Sources of Candidate Fundraising Affect Perceptions of Electability

Abstract

The amount of money flowing in political campaigns in the U.S. has increased dramatically in recent years. Campaign fundraising can provide several signals to voters—especially in primary elections where voters may be considering the electability of candidates in the general election. We conduct a survey experiment to distinguish whether funds influence electability perceptions because they are a war chest of resources that a candidate can use in the general election or because funds signal the candidate’s ability to draw in public support. The results of our survey experiment show that self-funding is less effective than money from donors at increasing primary voters’ perceptions of candidate electability. Our results provide insights into how news coverage on the sources of fundraising might mitigate advantages that self-funded candidates might otherwise have.

Keywords

electability fundraising elections public opinion primary elections

In 2024, each of the top ten Senate candidates in terms of fundraising raised between $47.3 million and $93.7 million dollars (Open Secrets, 2024b). In total, Senate candidates raised $1.6 billion (Open Secrets, 2024a). This campaign funding can affect how people vote by increasing name recognition and persuading voters through advertising (Hewitt et al., 2024) or by shaping how individuals expect the candidate to perform if elected (Ashworth, 2006). Successful fundraising can also provide signals about the ability to compete in the general election (Gerber, 1998; Green & Krasno, 1988, 1990; Jacobson, 1990) as well as about candidate quality (Bond et al., 1985). In part because partisanship cannot be used to differentiate candidates, campaign finance plays a larger role in primary than in general elections (Bonica, 2017; Thomsen 2023, 2025)—a differential that might be increasing (Porter & Treul, 2024). Within primary elections, fundraising figures potentially serve an important signaling function regarding which candidate is the most electable (Anderson et al., 2025)—which candidate has the best chance to carry their party to victory in the general election.¹

For some candidates, a significant portion of their funds come from self-funding; 65 House, Senate, or presidential candidates contributed more than $1 million to their own campaigns in 2024 (Open Secrets, 2024c). For instance, Wisconsin Republican Senate candidate Eric Hovde self-funded $20 million (63%) of his $31.6 million total. Sometimes media coverage highlights the source of funds, but other times it does not. In Eric Hovde’s case, for example, two posts on X about 2024 first quarter fundraising reported that he had raised $9.1 million in the first quarter; one provided the additional detail that this included an $8 million loan from himself while the other just reported the total (see SI 7 for screen shots of these posts).

The variability in self-funding and the coverage of it raises the question of whether voters respond to the specific sources of a candidate’s campaign funding in assessing electability. We identify two ways fundraising can shape perceptions of electability. First, voters may value the sheer amount of resources that a candidate has because these resources can be used to boost their performance in the general election (Gerber, 1998). In this view, it does not matter where the “war chest” comes from; it only matters that it exists. Second, voters might conceptualize successful fundraising primarily as an indicator of a candidate’s support among the public and political elites (Hassell 2018, 2023)—support that will help them secure both donations and votes in the general election. In this case, voters might discount self-funding that comes from the candidates themselves (Brown, 2013) and award credit only for money raised from political contributors.

We use a survey experiment that varies the degree of self-funding to assess the effect of self-funding on electability. The results show that, dollar-for-dollar, self-funding is less effective than money from donors at increasing primary voters’ perceptions of candidate electability. These results are most consistent with the hypothesis that voters are using campaign fundraising as a signal of the public support the candidate enjoys.

The survey experiment and estimation offer an additional methodological contribution to the literature on campaign finance because they demonstrate that the inherent multicollinearity between total fundraising and its constituent sources can raise challenges with interpretation. Given that total fundraising is the sum of fundraising from each source, it is impossible to estimate models that include each type of funding and the total amount raised in the same regression. We demonstrate how the coefficients from a survey experiment can be correctly interpreted to test if primary election voters value campaign fundraising as a signal of electability because they increase the sheer resources the candidate has or because they are sign of the support the candidate enjoys.

By revealing the mechanisms by which fundraising affects electability, we provide cautious optimism that candidates who lack independent wealth may still be able to compete with wealthier candidates under some circumstances. The amount of funding needed to run a modern campaign creates barriers to entry for candidates not from the wealthy segments of society or those without wealthy professional networks (Bonica, 2020; Carnes, 2020), especially as many wealthy candidates self-fund major parts of their campaigns (Ovtchinnikov & Philip, 2023; Steen, 2009). The costs of running for office and the potential power of self-funding raise normative questions about who can afford to run for office and whose voices are represented (Page & Gilens, 2020). However, our results suggest that although personally wealthy candidates have an advantage in total campaign fundraising, some of that advantage might be mitigated if news outlets more regularly report on the source of candidate fundraising.

How Sources of Funding May Affect Electability

In the two-stage election process of the United States, voters have reason to assess and vote for primary election candidates based on how much the candidate appeals to them and based on their perception of the candidates’ electability in the general election (Bartels, 1988). In fact, primary voters are more likely to vote for a more ideologically distant candidate when told that candidate is more electable (Simas, 2017), particularly for voters with negative views of the opposing party (Albert & Costa, 2024). These electability considerations appear to have been in play in many presidential elections, including the Mondale/Hart 1984 Democratic primary (Bartels, 1987) and the 1988 and 2004 presidential primaries (Abramson et al., 1992; Rickershauser & Aldrich, 2007). Recent research focuses on the importance of electability and what drives perceptions of electability in congressional elections (Anderson et al., 2025; Green et al., 2023; Hassell & Visalvanich, 2024).

To some extent, voters appear to develop expectations about the electability of candidates that are largely in line with expert assessments, with perceptions of electability increasing for more moderate candidates, those with prior elected office experience, and those who raised more money for their campaign (Anderson et al., 2025). Voters also seem to judge electability on the basis of demographic features such as gender and race, with women and candidates of color being assessed as less electable (Green et al., 2023). Of course, it is also likely that voters base their perceptions on their own ideological preferences and identities (Hassell & Visalvanich, 2024), and they may have inflated views of the electability of their preferred candidates.

Campaign finance information can provide one indicator of electability via its role in signaling a candidate’s quality, resources, and potential for support. Campaign finance provides indications of the quality of a candidate, in part because quality candidates are likely to raise more funds (Brown, 2013). Fundraising may also be an indicator of the candidate’s skills and support or the resources they could bring to the general election. Early fundraising in primary elections can pay dividends for future fundraising efforts (Biersack et al., 1993; Case & Porter, 2025; Magleby et al., 2018; Smidt & Christenson, 2012). In fact, early fundraising can boost a candidate’s likelihood of prominent endorsements (Blum et al., 2024), may signal strength to donors, media, and party organization (Thomsen, 2025), and is associated with electoral success in primary elections (Bonica, 2017; Porter & Steelman, 2023). Fundraising success might also signal greater electability in the general election (Anderson et al., 2025).

More specifically, the sources of campaign funds may affect voters’ perceptions because voters can use the source to make inferences about the candidate. There are many features of fundraising that voters might consider. For instance, voters might dislike candidates who receive money from Super PACs (Goodliffe & Townsend, 2024) while preferring candidates who rely on small dollar donors, since this fundraising source might signal that the candidate has broad backing from the public² or reduce concerns about potential corruption in the system (Malbin, 2013; Culberson et al., 2019; Albert & Raja, 2020). The prevalence of small dollar donors might also signal other features about the context of the race or candidate, as research has shown that small donor contributions are more common in competitive races and for ideologically extreme incumbents (Albert & Raja, 2020; Culberson et al., 2019).

Voters might also consider whether funds come from outside the district or state versus from within the district or state. While early money from inside the district historically predicted future fundraising and electoral success in the primary, a shift in more recent years reveals that political amateurs are more successful in future fundraising and electoral success when they raise money from outside the district (Porter & Steelman, 2023). Outside funding might increase the overall resources a candidate has and signal grassroots support among partisans, while funding from within the district might signal the appeal of the candidate to prospective voters.

Finally, and the focus of this paper, voters might consider self-funding from a candidate differently from funds raised from other sources. We focus on how the source of fundraising affects electability perceptions in the context of a primary election. We specifically contrast self-funding against campaign contributions from other sources. Campaigns that self-finance can potentially signal two things to voters: the potential and willingness to put more personal resources into campaigns but also that they might be in need of an infusion of campaign funds. Our survey experiment largely precludes the possibility that respondents will see self-donation as an indication that the campaign is struggling, because we describe them as one of the “leading” candidates in the primary race. Significantly, by precluding the negative aspect of the signal, our test likely gives an upper-bound on the impact of self-funding.

Our approach has the upside of focusing more directly on the resource aspect of self-funding, which allows us to distinguish between two mechanisms via which fundraising might signal electability: (1) because the funds reveal a resource that will be useful to a candidate in the general election or (2) because the funds demonstrate the ability of the candidate to secure public support. If fundraising primarily signals the volume of resources available to run a robust campaign in the general election, then whether the money comes from the candidate themselves or other contributors should not matter: greater fundraising can increase voters’ perception that the candidate will have lots of resources to use in the general election irrespective of whether the funds come from donors or the candidate’s own wealth. The bigger the candidate’s war chest, the more the candidate can spend to win (Gerber, 1998). In general, fundraising can signal a strong candidate (Thomsen, 2025) and even the self-funding portion may signal to voters that the candidate is financially competent and successful, and thus potentially of higher quality. The literature suggests that fundraising may not operate on electability via deterrence of challengers, as the findings in this area are mixed (Box-Steffensmeier, 1996; Epstein & Zemsky, 1995; Goodliffe, 2001). Instead, the resources a candidate has may indicate their ability to run a robust campaign, connect with voters, and help them campaign in a competitive general election.

By contrast, money from donors (other than the candidate themselves) might provide information that self-funding does not. Voters might use funding from donors as a stronger signal about a candidate’s quality than the total amount of campaign resources on its own. High-quality candidates might use their charisma, their personal networks, or their campaign skills to secure more donations. They might be able to use those same skills to secure more votes. If fundraising amounts indicate the potential for public support, then self-funding will not improve perceptions of electability because only funds raised from other individuals or PACs show enthusiasm from supporters outside the campaign.³

There are a variety of reasons that self-funding is likely to be less effective at improving perceptions of a candidate’s electability than overall fundraising. Consider the observed relationship between self-funding and electoral success. At best, money that comes from self-funding has little to no relationship with vote share (Brown, 2013; Steen, 2009); at worst, self-funding has a negative impact on vote share (Alexander, 2005). Moreover, wealthy candidates do not deter strong challengers or win more often (Milyo & Groseclose, 1999). By contrast, the amount coming from externally-financed spending is associated with a higher vote share, with the logic being that money flows to the strongest candidates (Alexander, 2005; Brown, 2013). The same logic could apply when voters evaluate electability; externally-financed spending signals quality while self-funding does not. While self-funding is somewhat predictable (e.g., wealthy candidates are more likely to do it, see Steen (2009)), it is unlikely to be strategic, conditional on entering the race (Brown, 2013, p. 31) and thus not a strong signal of viability or electability. A candidate’s personal contributions to the campaign could matter in other ways as well, some positive and some negative. On the positive side, primary voters might prefer a candidate who can self-fund because it signals they are financially competent, successful business leaders, or are less likely to bought be special interests backing their campaign. On the negative side, primary voters might fear that a candidate who heavily self-funds a campaign could be perceived by general election voters as elitist and out of touch or less accountable to constituent interests. For instance, Steen (2009, p. 15) finds that self-funded members of Congress are less likely to engage in communication formats with constituents that allow them to learn and respond to constituent concerns. These perceptions could contribute to public cynicism about campaign finance (Primo & Milyo, 2020).

These considerations lead us to two hypotheses related to the two potential mechanisms for how campaign funding could influence perceptions of primary election candidates’ perceived electability in the general election:

Fundraising as a sign of resources. Information about higher levels of campaign funding, regardless of its source, will increase primary voters’ perceptions of a candidate’s electability.

Fundraising as a sign of ability to gain public support. Only information about higher levels of campaign funds raised from contributors will increase primary voters’ perceptions of a candidate’s electability. As a result, self-funding will not increase perceptions of electability.

These hypotheses conceptualize self-funding in simple dollars-and-cents terms. That is, H1 predicts $1 million of self-funding to increase electability by the same amount, irrespective of whether $1 million represents a small or large percentage of a candidate’s total pool of funds. After we present our main results, we discuss and consider self-funding in percentage terms.

Survey Experiment on Campaign Fundraising and Electability

We test these hypotheses with a pre-registered survey experiment in which respondents were asked to imagine themselves voting in their preferred party primary for U.S. Senate elections and to evaluate candidates based on a mock news story in which the amounts and types of funds raised by candidates were manipulated. We included our experiment in a survey of 1,516 respondents, fielded by Qualtrics in July and August 2023. Although not perfectly representative of the U.S. electorate, the sample was targeted to meet Census benchmarks for gender, age, race, and education. Table SI-1 in the Supporting Information reports a demographic breakdown for the sample and Table SI-2 demonstrates balance in demographics across the survey conditions.⁴

Survey respondents were presented with the vignette and question shown in Figure 1. In the vignette, respondents read a news story about two candidates competing in a primary election. The vignette was designed to mimic the style of the many media stories we reviewed about candidate fundraising in primary campaigns. We assigned people to be asked about the party whose primary each respondent indicated voting in during the 2022 midterms. If the respondent did not vote in a particular party’s 2022 primary election—e.g. they participated in a blanket primary or did not vote in the primaries at all—they were presented with a randomly assigned primary.

Candidate and opponent names were randomized across four names. Candidate 1 was Amy Mueller, Garrat Novak, Jack Evans, or Molly Kruger. The opponent was Wyatt Smith, Sarah Miller, Kristen Clark, or Greg Adams. The text used pronouns matched to the stereotypical gender for each name. Using multiple names ensured that the results we found were not the result of a specific pair of names we chose. We found no evidence that names or the genders they imply affected the results and so we pool the analyses across names.

The key for the study is that we varied how much money Candidate 1 raised in total and how much was self-funded. As Figure 1 shows, we reported that candidate 1 raised $Total million during the second quarter of the election year, a value randomly drawn from between $4.5 million and $11.2 million (with $0.1 million increments). Candidate 1 was juxtaposed with an opponent (Candidate 2) who raised $4.3 million. We held constant how much Candidate 2 raised in the reporting period. Respondents also read that one difference between the candidates is that $Self-funded million of Candidate 1’s fundraising total came from a personal loan the candidate made to her campaign. This wording encourages respondents to infer that self-funding is a distinct aspect of Candidate 1’s fundraising.⁵ We randomized the amount of self-funding to be between 10% and 90% of the total amount they had raised (which gives a range of about $0.5 million to $10 million for the self-funded amount). In the vignette, we reported both a dollar amount that Candidate 1 had given to their own campaign and what percent of the total that represented for their campaign.^6,⁷

Following the vignette, we asked: “For each candidate, what is the percent chance that they will win the November general election if they win the [party] primary?” Respondents indicated their response via sliders for each candidate that ranged from 0 to 100. Respondents were told that a value of 0 meant the candidate has no chance of winning the general election, a value of 50 meant that the winners of the Democratic and Republican primaries both has an equal chance to win the general election. A value of 100 meant that the candidate will definitely win the general election. We reminded respondents that the question was about how likely each candidate is to win the general election against the out-party nominee, not about who would win the primary between Candidate 1 and Candidate 2. While this language stressed that respondents should think about electability, we cannot rule out that primary viability considerations still play some role in respondents’ electability perceptions. We define Candidate 1’s electability advantage—our main dependent variable—as the difference between the two scores each respondent entered (Candidate 1 – Candidate 2). This variable is a continuous measure running from −100 to +100.⁸

Testing the Competing Hypotheses

To identify the effects of different sources of campaign funding, we begin with a psychological model of how fundraising information might affect a voter’s electability perceptions:

E l e c t a b i l i t y A d v a n t a g e = α + t T o t a l + d D o n o r s + s S e l f ‐ f u n d e d + ε

(1)

In this model, a candidate’s perceived electability advantage is a function of Total funds raised, funds raised from Donors, and Self-funding. At first blush, this model has a quirk: because our setup only includes two sources of campaign funding, Total is simply the sum of Donors and Self-funded. As such, perhaps our starting point should simply be:

E l e c t a b i l i t y A d v a n t a g e = α + d D o n o r s + s S e l f ‐ f u n d e d + ε

(2)

We prefer the more flexible equation (1) because it is more psychologically realistic. Although Total is simply the sum of Donors and Self-funded, it is nonetheless plausible that each perception might have distinct psychological relevance. That is, the same quantitative information could influence people in different ways under different mathematical transformations, as in when support for a medical intervention changes depending on whether it is described as lives “saved” versus “lost” (Tversky & Kahneman, 1981). In our case, Total funds could have distinct psychological effect on perceptions of electability above and beyond the values of the two underlying components. An advantage of using equation (1) as the starting point for our analytical framework is that, because we start from a less restrictive model, our results can be understood in different terms by applying appropriate assumptions. For instance, to consider how to interpret our results under the supposition that equation (2) is the correct framework, one can simply assume that t = 0 in the derivations we describe next.

Although equation (1) is a plausible psychological model, it is not directly estimable, due to the interdependence described above: Total is the sum of Donors and Self-funded. This difficulty does not invalidate equation (1). Rather, it places us in a situation similar to when demographers seek to simultaneously estimate independent effects of age, time period, and birth-cohort. All three components are theoretically important. But because any two of them mathematically define the third, it is not possible to estimate effects for all of them in a standard model (Fosse & Winship, 2019).

In our case, the remedy is careful attention to how the results our regression models can produce relate to theoretical quantities of interest. Fortunately, our objective is not to estimate independent effects of Total, Donors, and Self-funded. Rather, we seek to recover quantities for which the hypotheses about sheer resources (H1) and public support (H2) generate different testable implications. Such quantities are estimable in our data. To see how, use the identity, valid in our experiment:

D o n o r s = T o t a l - S e l f ‐ f u n d e d

to make substitutions that eliminate Donors from equation (1), as in:

E l e c t a b i l i t y a d v a n t a g e = α + t T o t a l + d (T o t a l - S e l f ‐ f u n d e d) + s S e l f ‐ f u n d e d + ε

(3)

Algebraic rearranging then yields:

E l e c t a b i l i t y a d v a n t a g e = α + (t + d) T o t a l + (s - d) S e l f ‐ f u n d e d + ε

(4)

Next, compare equation (4) to our pre-registered regression model:⁹

E l e c t a b i l i t y a d v a n t a g e = β_{0} + β_{1} T o t a l + β_{2} S e l f ‐ f u n d e d + ε

(5)

Comparing equations (4) and (5) clarifies that the coefficient $β_{1}$ should not be interpreted as the independent effect of the total amount of campaign funds.¹⁰ Instead, $β_{1}$ gives the combined effect of a one unit increase in the total amount raised and a one unit increase in the amount raised from donors. This interpretation makes intuitive sense when one realizes that, because equation (5) holds constant the amount of money that is self-funded, any increase in the total amount raised must come from an increase in the amount raised from donors.

Although equation (5) requires careful interpretation, it leads to clear predictions for H1 and H2. Both hypotheses predict a positive coefficient for $β_{1}$ (the coefficient on Total fundraising in equation (5)). The “sheer resource” hypothesis (H1) does so because it expects the t component of equation (4) to be positive (money helps), and the d component to be non-negative (no per se penalty for money that comes from donors). Thus, t + d (in equation (4)) will be positive. For slightly different reasons, the “public support” hypothesis (H2) also predicts $β_{1}$ to be positive. The hypothesis predicts t to be non-negative (no per se penalty for total funds raised) and it predicts d to be positive (money from donors helps). As a result, t + d will again be positive.

Our two hypotheses lead to different predictions for

β_{2}

(the coefficient on the self-funded amount in equation (5)). Comparing equation (5) with equation (4) shows that the coefficient

β_{2}

estimates the effect of the s – d. The “sheer resource” hypothesis (H1) implies that voters are indifferent to the source of funds. Thus, s and d should be the same size. Thus, under H1, s – d will be 0 and so

β_{2}

is predicted to be 0. In contrast, the “public support” hypothesis (H2) is premised on the idea that voters discount self-funding: that d is larger than s. Thus, under H2, the quantity of s – d will be negative, so

β_{2}

is predicted to be negative. Table 1 summarizes our expectations.

Table 1.

Expectations for Hypotheses

Variable	Coefficient in equation (5)	Theoretical quantity in equation (4)	Expectations
Variable	Coefficient in equation (5)	Theoretical quantity in equation (4)	Hypothesis 1 (sheer resources)	Hypothesis 2 (Public support)
Total funds	$β_{1}$	t + d	+	+
Self-funded $	$β_{2}$	s – d	0	-

Deriving how researchers can test specific theoretical relationships despite the dependencies that are characteristic of fundraising information is one of our manuscript’s contributions. For completeness, in the appendix, we show how our study leads to identical conclusions about H1 and H2 under two alternative estimation approaches—substituting for either Total or Self-funded (Table SI-3).

The Differential Impact of Funding Sources on Electability Perceptions

Table 2 presents the main regression models, estimated both for all respondents and separately by party identification with “leaners” coded as partisans.¹¹ In our pre-analysis plan, the primary model specification includes all respondents. We noted that other exploratory analyses would examine partisan differences. While we did not have a priori expectations that the results would differ across partisan groups, extant literature points to some evidence of partisan differences in how people assess electability or use electability in their vote choice (Anderson et al., 2025; Hassell & Visalvanich, 2024; Kelly & Eric, 2022), though other research points to similarities in how voters weigh electability (Simas, 2017).¹² We focus on the results in column 1 that use all the respondents in the sample.¹³ Recall that

β_{1}

measures a one-unit increase in Candidate 1’s total fundraising amount, which gives the effect of “t + d” (see Table 1). The associated estimate is 1.19. Because the Total variable ranges from 4.5 to 11.2, movement across its full range would be associated with a change of approximately 8.0 in the dependent variable, which given the DV’s standard deviation of 28.9 is an effect of 0.28 in standardized terms. This effect is statistically significant. Both H1 and H2 make directional predictions about

β_{1}

(Table 1), so one-tailed hypothesis tests would be appropriate. But the estimate is significant by either standard (one-tailed p = 0.004, two-tailed p = 0.008) and consistent with both hypotheses (which each predict a positive coefficient on this variable in this model).

Table 2.

Effects of Self-Funding on Electability

DV = candidate 1’s relative electability advantage
	All	Democrats	Republicans	Independents
Total funding (dollars)	1.19	0.74	1.60	1.66
	(0.45)	(0.68)	(0.76)	(0.83)
Self-funding (dollars)	−0.74	−0.74	−0.61	−1.06
	(0.41)	(0.63)	(0.69)	(0.73)
Constant	−3.13	1.85	−8.40	−6.17
	(3.24)	(4.87)	(5.52)	(5.96)
N	1,389	645	538	206
R ²	0.005	0.003	0.008	0.021

Note. Standard errors are in parentheses.

The coefficient on Self-Funding $ ( $β_{2})$ provides clear support for the public support hypothesis (H2) and comes close to rejecting the sheer resources hypothesis (H1). The coefficient on Self-Funding is negative as predicted by H2 and equal to −0.74. In terms of substantive significance, Self-funding ranges from 0.5 to 10.0, so movement across the full range amounts to an electability difference of −7.1 (−0.24 standardized). In terms of statistical significance, H2 makes a directional prediction that $β_{2}$ will be negative, suggesting that a one-tailed test is appropriate. We can reject the null hypothesis that the $β_{2}$ coefficient is non-negative with a one-tailed test (p = 0.035). While our results are consistent with H2, we fail to reject H1 in this test using the conventional 0.05 level because H1 predicts that the coefficient would be zero, implying that a two-tailed test should be used for H1. The p-value for that two-tailed test is 0.070, which is above the standard 0.05 level (though close).

The models in columns 2 through 4 separate the sample by partisan affiliation. These models have lower statistical power due to smaller sample sizes. Still, on balance the results more closely align with Hypothesis 2 than Hypothesis 1. It seems that respondents see campaign resources that are not self-funded as a more meaningful indicator of electability because they convey broader support for a candidate that will presumably carry over to the general election. Self-funding might be helpful to a candidate’s chances and allow them to build other aspects of an effective campaign (Biersack et al., 1993; Smidt & Christenson, 2012), but our survey respondents placed more weight on funding from other sources when assessing electability.¹⁴

As outlined in the pre-registration plan, we also estimate the baseline model after restricting the sample to respondents who indicate they have voted in recent primary elections. Tables SI-4a and SI-4b in the Supporting Information report models for individuals who say they voted in a primary in 2022 and for people who say that “always” or “nearly always” vote in congressional primaries. In these analyses, which exclude approximately 40% of our observations, $β_{2}$ , key for adjudicating our hypotheses, is statistically significant in only one case (for independents who regularly vote in primary elections, one-tailed p < 0.03). However, it is directionally consistent with H2 in seven out of eight cases (the exception being the model for Republicans who regularly vote in primaries).

Robustness Check: Examining Self-Funding as a Percentage of Total Fundraising

To this point we have examined self-funding in absolute dollar terms. This approach implicitly assumes that, in developing electability perceptions, voters would attend to the sheer amount of money that a candidate contributes to their own campaign. An alternative psychological model of electability perceptions might suppose that voters attend to the percent of that total that was self-funded—perhaps because percentages more strongly convey how reliant a campaign is on the candidate’s personal wealth. Our design allows us to examine how electability perceptions respond to the percentage of self-funding, because such percentages can be derived from the fundraising values we randomly assigned and indeed were explicitly presented to respondents (see Figure 1), increasing the possibility that respondents were thinking in percentage terms. The only difference between the analyses in the previous section and the ones we report here is that self-funding will now enter our models as a percentage of the total, rather than a dollar amount:

Y = γ_{0} + γ_{1} T o t a l + γ_{2} P e r c e n t S e l f ‐ f u n d e d + ε

(6)

While the analyses in the previous section were pre-registered, the analyses here were not.

Before turning to the results, note that although $γ_{1}$ and $γ_{2}$ in equation (6) do not map onto theoretical quantities the way $β_{1}$ and $β_{2}$ did in Table 1, the directional predictions in columns 4 and 5 of Table 1 remain the same even when Self-funding is operationalized in percentage terms. If H1 is correct, $γ_{1}$ will be positive because if the candidate has more money, then people will perceive them to be more electable. Further, $γ_{2}$ will be equal to 0 because if candidates’ resources are all that matters to voters’ perceptions about electability then it does not matter where those resources come from and how much or little of those resources are self-funded. If H2 is correct, and perceptions of electability are influenced only by how much money comes from donors, with self-funding not having an effect, then $γ_{1}$ will be positive and $γ_{2}$ will be negative. The coefficient $γ_{1}$ will be positive because when the percent of funding that is self-funded is held constant, increasing the total funding means that you are increasing the amount of funding coming from donors. For example, if a candidate had raised $2 million and 50% of that was self-funded then that would indicate that $1 million had come from donors. If a different candidate had $4 million and 50% of that was self-funded then that would indicate that $2 million had come from donors. If H2 is correct, $γ_{2}$ will be negative because an increase in the percent that is self-funded when holding constant the total amount of funding means that there is a decrease in the amount of money coming from donors. And decreasing the amount of money from donors will lead to a decreased perception of electability if H2 is correct.

Table 3 reports the relevant results. As before, we focus primarily on the results for all respondents in Column 1. The coefficient of −6.89 associated with self-funding (percentage) implies that, holding the amount of fundraising constant, shifting from an entirely donor-supported campaign to an entirely self-funding campaign would decrease Candidate 1’s electability advantage by 6.89 points. This relationship is significant under either a one- or two-tailed hypothesis test (one-tailed p = .022; two-tailed p = .043). Thus, this is again consistent with the “public support” hypothesis (H2). In addition, the results in Table 3 lead us to reject the “sheer resources” hypothesis (H1), which predicts this coefficient should be zero, which it is not (two-tailed p = 0.038). As before, the same basic pattern arises in each partisan subgroup, though Republicans appear more responsive to total funding than Democrats.

Table 3.

Effects of Percentage Self-Funded on Electability

DV = candidate 1’s relative electability advantage
	All	Democrats	Republicans	Independents
Total funding (dollars)	0.80	0.35	1.28	1.12
	(0.40)	(0.60)	(0.67)	(0.73)
Self-funding (percentage)	−6.89	−5.18	−7.44	−10.38
	(3.31)	(5.06)	(5.58)	(5.91)
Constant	0.50	4.59	−4.45	−0.89
	(3.68)	(5.57)	(6.29)	(6.58)
N	1,389	645	538	206
R ²	0.006	0.002	0.010	0.025

Note. Self-funding is scaled from 0 to 1. Standard errors are in parentheses.

News Coverage of Candidate Self-Funding

It is important to understand how voters respond to information about self-funding because they are exposed to variable information through news reports. More thorough reporting on self-funding is one mechanism by which candidates without personal wealth could close the electability gap that may arise from unreported self-funding. To see how often news coverage mentioned the amount of self-funding, we examined coverage of candidates in the state newspapers (the top-two by circulation with online accessibility) in the three months leading up to 19 competitive statewide primary elections in 2022. Our search identified 127 articles (about 25% of all articles) that mentioned the campaign fundraising of one or more candidates in the race.

To see how often the campaign fundraising articles discussed the candidates self-funding, we further restricted the sample to U.S. Senate races where one or more candidates self-funded at least $1 million.¹⁵ There were 87 articles about campaign fundraising in these races, with one-third of those articles (29) mentioning self-funding. Table 4 gives the results by primary election and highlights the considerable variability across these contests. The news articles about fundraising in the Democratic Wisconsin primary and the Republican Ohio primary, in which there was significant self-funding, mentioned the personal funding of the candidates most frequently. In Wisconsin, Alex Lasry put more than $14 million of his own funds towards the Democratic primary; and in Ohio four different candidates put millions of dollars of their own wealth into the primary: Gibbons $18 million, Dolan $10 million, Timken $4 million, and Moreno $3 million. Personal campaign funds were big news in both primaries. Surprisingly, there was much less coverage of self-funding in the other primaries that also included candidates using millions of dollars of their own money in the campaign. In Pennsylvania, this included large amounts of funding from Oz, McCormick, and Sands, while Lamon put more than $18 million of his own funds on an unsuccessful primary campaign in Arizona. This analysis demonstrates the variability in reporting about self-funding, suggesting both that many voters are exposed to information about self-funding and that more consistent news reporting about self-funding could close the electability gap that might otherwise arise from the absence of coverage paired with the ability of wealthy candidates to self-fund.

Table 4.

Newspaper Coverage of Competitive US Senate Primaries in 2022

	Articles discussing campaign funds…
Senate Primary race	But NOT personal funding	AND personal funding
Alabama–Republican	8	3
Arizona–Republican	2	0
New Hampshire–Republican	11	1
Ohio–Republican	18	10
Pennsylvania–Republican	12	1
Wisconsin–Democratic	7	14
Total	58	29

Notes. Count of articles from six U.S. Senate primaries with self-funding, where at least one candidate self-funded at least $1M.

Conclusion

We explored two ways in which campaign finance funding could influence primary voters’ perceptions of candidate electability in the general election: a sheer resource model and a signal of support model. Our survey experimental evidence is most consistent with voters trying to use money as a signal of support when deciding their perceptions of electability. While candidates who self-fund their campaigns may still be able to use such money to successfully execute their primary and general election campaigns (though see Brown, 2013), raising funds from donors is a better signal of electability to the primary voters whose support candidates need at that stage.

Future research might further investigate how voters respond to other aspects of campaign finance. For example, how does the public view candidates who take advantage of the Supreme Court decision in FEC v. Ted Cruz allowing elected officials to loan money to their campaign but pay themselves back post-election from donations? Also, do voters differentiate money from small-money donors versus candidate committees, party committees, PACs, or party leadership? What about money from out-of-state donors versus in-state donations? And does the ideology of the donor shape voters’ perceptions of the candidates? Further, researchers could look further at how candidates and the media talk about campaign finance. When and how do journalists cover the amount of self-funding used in political campaigns? Does the coverage of self-funding depend on the partisanship of the candidate or the amount that a candidate donated to their campaign? Does that coverage depend on the context of the campaign? Is it more likely to be covered in primary elections because there are potentially fewer other differences between the candidates running?

Our results showing that not all campaign funding is equal, at least in the eyes of voters, has implications for both candidates and the media. For candidates, there is value in fundraising from donors – and highlighting that fundraising – even if they could completely self-fund their own campaign. Funding from donors improves voters’ perceptions of the candidate’s electability, likely because it helps indicate to voters that the candidate can secure more public support, which will help them in the general election. For journalists, our results highlight the value to voters of giving both the total a campaign has raised and the sources of that money. Although content analysis of news stories shows that reporters sometimes highlight self-funding by candidates as distinct from overall fundraising (see discussion above), this is not always the case as shown even by our opening example of X posts about Republican Eric Hovde’s first quarter fundraising in 2024. The journalistic practice of isolating fundraising due to personal loans versus traditional solicitation of contributions thus provides information that voters use to determine who will be the most formidable nominee for the general election. This information can be conveyed very simply as we have done in our experiment.¹⁶ Information about sources of funding may also help address a small part of the inequality that comes from the increase in self-funding of campaigns (Bonica, 2020; Carnes, 2020). When self-funding is reported separately from donor funding, self-funding ceases to help candidates be viewed as more electable by primary voters.

Supplemental Material

Supplemental Material - Sources of Candidate Fundraising Affect Perceptions of Electability

Supplemental Material for Sources of Candidate Fundraising Affect Perceptions of Electability by Sarah E. Anderson, Barry C. Burden, Daniel M. Butler, Laurel Harbridge-Yong, and Timothy J. Ryan in American Politics Research

Footnotes

Acknowledgments

The survey used in this study was part of a wave of the Politics in the Field at UNC (PFUNC) survey. We are grateful to Marc Hetherington for the chance to include our questions on the PFUNC survey and to J Ehlinger for administrative assistance.

ORCID iDs

Sarah E. Anderson

Barry C. Burden

Daniel M. Butler

Laurel Harbridge-Yong

Timothy J. Ryan

Ethical Approval

The survey used in this study was approved by the University of North Carolina at Chapel Hill’s IRB (Study #23-0389).

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Politics in the Field at UNC (PFUNC) survey is funded by the Raymond Dawson Distinguished Professorship at the University of North Carolina at Chapel Hill.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Code and data to replicate all of the figures, tables, and appendices are available at https://doi.org/10.7910/DVN/F00NG4 (Ryan, 2025). *

Supplemental Material

Supplemental material for this article is available online.

Notes

References

Abramson

P. R.

Aldrich

J. H.

Paolino

Rohde

D. W.

(1992). “Sophisticated” voting in the 1988 presidential primaries. American Political Science Review, 86(1), 55–69. https://doi.org/10.2307/1964015

Albert

Costa

(2024). Winning at all costs? How negative partisanship affects voter decision-making. Political Behavior, Published Online, 1–27. https://doi.org/10.1007/s11109-024-09974-2

Albert

Raja

R. J.

(2020). Small donors in US elections. Politique Americaine, 35(2), 14–45.

Alexander

(2005). Good money and bad money: Do funding sources affect electoral outcomes? Political Research Quarterly, 58(2), 353–358. https://doi.org/10.1177/106591290505800214

Anderson

S. E.

Burden

B. C.

Butler

D. M.

Harbridge-Yong

Ryan

T. J.

(2025). Information and perceptions of electability in primary elections. Political Behavior, Published Online, 1–28. https://doi.org/10.1007/s11109-025-10054-2

Ashworth

(2006). Campaign finance and voter welfare with entrenched incumbents. American Political Science Review, 100(1), 55–68. https://doi.org/10.1017/S0003055406062009

Bartels

L. M.

(1987). Candidate choice and the dynamics of the presidential nominating process. American Journal of Political Science, 3(1), 1–30. https://doi.org/10.2307/2111322

Bartels

L. M.

(1988). Presidential primaries and the dynamics of public choice. Princeton University Press.

Bateson

(2020). Strategic discrimination. Perspectives on Politics, 18(4), 1068–1087. https://doi.org/10.1017/S153759272000242X

10.

Biersack

Herrnson

P. S.

Wilcox

(1993). Seeds for success: Early money in congressional elections. Legislative Studies Quarterly, 18(4), 535–551. https://doi.org/10.2307/439854

11.

Blum

R. M.

Cowburn

Masket

(2024). Who decides? Media, MAGA, money, and mentions in the 2022 republican primaries. Political Research Quarterly, 77(4), 1314–1332. https://doi.org/10.1177/10659129241268820

12.

Bond

J. R.

Covington

Fleisher

(1985). Explaining challenger quality in congressional elections. The Journal of Politics, 47(2), 510–529. https://doi.org/10.2307/2130894

13.

Bonica

(2017). Professional networks, early fundraising, and electoral success. Election Law Journal: Rules, Politics, and Policy, 16(1), 153–171. https://doi.org/10.1089/elj.2016.0413

14.

Bonica

(2020). Why are there So many lawyers in congress? Legislative Studies Quarterly, 45(2), 253–289. https://doi.org/10.1111/lsq.12265

15.

Box-Steffensmeier

J. M.

(1996). A dynamic analysis of the role of war chests in campaign strategy. American Journal of Political Science, 40(2), 352–371. https://doi.org/10.2307/2111628

16.

Brown

A. R.

(2013). Does money buy votes? The case of self-financed gubernatorial candidates, 1998–2008. Political Behavior, 35(1), 21–41. https://doi.org/10.1007/s11109-012-9193-1

17.

Carnes

(2020). The cash ceiling: Why only the rich run for office - And what we can do about it. Princeton University Press.

18.

Case

C. R.

Porter

(2025). Conceptualizing and measuring early campaign fundraising in congression elections. Political Science Research and Methods. https://doi.org/10.1017/psrm.2025.10014

19.

Cole

(2024). Sen. Bob casey and david McCormick raise over $8 million each in Q2. Pennsylvania Capital-Star. https://penncapital-star.com/briefs/pennsylvania-sen-bob-casey-and-david-mccormick-raise-over-8-million-each-in-q2/ (Accessed September 27, 2024).

20.

Culberson

McDonald

M. P.

Robbins

S. M.

(2019). Small donors in congressional elections. American Politics Research, 47(5), 970–999. https://doi.org/10.1177/1532673x18763918

21.

Epstein

Zemsky

(1995). Money talks: Deterring quality challengers in congressional elections. American Political Science Review, 89(2), 295–308. https://doi.org/10.2307/2082426

22.

Evans

(2024). U.S. Sen Sherrod brown leads fundraising ahead of crucial senate race in Ohio. CityBeat. https://www.citybeat.com/news/us-sen-sherrod-brown-leads-fundraising-ahead-of-crucial-senate-race-in-ohio-17818015 (Accessed September 27, 2024).

23.

Feigenbaum

J. J.

Cameron

A. S.

(2013). The vicious cycle: Fundraising and perceived viability in US presidential primaries. Quarterly Journal of Political Science, 1(1), 1–40.

24.

Fosse

Winship

(2019). Analyzing age-period-cohort data: A review and critique. Annual Review of Sociology, 45(1), 467–492. https://doi.org/10.1146/annurev-soc-073018-022616

25.

Gerber

(1998). Estimating the effect of campaign spending on senate election outcomes using instrumental variables. American Political Science Review, 92(2), 401–411. https://doi.org/10.2307/2585672

26.

Goodliffe

(2001). The effect of war chests on challenger entry in U.S. house elections. American Journal of Political Science, 45(4), 830–844. https://doi.org/10.2307/2669327

27.

Goodliffe

Townsend

(2024). Examining public perceptions of US campaign finance over time through survey experiments. Electoral Studies, 90(August), Article 102813. https://doi.org/10.1016/j.electstud.2024.102813

28.

Green

D. P.

Krasno

J. S.

(1988). Salvation for the spendthrift incumbent: Reestimating the effects of campaign spending in house elections. American Journal of Political Science, 32(4), 884–907. https://doi.org/10.2307/2111193

29.

Green

D. P.

Krasno

J. S.

(1990). Rebuttal to Jacobson’s “Nnw Eeidence for Ood Aaguments”. American Journal of Political Science, 34(2), 363–372. https://doi.org/10.2307/2111451

30.

Green

Schaffner

Luks

(2023). Strategic discrimination in the 2020 democratic primary. Public Opinion Quarterly, 86(4), 886–898. https://doi.org/10.1093/poq/nfac051

31.

Hassell

H. J. G.

(2018). The party's primary: Control of congressional nominations. Cambridge University Press.

32.

Hassell

H. J. G.

(2023). Party elite engagement and coordination in house primary elections: A test of theories of parties. American Journal of Political Science, 67(2), 307–323. https://doi.org/10.1111/ajps.12680

33.

Hassell

H. J. G.

Visalvanich

(2024). Perceptions of electability: Candidate (and voter) ideology, race, and gender. Political Behavior, 46(4), 2075–2098. https://doi.org/10.1007/s11109-023-09909-3

34.

Hewitt

Broockman

Coppock

Tappin

B. M.

Slezak

Coffman

Lubin

Hamidian

(2024). How experiments help campaigns persuade voters: Evidence from a large archive of campaigns’ own experiments. American Political Science Review, 118(4), 1–19. https://doi.org/10.1017/S0003055423001387

35.

Jacobson

G. C.

(1990). The effects of campaign spending in house elections: New evidence for old arguments. American Journal of Political Science, 34(2), 334–362. https://doi.org/10.2307/2111450

36.

Kelly

J. T.

Eric

(2022). Party over preference? Strategic primary voting in the age of outsiders. Social Science Quarterly, 103(3), 607–621. https://doi.org/10.1111/ssqu.13138

37.

Lightman

(2024). Adam schiff leads the California senate money race. Will that matter in tuesday's primary? The Sacramento Bee. https://www.sacbee.com/news/politics-government/election/california-elections/article286138356.html (accessed September 27, 2024).

38.

Magleby

D. B.

Goodliffe

Olsen

J. A.

(2018). Who donates in campaigns? The importance of message, messenger, medium, and structure. Cambridge University Press.

39.

Malbin

M. J.

(2013). Small donors: Incentives, economies of scale, and effects. The Forum, 11(3), 385–411. https://doi.org/10.1515/for-2013-0050

40.

Milyo

Groseclose

(1999). The electoral effects of incumbent wealth. The Journal of Law and Economics, 42(2), 699–722. https://doi.org/10.1086/467439

41.

Minozzi

Woon

(2023). Electability salience can bias voting decisions. Research & Politics, 10(1), Article 20531680231159074. https://doi.org/10.1177/20531680231159074

42.

Open Secrets . (2024a). Elections overview. https://www.opensecrets.org/elections-overview (accessed April 30, 2025).

43.

Open Secrets . (2024b). Fundraising totals: Who raised the Most? https://www.opensecrets.org/elections-overview/fundraising-totals (accessed April 30, 2025).

44.

Open Secrets . (2024c). Top self-funding candidates. https://www.opensecrets.org/elections-overview/top-self-funders (accessed April 30, 2025).

45.

Ovtchinnikov

A. V.

Philip

(2023). Self-funding of political campaigns. Management Science, 69(4), 2425–2448. https://doi.org/10.1287/mnsc.2022.4404

46.

Page

B. I.

Gilens

(2020). Democracy in America? What has gone wrong and what we can do about it. University of Chicago Press.

47.

Peacock

Dugger

Knox Fanelli

Jackson Harris

Bert McLelland

Richardson

L. A.

(2021). Choosing a candidate: Traits, issues, and electability. American Behavioral Scientist, 65(3), 540–557. https://doi.org/10.1177/0002764220978458

48.

Porter

Steelman

T. S.

(2023). No experience required: Early donations and amateur candidate success in primary elections. Legislative Studies Quarterly, 48(2), 455–466. https://doi.org/10.1111/lsq.12396

49.

Porter

Treul

S. A.

(2024). Evaluating (in)experience in congressional elections. American Journal of Political Science, 69(n/a), 284–298. https://doi.org/10.1111/ajps.12854

50.

Primo

D. M.

Milyo

J. D.

(2020). Campaign finance and American democracy. University of Chicago Press.

51.

Raja

Raymond

Schaffner

B. F.

(2015). Campaign finance and political polarization: When purists prevail. University of Michigan Press.

52.

Rickershauser

Aldrich

J. H.

(2007). “It’s the electability, stupid” − or maybe not? Electability, substance, and strategic voting in presidential primaries. Electoral Studies, 26(2), 371–380. https://doi.org/10.1016/j.electstud.2006.09.003

53.

Ryan

Anderson

Butler

Burden

Yong

(2025). Replication Data for: Sources of Candidate Fundraising Affect Perceptions of Electability (Version V1) [dataset]. Harvard Dataverse. https://doi.org/10.7910/DVN/F00NG4

54.

Simas

E. N.

(2017). The effects of electability on US primary voters. Journal of Elections, Public Opinion, and Parties, 27(3), 274–290. https://doi.org/10.1080/17457289.2016.1270287

55.

Smidt

Christenson

(2012). More bang for the buck:campaign spending and fundraising success. American Politics Research, 40(6), 949–975. https://doi.org/10.1177/1532673x12444308

56.

Steen

J. A.

(2009). Self-financed candidates in congressional elections. University of Michigan Press.

57.

Thomsen

D. M.

(2023). Competition in congressional elections: Money versus votes. American Political Science Review, 117(2), 675–691. https://doi.org/10.1017/S0003055422000764

58.

Thomsen

D. M.

(2025). The money signal: How fundraising matters in American politics. University of Chicago Press.

59.

Tversky

Amos

Kahneman

Daniel

(1981). The framing of decisions and the pyschology of choice. Science, 211(4481), 453–458.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.65 MB