Gender, Issue Stereotypes, and the Electoral Returns to Distributive Politics in the United States

Abstract

Elected officials can often successfully increase voter support in their district by “bringing home the bacon,” yet theory suggests that the electoral effects of such efforts may depend on the legislator’s gender and whether the legislator delivered benefits in a stereotypically feminine (e.g., healthcare) or masculine (e.g., agriculture) issue area. Using both observational and experimental data in the United States, we find weak, limited evidence that issue area conditions the electoral impact of credit claiming for legislators of either gender. In addition, we show that men and women are rewarded comparably when they secure benefits for their district, regardless of issue area. Our findings suggest that women legislators — typically more effective than men at securing these benefits — can use distributive politics and credit claiming as an effective electoral strategy without concern that issue-based gender biases in the electorate will get in the way.

One way that legislators can show that they are working to support their constituents is by securing government benefits for their district (Cain, Ferejohn, and Fiorina 1987; Mayhew 1974). Doing so can pay electoral dividends. Indeed, scholars have consistently shown that members of Congress (MCs) who bring back federal funds to their districts often secure more electoral support in their next election (e.g., Alvarez and Saving 1997; Lazarus and Reilly 2010; Levitt and Snyder 1997; Stein and Bickers 1994), particularly when they actively claim credit for delivering those benefits as part of their campaign message (Grimmer, Messing, and Westwood 2012).

Congresswomen have particular reason to credit-claim: they are more effective at securing federal funds for their districts than their male counterparts (Anzia and Berry 2011; Volden, Wiseman, and Wittmer 2013). The obvious question then is whether voters reward women for these legislative successes. Such rewards from voters can be critically important for women’s electoral successes because women face more competitive elections and more high-quality challengers, and they must often be more qualified than their male counterparts to win reelection (Barnes, Branton, and Cassese 2017; Fulton 2012; Lazarus and Steigerwalt 2018; Milyo and Schlosberg 2000). Despite a wealth of research suggesting that women face higher evaluation and qualification standards than do men (Bauer 2020; Butler, Naurin, and Öhberg 2022; Fulton 2012), recent work suggests that the electoral returns to credit claiming for district spending do not depend on legislator gender (McLaughlin 2023). Thus, women appear to be rewarded for the money they bring back to their district — that is, just as they would if they were a man.

Our paper examines whether the returns to pork barrel politics for women and men depend on whether they are securing funds for projects that fit into feminine or masculine stereotypic strengths. Voters often see women legislators as well suited to tend to issues such as education or healthcare because these issues fit into broader stereotypes about women as caring and compassionate (Bauer 2021; Huddy and Terkildsen 1993). At the same time, voters see women legislators as lacking the expertise needed to legislate on masculine issues such as crime or national security (Holman et al. 2019). We examine how voters might respond to women who credit-claim on issues that reinforce their stereotypic strengths or overcome their stereotypic weaknesses. The literature offers ambiguous conclusions on this point, with some scholars finding that women receive the greatest electoral benefit from highlighting feminine strengths (Anzia and Bernhard 2022; Bernhard 2021) and other research finding that women benefit from overcoming masculine weaknesses (Bauer 2017). We also compare across genders to assess whether women and men receive a differential boost for bringing back district benefits and credit claiming in feminine versus masculine issue areas.

We test the gendered, issue-based dynamics of credit claiming with two empirical tests, both in the United States. First, we use observational data tracking the funds that MCs have secured for their districts for projects that both fit into and counter women’s stereotypic issue strengths. We complement this analysis with a survey experiment, where we ask respondents to evaluate a fictitious member of Congress who brought back federal spending after randomizing the legislator’s gender and the issue area that they delivered spending in. Across both analyses, we find that, among legislators of either gender, issue area does not clearly and consistently condition the electoral impact of credit claiming, suggesting that women (men) can potentially net the same number of votes when they deliver spending in feminine issues areas as opposed to masculine issue areas. In addition, we find that men and women are rewarded comparably when they secure benefits for their district, regardless of issue area. That is, for example, men and women in Congress receive similar voter support when both deliver new education-focused spending — and demonstrate strength in feminine issue area — for their district.

Our research has important implications for understanding women’s ability to secure reelection and the quality of representation constituents might receive based on the gender of their representative. Like McLaughlin (2023), we conclude that credit claiming for distributive politics can be an effective electoral strategy for men and women in the US Congress. Our new and unique contribution is to show that this holds true no matter the issue or policy area. In doing so, we are the very first to apply an issue stereotypes framework to elections and pork barrel politics. All told, our findings are particularly noteworthy for women, who are both more effective at securing district benefits and often subject to gender stereotypes in the ballot box.

The Electoral Effects of District Spending and Credit Claiming

Voters have mixed feelings about pork barrel spending. Although they oppose earmarks in general, they also want their representatives to work for them. In other words, voters hate earmarks but like it when their own MC secures money for their district.¹ As a result, bringing government dollars back to one’s district can help MCs win reelection (Cain, Ferejohn, and Fiorina 1987; Mayhew 1974). From a theoretical perspective, government aid is thought to help incumbents develop a personal vote independent of their partisanship. Put differently, it enables incumbent politicians to secure votes based on their performance while in office.

Consistent with these claims, empirical research routinely reports a positive relationship between the amount of federal spending brought to a district and support for incumbent legislators and executives in the US (e.g., Alvarez and Saving 1997; Kriner and Reeves 2012; Lazarus and Reilly 2010; Levitt and Snyder 1997). Whether these are direct effects, of course, hinges on whether local voters are aware of federal spending. If voters are unaware of what legislators have brought home, they surely cannot reward them for it. Given low levels of political knowledge in the electorate (Delli Carpini and Keeter 1996; Jerit and Barabas 2017), this may seem unlikely. At the same time, Braidwood (2013) shows that there is a positive relationship between district spending levels and constituents’ ability to recall spending projects, even when adjusting for a variety of other factors including general political interest and knowledge.

Another possibility is that spending affects votes shares indirectly, or conditionally. For example, when incumbents procure more federal dollars for their district, potential challengers do not run (Bickers and Stein 1996). This suggests that pork increases support for incumbents because it weakens the quality of their opposition, netting increased support for the incumbent. Others demonstrate that whether earmarks increase support depends on an incumbent’s credit-claiming activities (Grimmer, Messing, and Westwood 2012). Here, it is not whether incumbents bring back pork, but whether they effectively associate themselves with that spending on the campaign trail.

The effects of federal dollars on vote share may also be conditional on characteristics of the incumbent. Sellers (1997), for example, argues that only fiscally consistent incumbents will see electoral returns to district spending; it is these legislators for whom pork provides a consistent signal of their record and ideological commitments. In support of this claim, he shows that fiscal liberals who bring back federal resources, as well as fiscal conservatives who bring back no money, fare the best electorally. These findings suggest that incumbent legislators may often face constraints in whether they can win votes through federal resources and credit-claiming activities. Despite Sellers’s intriguing argument, there has been limited research focused on whether personal, incumbent traits beyond ideology affect voter responses to increased federal spending in the district.

One especially interesting trait to consider is gender. Women, on average, tend to be highly productive legislators. Congresswomen are more successful at bringing home federal dollars to their districts (Anzia and Berry 2011), and they see more of their bills passed into law compared to men, in part because women legislators are more willing to collaborate and cosponsor legislation (Holman and Mahoney 2019; Volden, Wiseman, and Wittmer 2013). It is thought that the high level of productivity from women legislators comes from a worry about gender bias from voters (Lazarus and Steigerwalt 2018). Research suggests that women are right to worry that voters will hold them to higher standards (Bauer 2020; Fulton 2012), and these higher standards make it likely that women’s records will be evaluated differently than men’s records, even if women have comparable records or better records than men. Along these lines, approval ratings for women legislators are more closely linked to their policy records than they are for men (Kaslovsky and Rogoswki 2022).

McLaughlin (2023)’s experimental analysis, however, finds that there is no effect of legislator gender on voter responses to credit claiming for district spending. In other words, credit claiming for district spending had an equally positive effect on support for male and female legislators. Of course, even if voters do not evaluate the credit-claiming messages of women and men differently, women may face more of a punishment for not credit claiming compared to men given that voters assume that women lack the qualifications and experience needed for political office even if those women are incumbents (Bauer 2020). Consequently, credit claiming for legislative productivity and accomplishments may be one way that women legislators specifically can boost their electoral support.

We build on this recent research in several ways. First, we use both observational data on actual federal, district-level spending over a more than 20-year period, and an experiment, to assess how women and men might differentially benefit from district spending and credit claiming. Second, and most importantly, we expand on existing work to test whether the stereotypic nature of the issue on which a legislator delivers spending affects how voters evaluate women and men.

Will Voters Reward Women for District Spending?

Gender stereotypes shape how voters perceive and evaluate female leaders and candidates (Bauer 2024; Kaslovsky and Rogoswki 2022). Voters view political leadership roles through the lens of masculinity (Sweet-Cushman 2022) but do not readily see women as having the masculine qualities associated with political leadership (Schneider and Bos 2014). Voters also use feminine and masculine stereotypes to form perceptions about women’s and men’s issue expertise (Holman, Merolla, and Zechmeister 2021; Schneider 2014). Feminine stereotypic issues are those that reinforce women’s stereotypic caregiving roles and feminine traits, such as education (Huddy and Terkildsen 1993). Masculine stereotypic issues reflect the power-seeking roles often filled by men and are associated with masculine traits, such as national security and crime (Holman, Merolla, and Zechmeister 2021).

There are competing views on how voters should respond to women legislators based on these perceptions of issue expertise. On one hand, voters may reward women who meet and fulfill stereotypic expectations (Iyengar, Valentino, and Ansolabehere 1996). People in general prefer it when members of stereotypic groups conform to the group stereotype in their behavior (Rudman et al. 2012; Sinclair and Kunda 1999). From this perspective, voters should reward women who deliver for the district on stereotypic women’s issues because these legislators are fulfilling stereotypic expectations. On the other hand, because women are already attributed strengths on stereotypically feminine issues, they may benefit the most from bolstering their credentials on stereotypically masculine issues (Holman, Merolla, and Zechmeister 2011; Schneider 2014). In this case, delivering goods and credit claiming in stereotypically masculine issue areas overcomes the perception that women lack the masculine traits and masculine issue expertise that voters tend to associate with political leadership (Bauer 2020; Schneider and Bos 2014). Our data and research design allows us to test these two alternatives.

H_1a: Women legislators will receive more positive evaluations when they secure funds in stereotypically feminine issue areas relative to stereotypically masculine issue areas.

H_1b: Women legislators will receive more positive evaluations when they secure funds in stereotypically masculine issue areas relative to stereotypically feminine issue areas.

For men, our expectations are more straightforward. We do not expect differences in how voters respond to men who credit-claim on feminine issues or masculine issues. Because voters already see men as experts on masculine issues (Holman et al. 2019), men can credit-claim on feminine issues while still being seen as strong on masculine issues. Men who secure funds and credit-claim on masculine issues will not necessarily lose on feminine issues because voters do not prioritize feminine issues as the most important political issues for legislators to consider (Holman et al. 2019). As such, we expect no differences in how voters respond to men who “bring home the bacon” in feminine issues or masculine issues.

We also outline and test within-issue, across-gender predictions about the effects of credit claiming. Specifically, although voters may respond positively to women who secure funds in masculine issue areas, we do not expect voters to rate women more positively than men who also do so in masculine issue areas. Women candidates and legislators have a much higher bar to overcome in proving their masculine credentials relative to men (Bauer 2020; Ditonto 2017; Fulton 2012). Consequently, women must perform exceptionally better than men in a masculine area to overtake men in their stereotypically masculine strengths.

We also expect men to outperform women when they credit-claim on feminine issues. We base this expectation in social psychology research documenting how people tend to over-reward men who excel at feminine tasks or engage in feminine behaviors (Moss-Rascusin, Phelan, and Rudman 2010). Indeed, recent research finds that male legislators who use a feminine tone or style to credit-claim receive more positive evaluations than female legislators who also credit-claim in a feminine way (Bauer 2024). We expect this feminine boost for men to carry over into how voters evaluate men who credit-claim in feminine issue areas. In sum, we predict that men will receive better evaluations than women on both feminine and masculine issue areas but note that this gain for men occurs on feminine issues for different reasons than it occurs on masculine issues.

H_2a: Men should receive more positive evaluations when they secure funds in stereotypically feminine issue areas relative to women legislators.

H_2b: Men should receive more positive evaluations when they secure funds in stereotypically masculine issue areas relative to women legislators.

Empirical Tests

We conduct two empirical tests on the dynamics between legislator gender, the stereotypic expectations for women and men, and how voters reward legislators who secure distinct benefits for their districts. We combine observational analyses with an experimental approach to maximize both the external and internal validity of our research (McDermott 2002; Morton and Williams 2010). Our first test relies on observational data over a more than 20-year period on the amount of funds that US House members secured for their districts across several stereotypic issue areas in which legislator’s secure funds, and the vote support these legislators receive in the following election. Our second empirical test uses an experimental approach to track whether there are differences in how voters reward women and men who secure benefits that fit into stereotypically feminine and stereotypically masculine issue areas. Taken as one, these analyses offer a strong test of whether and how legislator gender directly affects the ability of legislators to translate district spending and credit-claiming activities for that spending into voter support across different spending and issue areas.

Observational Analysis

We begin with data on district-level federal spending and US House election results from 1984 to 2008. Specifically, we test whether there is a relationship between delivering a particular portfolio of goods and the incumbent’s vote share in the next election. Our analysis therefore relies on the fact that actual district spending and citizen recall about spending are positively correlated (Braidwood 2013). As such, we can view spending measures as a rough proxy for what citizens know about their legislator’s activities and accomplishments, and thus what they could be using to evaluate their incumbent when voting.

Following others, we measure district spending primarily using the Federal Assistance Award Data System (FAADS) database (Anzia and Berry 2011; Berry, Burden, and Howell 2010; Dynes and Huber 2015; Stein and Bickers 1994). The FAADS data provides information on every federal financial award or payment made to domestic nongovernment organizations and individuals. In other words, the data include information on spending for every federal program outside of defense. Each award or payment includes information on where the funds were spent, allowing us to place each award into a congressional district. As in Napolio (2023), we exclude direct payments, as well as formula, need-based grants for which funds are allocated based on well-defined criteria. Instead, we limit our analysis to project grants, as these tend to be more discretionary and thus subject to congressional influence.² Previous research has found substantial variation across districts in discretionary spending, as captured by the FAADS data. Indeed, female (Anzia and Berry 2011), electorally vulnerable (Lazarus 2009), and moderate legislators (Alexander, Berry, and Howell 2016), as well as those in the president’s party (Berry, Burden, and Howell 2010), tend to deliver more spending to their constituents than their counterparts. For women, this is even true in more masculine spending areas, like transportation. Put simply, not every legislator delivers in the same way for their district. We use that variation to test whether, electorally, “delivering more” depends on one’s gender.

We created four issue-specific measures of district spending. These policy-specific measures allow us to assess our primary question of interest: whether stereotypes associated with certain issue areas — that is, whether they are feminine or masculine issues — affect the electoral boost that legislators receive, conditional on their gender. To measure feminine policy spending, we calculate the amount of project grant dollars delivered from the Department of Health and Human Services and the Department of Education. For masculine spending, we use the grants awarded by the Department of Agriculture and Transportation. As an additional masculine issue area, we also draw on a non-FAADS dataset of discretionary military construction projects (Hammond and Rosentiel 2020).³ We chose these stereotypic issues based on past work about how voters perceive the issue expertise of women and men in politics (Bauer 2021; Huddy and Terkildsen 1993). To be sure that people see these issues in these gendered ways, we conducted a small pretest of N = 100 respondents on Amazon’s Mechanical Turk (MTurk) that asked people which gender would be better at legislating on military or health-related issues. We found that 75% of respondents believe that men are better at military issues, and 85% say that women are better at health-related issues.

Within each issue area, we calculate the amount of project grant spending in the district during the election year. Because of right-skew in the data, we log-transform these measures. We then match these spending data with data on US House election outcomes. Our primary dependent variable is the incumbent’s share of the two-party vote in the election. We restrict our analyses to elections where the incumbent served through the entire term (i.e., the incumbent did not win a special election to fill a vacancy) and thus was solely responsible for the federal spending in their district. We exclude members who ran unopposed, as well as races featuring two Republicans or two Democrats.⁴

For each of the five policy areas, we estimate an equation of the following form using ordinary least squares:

$ {Y}_{it}={\beta}_0+{\beta}_1{Spending}_{it}+{\beta}_2{Male}_i+{\beta}_3 Spending\times {Male}_{it}+{X}_{it}+{\theta}_{it}+{\mu}_{it} $

where Spending_it is the log-transformed amount of federal spending incumbent i brought to their district in election year t. Male_i is an indicator variable equal to 1 if incumbent i is male, and 0 otherwise. β₁ gives the effect of spending on vote share for female legislators, whereas β₃ gives the additional increase or decrease in vote share that male MCs receive for the same amount of spending. θ_it are year fixed effects that account for shocks to electoral support common to all legislators. μ_it is an unobserved error term.

Χ_it represents a vector of candidate- and election-level control variables known to affect incumbent vote share in congressional elections. First, we control for the incumbent’s share of the two-party vote in the previous election, meaning that our estimates reflect the boost an MC receives for federal spending relative to how they fared in the previous cycle. In addition, we control for district partisanship using the incumbent party’s share of the two-party vote in the previous presidential election. Combined, these variables account for party strength and the legislator’s typical personal vote above and beyond what their partisanship provides.

We also account for several candidate-level traits and characteristics. Ideologically extreme MCs tend to receive fewer votes (Canes-Wrone, Brady, and Cogan 2002; Hall 2015); we control for this using the absolute value of the MCs’ cycle-specific Nokken-Poole DW-NOMINATE score. Likewise, when incumbents face a quality challenger, they receive fewer votes. Following Jacobson and Kernell (1981), we therefore include in the model a dummy variable equal to 1 if the challenger candidate has held office before. Moreover, when challenger candidates outspend the incumbent, incumbent vote share tends to decrease (Jacobson 1978; Krasno and Green 1988). Using data from the Federal Election Commission, we calculate the spending gap by subtracting incumbent spending from challenger spending, such that positive values indicate the challenger outspent the incumbent.⁵ Finally, we control for the whether the MC is a freshman seeking reelection for the first time, as these members often experience a particularly large “sophomore surge” (Alford and Brady 1989).

The national electoral environment and context also impact the fortunes of incumbents. For instance, in midterm elections, it is well documented that the president’s party tends to lose seats (e.g., Tufte 1975). Moreover, when economic conditions are poor, support for the party in power is lower (Abramowitz 1985). To account for these factors, we control for whether it is a midterm election, whether the MC is a member of the president’s party, as well as presidential approval and the change in real disposable income per capita over the year prior to the election. Importantly, each variable beyond party-in-power is also coded by in-party. Presidential approval (real disposable income) is equal to presidential approval (real disposable income) times -1 for members of the non-presidential party. Likewise, midterm is equal to 0 in a presidential cycle, -1 for the out-party in a midterm year, and 1 for the in-party in a midterm year. Coding these variables by in-party is necessary because the direct effects of these three variables depends on whether the MC is in the president’s party or not.⁶

Finally, we also control for variables that may predict the distribution of federal spending, including whether the MC (1) is a member of the House majority party (Albouy 2013; Balla et al. 2002; Carroll and Kim 2010); (2) sits on powerful committees such as Appropriations or Ways and Means (Evans 1994; Lee 2003); (3) is a committee chair or ranking member (Hammond and Rosentiel 2020); and (4) is a party leader.⁷

Table 1 presents the results.⁸ Columns (1) and (2) present the results for the two feminine issue areas — health and education — whereas columns (3) through (5) report results for the three masculine issue areas of agriculture, transportation, and military construction. We start by assessing H1a and H1b, which concern whether women legislators receive greater voter support when they secure spending in stereotypically feminine issue areas relative to masculine issue areas, or vice versa. We evaluate these hypotheses by comparing the coefficient for spending in Table 1 across issue areas. Because we include an interaction term between spending and gender with gender coded 1 for males and 0 for women, the base term spending then gives the effect of spending within the given issue area among women legislators.

Table 1. Effects of district spending on vote share by gender and policy area

Overall, we find no statistically significant support for either prediction. For example, consider a comparison between the effects of education spending and healthcare spending. The point estimates indicate that for each 1% increase in education spending brought back to the district, women can expect to receive a 0.0008 percentage point boost in their vote share. In contrast, for a similar-sized increase in agriculture spending, women will lose 0.0007 percentage points in the next election. These differences are suggestive that women may benefit more when they deliver spending in feminine issue areas relative to spending in masculine issue areas. Constructing a 95% confidence interval for the difference between these two estimates, though, we find that the difference is not statistically distinguishable from zero (95% = [-0.128, 0.418]). We find a similar (lack of) difference when comparing other feminine and masculine issue areas.

Next, we turn to H2a and H2b. These state that men should receive more positive evaluations than women when they secure benefits in feminine or masculine policy areas. We assess both predictions through the interaction term between spending and legislator gender, coded as 1 for men and 0 for women. The interaction term therefore indicates whether men receive different levels of support than women for bringing back more money to the district in the given issue area.

Across all five issue areas, we find no statistically significant differential effects of spending on vote share by gender — men and women appear to receive the same levels of support for spending, and this is true for both stereotypic feminine and masculine issue areas. It is worth noting that, considering only the point estimates, we sometimes find suggestive evidence that women are rewarded more than men for spending on women’s issues. For instance, for education spending, the interaction is negative, However, the interaction effect for health spending — another feminine issue area — is positive, suggesting that men get more electorally from such spending. Likewise, looking at the masculine issue areas where women are perceived to be at a deficit, the estimates alone suggest that men gain more from agriculture and transportation spending than women. At the same time, women might garner more votes from delivering military construction projects than do men. All told, we find no statistically significant evidence, or even estimates that systematically run in a similar substantive direction within stereotype categories but across issue areas, to suggest that voters reward men and women legislators differently within issue areas.

We also report several additional analyses of different datasets. Hammond and Rosentiel (2020) point out that spending data — i.e., the FAADS data — can lead to timing issues: measures of spending in a given year are a function of spending from new appropriations and spending from prior-year appropriations. Our analyses of the stereotypically masculine military construction projects (column 5 of Table 1) obfuscate this issue, but we are left without data or analyses on feminine issue area projects unaffected by it.

We address this problem with an analysis of earmarks distributed prior to the 2022 election. The 117th Congress (2021–2023) reestablished a method for members to procure pork for their district. Legislators are allowed to request funding to support specific community projects as part of the annual appropriations process. They can make up to 10 requests. Using a dataset of approved appropriation requests compiled by the Bipartisan Policy Center,⁹ we calculate the amount of money each member procured in education and health (feminine issue areas), as well as transportation (a masculine issue area). We reestimate the models in Table 1 using these measures and with vote share as the dependent variable, with one notable caveat: due to redistricting, we do not control for the incumbent’s vote share in the 2020 election. We do, however, control for district partisanship using the 2020 election results recalculated based on the 2022 district boundaries. Given the tighter connection between presidential and congressional voting today (Jacobson 2015), controlling for an incumbent’s vote share is less important than in analyses of the 1980s, for example, when the incumbency advantage was much larger.

The results can be found in Table A2. Overall, our results are similar to those over the longer time period using the FAADS data. We do not find any significant interactions between the type of spending and the gender of the incumbent. We do find a positive and significant effect of health spending for women legislators — an effect that is also distinguishable from the effect of transportation spending (a masculine issue area) among women legislators at the 95% confidence level ([0.028, 0.259]). Thus, in 2022 at least, we find some evidence consistent with H1a. But again, in our model on health spending, the interaction term between spending and male, while negative, is not statistically significant. We thus cannot conclude that the returns to health projects are different for men and women.

Using the 2022 earmarks data, we also assess approval ratings instead of vote share.¹⁰ Other research — including studies focused on gender — has used both favorability and vote support to track democratic accountability (e.g., Ansolabehere and Jones 2010; Ansolabehere and Kuriwaki 2022; Bauer, Harbridge Yong, and Krupnikov 2017; Kaslovsky and Rogoswki 2022). Approval characterizes how district goods and gender affect overall assessments of job performance, whereas vote choice picks up on the downstream electoral impact of goods and gender. In this way, they are measuring different types of evaluations. Indeed, favorability can affect whether a person will vote for a candidate, yet people can also rate candidates unfavorably for whom they still vote (Elis, Hillygus, and Nie 2010). Using these dual outcomes is also particularly important in research focused on gender, as experimental research suggests that social desirability pressures can overestimate vote support for women candidates (Krupnikov, Piston, and Bauer 2016). What is more, an analysis of approval also allows us to include in our analysis those legislators who run opposed — those MCs whom others have shown tend to be most effective at delivering public spending (Bickers and Stein 1996).

We merge the information on earmarks with the 2022 Cooperative Election Study (CES). The CES is useful for this analysis because it asks respondents whether they approve of their legislator and is large (60,000 respondents), such that there are respondents in every congressional district. The results are in Table A3; in addition to spending, gender, and the interaction of the two, we also include as regressors some of the member-level factors used in Table 1 (e.g., ideological extremity, committee memberships, etc.) as well as whether the survey respondent and their representative are copartisans. Table A3 shows the results. Overall, looking at each of education, health, and transportation spending, we find no interaction effects between spending and gender. Although we find a positive and significant effect of health spending for women legislators, it is not statistically different from the effect of transportation spending at typical levels. On approval in 2022, there is no support for H1a.

Our final robustness check considers how individual characteristics might affect whether people reward men and women for different kinds of district spending. Research suggests that individuals who pay less attention to politics are more likely to use stereotypes to negatively evaluate women candidates (Bauer 2015). We therefore reestimate Table 1 but introduce a triple interaction between district spending, legislator gender, and the share of the citizens in the district with a college degree (sourced from Hunt 2022). The results are in Table A4. The coefficient on this three-way interaction indicates whether the extent to which the electoral dividends to district spending are gendered depend on the education level in the district. We find little substantive or significant conditional effect by education level — except for transportation spending, where it appears that high-education districts are particularly likely to reward men for bringing back spending in this masculine policy area.

Experimental Analysis

Our observational analysis reports very weak, if nonexistent, effects. One reason may be because the effects of district spending depend on whether elected officials use district spending as part of their campaign message (Grimmer, Messing, and Westwood 2012), information about which is, of course, not readily available about every congressional incumbent. In other words, our observational analysis is not able to test what happens when legislators actually claim credit for district benefits. We therefore also conducted a survey experiment. Our experiment varied the gender of the legislator and whether the legislator secured funds for a military base (a stereotypically masculine issue) or a hospital (a stereotypically feminine issue) in their district.¹¹

We manipulated legislator gender with the names Carol Hartley or Chris Hartley, which come from past research on gender and voter evaluations (Bauer 2020). Past scholarship uses these names because they are relatively common names across different generations of legislators in office, they do not have a strong association with one racial or ethnic over another, and they are names that are common enough to be those of a legislator (Elder and Hayes 2023). We did not use photos because photos, especially those of women, can affect how people process gender cues based on the attractiveness or perceived age of the candidate (Carpinella and Johnson 2016). Table 2 outlines the conditions and sample sizes for the experiment.¹²

Table 2. Experimental design, October 2022, N = 503

We set up our treatments to appear as a press release or short newspaper blurb recounting a legislator’s recent accomplishments, and within it, we varied the key factors manipulated in our experiments — gender and the type of spending secured. Participants read the following text:

Hartley Brings Funds Home to District

Representative Chris/Carol Hartley, a Democrat/Republican, is proud to announce he has secured $50,000 for Smith Hospital/Military Base. The funds will create new jobs in the district and boost the economy.

“This is great news for our local community,” said Representative Chris/Carol Hartley. “With these funds, our local hospital/military base will have the resources it needs to serve our district.”

We controlled for legislator party so that participants in our study saw a legislator with whom they shared political party (Fridkin, Kenney, and Woodall 2009). Thus, Democratic participants saw Democratic candidates, and Republican participants saw Republican candidates. Those who identified as independents on our party screening question were asked to select the party they leaned more closely toward and were then sorted into a shared-partisan condition.

We conducted our experiment on Amazon’s Mechanical Turk (MTurk), a commonly used online survey platform where subjects receive a payment for their participation. Table A6 provides demographic information on our sample. The sample skews Democratic, liberal, young, and a bit more highly educated than the general population of the US. This sample does tend to be more supportive of voting for women candidates. MTurk, however, is particularly beneficial because studies show these samples tend be less likely to exhibit social desirability pressures that can lead to overestimates in evaluations of women (Krupnikov, Piston, and Bauer 2016). Another benefit of our sample is that most women in elected office are Democrats elected in Democratic-leaning districts. Our sample represents the type of constituent that most women face when they run for reelection. Data quality is always a concern with online survey platforms (Peer et al. 2022). We took several measures to ensure we did not collect low-quality data. First, we chose settings in MTurk and in Qualtrics to detect and eliminate bots. Second, we checked our data for unusual responses such as a large number of participants coming from a single location or participants with unusual personal characteristics (Douglas, Ewell, and Brauer 2023; Kennedy et al. 2020), and we found no suspicious activity.

Following our observational analyses and accompanying robustness checks, we use two key outcome variables to test the rewards that women and men receive when they secure financial benefits for projects in their districts. First, we use favorability, asking participants to rate each candidate’s favorability on a scale of 1 to 7, where higher values indicate a more favorable rating. And second, we ask about vote support using a multiple-choice question with the options of very likely, somewhat unlikely, somewhat likely, and very unlikely to vote for the legislator in the next election. We refrain from including a neutral option, as this can trigger social desirability pressures (Krupnikov, Piston, and Bauer 2016). We recoded both variables to range from 0 to 1, with higher values still indicating more positive evaluations of the candidates.

To reiterate, our first set of predictions examines whether women will fare better when they emphasize feminine over masculine issues (H1a) or when they emphasize masculine over feminine issues (H1b). Our main predictions in H1 center around differences for the woman candidate, but we also test differences for the man using a similar approach.¹³ For our second set of predictions, we focus on differences between women and men and expect to see that men receive more positive evaluations than women when they credit-claim on stereotypically feminine (H2a) and stereotypically masculine issue areas (H2b). That is, we expect men to be rewarded more than women regardless of the stereotypic issue area in which the legislator credit-claims.

Figure 1 displays the mean ratings for the legislators across the four conditions, with panel (a) showing the results for favorability and panel (b) showing the results for vote choice.¹⁴ As before, we begin with H1a and H1b. We find that women appear to be rewarded more for conforming to stereotypic expectations (H1a) rather than countering stereotypic expectations (H1b). In particular, we find she is rated 6.6% more favorably when she brings home funds for a hospital as opposed to a military base, an effect that is statistically significant at conventional levels. We therefore find support for H1a. These higher favorability ratings, however, do not translate into more vote support for the woman, as we find no differences in the likelihood of voting for the woman across the hospitals and the military base conditions, at least at conventional standards of significance (p = 0.177). The fact that higher favorability does not translate into increased vote support may mean that distributive politics and credit claiming are not determinative of vote choice on their own; other factors may simply matter more in the ballot box, even as they work to shape broad assessments of a representative’s job performance. In part, this may also explain the largely null results in our observational analysis — support for H1a in our 2022 vote share analysis notwithstanding.

Figure 1. Candidate evaluations by experimental conditions.

We also conducted these same sets of comparisons for the man, but again, we expect him to do just as well across either issue area. The male legislator’s favorability rating improves slightly by 0.042 points (SE = 0.025), or 4.2%, when he secures funds for the hospital as opposed to the military base. Combined with the findings for women, it appears that voters may over-reward men for credit claiming in a counter-stereotypic issue area, whereas voters reward women for fulfilling stereotypic expectations. We find no significant differences on the vote support outcome, p = 0.347. H2a argues that men will fare better than women when they credit-claim in a feminine issue area. We reject H2a: in the hospital condition, there are no significant differences across gender on favorability ratings (p = 0.774) and no differences on vote support (p = 0.359). We also predicted that men might receive a greater boost when they credit-claim in a masculine issue area relative to women (H2b). Here again, we found no differences across gender in the military base condition on favorability or vote choice.

We also conducted two robustness checks focused on how individual characteristics might affect evaluations. As in our observational study, we assess whether the effects also depend on education levels, as well as political interest. We measured political interest and sophistication in several ways. For political interest, we asked participants, separately, how closely they followed 2022 midterm election news and how frequently individuals followed the news. Higher values on both variables indicate lower levels of interest. We also measured individual levels of education, which we coded as 1 for whether a respondent had a college degree, and 0 otherwise. We estimated models with three-way interactions between each of our interest and education variables with a variable for whether the experimental condition was a woman and whether the legislation brought home funds for hospitals. Our results are in Table A8, and like in our observational data, we find null results across the board.

Table A9 breaks down our results by participant party. We find that Democrats rewarded the woman more for hospital spending than military spending on both favorability and vote support. In contrast, on our favorability outcome, Republicans reward the man for hospital spending more than military spending.¹⁵ Members of both parties prefer when legislators deliver hospital spending, suggesting that healthcare funding is an issue with more diffuse public support relative to military base spending. This finding makes theoretical sense. Both hospitals and military bases can benefit a district through increased employment opportunities for those living in the district, but the benefits of a hospital may be considered more widespread. After all, not everyone, especially civilians, will get employment on a military base or related industry, but everyone may, at some point, need to use a hospital. Hospitals may also be a higher priority for voters in the wake of the COVID-19 pandemic. Members of the two parties differ, however, in one important way: Democrats reward women for reflecting feminine stereotypes, whereas Republicans reward men for doing the same. Future research should explore why.

Discussion and Conclusion

Women in Congress are much more effective at passing legislation and securing benefits for their districts than are men (Anzia and Berry 2011; Lazarus and Steigerwalt 2018) — in part because they have to. Indeed, women incumbents typically face stronger electoral challengers (Branton et al. 2018) and thus are potentially more likely to be defeated otherwise. Canonical political science theory and evidence suggests that women may be able to overcome these challenges by campaigning based on their accomplishments in office and, perhaps especially, the project funding that they have delivered for the district. Recent research suggests that this is indeed an effective strategy, as women face none of the expected biases and instead are rewarded for their “bringing home the bacon” to the same degree as men (McLaughlin 2023). Our paper confirms these effects but, for the first time, shows very weak, limited evidence that they depend on whether the policy area in which those benefits are delivered is a stereotypically feminine or masculine one. Put simply, women can turn district benefits into electoral benefits no matter the kinds of benefits they are able to procure through the legislative process. Our confidence in these findings is enhanced by our use of our both observational and experimental data.

Our work is not without limitations, both empirical and theoretical. Theoretically, we are unable to shed light on why there is a gender bias on roll-call voting (Kaslovsky and Rogowski 2022), but no gender bias for credit claiming on district spending. Second, our study does not directly test whether credit claiming on masculine issues overcomes the perception that women lack the masculine traits and masculine issue expertise that voters see as necessary for good leadership (Bauer 2020; Schneider and Bos 2014). Further, we used issues that broadly fit into feminine and masculine stereotypic strengths, and we did not use women’s equity issues that can disproportionately affect women as a group. Voters may respond differently to women who credit-claim on issues that benefit women more so than men.

Our study controlled for shared partisanship between our hypothetical candidates in the experiment and our respondents. We did not investigate whether credit claiming is a tool that legislators can use to secure support from out-partisans in their districts, especially if a legislator represents a competitive district. This is especially important for women candidates of either political party who serve in competitive districts, as out-partisans can be much more motivated than in-partisans to use gender stereotypes to negatively rate women (Krupnikov and Bauer 2014).

We also did not test the intersection of race/ethnicity and gender. Voters may rate Black women legislators, or other women of color legislators, differently than white women or their male counterparts when they credit-claim on feminine issues or issues perceived to disproportionally affect minority communities. Research suggests that legislators of color support legislation that benefits communities of color (Grose 2010), and when these legislators represent majority-minority districts, they can benefit from providing this type of substantive representation. A slow but steady number of legislators of color represent majority-white districts, and these legislators may face different trade-offs when they credit-claim on issues perceived to disproportionately benefit minority communities — trade-offs that also may be compounded by their gender.

Finally, we note that we examined legislator evaluations for bringing home district funds in the context of American congressional elections. Research in other countries, though, does corroborate our own findings. For instance, research on women executives in Australia suggests an absence of gender differences between women and men in credit-claiming success (de Geus, McAndrews, and Loewen 2020). It is possible that the effects may differ for nonexecutive elected officials, or in countries with parliamentary systems and multimember districts. Although comparative research shows that women in legislatures work to secure benefits for their districts and often do so through collaborative efforts (Barnes 2016), it is not clear how voters evaluate those benefits.

Although we show an absence of gender differences in how voters respond to the credit-claiming behaviors of women and men in politics, we would be hesitant to conclude that gender bias no longer poses a problem for women legislators and candidates. Still, we do find evidence of a more even playing field when it comes to credit claiming for incumbent legislators. Women legislators can, based on our research, mitigate any potential for gender biases through distributive politics and credit-claiming strategies, and they can do so across issue areas. In this way, credit claiming can provide women legislators with a type of insurance against voters relying on feminine stereotypes to discount their qualifications.