As political institutions diversify, it is critically important to understand better how perceptions of descriptive representation within government shape Americans’ political attitudes. Building upon a sociotropic model of representation (e.g., Schwindt-Bayer and Mishler Reference Schwindt-Bayer and Mishler2005), Stauffer (Reference Stauffer2021a, 1228) posits that gender diversity signals institutions are more open and responsive, explaining why Americans who perceive that more women serve in federal and state legislatures exhibit greater external political efficacy.
Theoretically, the same model of representation ought to apply to other minority groups. Like women, Blacks bring different experiences, perspectives, and interests to bear in elected office (Phillips Reference Phillips1998; Reingold, Haynie, and Widner Reference Reingold, Haynie and Widner2020). Blacks serving in office then potentially signals institutions are open and responsive. However, it is empirically unclear whether perceptions of racial descriptive representation also relate to Americans’ efficacy toward political institutions.
This letter provides three key contributions. First, we use an independent survey to reaffirm Stauffer’s findings that perceptions of gender descriptive representation relate to citizens’ attitudes toward Congress and their state legislature. Such validations are essential best practices as the social sciences face a “replication crisis” (Camerer et al. Reference Camerer, Dreber, Holzmeister, Ho, Huber, Johannesson and Kirchler2018). Second, we test an extension of Stauffer’s argument using novel survey questions, finding that perceptions of Black descriptive representation positively relate to external efficacy. Third, we newly show that Americans’ perceptions of descriptive representation relate to their trust in political institutions. Together, our and Stauffer’s studies point to citizen perceptions of gender and racial descriptive representation as critical components of whether individuals have faith in democratic institutions.
DESCRIPTIVE REPRESENTATION AND ATTITUDES TOWARD GOVERNMENT
Descriptive representation matters for understanding the political behavior and attitudes of political minorities (e.g., Mansbridge Reference Mansbridge1999). Focusing on gender, women’s political engagement increases when women candidates seek high-profile office (Atkeson Reference Atkeson2003) or are in competitive elections (Dolan Reference Dolan2006). Women are more supportive of women candidates (Martin Reference Martin2019) and have higher evaluations of women members of Congress (Lawless Reference Lawless2004; see also English, Pearson, and Strolovitch Reference English, Pearson and Strolovitch2019).
Building on this research, Stauffer (Reference Stauffer2021a) argues that women’s descriptive representation shapes both majority and minority group members’ attitudes largely through two mechanisms. One mechanism draws upon the aforementioned sociotropic model of representation, where descriptive representation signals that political “institutions are open to a wide array of viewpoints and perspectives” (Stauffer Reference Stauffer2021a, 1228–9). The second mechanism asserts descriptive representation signals that decision-making and broader political processes are fair. As characterized by Dovi (Reference Dovi2007, 308), “an all-female Congress could not legitimately represent U.S. men and an all-male Congress could not legitimately represent women.” With gender descriptive representation leading to views of more open and legitimate institutions, people should feel more politically efficacious toward institutions where women are more prevalent.
Stauffer (Reference Stauffer2021a, 1227) also highlights that “it is citizen perceptions of women’s representation that shape their evaluations of institutions, rather than the objective reality of women’s presence.” Although actual descriptive representation can produce indirect signals, the direct effect of voters’ perceptions of descriptive representation is likely more consequential for voters’ political attitudes. Using actual levels of descriptive representation as a proxy for perceptions may also be limiting, as Americans often misperceive the size of groups or who serves in government (Ahler and Sood Reference Ahler and Sood2018). Misperceptions lead some voters to electorally punish those they perceive to be in power (Rogers Reference Rogers2023) and impact Americans’ trust in political institutions (Tate Reference Tate2002).
If citizens’ perceptions of greater gender descriptive representation increase citizens’ political efficacy, it begs the question of whether citizens’ perceptions of other groups also affect attitudes toward political institutions. For instance, Blacks, like women, bring different perspectives to bear in government. People may feel that institutions where Blacks are more prevalent are places where “all points of view will be represented” (Stauffer Reference Stauffer2021a, 1228), consistent with Stauffer’s first theoretical mechanism. Consistent with Stauffer’s second mechanism, an all-white legislature could unlikely legitimately represent Blacks, echoing arguments that “to be viewed as legitimate by all segments of society, a political institution must mirror the image of that society” (Scherer and Curry Reference Scherer and Curry2010, 92; see also Mansbridge Reference Mansbridge1999, 650).
While Stauffer’s argument may theoretically apply to race, the empirical evidence is unclear. On the one hand, racial and ethnic minorities value descriptive representation (Casellas and Wallace Reference Casellas and Wallace2015), with descriptive representation decreasing political alienation among Latinos (Pantoja and Segura Reference Pantoja and Segura2003), increasing trust in government among Blacks (Heideman Reference Heideman2020), and improving how Blacks evaluate courts (Scherer and Curry Reference Scherer and Curry2010) or members of Congress (Tate Reference Tate2002). Descriptive representation is also valued from an intersectional perspective, with people feeling individuals who share their race or ethnicity and gender would best represent their interests (Gershon et al. Reference Gershon, Montoya, Bejarano and Brown2019). People view international organizations as fairer and more legitimate when panels have greater racial and gender diversity (Chow and Han Reference Chow and Han2023), and whites and Blacks see policymaking decisions as fairer when more Blacks serve in local government (Hayes and Hibbing Reference Hayes and Hibbing2017). These studies provide optimism that racial diversity, like gender diversity, positively impacts citizen political efficacy.
On the other hand, considerable evidence suggests not everyone equally values racially diverse political institutions. Black descriptive representation has neutral or negative effects on white attitudes toward members of Congress (Gay Reference Gay2002), local governments (Abney and Hutcheson Reference Abney and Hutcheson1981), mayors (Heideman Reference Heideman2020), and courts (Scherer and Curry Reference Scherer and Curry2010). Thus, while Americans may view gender-inclusive institutions as open and politically legitimate, prior work suggests that some may view racially inclusive institutions as more closed off and thereby feel less efficacious toward those institutions.
Replication and Validation of Stauffer’s Study
We first replicate Stauffer’s study by following her research designs as closely as possible using an independent survey. Each study relies on Cooperative Election Studies (CES), conducted online by YouGov in two waves, where respondents are compensated (Schaffner, Ansolabehere, and Luks Reference Schaffner, Ansolabehere and Luks2019). Our study employs a 2018 module of approximately 1,000 respondents with two waves from September 27 to November 5 and November 7 to December 3 (Rogers Reference Rogers2022), and Stauffer employs 2016, 2017, and 2018 modules. Not all respondents completed both waves, resulting in sample sizes smaller than 1,000 (Stauffer Reference Stauffer2021b).
Differences between our surveys, such as question wording, merit attention as they could impact responses (Gideon Reference Gideon and Gideon2012).Footnote 1 Table 1 lists questions from our studies. Stauffer’s question wordings are more general but align conceptually with ours. For example, our Question 1 uses the term “represents,” while Stauffer’s Question 4 uses “care.” “Represents” is consistent with addressing whether “citizens feel that they are being “fairly and effectively represented” (Stauffer Reference Stauffer2021, 1226). Similarly, Stauffer’s Questions 4 and 5 reference “government officials” or “government.” Meanwhile, we ask about specific government officials of interest: members of Congress and state legislators. Our measures are empirically similar. Following Stauffer (Reference Stauffer2021a, 1232), we create a zero-to-one Congressional efficacy measure by rescaling the sum of Questions 1 and 2 responses. The means of Stauffer’s and our Congressional efficacy measures are 0.36 and 0.34 and standard deviations are 0.25 and 0.23, giving us confidence that we are capturing similar concepts.
Table 1. Questions Used to Measure Political Efficacy
Note: Questions 1 to 5 have five response options from “Strongly Disagree” to “Strongly Agree.” Question 6 has four response options from “Not at all responsive” to “Very responsive.”
We estimate the relationship between efficacy and citizens’ beliefs about gender descriptive representation with linear and ordered probit models nearly identical to Stauffer’s, with the only difference being that Stauffer’s includes year fixed effects. We refer readers to Stauffer (Reference Stauffer2021a) for details about measurement and statistical models. The appendix provides summary statistics and replications of Stauffer’s tables and figures.
Figure 1 illustrates Stauffer’s and our central results. The first row of plots shows comparable coefficients on Americans’ perceptions of descriptive representation variables. Stauffer’s (circles) and our (triangles) estimates are each positive and statistically significant. In both studies, a shift in a respondent’s beliefs that women comprise 35 instead of 25 percent of Congress correspond to approximately a 0.03 increase in the zero-to-one efficacy measure (Table A-1). Focusing on state legislatures, Stauffer finds that increasing a respondent’s beliefs about the percentage of women in their state legislature from 25 percent to 35 percent increases the probability that a respondent believes their state legislature is “moderately responsive” by 0.019 and “very responsive” by 0.009 (Table A-2). Similarly, in our study, this change increases the probability that a respondent “somewhat agrees” or “strongly agrees” that calling or emailing their state legislators would make a difference in shaping policy by 0.025 and 0.011 (Table A-3). Neither Stauffer nor our analyses find that these relationships vary by the respondent’s gender. Our study further reaffirms Stauffer’s findings that younger Americans view their state legislature as more responsive, and conservative Americans perceive their state legislature as more responsive, particularly in conservative states.
Figure 1. Coefficients Plots for Descriptive Representation Measures
Note: Coefficients capturing the relationship between political efficacy and perceived (top row) or actual (bottom row) gender descriptive representation in Congress (first column) or state legislatures (second column). The third and fourth columns are similar but study Black descriptive representation. Congressional analyses employ a linear model to estimate the zero-to-one Congressional efficacy measure, as described in the main text. State legislative analyses employ an ordered probit model, where Stauffer’s and the current study, respectively, have four or five response options.
Recall that Stauffer makes the critical distinction that citizens’ beliefs about descriptive representation are more important than actual levels of descriptive representation. Since the actual percentage of women in Congress remains constant during a cross-sectional survey, Stauffer leverages differences across states to examine how the actual gender composition of state legislatures relates to citizens’ efficacy toward their legislature. Stauffer’s and our estimates capturing this relationship are near and statistically indistinguishable from zero (Figure 1, second row, leftmost plot), suggesting little relationship between actual descriptive representation and efficacy. These relationships, again, do not vary by gender.
Extending Stauffer’s Study to Race
Our study reaffirms Stauffer’s findings that perceptions of descriptive representation relate to citizens’ efficacy toward these institutions, but is this relationship exclusive to gender descriptive representation? To address this question, we first study citizens’ awareness of racial descriptive representation. Second, we mirror Stauffer’s study of gender but replace respondents’ estimates of the percentage of women lawmakers with their estimates of the percentage of Black lawmakers.
Do Citizens Know Who Represents Them?
Stauffer argues that perceptions of descriptive representation should matter more than “objective reality” and shows that Americans’ perceptions about actual gender descriptive representation are often inaccurate. Using our survey, left panels in Figure 2 illustrate the distribution of respondents’ estimates of the percentage of women and Blacks in Congress. The right panels plot their estimates of women and Blacks in their state legislature compared to the true percentage. On average, Americans overestimated how many women served in Congress and their state legislature by 12 and 8 percentage points, respectively. The comparable figures for Black descriptive representation are 17 and 14 percentage points. Statistical analyses accounting for respondent characteristics (e.g., age, education, income) show that the predicted probabilities of overestimating descriptive representation of women and Blacks in state legislatures are, respectively, 0.12 and 0.08 less for Americans with a college degree (Table A-16). Neither women nor Black respondents differ meaningfully from other groups when overestimating descriptive representation. However, Americans exhibiting higher levels of hostile feminism or racial resentment overestimate more often, except when approximating the percentage of Blacks in the state legislature.
Figure 2. Respondent Estimates of Women and Black’s Representation
Note: Left panels illustrate the distribution of respondents’ estimates of the percentage of women and Blacks in Congress. In 2018, approximately 21 and 10 percent of members of Congress were, respectively, women or Black. Right panels plot respondents’ estimates of the percentage of women and Blacks in their state legislature compared with the true percentage.
Americans’ overestimates lead to inaccurate beliefs about descriptive representation. Figure 3 shows the percentages of Americans who “accurately” identified descriptive representation within five percentage points (Dolan Reference Dolan2011). Only 28 percent of Americans “accurately” identified gender and racial descriptive representation in Congress. The comparable figures for state legislatures are 29 and 25 percent. Those more knowledgeable about federal or state politics are more likely to be “accurate,” and the predicted probability that a respondent’s beliefs were accurate about the actual percentage of Blacks in Congress was 0.09 higher for Americans with a college degree. Neither women nor Black respondents were more accurate, but racially resentful Americans were less likely to be accurate about Black descriptive representation in state legislatures (Table A-19).
Figure 3. Percent of Americans with Correct Perceptions of Descriptive Representation
Note: Weighted percentages of respondents who correctly estimated levels of descriptive representation within five percentage points.
Perceptions of Black Descriptive Representation and Political Efficacy
To evaluate the relationship between perceptions of Black descriptive representation and political efficacy, we substitute variables from Stauffer’s study focused on women with comparable variables focused on Blacks. We replace respondents’ perceptions of gender descriptive representation with their perceptions of Black descriptive representation. We control for whether respondents were represented by a Black Senator or U.S. House member and the percentage of Blacks that served in the respondent’s state legislature instead of comparable gender variables. We also replace Stauffer’s white respondent indicator variable with a Black indicator variable.
Citizens who believe that more Blacks serve in office are more efficacious toward Congress or their state legislature (Figure 1: top row, rightmost two plots; Tables A-1 and A-4). Increasing a respondent’s beliefs about the percentage of Blacks in Congress or state legislature from 25 percent to 35 percent increases one’s efficacy toward Congress by approximately 0.024 or increases the probability that someone “strongly agrees” that calling or emailing their state legislators would make a difference in shaping policy by 0.008. Consistent with Stauffer’s and other research (Wolak Reference Wolak2018; Lawless Reference Lawless2004), we also find little relationship between the actual levels of descriptive representation and citizens’ views about the responsiveness of their state legislature (Figure 1, bottom right panel). We again find that younger and conservative Americans see their state legislature as more responsive, particularly in conservative states, and these relationships do not vary by the race of the respondent. Supplementary analyses (Online Appendix: Section 4), however, provide evidence that racially resentful Americans who believe there are more Blacks in their state legislature are more (less) likely to have negative (positive) attitudes toward their state legislature (Table A-14).Footnote 2
Descriptive Representation and Trust in Government
Overall, Stauffer (Reference Stauffer2021a, 1228) argues that perceptions of descriptive representation can “engender greater feelings of trust, efficacy, and legitimacy.” Stauffer does not test the relationship between perceptions of descriptive representation and trust, but evaluating this relationship can illuminate why perceived inclusion and efficacy are connected. We take advantage of unique questions on our survey and substitute the above efficacy measures as our dependent variable with respondents’ answers to “How much of the time do you think you can trust the following levels of government to do what is right?” for Congress and their state legislature. Respondents could choose “Never,” “Some of the time,” “About half of the time,” “Most of the time,” and “Always.”
Figure 4 is organized similarly to Figure 1 and illustrates coefficients from ordered probit analyses of citizen trust, where the above response options are our dependent variable (Tables A-20–A-22). An American believing women comprise 35 percent instead of 25 percent of Congress or their state legislature increases the predicted probability of trusting these respective institutions “most of the time” by 0.010 or 0.017 (Figure 4, top row, first two columns). Meanwhile, actual levels of gender descriptive representation have little relationship with attitudes toward legislative institutions (Figure 4, bottom row, leftmost panel).
Figure 4. Descriptive Representation and Trust in Government
Note: Ordered probit coefficients reflecting the relationship between trust and perceived (top row) or actual (bottom row) gender descriptive representation in Congress (first column) or state legislatures (second column). The third and fourth columns are similar but study Black descriptive representation.
Focusing on Black descriptive representation, our findings are in the expected direction but do not meet conventional levels of statistical significance. As illustrated by the top right panels of Figure 4, an American believing Congress is 35 instead of 25 percent Black increases the predicted probability that an American trusts Congress “most of the time” by 0.007 (t-statistic of difference 1.81). In state legislatures, the comparable increase is 0.009 (t-statistic of difference 1.48; Table A-22). However, we find that racially resentful Americans are less likely to trust their state legislature if they believe there are more Black state legislators (Table A-15). In state legislatures, we also find a negative relationship between actual Black descriptive representation and political trust (Table A-22), suggesting that actual descriptive representation can shape Americans’ attitudes.
DISCUSSION
We validate Stauffer’s findings concerning the relationship between political efficacy and Americans’ beliefs about the gender composition of Congress and state legislatures. We newly argue and show that Americans who believe more Blacks serve in their legislature also believe their legislature is more responsive. We additionally demonstrate Americans’ perceptions of gender descriptive representation relate to their trust in government. Together, our replication and extensions strengthen support for Stauffer’s argument that Americans’ perceptions of descriptive representation are critical for understanding their attitudes toward government institutions.
A limitation of our research is that we study the descriptive representation of women and Blacks as separate categories. However, more Black women are serving in federal and state legislative offices (Brown, Clark, and Mahoney Reference Brown, Clark and Mahoney2022; Hardy-Fanta et al. Reference Hardy-Fanta, Lien, Pinderhughes and Sierra2016) and have distinct policy preferences (Reingold and Smith Reference Reingold and Smith2012). We thus encourage future studies to employ an intersectional approach to assess how citizen perceptions of the race and gender of lawmakers shape political efficacy and trust.
We hope our extension to racial descriptive representation lays the groundwork for promising research avenues. We implore scholars to investigate how perceptions of other minority groups in elected office, such as ethnic, sexual, and religious minorities, shape Americans’ attitudes toward government. Other avenues to examine are how the presence of multiracial elected officials—a group growing in number—affects perceptions of government (Lemi Reference Lemi2018) along with studying differences across Americans themselves, such as those who hold intersectional identities or biases against certain groups. Such investigations will potentially extend Stauffer’s theory to other groups and improve our understanding of diversity’s important implications for Americans’ faith in their representative institutions.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit http://doi.org/10.1017/S0003055424001308.
DATA AVAILABILITY STATEMENT
Research documentation and data that support the findings of this study are openly available at the American Political Science Review Dataverse: https://doi.org/10.7910/DVN/QBFD2S.
ACKNOWLEDGMENTS
We would like to thank Kaitlin Klasen, John Barrett, Abigail Faust, Thomas Holland, Cher Hu, Emily Johansson, and Sequoyah Lopez for their research assistance. We are additionally grateful to Lauren Davenport, Andrew Engelhardt, Tim Ryan, Katelyn Stauffer, Jennifer Wolak, attendees of the 2019 State Politics and Policy Conference, the UNC State Politics Working Group, and the anonymous reviewers for their helpful feedback on previous versions of this project and manuscript.
FUNDING STATEMENT
The Saint Louis University College of Arts and Sciences provided the funding for the Saint Louis University CES Module.
CONFLICT OF INTEREST
The authors declare no ethical issues or conflicts of interest in this research.
ETHICAL STANDARDS
The authors declare that the human subjects research in this article was reviewed and approved by Saint Louis University’s IRB, Protocol number #29440. The authors affirm that this article adheres to the principles concerning research with human participants laid out in APSA’s Principles and Guidance on Human Subject Research (2020).
Comments
No Comments have been published for this article.