decide whether to participate in politics based on perceptions of the candidate's strengths. Several studies demonstrate that potential voters are more active when the cost of acquiring information about the election is small (Blais Reference Blais2000; Downs Reference Downs1957). Given that individuals can make inferences about candidate traits with relatively little information, trait assessments require less sophistication and effort than an understanding of a candidate's policy platform (Funk Reference Funk1999; Hayes Reference Hayes2010). The low cost of acquiring information on candidate traits provides a shortcut to voters and may incentivize turnout when individuals view the candidate as holding positive qualities.
Moreover, individuals often make decisions about their levels of political activity based on the perceived benefits they may receive should their preferred candidate be elected (Blais Reference Blais2000; Downs Reference Downs1957). When voters perceive neither candidate as capable, nor as working hard for individuals like themselves, they are much more likely to avoid voting (Piven and Cloward Reference Piven and Cloward1988). Additionally, research demonstrates that when voters are exposed to negative information about their preferred candidates, their evaluation of the candidate decreases and this diminishes turnout (Popkin Reference Popkin1994). In combination, when voters perceive candidates as holding more negative candidate traits they may feel less interested in participating in elections to improve the candidate's standing. Conversely, when they perceive the candidate as being strong in a number of areas, they may feel an increased incentive to participate in the electoral process.
WHY DOES DESCRIPTIVE REPRESENTATION MATTER? A CANDIDATE CENTERED PERSPECTIVE
There are some presumed traits of black candidates which would make them more appealing to the black electorate than white politicians (Casellas and Wallace Reference Casellas and Wallace2015; Lewis-Beck et al. Reference Lewis-Beck, Tien and Nadeau2010; Piston Reference Piston2010; Reeves Reference Reeves1997; Williams Reference Williams1990). McDermott (Reference McDermott1998), for example, showed that black candidates, regardless of their stated positions, are generally perceived as being more concerned with advancing minority rights than their white counterparts (see also Tate Reference Tate2001). Similarly, Sigelman et al. (Reference Sigelman, Sigelman, Walkosz and Nit1995) and Williams (Reference Williams1990) demonstrate that black candidates are perceived as being more compassionate and moral than similarly situated white candidates. Black candidates are also perceived by members of the black community as working harder for their political interests and as being more competent (Weaver Reference Weaver2012).
Black voters are generally more approving of co-racial candidates even if these politicians consciously distance themselves from their race (see Harris Reference Harris2012). This is because even if black candidates do not actively discuss racial policies, it is largely assumed that black politicians will address racial inequality more than white candidates (Harris Reference Harris2012). Given perceptions of their efforts to improve conditions in the black community, many blacks may perceive co-racial candidates as being more competent and empathetic (Harris Reference Harris2012; Lerman, McCabe, and Sadin Reference Lerman, McCabe and Sadin2015).
These positive perceptions about black elected officials and candidates may explain why blacks are so much more politically active when a co-racial candidate is on the ballot. Similar to the studies discussed in the previous section, there is evidence which suggests that blacks make decisions about whether to participate based on the level of concern candidates/parties show for individuals like themselves and these candidates’ abilities to implement supportive policies (Dawson Reference Dawson1995; Hajnal and Lee Reference Hajnal and Lee2011; Kaufmann Reference Kaufmann2004; Tate Reference Tate1991). Tate (Reference Tate1991), for example, argued that black voter turnout decreased significantly in the 1988 general election because blacks felt that the Democratic Party did not show enough concern for black voters because of its poor treatment of Jesse Jackson.
However, black voters do not bestow their favor indiscriminately on black candidates. Those who are perceived as being less competent generally fail to increase black participation. Keele et al. (Reference Keele, Shah, White and Kay2017), for example, demonstrate that descriptive representation only increased black political participation when voters felt that the candidate was a capable leader and a competitive candidate (see also Bositis Reference Bositis2011). This research indicates that descriptive representation's effect on black political participation only occurs when voters perceive co-racial candidates as being strong in various candidate traits.
Finally, it is possible that black voters will respond to perceptions of the morality of the candidate when deciding whether to participate in electoral activities. Research demonstrates that voters often view immoral candidates more negatively than politicians who are not marred by political scandals (Doherty, Dowling, and Miller Reference Doherty, Dowling and Miller2011; Funk Reference Funk1996). Perceptions of moral failings in an individuals’ preferred candidate may then serve to depress enthusiasm and turnout. This decline in enthusiasm for perceived immoral candidates may be greatest among African Americans who tend to be more religious than non-blacks and place a premium on the morality of individuals (Lewis Reference Lewis2013). As a result, blacks may be less enthusiastic and willing to participate in politics when their preferred candidate is perceived as being less moral. Based on this research, I hypothesize:
H1
Blacks will view co-racial candidates more positively across a variety of candidate traits than white elected officials. This greater level of confidence in black candidate's traits can explain why black voters are more participatory when there is a co-racial candidate on the ballot.
DATA
To test the candidate trait hypothesis, I use a case study of the two most recent Democratic Presidents, Barack Obama and Bill Clinton. While these candidates differ in some ways, they provide a good comparison for several reasons. First, the two candidates represent the only Democrats elected to the White House in the previous three decades. Second, both candidates followed unpopular Republican incumbents. Third, and possibly most important for the purposes of this study, both candidates are well-liked in the black community.Footnote 3
Using two Presidents who are held in high-esteem in the black community and the ANES Surveys, I examine various forms of political participation across four presidential election cycles (1992, 1996, 2008, and 2012). The ANES asks a number of questions about respondent's political behaviors including whether the respondent voted, attended a campaign rally or event, displayed campaign paraphernalia (such as a bumper sticker, yard sign, or T-shirt), made a monetary campaign contribution, or tried to convince others to support a political party or specific candidate during the election season (i.e. proselytize). I use these measures as dependent variables for my analysis. All five of these political participation variables were coded separately and dichotomously. Respondents were given a score of one for each variable if they engaged in the political activity and a score of zero if they did not (see Supplementary Appendix for wording of each question).
It is important to note that the dependent variables are self-reported measures of various forms of political participation. The self-reported nature of the ANES combined with incentives for respondents to overstate their political activity has often led to large gulfs between reported and actual behavior (see Ansolabehere and Hersh Reference Ansolabehere and Hersh2012; Cassel Reference Cassel2003; Silver and Anderson Reference Anderson and Silver1986). Moreover, many political activities cannot be validated such as attending a rally or displaying campaign paraphernalia. Nonetheless, there are reasons, stated below, to believe that the examination of differences in political participation in this analysis will not be significantly diminished by problems with social desirability response bias.
My dependent variables measure change in political behavior over time using the same survey (i.e. the ANES). While blacks may be more likely to overstate their levels of political participation, a problem would only arise if they were systematically more likely to do so in the 2008 and 2012 surveys. Using voting as an example, McKee, Hood, and Hill (Reference McKee, Hood and Hill2012) demonstrate that the growth in black turnout between 2004 and 2008 using voter-validated turnout data in Georgia showed higher levels of growth in black turnout than non-voter-validated surveys such as the ANES. The authors attribute this result to previous overestimates of black turnout in non-voter-validated surveys. Thus, it is possible that the difference in black political participation when Obama is on the ballot may be under rather than overstated because black respondents in 1992 and 1996 overestimated their participation rates.
Along these lines, Stout and Martin (Reference Stout and Martin2016) demonstrate that blacks were not systematically more likely to overstate voting when they had the opportunity to vote for someone who shares their race. These authors argue that blacks’ may face universal pressures to overstate participation given the sacrifices of the Civil Rights Movement to ensure blacks can vote. This pressure remains whether there is a black candidate on the ballot or not.
Moreover, vote-validated studies confirm that black political activity as measured by voting was substantially greater in 2008 and 2012 than it had been in previous years (McKee, Hood, and Hill Reference McKee, Hood and Hill2012). As a result, if we find blacks participated at much higher rates in 2008 and 2012, it is unlikely that these changes could be reduced to higher levels of over-reporting of political participation. Finally, given that citizens are less expected to engage in political activities beyond voting, political participation measures such as proselytizing, wearing campaign gear or displaying a bumper sticker, donating to a political campaign, and attending a campaign event or rally should be less susceptible to problems with social desirability response bias.Footnote 4 Thus, my wider definition of political participation may help guard from the possibility that changes in political participation are solely due to individuals overstating their voting rates.
Candidate Characteristics
My mediating variables of interest measure respondents’ attitudes about the Democratic Presidential candidate's leadership skills, knowledge, empathy, intelligence, and morality. For each characteristic, the ANES first asks, “In your opinion, does the phrase ‘….’ describe Democratic candidate [Bill Clinton/ Barack Obama].” To measure leadership, the ANES uses the phrase “provides strong leadership.” The phrase “cares about people like you” is used to measure empathy. To measure intelligence and knowledge, and morality, the ANES uses the phrases “is intelligent,” “is knowledgeable,” and “is moral,” respectively. The scores on these measures range from “not well at all” to “very well/extremely well.” Given that I argue that blacks decide to participate based on the overall strengths of the candidate, I first examine the influence of a combined trait scale, which is created by adding each of the five traits together (Cronbach's α = .92) on the relationship between co-racial candidates and participation. I then examine whether any individual trait is a stronger mediator between descriptive representation and political activism through the analysis of all five traits separately.Footnote 5
I control for a number of variables to minimize potential confounding effects between Obama's presence on the ballot, perceptions of candidate empathy, and higher levels of political participation. First, I control for socio-demographic factors such as the respondents’ income, education, gender, marital status, time lived at current residence, church attendance, and age. All of these variables have long been shown to be strong predictors of political participation and perceptions of candidate traits (Verba, Schlozman, and Brady Reference Verba, Schlozman and Brady1995; Wolfinger and Rosenstone Reference Wolfinger and Steven1980). I also control for the respondent's partisanship as those who identify as Democrats may be more likely to rate both Clinton and Obama as being better across different candidate traits than respondents who identify with a different political party. Moreover, individuals who identify with any political party are much more likely to participate in politics (Wattenberg Reference Wattenberg1998).
Additionally, I control for the percent of the African American population in the respondent's congressional district. According to numerous studies, blacks in majority–minority districts tend to be more empowered and participate at greater rates (Fraga Reference Fraga2016). Moreover, it is possible that these individuals perceive candidates as being more receptive to their needs given their electoral power. I also control for the latest date one could register to vote within the state and whether the black respondent resided in the South. Both of these variables have been shown to influence turnout (Dawson Reference Dawson1995; Wolfinger and Rosenstone Reference Wolfinger and Steven1980).
METHODS
While standard regression analysis allows us to examine the effect of a given independent variable (X) on a dependent variable (Y), mediation analysis examines how much of an independent variables (X) effect on a dependent variable (Y) can be attributed to a mediating variable (Z). Specifically, the models estimated in this analysis examine how much of Obama's (X) effect on black political participation (Y) can be attributed to differences in perceptions of candidate traits between Obama and Clinton (Z). Thus, mediation analysis is useful for this project because it allows for an empirical examination of why descriptive representation (X) influences black participation (Y).
To assess the mediating effect of candidate traits on black political participation in elections with black and white candidates, I use the mediation analysis approach suggested by Imai, Keele, and Tingley (Reference Imai, Keele and Tingley2010) and Imai et al. (Reference Imai, Keele, Tingley and Yamamoto2011). This method (Imai et al. Reference Imai, Keele and Tingley2010; Reference Imai, Keele, Tingley and Yamamoto2011) examines the mediating effect of a variable of interest between a specified treatment and dependent variable through the estimation of two separate regression models. For the purposes of this study, the first regression model estimates the effect of having Obama on the ballot (i.e. treatment variable) on perceptions of the Democratic candidate's traits (i.e. mediating variable). Following this regression with the appropriate controls, the mediation analysis then estimates a second regression predicting various forms of political participation separately using the treatment (i.e. R in year with Obama on the ballot), the mediator (i.e. perceptions of Democratic candidate traits), and the specified pre-treatment control variables as predictors.
Using information from these two regression models, the mediation analysis creates a predicted value of perceptions of candidate traits based on the candidate's race. The mediation analysis then examines the change in the different forms of political participation separately for blacks in the Obama and Clinton years. Using perceptions of empathy as an example, when predicting the mediation effect for the treatment group on voting, the model estimates change in predicted turnout if the perceived levels of candidate empathy changed from their observed value to the levels of perceived empathy; the respondents in the treatment group would have had if everything else was the same but Clinton was the Democratic Party's nominee in 2008 and 2012. The method then repeats this process for blacks in control condition. Finally, the model takes the average of these two counterfactual analyses to provide an estimate of the average mediation effect.
The analysis also provides information on the direct effect which is the proportion of the variance in different forms of political participation which are associated with Obama's presence on the ballot, but are not explained by the mediating variable (candidate traits). The mediation analysis calculates the mediation effects and direct effects separately for individuals in the Obama years (Med Effect-T, Dir Effect-T) and Clinton years (Med Effect-C, Dir Effect-C). Finally, each model provides an estimate of the total effect (Tot. Effect) which is the effect of Obama on the ballot on various forms of political activity. These effects can be interpreted as predicted probabilities.
To ensure that the treatment is exogenous to other factors in observational studies, such as this one, researchers control for several independent variables which theoretically should not be influenced by the treatment (i.e., are not influenced by Obama being on the ballot). For the purposes of this study, I have controlled for several socio-demographic and contextual variables which should not be influenced by Obama's presence on the ballot, but are meaningful predictors of both perceptions of candidate traits and political participation. As a result, I have attempted to ensure that exposure to the treatment is not biased by exogenous factors.Footnote 6
MEDIATION ANALYSIS RESULTS
Table 1 presents the results of 30 separate mediation models. The models estimate perceptions of each candidate characteristic combined and separated (A: all characteristics combined, B: leadership, C: empathy, D: knowledge, E: intelligence, and F: morality) across a range of political activities (i.e. voting, proselytizing, wearing campaign gear, attending a campaign rally/event, and campaign contributions). For each mediation model, Table 1 presents the mediation and direct effects for the treatment (Obama years) and control groups (Clinton years), the total effects, and the percent of the relationship between Obama's presence on the ballot and change in political participation that is associated with changes in perceptions of candidate traits. The models presented in Table 1 only include black respondents.
Table 1. Mediation Analysis Assessing the Effects of Candidate Traits on the Relationship between Obama and Participation
*Significant at .05. NS, not significant at .05. Results are derived from 30 separate mediation analyses predicting the mediating effect of candidate traits in explaining the relationship between Obama and changes in political activity. Total Effects present min–max probabilities. The models include controls for gender, age, marital status, income, education, partisanship, R’s time at residence, church attendance, % black in R’s congressional district, south, and days to register before the election.
Only blacks are included in the analysis above.
The total effects rows in Table 1 demonstrate that even when controlling for a number of factors such as age, income, education, and partisanship, blacks were significantly more likely to vote, proselytize, display campaign paraphernalia, and donate to a campaign during the Obama years than they were in the Clinton years. Not only were blacks significantly more likely to participate during the Obama years, but this difference was substantial. For example, blacks were about 7.5% more likely to vote in the Obama years than when Clinton was on the ballot. Moreover, blacks were almost 20% more likely to proselytize about politics and display campaign paraphernalia when Obama represented the Democratic Party than in the 1990s. The only form of political participation in which blacks did not differ between the Obama and Clinton years was attending a campaign rally.
Not only did Obama's presence on the ballot boost black participation, I find support for my hypothesis that blacks’ perceptions of Obama as stronger across candidate traits are consistently associated with blacks being more politically active when he was on the ballot. Table 1 demonstrates that a large percentage of the mobilizing effect that Obama had on the black community is associated with perceptions that Obama was better across several traits than Clinton. In fact, holding all else equal, almost 40% of Obama's effect on influencing black turnout in 2008 and 2012 (about 3% of the 7.5% increase in turnout) is mediated by differences in perceptions of the combined candidate trait scale between these two candidates. When examining specific traits, it appears that perceptions of Obama as being more empathetic and moral mediate the largest percentage of Obama's influence on black turnout. Morality and empathy mediate about 33% and 26% of the total effect of descriptive representation on increasing black voting during 2008 and 2012.
In addition to turnout, higher levels of black's proselytizing activities also appear to be significantly mediated by perceived differences in candidate traits between Obama and Clinton. As with turnout, blacks were significantly more likely to try to convince at least one other person about their voting decision in years in which Obama was the Democratic Party's presidential nominee. Of the roughly 17% growth in black proselytizing in 2008 and 2012, a little more than 7% of this increase is associated with Obama's higher score on the combined trait scale relative to his predecessor Clinton. As a result, 37.2% of the total effect of descriptive representation on increasing proselytizing activities is mediated by differences in perceptions of candidate strengths between the two Democratic Presidents. While all candidate traits individually also significantly mediate the relationship between descriptive representation and proselytizing, differences in perceptions of morality (26.4% mediated) and empathy (24.2% mediated) are the strongest individual mediators.
The same pattern appears to hold for displaying campaign paraphernalia and making campaign contributions. About a fifth of the 10% increase in campaign contributions from the black community during the Obama years are associated with higher scores on Obama's candidate trait scale. Not only are blacks more likely to make campaign contributions to co-racial candidates, this increased generosity is partially mediated by perceptions that descriptive candidates are stronger across several candidate traits. Additionally, differences in perceptions of candidate strength explained about 2.2% of the 11% growth in blacks wearing or displaying campaign paraphernalia in years where Obama was on the ballot.
As with voting and proselytizing, each trait individually mediates a significant amount of the variation between Obama being on the ballot and higher levels of political participation; however, it is morality and empathy that have the largest mediating effects. Empathy alone mediates 16% and 20.2% of descriptive representation's total influence on increasing blacks’ displaying campaign paraphernalia and donating to a campaign. Similarly, morality explains 18% and 16% of Obama's total influence on blacks displaying campaign paraphernalia and donating to a campaign. In sum, the role of candidate traits appears to explain a large and important portion of why Obama's candidacy is associated with higher levels of black participation.Footnote 7 Moreover, morality and empathy appear to play the largest individual mediating roles in the relationship between descriptive representation and higher levels of black political activity.Footnote 8
SURVEY EXPERIMENT
While the analysis in the previous section provides support for my hypothesis, more research is necessary to better ensure that the effect of the race of the candidate is driving both more positive trait assessments of co-racial candidates and higher levels of black political participation. In particular, there might be something specific about the Obama years beyond Obama's race which are not accounted for in my model that both improve blacks’ perceptions of Obama's strengths relative to Clinton and increase turnout. For example, Obama facing a more conservative candidate in 2012 than Clinton did in 1996 may have made blacks view Obama more positively and increased participation. Moreover, Obama's more sophisticated campaign strategy which more explicitly targeted blacks may in part explain some of these differences. Given the inability to control for all potential confounding factors in any real-world assessment of the relationship between descriptive representation, candidate traits, and political participation, an experiment which holds constant everything but the race of the candidate would provide more internal validity for my study. Moreover, if I find the same result across a real-world analysis and an experiment which isolates the influence of the race of the candidate, there should be a strong support for my candidate trait hypothesis.
To further isolate the mediating effects of candidate characteristics on the relationship between descriptive representation and higher levels of turnout, I conducted a survey experiment using a Qualtrics panel between July 10 and July 13, 2017. Qualtrics interviewed a total of 140 black respondents from their panel. While the sample was not collected to be nationally representative, the differences in key factors in the sample and in the population are small. For example, blacks in my sample had a median age of 33 (compared with 31 in the U.S. black population), a median household income of 30,000 (compared with 33,000 in the U.S. black population), and about 26% had a graduated college (compared with 15% in the U.S. black population).Footnote 9
The survey presented respondents with one of two candidate webpages. The respondent was told that the candidate is a U.S. Senator that many are considering as a viable candidate to win the 2020 Democratic presidential primary (see Supplementary Appendix for screenshots of the experiment). The topic covered on the webpage is economic justice and the text focuses on issues such as worker's rights, a living wage, and union bargaining. The webpage was created to simulate what a real-world Democratic presidential candidate would discuss and was modeled after information presented by both Hillary Clinton and Bernie Sanders on their webpages in 2016. The only difference between the control and treatment groups was whether the respondent reviewed a webpage with a candidate who had a black sounding name (Malik Johnson––treatment) or a white sounding name (Wyatt McHarris––control). This experimental treatment serves as a proxy for descriptive candidates.
To measure the effect of the candidate's race on turnout, I asked respondents how likely they would be to vote in the 2020 presidential election if hypothetical candidate Wyatt McHarris/Malik Jackson represented the Democratic Party. The dependent variable was scaled from 1 (extremely unlikely) to 5 (extremely likely). In addition to asking about turnout, I also asked respondents to rate the candidate's levels of leadership, empathy, intelligence, and honesty. In particular, these questions ask, “If you had to guess, how well does the phrase ‘provides strong leadership’/‘really cares about people like me’/‘is intelligent’/‘is honest’ describe Malik Jackson/Wyatt McHarris.” These are all measured on a five-point scale which range from 1 (not well at all) to 5 (extremely well). While random assignment ensures that key variables like gender, age, income, education and partisanship are not statistically different between the control and treatment groups, I include these variables in my model to completely control for these factors. Moreover, I remove respondents who were unable to name the candidate after reviewing the webpage. This serves as an attention check.
Table 2 presents the results of five separate mediation analyses examining whether respondent's who received the black candidate treatment (X) viewed the candidate as being stronger on a host of characteristics (Z) and then if these assessments increase their stated willingness to turnout to vote in the 2020 presidential election (Y).Footnote 10 The results in Table 2 provide strong support for my hypothesis. Namely, blacks who were presented the opportunity to potentially vote for a co-racial candidate in 2020 professed a much greater likelihood of voting. In fact, blacks who were presented with a candidate with a black sounding name were about half a point more likely to express interest in voting in 2020 than blacks who reviewed the same information but with a candidate who had a white sounding name.
Table 2. Mediation Analysis Assessing the Effects of Candidate Traits on the Relationship between a Hypothetical Black and White Candidate and Intention to Vote in 2020
*Significant at .05. NS, not significant at .05. Results are derived from five separate mediation analyses predicting the mediating effect of candidate traits (Z) in explaining the relationship between being presented with a hypothetical black or white candidate (X) and changes in estimated turnout (Y). The models include controls for gender, age, income, education, and partisanship. Only blacks are included in this analysis.
Moreover, most candidate traits explained a significant amount of the relationship between the presence of a black candidate in the experiment and higher levels of expressed interest in turning out to vote in 2020. For example, differences between the control and treatment groups on a combined trait scale, which was calculated by adding individual's perceptions of the candidate's leadership abilities, empathy, intelligence, and honesty, explain almost half (49%) of why black respondents expressed more enthusiasm in voting when they were presented with a black sounding name over a white sounding name.
Empathy by itself plays an even larger role in explaining why descriptive representation spurs greater levels of turnout. Black respondents’ perceptions of the co-racial candidate as being more empathetic than the same white candidate explains 60% of their higher expressed intention to turnout in 2020 when the black candidate is on the ballot. Both intelligence and honesty play a lesser, but still substantial role in explaining why descriptive candidates motivate black turnout. Perceptions of the hypothetical black candidate as being more intelligent and honest compared with the same hypothetical white candidate explain a little over 40% of blacks’ higher interest in voting when a descriptive candidate is on the ballot.
The only candidate trait which did not explain differences in the presence of a descriptive candidate on the ballot and higher levels of expressed intention to vote in 2020 was leadership. While this differs from the previous analysis with the ANES, differences in perceptions of leadership may matter less to black voters as all politicians, regardless of race, may be perceived as being strong in this area. Leadership withstanding, the results of this analysis in combination with the results for the comparison of Obama and Clinton provide both internal and external validity for my candidate trait hypothesis.
CONCLUSION
While numerous scholars’ debate what Obama's Presidency has meant for black politics, one thing is certain; Obama's historic bid for the White House had a dramatic effect on increasing black political participation. As this study details, not only were African Americans more likely to vote in years where Obama was on the ballot, but they also became more involved in a host of other political activities. While there are several reasons why black candidates like Obama increase black participation, the results of this paper demonstrate that much of the surge in black political activity when a descriptive candidate is on the ballot is correlated with blacks more positive assessment of these candidates’ attributes. Namely, blacks perceive Obama and black candidates like him as being more empathetic, intelligent, moral, and honest than comparable white candidates. The perceptions of black candidates as being better politicians inspire blacks to become more politically involved.
The results of this analysis make an important theoretical contribution to studies of descriptive representation by detailing why descriptive representation motivates minorities to participate in politics. This study demonstrates that traits specific to the candidates’ matter. Blacks are partially motivated to become more politically active when there is a black candidate on the ballot because of perceptions that these candidates are uniquely concerned about individuals like themselves, are more moral, and are more capable in terms of intellectual ability to perform in office. This connection leads many blacks to believe that co-racial candidates are more apt to advance black politics than their white Democratic counterparts. Thus, the simple substitution of a white candidate for a black candidate is unlikely to yield the same increases in black political participation regardless of the context or monetary expenditures.Footnote 11 This is particularly true for blacks who see their race as influencing their opportunities for success.
Of the individual traits, empathy appeared to play the largest roles in linking descriptive representation and political participation. Given the tradition of white politicians being perceived as taking the black vote for granted (see Frymer Reference Frymer2010), it is not surprising that perceptions of black candidates as being more empathetic would play a large role in explaining higher levels of black participation when they are on the ballot. While candidate characteristics such as leadership and intelligence may be present in white politicians, their inability to experience the problems blacks face in the United States may lead black voters to believe that they care less about individuals like themselves regardless of the campaign rhetoric.
While this study improves our understanding of the link between descriptive representation and black political empowerment, it is not without its shortcomings. First, this study uses only one real-world candidate, Barack Obama, to examine the mediating influence of candidate traits on the relationship between descriptive representation and political participation. To ensure that my results are not Obama-specific, I re-estimated the analysis in Table 1 comparing 1988 Democratic Presidential Primary candidate Jesse Jackson to 1992 Democratic Presidential Primary candidate Bill Clinton. Like Obama in Table 1, the results of this analysis demonstrate that Jackson's ability to increase black turnout in the primary is largely explained by more positive feelings about Jackson's candidate traits relative to Clinton (see Supplementary Appendix).
This finding demonstrates that the mediating influence of perceptions of candidate traits on descriptive representation's ability to increase participation matters beyond message, as Obama and Jackson are on the opposite end of the spectrum in terms of their racial outreach (see Stout Reference Stout2015). Nonetheless, both candidates’ ability to increase black participation is substantially and significantly explained by perceptions of their candidate traits. This provides more confidence in my hypothesis that descriptive representation works through candidate characteristics to increase political participation.
While these analyses demonstrate that black presidential candidates’ ability to increase black participation can be explained by more positive perceptions of their candidate's traits, future research should re-examine these results for other black candidates at different levels of government. Such an analysis would shed more light on whether the relationships examined in this study holds for black candidates with varying characteristics (party, viability, campaign strategy, level of elected office). However, given the substantial policy disconnect between the average black voter and black Republicans, such an analysis is unlikely to produce the same results as the one completed in this paper.
Second, the results of this analysis demonstrate that black voters uniquely view Obama as being better across a number of traits and this increases black political participation. However, it is possible that perceptions of candidate traits are not driven by confidence in Obama, but rather black voters like him more and rate him higher as a result. If this is the case, the results in Table 1 may not be driven by candidate traits; rather they may just be a proxy measure for overall likability of descriptive representatives. I assess this possibility by controlling for differences in democratic candidate feeling thermometer scores as post-candidate controls and present the results in the Supplementary Appendix.Footnote 12 While controlling for democratic candidate feeling thermometer scores diminished the explanatory effect of candidate traits on the relationship between descriptive representation and political participation slightly, the relationship remains substantial and significant.
Third, this study largely relies on self-reports of political activity which may be problematic due to social desirability response bias. While I earlier discussed why this problem should not unduly hinder confidence in my results, future research should re-examine my findings with validated data where possible. In spite of these shortcomings, the consistent explanatory power of candidate traits as mediators of the relationship between descriptive representation and political participation provides an important addition to the racial/ethnic politics literature which should be further refined as better data become available.
SUPPLEMENTARY MATERIAL
The supplementary material for this article can be found at https://doi.org/10.1017/rep.2018.20
