Working-Class Legislators and Perceptions of Representation in Latin America

Descriptive Representation and Workers’ Perceptions of Representation

Legislatures are the primary institutional vehicle for achieving democratic representation (Mainwaring 2006; Taylor-Robinson 2010), and interparty politics within the legislature is key to the success of programmatic policy goals (Crisp 2006). Indeed, it is within the legislature that descriptive representation generally yields substantive outcomes (Mansbridge 1999, 2015; Pitkin 1967). Despite legislatures’ representative function, Latin America has been characterized by a significant disenchantment with and mistrust of the legislature in recent decades (Mainwaring 2006; Mainwaring, Bejarano, and Leongómez 2006).

We posit that the exclusion of the working class—a group that constitutes a sizable majority of the population in Latin America—contributes to this mistrust of political institutions. Whereas workers make up the vast majority of the labor force across Latin America, they only hold a small share of legislative seats in the region. When white-collar representatives are left to act on behalf of the working class, this signals to workers that their participation in politics is not valued. By contrast, the mere presence of working-class representatives in decision-making bodies may have a direct effect on symbolic representation—improving citizens’ feelings about representation (Schwindt-Bayer and Mishler 2005). Moreover, when descriptive representation leads to improved policy representation, it may have an indirect effect on citizens’ evaluations of the government. In this section, we argue that representation of the working class may help restore trust in and satisfaction with legislatures among working-class citizens. In the next section, we explain why higher levels of working-class representation may improve perceptions of representation in society more broadly.

Descriptive Representation Improves Everyone’s Perceptions of Representation

Descriptive representation of the working class may also have profound implications for symbolic representation in society more broadly. Although it may seem obvious that descriptive representation of the historically excluded improves perceptions of representation among members of that group, theoretical and empirical insights from previous research on representation suggest that these benefits should extend to all groups. As we elaborate below, workers’ descriptive representation and policy responsiveness should appeal to blue-collar and white-collar citizens alike.

Greater descriptive representation may enhance the de facto legitimacy of the political system, not only for historically marginalized groups but for the powerful ones as well (Mansbridge 1999). In the same way that some white voters support the descriptive representation of racial and ethnic minorities (Hayes and Hibbing 2017) and some men support women’s numeric representation (Barnes and Córdova 2016; Morgan and Buice 2013), there may be a large number of white-collar professionals who—at least in principle—support the representation of working-class citizens. Such individuals are likely to view working-class representation as a sign that the legislature is representative and the democratic process is trustworthy—rather than dominated by the rich or overrun by corruption.

Indeed, beyond indicating that the legislature is more representative of the citizenry, workers’ political inclusion may signify that the democratic process is functioning properly by delivering transparency and impartiality in the exercise of power. In particular, working-class representatives could signal a general lack of corruption and elite dominance in government that appeals to white-collar and working-class citizens alike. Criminal activities such as corruption and bribery often operate through “networks of collusion” (Barlow 1993), and political outsiders are often perceived to be less corrupt because they lack the networks and political connections necessary to successfully engage in corruption (Barnes, Beaulieu, and Saxton 2018). Moreover, when the electoral process produces representative outcomes, it signals that all citizens—not merely political elites—have an influence in the policy-making process, and that ordinary citizens are capable of governing.

Research on elite cues offers further theoretical insights into how and why descriptive representation of marginalized groups may enhance everyone’s perceptions of representation. Elite behavior—such as nominating more working-class candidates to run for office—sends important pro-egalitarian signals, promoting support for working-class representation (Beaman et al. 2009; Hansen 1997). And, although elite behavior likely shapes most citizens’ attitudes, it particularly influences citizens with weak (or no) personal investment in, or uncertain views on, the issues (e.g., Barnes and Córdova 2016; Clayton, O’Brien, and Piscopo 2019). The presence of women political leaders, for instance, reduces gender stereotypes among men (Alexander 2015; Alexander and Jalalzai 2018; Kerevel and Atkeson 2015) and is associated with increases in men’s satisfaction with democracy and legislative trust (Barnes and Taylor-Robinson 2017; Schwindt-Bayer 2010; Schwindt-Bayer and Alles 2018). Morgan and Buice (2013) explain that because men are less likely than women to hold firm beliefs about women’s role in leadership, they are more susceptible to cues transmitted via elite behavior. In this vein, they demonstrate that men are more supportive of women’s leadership in countries where elites nominated women to serve as cabinet ministers. This argument implies that citizens who do not have a strong self-interest in working-class representation may rely heavily on elite cues from party leaders and Congress when assessing the quality of representation. Thus, where party leaders have incorporated members of the working class, this pro-egalitarian message is likely to extend beyond workers—shaping the mass-public opinion of all citizens.

Working-class lawmakers may also foster greater trust in legislatures by representing policy preferences that appeal to a broad swath of citizens, and not exclusively to workers. There is some evidence to suggest that legislators with working-class backgrounds are more economically progressive (Carnes and Sadin 2015; Grumbach 2015). As explained above, working-class legislators in Latin America espouse preferences for more statist economic policies, increased government spending, and more expansive social welfare policies (Carnes and Lupu 2015). These stances are more in line with average citizens’ preferences in Latin America (Barnes and Córdova 2016) than are white-collar legislators’ preferences, which favor the economic interests of the wealthy. As such, the presence of working-class lawmakers may indicate the potential for the government to advance progressive economic policies that may appeal to economically vulnerable white-collar professionals, progressive voters, and average citizens more generally. As a result, higher levels of working-class legislators may enhance all citizens’ evaluations of legislatures through policy responsiveness.

The Visibility of Working-Class Legislators in Latin America

Working-class status is arguably more difficult for citizens to observe than characteristics such as race, ethnicity, or gender. Importantly, however, there are at least three good reasons to believe that working-class legislators are visible and that citizens are generally aware of the extent to which workers are represented in the legislature. First, parties and individual candidates/deputies have incentives for voters to know about politicians’ personal backgrounds. In campaign communications, candidates’ job experience is the most frequently cited personal quality (Shyles 1984), as it provides an extremely useful information shortcut for voters (Campbell and Cowley 2014; McDermott 2005; Mechtel 2014).

In the United States, the “mill workers son” heuristic is often employed in political rhetoric to refer to candidates who grew up working class (Carnes and Sadin 2015; Vandello, Goldschmied, and Richards 2007). Individual candidates in Latin American countries with both low (e.g., Argentina) and high (e.g., Bolivia and Brazil) levels of personal vote-seeking incentives (Johnson and Wallack 2012) make similar appeals to their working-class backgrounds. National Argentine Deputy Facundo Moyano employs this tactic, telling constituents he was not the “hijo de” (son of) anyone powerful, rather he was raised by a working-class mother. ¹ Likewise, Bolivian Deputy Herber Choque Tarqui, from La Paz, describes himself as “De padres Mineros” (of miner parents). ² In Brazil, candidates frequently use professional nicknames, such as “Doctor Carlos,” on official ballots to stand out in crowded open-list races (Boas 2014).

Whereas individual candidates may have a stronger incentive to emphasize their class in candidate-centered electoral systems, political parties have incentives to emphasize individuals’ occupational backgrounds in party-centered electoral systems. This is because parties likely assume candidates who share the majority of the population’s life experiences will fare better electorally than members of the political elite as they can personally connect with voters (Castañeda and Navia 2007). There is ample evidence of such tactics in countries with low personal vote-seeking incentives. In Argentina, beyond providing clientelistic benefits, one of the primary roles of punteros (political brokers) includes organizing rallies and neighborhood meetings so that voters can meet candidates face-to-face and personally connect with them (De Luca, Jones, and Tula 2002; Zarazaga 2014). In Peru, clientelism is used to lure voters to campaign events, meetings, and rallies so that candidates can introduce themselves (Muñoz 2014). Evidence from Costa Rica indicates that political parties in party-centered electoral systems know that they benefit when their legislators build close rapport with constituents in neighborhoods and communities (Taylor 1992).

Second, news coverage and official government websites across Latin America (regardless of parties’ and candidates’ electoral incentives) routinely report broader trends about politicians’ occupational backgrounds. Consider for instance an article published prior to the 2014 election by La Razón, a national Bolivian newspaper out of La Paz, explaining that the “lists of candidates for national deputies include laborers, miners, micro-entrepreneurs, peasants, indigenous, truck drivers and merchants, among others.” ³ In Honduras, La Tribuna ran an article after the 2015 election explaining that although Congress was historically dominated by priests, large property owners, and ranchers, today lawyers, engineers, health professionals, and teachers, as well as some union leaders and peasant leaders hold office. ⁴ Ahead of the 2017 Argentine midterm elections, Ambito reported that of eleven deputies from labor unions currently serving in the Lower House, five of them will complete their term, and only two will run for reelection. ⁵ The article detailed each deputy’s district and union affiliation.

In other instances, news reports point to more systematic coverage of deputies’ backgrounds. Costa Rica Hoy, a national Costa Rican newspaper, launched a website “Yo Voto Elecciones 2018” where it reported information about all candidates competing for national deputy prior to the election and the winners, along with their political party, age, and occupation. ⁶ In February 2018, El Mundo, a national paper in El Salvador, ran a story listing all candidates for national deputy along with brief biographies that included occupational information when available. ⁷ Leading up to the 2018 Mexican election, a headline read “Do you know your candidates for senators and deputies? This page tells you who they are and shows their CV,” and directed readers to a website cataloging candidates’ occupational backgrounds. ⁸ Comparable articles ran in Uruguay profiling legislators after the 2015 election. ⁹ Beyond news coverage, many government websites include short biographies of national deputies.

Third, research demonstrates citizens are remarkably adept at inferring class status from facial images (Bjornsdottir, Alaei, and Rule 2017; Bjornsdottir and Rule 2017) ¹⁰ and speech (Kraus and Keltner 2009; Kraus, Park, and Tan 2017). Voters across Latin America (regardless of personal vote-seeking incentives) are regularly exposed to politicians’ facial images and speech. With respect to facial images, a large number of countries include candidates’ pictures on ballots (Tchintian 2018), most legislatures include photographs of deputies on their websites, and images of deputies and candidates frequently appear in newspapers. Bjornsdottir and Rule (2017) explain that because people are attuned to social class, they can use stereotype-related impressions to assess individuals’ class based on facial images alone. In a series of experiments, subjects were shown facial images and were able to categorize images by class at a rate significantly better than chance. Thus, even naïve judgments of politicians (i.e., judgments made by relying on headshots or pictures alone) inform voters’ assessments of working-class representation.

Voters in Latin America are also exposed to politicians’ speech when they hear them at campaign rallies, in neighborhood meetings, and (albeit less often) on the radio (e.g., De Luca, Jones, and Tula 2002; Taylor 1992; Zarazaga 2014). A growing body of research indicates that people can accurately perceive other’s social class (as measured using occupational status) based on minimal exposure to speech patterns, dialects, and accents (e.g., Giles and Sassoon 1983; Kraus and Keltner 2009). Kraus, Park, and Tan (2017), for example, argue that speech style is an accurate signal of social class even when the content of speech is held constant. Using isolated speech recordings from seven spoken words, they find that subjects’ evaluations of the speaker’s social class (using both measures of educational attainment and occupational status) were significantly and positively correlated with the speaker’s known social class, leading the researchers to conclude that “social class is rapidly and accurately perceived in the early stages of social perception” (Kraus, Park, and Tan 2017, 426). Although this research is based on English speakers, there is evidence that linguistic markers also vary by class in Latin America (Lipski 2011). Hence, even minimal exposure to politicians’ speech informs citizens’ perceptions of working-class representation.

Taken together, social science research on Latin American politics, campaigns, and social psychology, combined with primary and secondary accounts from newspapers, government websites, and Latin American scholarship, indicates strong evidence that citizens can detect working-class representation. As with the share of women legislators, people may not be able to accurately guess the exact share of seats held by working-class representatives (Carnes and Lupu 2016b). Nonetheless, because parties and candidates have incentives to showcase class during campaigns, news coverage and government websites often report occupational backgrounds, and ordinary citizens can infer legislators’ class from facial images and speech, working-class legislators are visible in Latin America.

Evidence from Latin America

Latin America offers an ideal setting to evaluate our hypotheses. Latin America hosts substantial variation in the level of working-class representation in national legislatures. During the period under investigation, workers’ access to legislative office ranges from a low of 0 percent in Costa Rica to upward of 17 percent in Bolivia. This variation is necessary to evaluate our hypothesized link between workers’ numeric representation and citizens’ beliefs about representation. Equally important, workers make up the vast majority of the labor force in Latin America—between 60 and 85 percent. ¹¹ In many Latin American countries, poor and working-class citizens have suffered a history of social and political exclusion (Taylor-Robinson 2010) and they remain dramatically underrepresented (Carnes and Lupu 2015).

To evaluate the relationship between workers’ descriptive representation and citizens’ perceptions of legislative representation, we analyze three individual-level survey questions from LAPOP in eighteen countries from 2008 to 2010. ¹² In addition, we leverage elite-level data on legislators’ occupational backgrounds as a measure of working-class descriptive representation. Occupational data comes from waves 3 to 5 of the USAL survey of Latin American legislators. ¹³

Dependent Variables

We measure citizens’ feelings of being “fairly and effectively represented” (Schwindt-Bayer and Mishler 2005, 407) with three questions designed to tap citizens’ feelings and attitudes toward the legislature. LAPOP asks: “Thinking of Congress as a whole, without considering the political parties to which they belong, do you think members of Congress are performing their jobs” very poorly (coded 1), poorly (2), fair (3), well (4), or very well (5)? Respondents are also asked to indicate their level of trust in the legislature on a scale from 1 (none) to 7 (a lot). Third, respondents were asked, “To what extent does the Congress accomplish what you would hope for it to?” Responses range from not at all (1) to a lot (7). ¹⁴

Figure 1 plots the percentage of respondents across each of the outcome options for the three dependent variables. Here, the data are pooled across all of the country-waves in the analysis. Online Appendix Figures A1 to A3 show the distribution of each of the dependent variables for each country-wave in the analysis. For the first, Legislative Approval, the modal response is “neither good nor bad” (3), with nearly half of all responses in this category. The modal response for Legislative Trust is “some” (4), with slightly more than 20 percent of respondents selecting this response, and closely followed by “none” (1) with 18.8 percent. For Legislative Accomplishment, the modal response is “somewhat” (4) with 22 percent of respondents, followed closely by the lowest response outcome (1) with 20 percent.

OPEN IN VIEWER

Figure 1.

Perceptions of legislative representation in Latin America.

Independent Variables

Working-class representation

The key explanatory variable in this analysis is descriptive representation of the working class. We use data from USAL to construct a measure that captures the percentage of legislators that come from a working-class background. Following Carnes and Lupu (2015), we operationalize “working-class background” with a legislator’s occupation prior to getting elected. ¹⁵ As Figure 2 shows, the percentage of workers in the legislature varies dramatically across Latin American countries, from 0 percent in Costa Rica in 2008 to 17 percent in Bolivia in 2008. The occupation-based conceptualization of class is fundamentally distinct from socioeconomic status as defined by income (Carnes and Lupu 2015). In sharp contrast to income or education, how individuals earn a living most accurately reflects people’s place in society (Brooks and Manza 2007), and individuals in different professional strata, even those with the same income, face dramatically different life chances (Carnes and Lupu 2015). Occupational background reflects class better than income because “the dividing line between social classes in most societies revolves around the labor market, that is, how people earn a living” (Carnes 2013, 3).

OPEN IN VIEWER

Figure 2.

Distribution of percent workers in the legislature.

Working-class citizens

To test our first hypothesis, that the relationship between working-class legislators and positive evaluations of the legislature is stronger for working class citizens, we create a dummy variable that measures whether or not individuals belong to the working class. Following Carnes and Lupu (2015), we use occupational data from LAPOP and code individuals as working-class if they indicate that their main occupation or type of work is as a skilled worker, office worker, employee in the service sector, food vendor, farmhand, domestic servant, or servant. ¹⁶ All other occupations are coded as non working-class.

Analyses and Results

Working-Class Legislators and Perceptions of Representation

We begin by looking at the correlation between the percent workers and each of the three dependent variables. In Figure 3, we plot the correlations between the average response (i.e., the country-year mean) to each dependent variable (y axis) and the percent workers in the legislature (x axis). We also include the corresponding regression line and the correlation coefficient for the relationship. ¹⁸ For all three dependent variables, the slope of the regression line is positive and significant, indicating that more working-class representation is associated with better legislative approval, trust, and accomplishment.

OPEN IN VIEWER

Figure 3.

Correlation between working-class legislators and perceptions of representation.

Next, a one-way ANOVA reveals the three dependent variables vary significantly across the thirty-one country-years included in this analysis (p < .001), suggesting that factors measured at the country-year-level explain some of the variation in the individual-level dependent variables. To account for the nested nature of the survey—individuals i living in country-years j—as well as individual-level and country-level data, we estimate a series of multilevel ordered logit models to test our hypotheses about workers’ descriptive representation. Online Appendix Table B9 and Figure B5 demonstrate that the results are robust to a multilevel linear model specification.

Table 1 presents baseline and interactive models for each dependent variable. Models 1, 3, and 5 allow us to assess the direct relationship between working-class legislators and citizens’ perceptions of representation. Models 2, 4, and 6 include a cross-level interaction between percent workers and the individual-level variable for working-class respondent and allow us to investigate whether the relationship between descriptive representation and citizens’ perceptions is different for working-class compared with non working-class citizens. ¹⁹ The coefficient for percent workers is positive and significant in every model.

OPEN IN VIEWER

Table 1.

Working-Class Legislators and Perceptions of Legislative Representation.

	Approval		Trust		Accomplishment
	(1)	(2)	(3)	(4)	(5)	(6)
% Workers in Legislature	.07*** (.02)	.07** (.03)	.07*** (.02)	.07*** (.02)	.07*** (.02)	.06** (.03)
Working-class Citizen		.10** (.04)		.06(.04)		.08* (.05)
% Workers in Legislature × Working Citizen		−.00(.01)		−.01(.01)		−.01(.01)
Individual-level Controls
Education	−.03*** (.00)	−.03*** (.00)	−.02*** (.00)	−.02*** (.00)	−.01** (.00)	−.00(.00)
Political Interest	.16*** (.01)	.17*** (.01)	.19*** (.01)	.18*** (.01)	.19*** (.01)	.17*** (.02)
Income	−.23*** (.05)	−.20*** (.08)	−.04(.05)	−.20*** (.07)	−.07(.06)	−.18** (.09)
Female	.19*** (.02)	.17*** (.03)	.12*** (.02)	.07*** (.03)	.13*** (.02)	.12*** (.03)
Ideology = Left (1–3)	.10*** (.03)	.07(.05)	−.02(.03)	.00(.05)	.19*** (.04)	.19*** (.06)
Ideology = 4–5	.10*** (.03)	.03(.04)	.05* (.03)	.08* (.04)	.19*** (.04)	.18*** (.05)
Ideology = 6–7	.22*** (.03)	.20*** (.05)	.25*** (.03)	.24*** (.04)	.37*** (.04)	.35*** (.05)
Ideology = Right (8–10)	.30*** (.03)	.27*** (.05)	.45*** (.03)	.43*** (.04)	.48*** (.04)	.47*** (.06)
Age	−.01*** (.00)	−.01*** (.00)	−.00(.00)	−.00** (.00)	−.00(.00)	−.00(.00)
Rural	.11*** (.02)	.11*** (.03)	.23*** (.02)	.25*** (.03)	.14*** (.03)	.15*** (.04)
Economic Perception (worse)	−.50*** (.02)	−.54*** (.03)	−.35*** (.02)	−.35*** (.03)	−.41*** (.02)	−.42*** (.03)
Married	−.02(.02)	.00(.03)	−.07*** (.02)	−.05(.03)	−.05* (.03)	−.00(.04)
Divorced	−.08** (.04)	−.07(.06)	−.02(.04)	−.01(.05)	−.11** (.05)	−.00(.07)
Country-level Controls
Personal Vote-seeking Incentives	.03(.03)	.03(.03)	.02(.02)	.03(.02)	.02(.03)	.02(.03)
% Left Party in Legislature	−.00(.00)	−.00(.00)	−.01(.00)	−.01* (.00)	−.00(.00)	−.00(.00)
GNI Per Capita	2.20* (1.26)	1.53(1.35)	2.75** (1.10)	2.72** (1.27)	1.24(1.28)	1.47(1.32)
Observations	41,830	19,328	42,699	19,503	25,170	12,948
Country-year (N)	31	25	31	25	19	18
Wald χ2	1,318.01	697.53	1,469.75	653.88	837.45	401.64

Multilevel ordered logit coefficients (Stata 15.1). Political ideology = no response and marital status = single are excluded as reference categories. The large drop in observations between the baseline and interactive models is a result of the “working-class citizen” variable not being available for all country-years or respondents in the analysis. GNI = gross national income.

*

p < .10. **p < .05. ***p < .01 (standard errors).

Turning to the interactive models, the coefficient for the interaction between percent workers in the legislature and working-class respondents is insignificant. To assist with the interpretation of the coefficients associated with the interaction terms and their constituent parts, we calculate and graph the average marginal effect of this interaction result for the bottom two and top two response outcomes for each dependent variable (Brambor, Clark, and Golder 2006).

The marginal effects plotted in Figure 4 represent the average effect of an increase in the level of percent workers on the probability of observing each response outcome. Whereas a higher percentage of workers is associated with a decreased probability of observing negative evaluations of the legislature, it is associated with an increased probability of observing positive evaluations. That said, across the three dependent variables, the average marginal effect of a higher percentage of working-class legislators on the probability of observing a given outcome is the same for working-class and other citizens. As a result, we do find support for hypothesis 1, but only to the extent that we observe a significant relationship between higher levels of working-class representation and all citizens’ perceptions of the legislature—including working-class citizens. Importantly this relationship is not different for working-class citizens compared with other citizens. As this positive relationship is the same for working-class and non working-class citizens, we also find support for hypothesis 2.

OPEN IN VIEWER

Figure 4.

Marginal effect of percent workers on predicted satisfaction with representation.

To assess the magnitude of the relationship between percent workers and the outcome variables, Figure 5 plots the predicted probability of the bottom two and top two response categories for each dependent variable across the range of percent workers (i.e., 0% to 16.67%). ²⁰ Given that the marginal effect of the percent workers variable is the same for working-class and non working-class citizens, we use the results in models 1, 3, and 5 to calculate predicted probabilities.

OPEN IN VIEWER

Figure 5.

Working-class legislators and perceptions of legislative representation.

First for legislative approval, a move from the lowest to highest levels of working-class representation is associated with a .08 decrease in the probability of saying the legislature is performing “very badly” (=1), and a .14 decrease in the probability of saying “badly” (=2). Conversely, a move from the lowest to highest level of working-class representation is associated with a .18 increase in the probability of saying the legislature is performing “well” (=4), and a .03 increase in the probability of saying “very well” (=5). We observe the largest change in the probability of citizens saying the legislature is performing “well,” an increase from .12 at the lowest levels of working-class representation to .30 at the highest level.

As for legislative trust, a move from the lowest to highest level of descriptive representation is associated with a .14 decrease in the probability of indicating “no trust” in the legislature (=1), and a .06 decrease in the probability of indicating trust (=2). Conversely, this same increase in descriptive representation is associated with a .08 increase in the probability of indicating trust (=6) and .09 increase in the probability of saying “trust completely” (=7). The largest change in legislative trust is observed for the lowest response outcome—that is, the probability of indicating “no trust” declines from .23 to .09 as we move from the lowest level of descriptive representation to the highest level.

Finally, a similar pattern exists for legislative accomplishment: “how much do you agree or disagree that the legislature accomplished everything you hoped it would.” On a scale from 1 to 7, where 1 indicates not at all and 7 indicates completely, there is a modest decrease in the probability of observing the second response outcome. By contrast, there is a modest increase in the probability of observing the top two response outcomes when moving from the lowest to highest levels of working-class representation. As with the trust dependent variable, an increase in working-class representation is associated with the largest change in the probability of observing the lowest response outcome, a decrease from .27 to .10.

Combined, the results in Figures 4 and 5 lend support to both hypotheses. The average marginal effect of working-class legislators on the probability of observing high levels of legislative approval, trust, and accomplishment is positive and significant for all citizens, lending strong support for hypothesis 2. As the results in Figure 5 demonstrate, higher levels of working-class representation are associated with a decreased probability of expressing negative perceptions of representation, and an increased probability of expressing positive perceptions, on average, for all citizens.

Discussion and Conclusion

Widespread dissatisfaction with political institutions is evident in a number of Latin American countries (Mainwaring, Bejarano, and Leongómez 2006). Indeed, recent behavioral and attitudinal trends in Latin America, such as declining electoral turnout and increasing electoral volatility, led Mainwaring, Bejarano, and Leongómez (2006) to ask, if voters have free choice from an ample array of options, why do they remain so dissatisfied with representation? We argue that continued exclusion of the working class is a critical factor that erodes satisfaction with key agents of representation such as the legislature.

Leveraging elite-based surveys and mass-level public opinion data, we show that in Latin America, descriptive representation of the working class is associated with enhanced evaluations of the legislature. Where workers are better represented in the legislature, citizens are more likely to approve of the legislature’s job performance, trust the legislature, and say that the legislature accomplished everything they hoped it would. Moreover, the symbolic benefit of having more working-class legislators extends to all citizens, not just those from the working class. The observation that everyone benefits from descriptive representation is consistent with prior research findings that higher levels of women’s numeric representation are associated with both men’s and women’s satisfaction with democracy and legislative trust (Barnes and Taylor-Robinson 2017; Schwindt-Bayer 2010; Schwindt-Bayer and Alles 2018), beliefs about women’s ability to lead (Kerevel and Atkeson 2015), and perceptions of democratic legitimacy (Clayton, O’Brien, and Piscopo 2019).

That descriptive representation of the working class enhances all citizens’ evaluations of the legislature suggests multiple mechanisms may explain the link we observe between descriptive representation and citizens’ feelings about representation. First, it is possible that when democracies are more inclusive, they perform better on a number of measures that enhance citizens’ attitudes about representation. For instance, more inclusive legislatures may be more likely to produce distributive policies that improve the lives of multiple groups in society, beyond the working class. Second, the process through which working-class representatives are elected may also foster improved perceptions of representation. If the electoral process results in better descriptive representation, this may engender more trust in, satisfaction with, and positive evaluations of the legislature. Although we are unable to adjudicate between these different mechanisms with the current research design and available data, future research can examine the extent to which descriptive representation of historically marginalized groups, such as the working class, directly or indirectly enhances citizens’ perceptions of representation.

Although our empirical analysis focuses on Latin America, our main arguments are generalizable to other regions where inequality is high, trust in representative institutions is low, and wealthy elites dominate politics. In the United States, political trust has steadily declined since the 1960s (Hetherington 1998), economic inequality is escalating (Bartels 2008), and millionaires control all three branches of government (Carnes 2016). Similar trends are evident in the United Kingdom, with the wealth gap increasing ²¹ and the public desiring more workers in office (Carnes and Lupu 2016b). Our research suggests that increases in working-class representation could help improve perceptions and evaluations of legislatures in such contexts.

With respect to policy implications, the relationships uncovered in this research have important consequences for representative democracy. Our findings suggest that representative institutions would benefit from incorporating a more diverse array of class backgrounds. In Latin America, where class cleavages run deep and the gaps between the rich and poor are the largest in the world, political institutions such as the legislature draw their members from a very narrow set of elites. A common criticism of descriptive representation is that incorporating members of historically marginalized groups into representative bodies is dangerous for democracy, because it could bring unqualified representatives to power (Mansbridge 2015). Yet, our findings that higher levels of working-class representation are associated with better evaluations of the legislature contribute to a growing body of research demonstrating such fears are unfounded. Empirical research from Latin America and the United States, for example, finds that leaders without formal education perform at least as well as their college-educated colleagues—calling into question conventional beliefs about leadership qualifications (Carnes and Lupu 2016c). Similarly, research on Swedish municipal politicians and national legislators finds little trade-off between politicians’ competence and social backgrounds such as class (Dal Bó et al. 2017) and gender (Besley et al. 2017). Others have also demonstrated that descriptive representatives are at least as effective (Volden et al. 2013) and qualified (Murray 2015; Nugent and Krook 2015) as their colleagues.

In the same vein, our findings further rebut the criticism that workers’ descriptive representation could be dangerous for democracy by demonstrating that working-class descriptive representation actually strengthens democracy by enhancing trust in political institutions among working-class and non working-class citizens alike (Cleary and Stokes 2006). Not only are workers objectively qualified to serve as representatives, but citizens are positively disposed to working-class politicians (Carnes and Lupu 2016a; Wüest and Pontusson 2017). Moreover, we find that legislators from working-class backgrounds are associated with enhanced satisfaction with representation and improved perceptions of the legislature. Not only do workers exert a distinct effect on the policy-making process (Carnes and Lupu 2015; Micozzi 2018) but also our findings suggest working-class representation has the potential to wield a profound influence on citizens’ attitudes, thus bolstering the health of democratic institutions.

To address the challenges associated with underrepresentation of marginalized groups, our research suggests that political parties should adopt incorporation tactics such as recruiting representatives from underrepresented groups and working to strengthen representative linkages. Whereas most governments in Latin America have made major strides in the incorporation of women, other groups—such as marginalized urban and rural popular classes, indigenous communities, and the youth—remain on the margins of politics (Barnes and Rangel 2014; Hughes 2011, 2013; Morgan and Meléndez 2016). Although representative institutions can employ a number of different linkage strategies to strengthen their ties with constituents (Kitschelt and Wilkinson 2007), findings from this study imply that the incorporation of representatives from marginalized groups in society—for example, workers, indigenous groups, and the youth—has the potential to restore citizens’ attachments to formal mechanisms of representation.