INTRODUCTION
How do voters assess promise-keeping and hold elected officials accountable in a hyper-partisan environment? Because promise-keeping is theorized as a key way for voters to assess performance of elected officials (Arnold Reference Arnold1990; Reference Arnold1993; Mansbridge Reference Mansbridge2003), it is critical to understand how voters use information about promise fulfillment and if broken promises are used to hold elected officials accountable to their promises. Investigations into how partisans process information reveal that voters update their attitudes based on information despite partisanship (Bullock Reference Bullock2011; Coppock Reference Coppock2023). Yet, there may be some instances where voter information processing may follow partisan interests (Bayes et al. Reference Bayes, Druckman, Goods and Molden2020; Druckman, Levendusky, and McLain Reference Druckman, Levendusky and McLain2018; Klar Reference Klar2014b). In particular, it is unclear whether voters are able to update information in context of their evaluations of elected officials. For instance, voters appear to give approval ratings through a partisan lens (Donovan et al. Reference Donovan, Kellstedt, Key and Lebo2020). In this article, I focus on how and whether voters use promise-keeping in evaluations of elected officials as a process of accountability, by considering whether voters hold partisan elected officials accountable for broken promises.
Theories of representation suggest that broken promises are central to representation. The typical account of promissory representation describes candidates using promises to attract support from voters (Downs Reference Downs1957; Pitkin Reference Pitkin1967). In subsequent elections, voters assess fulfillment of those promises when determining continued support for elected officials (Arnold Reference Arnold1990; Reference Arnold1993; Mansbridge Reference Mansbridge2003). Much work investigates whether elected officials fulfill their promises (Pétry and Collette Reference Pétry, Collette and Imbeau2009; Royed Reference Royed1996; Thomson et al. Reference Thomson, Royed, Naurin, Artés, Costello, Ennser-Jedenastik and Ferguson2017)—and they largely do, in contrast to voter expectations for promises to be broken (Naurin Reference Naurin2009; Reference Naurin2011). Indeed, voters have a nuanced understanding of promises that is dependent on assessments of candidate attentions, trust in government, and anticipated success of policy interventions (Bonilla Reference Bonilla2021; Naurin Reference Naurin2011). Attention has focused on conditions in which parties keep their promises (e.g., Artes Reference Artes2011; Kostadinova Reference Kostadinova2013; Mansergh and Thomson Reference Mansergh and Thomson2017; Naurin Reference Naurin2014) and how the media understands promise fulfillment (Kostadinova Reference Kostadinova2017; Kostadinova and Dimitrova Reference Kostadinova and Dimitrova2012), there is less focus on what conditions inform voter perceptions of promise fulfillment. In particular, this work does not consider if assessments of promise fulfillment are contingent on a voter’s partisan leanings. Past evidence suggests partisan influence in decision-making (e.g., Bolsen, Druckman, and Cook Reference Bolsen, Druckman and Cook2014; Leeper and Slothuus Reference Leeper and Slothuus2014; Redlawsk Reference Redlawsk2002) may extend to determinations of accountability.
Naurin (Reference Naurin2014) demonstrates that voter assessments of promises are complex: voters may differently remember promises and outcomes. Thus, promise fulfillment may hinge on what information voters have and use to make decisions. However, voters view and interact with information differently based on their prior beliefs and partisan leanings (e.g., Carsey and Layman Reference Carsey and Layman2006; Gaines et al. Reference Gaines, Kuklinski, Quirk, Peyton and Verkuilen2007; Gunther et al. Reference Gunther, Edgerly, Akin and Broesch2012). It is clear that partisan affiliation has increased in intensity (Druckman, Peterson, and Slothuus Reference Druckman, Peterson and Slothuus2013; Rothschild et al. Reference Rothschild, Howat, Shafranek and Busby2019) and with increased salience as an identity (Mason Reference Mason2015; Reference Mason2018). Voter information processing often leaves voters to affirm existing beliefs and information (Druckman, Leeper, and Slothuus Reference Druckman, Leeper and Slothuus2018; Taber and Lodge Reference Taber and Lodge2006), flowing into decisions about candidates and policy. Partisan decision-making affects not only consumption of different information (Carsey and Layman Reference Carsey and Layman2006; Layman and Carsey Reference Layman and Carsey2002), but also affects accuracy of political judgments in some instances (Bullock and Lenz Reference Bullock and Lenz2019). I investigate if partisanship influences assessments of promise-keeping in retrospective candidate selection as well, similarly to how partisanship has shaped other components of representation that are critical to our normative understanding of elected officials’ performance (e.g., Donovan et al. Reference Donovan, Kellstedt, Key and Lebo2020). Bringing these two literatures into conversation helps extend both theories to better understand voter behavior. Because political judgments can be subject to partisan influence, it follows that partisan behavior should also matter in assessments of traditional political phenomena, such as promise-keeping and accountability. And, a broader understanding of the conditions under which voters act on promises as normative theory would predict should help to shape our understanding of how promise fulfillment retrospectively informs representation.
I hypothesize that assessments of promise fulfillment may differ based on partisanship. In theories of representation, voters should be able to assess if promises are kept or not without influence of partisanship. However, partisanship matters when voters make decisions (Iyengar and Krupenkin Reference Iyengar and Krupenkin2018), and assessments of promise fulfillment involve decision-making about elected official performance which may occur through a partisan lens. This suggests that in-partisans may penalize promise-breaking less than out-partisans, while out-partisans may be less likely to reward promise fulfillment than in-partisans.
To investigate these claims, I use two experimental studies to test how voters view promise fulfillment through the lens of partisanship. In the first study, I examine how partisan stances and party identification shifts responses to evaluations of candidate who keep or break their word on immigration. The second study uses a similar experimental design, but also examines perceptions of promise fulfillment through the issues of immigration and human trafficking. The two issues allow us to examine partisan perceptions of accountability when elected officials largely agree on an issue (human trafficking) compared to where elected officials largely disagree on an issue (immigration). The second study allows a test between whether voters react to a party label in determining promise fulfillment as much as they react to how an elected official engages on a particular issue.
Ultimately, the data indicate that respondents assess accountability through a partisan lens, even when there are fewer partisan divisions on the issue. Importantly, respondents distinguish between assessing promises in contexts of approval and performance (if officials fulfill their promise), suggesting that partisan responses may be strategic. The implications for these results extend to both work on partisan decision-making as well as promise fulfillment. First, these findings demonstrate that voter responses account for promise fulfillment more so when asked specifically about performance than when asked about approval. This suggests that studies may find that judgments about partisans may discount information unlike judgments about policies. Therefore, the motivated reasoning literature may need to differentiate between partisan responses to policy actions—where external information may make more of a difference—and electoral contexts—where external information may matter less. At the same time, respondents still differentiate between in- and out-partisans even through performance suggesting that partisanship still matters in assessments of promise fulfillment. Context, such as partial fulfillment and polarization on an issue, additionally affects how voters respond to information about fulfillment. Thus, the second contribution is to suggest that work on representation, accountability, and promise fulfillment may need to shift from focusing simply on promise-keeping itself, to incorporating how information environments may affect voter interest in holding elected officials accountable. Finally, partisan judgments of promise fulfillment suggest that democratic accountability may be unable to rely on assessments of elected officials keeping their promises moving forward.
PARTISANSHIP AND PROMISE FULFILLMENT
Scholars have long explored the role of parties in democracy and representation, with a general sense of the critical role that parties play in modern democracy and representation (Schattschneider Reference Schattschneider1942). Parties filter information, bridge cleavages, and act as an association collecting those with similar views, and speaking for interests that may otherwise be overlooked (Dahl and Lindblom Reference Dahl and Lindblom1953; Schattschneider Reference Schattschneider1942; Stokes Reference Stokes1999; Turman Reference Turman1951). However, parties themselves are also about social connections (Lazarsfeld, Berelson, and McPhee Reference Lazarsfeld, Berelson and McPhee1954). Both as entities that filter information and as social connection, parties may influence how voters perceive representation and accountability. This is particularly true when assessing accountability of elected officials based on their adherence to promise fulfillment. Here, I examine how these two streams of literature—partisan decision-making and representation—suggest that voters may be pulled by competing information sets when evaluating candidate accountability.
Promise Fulfillment and Accountability
What we understand of voter behavior in context of promise-keeping and accountability has yet to incorporate how polarized partisan politics may matter for assessing information and resulting behavior. Many conceptions of democracy and representation rely on campaign promises to signal how elected officials will behave in office (Mansbridge Reference Mansbridge2003). In the most basic and earliest form, there are two parts of this form of representation (Pitkin Reference Pitkin1967): a forward-looking judgment about how promises align with campaign promises and a backward-looking judgment about who well elected official actions (or inaction) align with campaign promises. Although it is important to consider candidate statements prospectively (Bonilla Reference Bonilla2021; Milita, Ryan, and Simas Reference Milita, Ryan and Simas2014), the likelihood of elected officials running for reelection or other political offices makes the promise fulfillment portion of this pathway particularly interesting (Aragones, Postlewaite, and Palfrey Reference Aragones, Postlewaite and Palfrey2007).
Investigations into promise fulfillment examine the specific pathway between promises of candidates and promise fulfillment (Royed Reference Royed1996). Two conclusions have been reached. First, parties (and elected officials) are more likely than not to fulfill their election promises (Fishel Reference Fishel1985; Krukones Reference Krukones1984; Pétry and Collette Reference Pétry, Collette and Imbeau2009; Thomson et al. Reference Thomson, Royed, Naurin, Artés, Costello, Ennser-Jedenastik and Ferguson2017). Across a variety of institutional contexts, countries, local or national elections, on average elected officials keep 67% of their promises (Pétry and Collette Reference Pétry, Collette and Imbeau2009). Although there are differences between parliamentary and presidential government (Royed Reference Royed1996) and whether or not parties are in power (Artes Reference Artes2011), promises are still overwhelmingly kept. Second, citizens are distrustful of candidate promise-keeping (Håkansson and Naurin Reference Håkansson and Naurin2016; Naurin Reference Naurin2011). Called the Pledge Paradox, the difference between voter expectations and reality seems partially determined by the difficulty in how voters define kept promises and skepticism over whether elected officials can successfully achieve outcomes rather than actions (Naurin Reference Naurin2011).
Theories of representation argue that promise-keeping should be of primary importance to voters when elected officials run for reelection (Mansbridge Reference Mansbridge2003). And indeed, empirical evidence demonstrates that both elected officials act to signal kept promises to their constituents (Fenno Reference Fenno1973; Mayhew Reference Mayhew1974) and that voters do penalize elected officials who do not follow through on their commitments (Bonilla Reference Bonilla2021). At the same time, there is evidence that voters identify promises in ways that differ slightly from academic conceptions of promises. First, voters attend to the expectation of commitment in candidate statements (Bonilla Reference Bonilla2021). Second, voters have a nuanced perspective of what it means to keep a promise. In some ways, Naurin (Reference Naurin2011) shows that voters very much mimic expectations that elected officials who break promises when they do not do what they said they would do as a candidate. However, this traditional sense of promise-keeping is contingent on voters knowing both the campaign promises and how elected officials have acted. When they are uncertain of the campaign promises, voters infer what officials have promised by expressing wishes about the state of the world they would like to see and then voters assess performance by how they view the current state of society or policy outcomes (Naurin Reference Naurin2011). Naurin (Reference Naurin2011) describes several individuals who had difficulty directly responding to questions about specific promises that elected officials had broken. Importantly, this work suggests a critical consideration for normative politics: while voters perceive promises as useful mechanisms to directly assess promise-keeping in theory, often voter perceptions of and aspirations for the status quo shift retrospective assessments of promises and voter evaluations. In essence, voter assessments of kept promises may be based on comparisons with achieved policy outcomes, but also may be associated with other issues (e.g., their bank account). However, the literature is not clear on how partisanship may cause voters to evaluate information about elected officials differently.
Partisanship and Decision-Making
Another stream of research examines how an increasing intensity of affiliation with a party alters how voters critically assess the world around them (Druckman, Peterson, and Slothuus Reference Druckman, Peterson and Slothuus2013; Mason Reference Mason2018; Rothschild et al. Reference Rothschild, Howat, Shafranek and Busby2019). What has prompted these analyses is a broader social trend of mass polarization—whether it is partisan sorting (Fiorina and Abrams Reference Fiorina and Abrams2008) or partisans increasingly identifying as partisan with greater separation on policy issues (Abramowitz and Saunders Reference Abramowitz and Saunders2008). Mason (Reference Mason2018) demonstrates that a better metric for determining polarization is to understand partisanship as an identity. As with other identities, partisans tend to be biased toward the in-group and biased against the out-group (Mason Reference Mason2018; Tajfel Reference Tajfel1974), thus it follows that partisanship can influence political decision-making.
These differences result in questions of whether partisan voters view and assess information differently. Studies of motivated reasoning view voter information processing as allowing potential for bias because voters make decisions using readily available information that leaves them highly susceptible to confirmation bias and prioritization of information consistent with prior beliefs (Druckman, Leeper, and Slothuus Reference Druckman, Leeper and Slothuus2018; Lodge and Taber Reference Lodge and Taber2013; Taber and Lodge Reference Taber and Lodge2006). Partisanship becomes a critical component behind voter support of public policies, and affects how elite framing matters to voters as well as changes which party cues voters will pick up (Bolsen, Druckman, and Cook Reference Bolsen, Druckman and Cook2014; Druckman and Levendusky Reference Druckman and Levendusky2019). Across several domains, motivated reasoning helps explain which information voters use to form policy preferences on scientific issues including genetically modified food, vaccinations, and climate change (Druckman and McGrath Reference Druckman and McGrath2019; Gaskell et al. Reference Gaskell, Bauer, Durant and Allum1999; Hornsey, Harris, and Fielding Reference Hornsey, Harris and Fielding2018; Sinatra, Kienhues, and Hofer Reference Sinatra, Kienhues and Hofer2014). Indeed, this evidence suggests partisanship can operate as an information shortcut to explain or defend attitudes (Lavine, Johnston, and Steenbergen Reference Lavine, Johnston and Steenbergen2012; Leeper and Slothuus Reference Leeper and Slothuus2014).
The utility, however, comes at a potential cost of ability to weigh substantive information (e.g., Druckman, Peterson, and Slothuus Reference Druckman, Peterson and Slothuus2013). The earliest arguments suggest that partisan information appears to motivate voters toward accuracy (Kruglanski Reference Kruglanski1989; Taber and Lodge Reference Taber and Lodge2006). But partisanship can also decrease accuracy (Bullock, Gerber, and Hill Reference Bullock, Gerber and Hill2015; Jerit and Barabas Reference Jerit and Barabas2012; Kim, Taber, and Lodge Reference Kim, Taber and Lodge2010; Lebo and Cassino Reference Lebo and Cassino2007) in part, because people are directionally motivated to uphold their beliefs (Bolsen and Palm Reference Bolsen and Risa2019). This is particularly true at higher levels of partisanship and higher commitment to prior attitudes (Lodge and Taber Reference Lodge and Taber2005; Taber and Lodge Reference Taber and Lodge2006). Yet, continued debates on motivated reasoning complicate if reasoning is motivated toward partisanship or other directional goals (Druckman and McGrath Reference Druckman and McGrath2019; Levendusky Reference Levendusky2023). And while partisans do take information into account and are persuaded to update their conclusions on new evidence (Coppock Reference Coppock2023), priming partisan goals can increase motivation to process information from a partisan lens (Bayes et al. Reference Bayes, Druckman, Goods and Molden2020).Footnote 1
This perspective then complicates how we might expect evaluations of promise-keeping to function as it introduces a tension in whether voters are motivated by partisan identity in their decision-making or by following traditional democratic values. The conflict between partisan identity and “political reality” can make “evaluating party performance objectively” difficult for partisans (Lavine, Johnston, and Steenbergen Reference Lavine, Johnston and Steenbergen2012, 9) because partisanship is key to how people approach decision-making (Iyengar and Krupenkin Reference Iyengar and Krupenkin2018; Mason Reference Mason2018). Evidence suggests that strengthened partisan identification actually has potential to change long-standing norms of accountability and how voters perceive the world and evaluate elected officials. Donovan et al. (Reference Donovan, Kellstedt, Key and Lebo2020) show that despite aggregate public opinion canceling out differences, co-partisans are more likely to give credit to the president for approving the economy and out-partisans are more likely to assign blame. In whole, this suggests that when evaluating accountability—a form of political decision-making that involves partisan judgments—there may be similar partisan evaluation of information.
Promise Fulfillment Accountability through a Partisan Lens
Given the partisan environment in the United States, where candidates and elected officials tend to focus on different issues and make an increasing number of commitments on those issues (Bonilla Reference Bonilla2021), it becomes important to examine accountability from a partisan lens. Namely, does a partisan perspective influence voter assessments of candidate promises? If so, the implications for what it means to hold elected officials accountable may then become a partisan enterprise. I argue that these perceptions build into how voters interpret accountability and consider what it means to recognize an elected official as successful in their term. Importantly, theories of promise fulfillment may predict different outcomes for how voters assess promise-keeping than theories of partisan decision-making, leaving several important unanswered questions about partisanship and accountability. In order to better understand how accountability operates in a polarized environment, I ask: Do voters evaluate broken promises of in-partisan and out-partisan candidates similarly? Are in-partisans more likely to punish broken promises among in- or out-partisans? Are in-partisans more likely to award kept promises among in- or out-partisans?
Prior literature on promise fulfillment and partisan decision-making have yet to engage with each other which means that each literature misses important discoveries from the other. Because theories of promise fulfillment rely on broad descriptions of mandate theory and promissory representation, there is a key assumption that voters will identify and punish broken promises (e.g., Thomson et al. Reference Thomson, Royed, Naurin, Artés, Costello, Ennser-Jedenastik and Ferguson2017). At the same time, it also maintains that voter evaluations may use different information sets (Naurin Reference Naurin2011), suggesting the need to better understand how voters use different information sets. This is particularly true given critical changes in polarization that may have consequences for how promises are interpreted or even what promises are made. Meanwhile, the literature on partisan decision-making suggests that persuasive information causes partisans to update their beliefs and allows for less partisan decision-making (Coppock Reference Coppock2023). Because promise fulfillment and evaluations of accountability are grounded in partisanship, evaluating promise fulfillment may in fact lead to partisan evaluations of promise-keeping (Bayes et al. Reference Bayes, Druckman, Goods and Molden2020), suggesting that voters will not punish broken promises for co-partisans nor will they reward promise-keeping for out-partisans. Importantly, voters grapple both with how they understand promise fulfillment and accountability while retaining partisan perspectives. Thus, research must work to engage how voters understand accountability in partisan contexts (Druckman Reference Druckman2022).
I also further complicate how we understand promises-keeping in two ways: by considering a more nuanced version of promise-keeping and by considering variance in issue polarization. First, it is important to allow nuance in promise fulfillment by presenting a case where promises may not be completely fulfilled, but also a promise may not be broken. Often, theory treats promise fulfillment as a binary: promises are broken or kept. In reality, however, while promises are fulfilled more often than not, elected officials devote more attention to issues that they promised on (Sulkin Reference Sulkin2011) and tend to not completely fulfill a promise, because they are blocked by partisan activity (Fishel Reference Fishel1985). Determining whether promises are fulfilled may seem a straight-forward exercise, but many promises are difficult to objectively measure as outcomes. In part, many promises do not have measurable outcomes (Royed Reference Royed1996) or they maybe perceived as aspirational (Naurin Reference Naurin2011).Footnote 2 Holding elected officials accountable is difficult, and voters may not have all the appropriate information (Arceneaux Reference Arceneaux2006), nor may they choose to use it. Further, given the difficulty for partisans to incorporate information into assessments on accuracy and policy, partial fulfillments of promises are inherently open to interpretation by those evaluating them.Footnote 3 Thus, I anticipate the most pronounced partisan differences will occur where elected officials may have partially fulfilled a promise.
Second, I analyze different types of issues that elected officials may promise on. If voters use partisanship to evaluate promises of elected officials, however, it is possible that these assessments may vary based on the subject matter because not all issues are treated the same by voters (Druckman and Leeper Reference Druckman and Leeper2012). In particular, many issues have become recognizably polarized, with increasing partisan divides (Abramowitz and Saunders Reference Abramowitz and Saunders2008; Hetherington and Weiler Reference Hetherington and Weiler2018; Mason Reference Mason2015). However, it is less clear if partisan evaluations spill over into less polarized issues. While bipartisan issues can be made partisan (Kahan et al. Reference Kahan, Jamieson, Landrum and Winneg2017), they can be used to mitigate polarization on partisan issues (Bonilla and Mo Reference Bonilla and Mo2018; Guay and Lopez Reference Guay and Lopez2021). Indeed, polarization on an issue could be a context under which partisan information processing is more likely than with on a less polarizing issue.
Testing Theories of Partisanship and Accountability
I examine these questions through the lens of immigration and human trafficking policy. Both issues are viewed as important in the minds of the public and are to some extent related, but immigration is a deeply polarized issue, while anti-trafficking efforts are broadly viewed as nonpartisan, and anti-trafficking efforts are broadly supported by both parties (Bouché, Farrell, and Wittmer-Wolfe Reference Bouché, Farrell and Wittmer-Wolfe2018).Footnote 4 More importantly, each issue was prominently discussed over the last few years, and discussed through the lens of promises and accountability.
Rhetoric on immigration played an undeniably important role in the 2016 election (Hooghe and Dassonneville Reference Hooghe and Dassonneville2018). More importantly, for electoral outcomes, the highly prominent narrative to reduce immigration from Latin America was a key factor in voters switching a 2012 Democrat vote to a 2016 Republican vote (Sides, Tesler, and Vavreck Reference Sides, Tesler and Vavreck2017). As a promise, the strong intent to reduce immigration was instrumental in Trump’s election as promissory representation would indicate it should be (Mansbridge Reference Mansbridge2003). However, two years into the Trump presidency, the wall was yet to be built (BBC News 2018), and by the end of his presidency, it existed only incompletely (Timm Reference Timm2021). At the same time, the strong anti-immigrant rhetoric continued, and several other steps had been taken to increase the difficulty for migrants to enter the country (Piere and Selee Reference Piere and Selee2017). Further, The White House had actively taken steps to signal that not only were they restricting immigration, they were also “positively” affecting change on separate, but closely related issues, such as human trafficking policy (Trump Reference Trump2020). The continued rhetoric around immigration, future promises, and signals of kept promises, suggests that immigration rhetoric (and especially anti-immigration rhetoric) would continue to play a role in the 2020 election cycle (as it did). For my purposes here, it suggested potential nuance with how voters might assess performance on these issues.
Attitudes toward human trafficking differ from those toward immigration, and subsequently, rhetoric differs too. Regardless of party, attitudes toward human trafficking elevate the importance of anti-trafficking efforts, express concern, and are opposed to trafficking (Bonilla and Mo Reference Bonilla and Mo2019). While there are some differences in partisan government actions to fight against trafficking (Farrell, Bouché, and Wolfe Reference Farrell, Bouché and Wolfe2019), anti-trafficking efforts still incorporate many of the same strategies, and receive broad, bipartisan legislative support (Bonilla and Mo Reference Bonilla and Mo2018). And while some tension exists between anti-trafficking efforts and immigration policy, the public may lack awareness of these connections and there are viable calls for bipartisanship in addressing this issue (e.g., Runde and Santoro Reference Runde and Santoro2017).
STUDY 1: PROMISES AND IMMIGRATION
In the first study, I examine how partisans react to promise fulfillment. This study examines only partisans on the partisan issue of immigration, which means that those responding will be both partisan and reacting to a partisan environment. As a result, this study tests the first two hypotheses through a partisan issue in a partisan context. Arguably, this may speak to the external validity of this test as the U.S. electorate increases partisan disaffection (Iyengar and Krupenkin Reference Iyengar and Krupenkin2018).
Description
I examine how partisans view promises based on immigration as fulfilled or unfulfilled. The structure of the experiment is a
$ 2\times 3 $
experiment that varies shared partisanship with an elected official and promise fulfillment (broken, partial, or kept). The promise fulfillment treatment was deployed in three parts. First, respondents were asked about demographic information and their opinions on several policy stances, including immigration.Footnote
5
, Footnote
6
Second, respondents were treated with the elected official’s party information and position. Although partisanship of the voter cannot be experimentally assigned, candidate partisanship was randomized to be a in-partisan or an out-partisan. The initial treatment presented the voters with an elected official’s party and stance on immigration as a campaigning candidate, allowing respondents to become acquainted with the candidate before making a retrospective judgment on the candidate’s actions. The campaign statement was made consistently with the typical partisan stance on immigration (Republicans opposing immigration and Democrats favoring immigration), and the candidate’s party was given during each stage of the vignette. As in Table 1, the campaign stance was issued as a promise, signaling a clear stance and strong commitment, which should make resulting differences of opinion less likely to be due to an ambiguous stance on the issue (Bonilla Reference Bonilla2021). At this point, the respondent was asked about the candidate prospectively, encouraging respondents to form an opinion about the elected official and more appropriately mimic an abbreviated set of electoral decisions. The respondent was asked where the official stands on immigration, how favorable the candidate was, if they would vote for a similar candidate and if they believed the candidate made a promise.
Prospective Treatment Text
Finally, the respondents were treated with the promise outcome in the second stage of the vignette. Here, respondents were presented randomly assigned one of three outcomes: the official clearly keeps their promise, clearly breaks their promise, or is unsuccessful in keeping their promise while still reiterating their stance on the issue.Footnote 7 The candidate’s stance is consistent between each stage, and the full treatment wording can be found in Table 2. As a result, there are six possible conditions in this experiment.Footnote 8
Retrospective Treatment Text
The respondents were then asked a series of questions about their opinions on the elected official. The first two questions ask about approval of the elected official, and is intended to mimic the question of partisan decision-making in context of electoral promises. If the information that voters have is about promise fulfillment and partisanship, how do voters make partisan decisions? These two questions asked “Is your opinion of this elected official favorable or unfavorable?” (with a four-point response scale from “very unfavorable” to “very favorable”) and “Would you vote for a candidate like this if this official ran for reelection?” (with a five-point response scale from “definitely no” to “definitely yes”). The third question asked “How successful do you think the official is on this issue?” (with a four-point response scale from “not at all successful” to “extremely successful”). Finally, I ask directly about whether the official acted to keep the promise: “Do you think the official has acted to keep his promise?” (with a four-point response scale from “not at all” to “completely”). To some extent, this last question acts as a manipulation check because it assesses if respondents notice differences between promise breaking and keeping. Together, the final two questions test the concept of representation and present a higher barrier to testing assessments of promise fulfillment because they ask about actions more than affect of a partisan. The different sets of questions allow us to examine if partisans notice promise-breaking and -keeping, and if they evaluate success and/or candidates based on that information.
Importantly, while data were collected across the entire seven-point partisan identification scale, I present only the partisan results here (including those who lean Republican or Democrat) and remove true independents—those who say they do not lean toward either party—from the analysis. Previous surveys have indicated that independents are not necessarily less partisan than those who identify as Republican or Democrat (Keith et al. Reference Keith, Magleby, Nelson, Orr and Westlye1992), indicating that independents who lean toward a party behave in ways that are partisan. Further, Klar, Krupnikov, and Ryan (Reference Klar, Krupnikov and Ryan2022) explain that the nuances embodied among those who identify in the independent category.Footnote 9 Although additional theoretical work is needed to best understand how independents are responding to promises—it is less clear if independents will treat all partisans as out-partisans or if they will be more inclined to judge every issue with out accounting for party.Footnote 10
Results
In general, there was a large partisan difference between in-partisan and out-partisan views of successful promise-keeping. And, it appears that partisanship alters patterns in respondent approval of candidates—but partisanship does not completely eliminate the importance of promise-keeping itself. (Demographic variables are reported in Section B of the Supplementary Material.)
The first experiment was fielded on October 15, 2019 to a sample of 547 U.S. adults through Amazon’s Mechanical Turk using the Qualtrics platform, with a VPN screener question to ensure human respondents. While Mechanical Turk offers a convenience sample, it has been demonstrated to be useful for experimental analyses (Berinsky, Huber, and Lenz Reference Berinsky, Huber and Lenz2012; Mullinix et al. Reference Mullinix, Leeper, Druckman and Freese2015; Paolacci and Chandler Reference Paolacci and Chandler2014). Because I am primarily interested in the responses of partisans, I perform the analyses with 237 Democrats and 140 Republicans in the sample, disregarding true independents. In all analysis, subjects were divided into whether they were grouped as in-partisan or an out-partisan candidate. All scales are transformed to a 0–1 scale.Footnote
11 To simplify the reporting of the results, I describe an index of each dimension—favorability and accountability. The first two questions compose the first dimension with a high Cronbach alpha scores (
$ \alpha =0.95 $
). I combine the later two questions into an index measuring accountability (
$ \alpha =0.78 $
).
Figure 1a presents the average respondent approval ratings by treatment.Footnote
12 In all cases, out-partisans receive a significantly lower favorability rating than in-partisans (even those who have broken their promise). When looking at in-partisans, respondents penalized candidates who broke a promise compared to those who partially kept a promise (
$ t(146)=4.21,\hskip0.5em p<0.001 $
). Between out-partisan officials, there is no significant difference between those breaking a promise and those partially fulfilling a promise (
$ t(120)=0.30,\hskip0.5em p=0.76 $
) or keeping a promise (
$ t(135)=0.002,\hskip0.5em p>0.99 $
). This measures suggests that while promise-breaking matters for approval of in-partisan candidates, it does not matter for approval for out-partisan candidates.
Study 1 Results
Note: This figure displays the mean response for candidate approval and perceptions of accountability, moderated by partisanship and promise fulfillment. The bands display the 95% confidence interval around the mean. A table with these results is in the Section C of the Supplementary Material.
In contrast, Figure 1b shows a very different pattern in responses to whether officials acted in accordance with their word. Measures asking about the performance of the elected official indicated an interesting alignment between in- and out-partisan responses. Here, all out-partisans were rated lower than corresponding in-partisans. The difference between out- and in-partisan candidates who broke a promise (
$ t(160)=1.99,\hskip0.5em p=0.05 $
) is smaller than those who partially broke a promise (
$ t(106)=3.05,\hskip0.5em p<0.01 $
) or kept a promise (
$ t(120)=4.85,\hskip0.5em p<0.001 $
). Among in-partisans, there the accountability index increases as the level of promise fulfillment increases, so that partial promises are rated higher than broken promises (
$ t(137)=4.35,\hskip0.5em p<0.001 $
), and kept promises are rated higher than partial promises (
$ t(149)=4.68,\hskip0.5em p<0.001 $
). A similar pattern exists among out-partisans, with one major difference. While broken promises are rated lower than partial promises (
$ t(136)=3.15,\hskip0.5em p<0.01 $
), kept promises are not rated higher than partially fulfilled promises (
$ t(121)=1.55,\hskip0.5em p=0.12 $
). As a whole, these differences suggest that respondents are attentive to promise-keeping when measuring success of elected officials. And there remains an in-party advantage for assessments of promise-keeping. Figure 1b demonstrates that in-partisans are more likely to indicate higher levels of performance for a promise-keeper than are out-partisans.
Next, I examine the results looking at the strength of partisanship of the respondent. Because the overall results are the same and to simplify the presentation, I examine a index of both dimensions—approval and favorability.Footnote
13 Figure 2 shows the results for in-partisans, and on the right column displays results for out-partisans. For in-partisans of either party, the penalty for candidates who broke their promises is most likely to be enacted by those with weaker partisanship (
$ {\beta}_R=-0.085 $
,
$ \hskip0.5em p=0.07 $
;
$ {\beta}_D=-0.067, $
$ p=0.05 $
) even while stronger partisans tend to be more supportive of the in-party candidate (
$ {\beta}_R=0.247 $
,
$ p<0.001 $
;
$ {\beta}_D=0.159 $
,
$ p<0.001 $
). Figure 3 shows the results where partisanship is inconsistent. In contrast, among out-partisans, there is little to no differentiation between how members from either party evaluate promise-breaking by the strength of identification (
$ {\beta}_R=-0.042, $
$ p=0.45; $
$ {\beta}_D=-0.06, $
$ p=0.14 $
), however, partisan strength does strongly matter for both Republican and Democrats (
$ {\beta}_R=0.176, $
$ p<0.001 $
;
$ {\beta}_D=0.111 $
,
$ \hskip0.5em p=0.02 $
). Thus, we actually see the opposite of Hypothesis 3: weaker partisans are more likely than stronger partisans to differentiate between kept and broken promises, but only if viewing in-partisan candidates. Respondents at all levels of partisan strength largely do not differentiate among out-partisans.Footnote
14
Study 1: Interactions Where Partisanship Is Consistent
Note: This figure displays the interaction between the strength of partisanship and an index of candidate approval. Republican respondents are on the left, and Democratic respondents are on the right. A table containing these results can be found in the Section D of the Supplementary Material.
Study 1: Interactions Where Partisanship Is Inconsistent
Note: This figure displays the interaction between the strength of partisanship and an index of candidate approval. Republican respondents are on the left, and Democratic respondents are on the right. A table with these results is presented in Supplementary Materials Section D.
In sum, this study indicates that assessments of fulfillment is informed by partisan affiliations, but in a nuanced way. First, respondents differentiated attitudes toward candidates on two dimensions: (1) approval and (2) assessments of performance. Out-partisan voters do not differentiate between elected officials who broke or kept their promises, even though in-partisans differentiated between broken and partially kept promises. However, in evaluating whether elected officials were successful in office or kept their promises, out-partisan opinions evaluated broken and partially kept promises relatively similarly to in-partisans, but out-partisans did not reward elected officials for keeping promises like in-partisans did. This suggests that on the partisan issue of immigration, partisanship does shift evaluations of accountability.
STUDY 2: PROMISES, IMMIGRATION, AND HUMAN TRAFFICKING
The second study follows the same pattern as Study 1, with a similar focus on immigration statements by candidates. In this version of the experiment, however, in addition to partisan immigration statements, I added a statement on human trafficking, an issue which has largely been considered as bipartisan (Bonilla and Mo Reference Bonilla and Mo2019; Farrell, Bouché, and Wolfe Reference Farrell, Bouché and Wolfe2019) and related to immigration (Bonilla and Mo Reference Bonilla and Mo2018). I also included candidates with no partisan descriptors attached to them, which allows me to separate the precise effect of partisan statements in understanding promise fulfillment. Study 2, then, tests all hypotheses by adding nine additional treatments to the six treatments from Study 1.Footnote 15
Description
Here, I examine partisanship and promise fulfillment against bipartisan statements through an experiment manipulating partisanship, candidate position, and whether the issue is bipartisan or partisan. It repeats the same three phases of the questionnaire design: measurement of demographic variables and general policy attitudes, treatment of partisanship and issue through a prospective description of candidate, and finally treatment of promise fulfillment as kept, broken, or partially fulfilled.
In the first treatment, respondents are again presented with candidate statements. The candidates presenting the statements are assigned to be Democrat, Republican, and only in the human trafficking condition, one candidate is not given a party. For those receiving statements on immigration, partisanship is assigned to be aligned with the traditional partisan stance (as described in Study 1). Those who received the human trafficking statement however, receive the exact same message since human trafficking is considered to be a non-polarizing issue.Footnote 16 Respondents were then asked questions about the candidate’s favorability, their commitment to the issue, and if they made a promise on their position or not.
In the second stage, respondents were presented with a second vignette about the (now) elected official’s performance on their issue. Respondents were randomly assigned to a candidate who kept their promise, broke their promise, or partially fulfilled their promise.Footnote 17 While the immigration statements remained consistent with Study 1, the full versions of the human trafficking statements can be found in Table 3. Respondents were then asked the same four questions about the elected official and their performance in office.
Human Trafficking Treatment Text
After displaying the second treatment, respondents were asked the same questions they were asked in Study 1. These included a five-point question on favorability toward the candidate and willingness to vote for a similar candidate. I also included a question with a four-point scale on how successful they thought the candidate was, whether the official acted consistently with their original position, and whether the official was representative of others in the party. All questions have been re-scaled to a 0–1 scale for ease of interpretation. Similarly to Study 1, partisanship was measured on a seven-point scale, and while independents are not included in the analysis here, these results are presented in Dataverse Appendix Section D.
Results
The results for Study 2 expand on and underline the findings from Study 1. This experiment was fielded prior to the 2020 presidential election season on October 29–30, 2020 to a sample of 2,303 U.S. adults through Lucid Marketplace. Lucid Marketplace offers a convenience sample that has been shown to be effective for experimental analysis (Coppock and McClellan Reference Coppock and McClellan2019). In all analyses, subjects were divided into whether they were in-partisan or out-partisan or received no partisan information about the candidate.Footnote 18 Additionally, I check for partisan differences on attitudes toward immigration and human trafficking. Indeed, I find that attitudes toward immigration vary by partisanship and are more polarized (with a wider range of attitudes) than that of human trafficking—which is largely consistent across different partisan identities. I present and discuss these results further in Section F of the Supplementary Material.
Immigration Results
As in the first study, I find the four sets of dependent variables examine a dimension of approval (
$ \alpha =0.94 $
) and accountability (
$ \alpha =0.85 $
), and I report the results as indices of these two dimensions. Figure 4 shows the mean rating for each dimension moderated by treatment, with the results for the candidate approval index in Figure 4a. In-partisans are all rated above all of the out-partisans. Among in-partisan candidates, respondents punish promise-breaking compared to partially fulfilled promises (
$ t(204)=-4.87,\hskip0.5em p<0.001 $
) and partially fulfilling compared to keeping a promise (
$ t(204)=-5.80,\hskip0.5em p<0.001 $
). Out-partisans, however, do not differentiate between broken promises and partially kept promises (
$ t(230)=-0.53,\hskip0.5em p=0.59 $
) or even broken promises and kept promises (
$ t(267)=-0.06,\hskip0.5em p=0.95 $
). For in-partisans, voters judge candidates by promise-keeping status. However, for in-partisans, promise-keeping does not matter for views of the candidate, suggesting that partisanship matters more in decision-making than promises.
Study 2: Immigration Results
Notes: This figure displays the mean response for each dependent variable, moderated by partisanship and promise fulfillment. The bands display the 95 percent confidence interval around the mean. A table with these results is in Section G of the Supplementary Materials.
The results for the second dimension, accountability contrast from that of the first dimension. Again, every in-partisan is rated higher than the corresponding out-partisan. Among in-partisans, broken promises are rated lower than partially kept promises (
$ t(199)=-3.30,\hskip0.5em p=0.001 $
), and partially kept promises are rated lower than kept promises (
$ t(243)=-3.11,\hskip0.5em p=0.002 $
). A similar pattern is seen for out-partisans: a broken promises penalized compared to a fulfilled promise (
$ t(238)=-2.48,\hskip0.5em p=0.01 $
), although partially kept promises are not significantly lower than kept promises (
$ t(217)=-1.32,\hskip0.5em p=0.19 $
). In this dimension, promise-keeping matters in respondent ratings even when given the out-partisan treatment. Notably, in-partisans still get a bump over out-partisans, even as voters are differentiating between promise-keeping status.
Now, I turn to the strength of partisanship. These results again examine an index of the four dependent variable measures described above (
$ \alpha =0.93 $
)—as the overall implications for partisanship are similar across all dependent variable measures. Figure 5a displays the results for Republicans viewing an in-partisan incumbent and Figure 5b displays the results for Democrats viewing an in-partisan incumbent. For in-partisans, those with weaker partisanship rate a candidate significantly lower than those with stronger partisanship (
$ {\beta}_R=0.86 $
,
$ p=0.001 $
,
$ {\beta}_D=0.54 $
,
$ p=0.03 $
), but there is no significant interaction for how promises are evaluated for Republicans (
$ {\beta}_R=0.07 $
,
$ p=0.46 $
,
$ {\beta}_D=0.06 $
,
$ p=0.75 $
), though promise-keeping itself is distinguished by Democrats (
$ {\beta}_D=0.15,\hskip0.5em p<0.001 $
). Results for out-partisans are found in Figure 5c and 5d. Among out-partisans, there is similarly an increased partisan reward for those with stronger partisanship (
$ {\beta}_R=0.19 $
,
$ p<0.001 $
,
$ {\beta}_D=0.80 $
,
$ p=0.03 $
), but no increased level of differentiation between those who broken or kept promises in promise fulfillment conditions across candidates (
$ {\beta}_R=-0.01 $
,
$ p=0.72 $
;
$ {\beta}_D=-0.25 $
,
$ p=0.32 $
). Unlike in Study 1, where increased partisanship caused a decreased likelihood to pay attention to promises, there is no significant distinction in the strength of partisanship in this study.Footnote
19
Study 2 Immigration Results Interacted with the Strength of Partisanship
Notes: This figure displays the mean response an index of the four dependent variables, moderated by the strength of partisanship and promise fulfillment. The bands display the 95 percent confidence interval around the mean. A table of results can be found in Section H of the Supplementary Materials.
Human Trafficking Results
I now turn to the human trafficking results, which offer some contrasts to immigration. As such, the human trafficking issue helps us to further differentiate the effects of partisanship on assessment of promises because it allows a viable opportunity to exploit variation in an issue where both Republicans and Democrats support anti-trafficking policies, and it is feasible for candidates from either party to take similar stances.
Figure 6 displays the results with the approval index on the left side (
$ \alpha =0.92 $
), and the accountability index (
$ \alpha =0.90 $
) on the right. Unlike in the immigration studies, on this issue, the two sets of dependent variables bear a striking resemblance to each other. When asked about candidate favorability, respondents rate in-partisan candidates significantly higher than out-partisan candidates within the same promise-keeping levels. Candidates who are not given a partisan position are more similar to out-partisans who have broken a promise (
$ t(241)=-0.27,\hskip0.5em p=0.78 $
) whereas in-partisans are rated higher (
$ t(237)=2.13,\hskip0.5em p=0.03 $
). And, nonpartisan candidates are more similar to in-partisans who have kept a promise (
$ t(254)=1.632,\hskip0.5em p=0.11 $
) as they receive an approval bump over out-partisans (
$ t(253)=3.10,\hskip0.5em p=0.002 $
). Within every partisan grouping, respondents rate broken promises lower than partially kept promises and partially kept promises lower than kept promises. These patterns largely hold over the accountability dimension as well. Out-partisans are rated the lowest across all levels of promise-keeping, and broken promises are rated lower than partial and kept promises. The take-aways are twofold. First, there is a partisan bump that exists outside of penalties and rewards for promise-keeping. Second, despite this partisan bump, in-partisans are more likely to reward out-partisans for keeping a promise on human trafficking than in the polarized issue area of immigration (
$ t(258)=4.48,\hskip0.5em p<0.001 $
).
Study 2: Human Trafficking Results
Note: This figure displays the mean response for each dependent variable, moderated by partisanship and promise fulfillment. The bands display the 95% confidence interval around the mean. A table with these results is in the Section I of the Supplementary Material.
Figure 7 shows the interaction between the strength of partisanship and promise-keeping, using the index of all dependent variables (
$ \alpha =0.95 $
). On this issue, there are some partisan differences between Republicans and Democrats in evaluating promises, and there are key differences between stronger and weaker partisans. First, among in-partisan candidates, stronger Republicans are more likely to reward-shared partisanship (
$ {\beta}_R=0.92 $
,
$ p=0.002 $
), but not Democrats (
$ {\beta}_D=0.28 $
,
$ p=0.37 $
). Importantly, there is no change in evaluating promise making by partisanship (
$ {\beta}_R=-0.10 $
,
$ p=0.63 $
;
$ {\beta}_D=0.03 $
,
$ p=0.88 $
). For out-partisans, both Republican and Democrat responses show a decrease in responsiveness to promise-keeping as partisanship grows stronger though (
$ {\beta}_R=-0.46, $
$ p=0.05; $
$ {\beta}_D=0.31,\hskip0.1em $
$ p=0.20 $
). Finally, candidates without a party saw the largest differentiation among responses to promises by the strength of partisanship for Republicans (
$ {\beta}_R=-0.66 $
,
$ p=0.007 $
) though Democrats were similarly likely to differentiate between promises at all levels of partisan strength (
$ {\beta}_D=0.15 $
,
$ p=0.50 $
). Combined with both Studies 1 and 2 on immigration, it is difficult to make a resounding conclusion by party. In all cases, I reject the hypothesis that stronger partisanship increases likelihood to judge based on promises. However, weaker partisanship does seem to increase the likelihood that individuals will attune to promise-keeping in candidate evaluations.Footnote
20
Study 2 Human Trafficking Results Interacted with the Strength of Partisanship
Note: This figure displays the mean response an index of the four dependent variables, moderated by the strength of partisanship and promise fulfillment. The bands display the 95% confidence interval around the mean. A table with these results is in the Section J of the Supplementary Material.
On the whole, this study underlines and replicates the findings from Study 1, and adds two additional pieces of information. First, the partisanship of an issue matters in how voters react to promise fulfillment. Namely, respondents differentiate between status of promise fulfillment in evaluations of out-party candidate favorability only in the instance of a nonpartisan issue. Second, when respondents prioritize promise fulfillment in their decision-making, they still differentiate between candidates with a partisan lens. Even if candidates are not routinely rated lower for breaking their promises based on partisanship, they are routinely rewarded for keeping their promises. In-partisans are always rewarded; out-partisans are only rewarded on nonpartisan issues. Thus, this study suggests that respondents are more likely to hold officials accountable for promise-breaking on an issue area that is less polarized (human trafficking).
CONCLUSION
A key theory of representation focuses on the accountability of elected officials. While other studies focus on the difficulties that voters may have in assessing elected official performance (e.g., Alvarez Reference Alvarez1997; Lenz Reference Lenz2013), here, I ask how partisan behaviors influence voter assessments. In particular, I draw into dialogue work on promises, representation, and partisanship, which suggests different outcomes of what promise fulfillment should mean for voting behavior. The results suggest that partisanship matters for voters when evaluating candidates based on promise fulfillment which complicates both the function of promises in representation and suggests additional context for when partisan information may be most relevant: (1) if evaluations are about approval or candidate actions and (2) polarization on the issue that candidates discuss.
First, responses to promise fulfillment are nuanced but partisanship clearly shifts how information is processed. Respondents do not reward promise-keeping among out-partisans even as they punish promise-breaking among in-partisans. However, when asked to evaluate promise fulfillment itself, respondents punished broken promises and rewarded kept promises even though they are less likely to recognize promise fulfillment among out-partisans and more likely to recognize promise-keeping among in-partisans.
Second, respondents are more likely to pay attention to promise-keeping when the issue is nonpartisan. Thus, a more polarized issue area seems to cue voters to assess information from a more partisan perspective than a less polarized issue area, suggesting that voters should be more likely to consider promise-keeping outside of partisan perspectives. However, it is not the case that there are no partisan differentiation on a less polarized issue: Study 2 demonstrates aggregate differences between in-partisans and out-partisans at each stage of promise fulfillment. This suggests that partisanship matters in evaluation of promise-keeping, confirming across issues that voters are more likely to assess that in-partisans as fulfilling their promises than out-partisan representatives, regardless of the actions of elected officials.
On the whole, the implications for these results matter for scholars of promise fulfillment, partisan information processing, and polarization. First, there is nuance in how voters interpret promise fulfillment which should shape future work on promise fulfillment. Voters and scholars agree that promises are designed as something a candidate has committed to do (Bonilla Reference Bonilla2021; Naurin Reference Naurin2014). But there is debate as to whether promise fulfillment should be considered as narrow—if the commitment must identify a specific target—or broad—if a commitment generally shifts policy (Thomson et al. Reference Thomson, Royed, Naurin, Artés, Costello, Ennser-Jedenastik and Ferguson2017). The data here suggest voters process distinctions between these categories due to the partial fulfillment treatment. The difference spotlights a tension on voter assessments of promise fulfillment: strict interpretations are less common among co-partisans, who are less likely to penalize partial promise fulfillment. This suggests, that even when a promise is broad, less specific, and potentially more difficult to measure, voters continue to engage with that promise beyond conventional metrics.
Second, the experiments in this study point to important contextual differences about when voters make more partisan decisions, suggesting that scholars of promises and representation should attend to other potential contextual differences. One obvious area for exploration is the effect of the changing environment of promises content. Candidates and elected officials tend to make partisan promises, focusing on issues that give themselves the advantage (Benoit and Hansen Reference Benoit and Hansen2004; Fagan, Kostadinova, and Tafolar Reference Fagan, Kostadinova and Tafolar2022; Petrocik Reference Petrocik1996). A second area where context may produce different partisan reactions is variation of promise content and speaker identity that may make decisions around accountability more or less opaque. This includes differences in promising intentions compared to outcomes (e.g., Kostadinova and Kostandinova Reference Kostadinova and Kostadinova2016). But what accountability means may also differ based on audience differences in understanding promises and meaning due to speaker identity (e.g., race, gender, party, etc.), audience identity, and promise content—for instance, on issues of race (e.g., Bonilla, Filindra, and Lajevardi Reference Bonilla, Filindra and Lajevardi2022; Stephens-Dougan Reference Stephens-Dougan2020).
Third, this study adds to the evidence that studies of polarization should separately analyze how voters process information when discussing policy issues and on candidate evaluations. A critical debate that has played out in this space is whether it is possible to demonstrate partisan directional reasoning (Druckman and Leeper Reference Druckman and Leeper2012). Because evidence on partisan information processing shows that in some cases partisans are persuaded to move against partisanship by information (Coppock Reference Coppock2023), context may be critical to determine when this is the case (Bayes et al. Reference Bayes, Druckman, Goods and Molden2020). This study suggests that even though voters recognize and differentiate between levels of promise fulfillment, there are partisan differences in information processing when voters are asked about direct candidate evaluations. Thus, studies of partisan information processing may see partisans as more likely to attend to new information when discussing policies (e.g., Coppock Reference Coppock2023) than when voters are asked to evaluate candidates (e.g. Bayes et al. Reference Bayes, Druckman, Goods and Molden2020).
Finally, this article adds to growing research investigating how partisanship informs voter interpretation of democratic accountability. Other work has demonstrated that partisanship has shifted presidential approval (Donovan et al. Reference Donovan, Kellstedt, Key and Lebo2020). Here, however, we see that voter interpretation of representation is subject to partisan beliefs.Footnote 21 This suggests that an important avenue for studies of partisan-motivated behavior should be how partisan information processing may shift understanding of fundamental democratic principles, an outcome that is critical to why partisan information processing ultimately matters and how democracy functions.
SUPPLEMENTARY MATERIAL
The supplementary material for this article can be found at https://doi.org/10.1017/S0003055424000807.
DATA AVAILABILITY STATEMENT
Research documentation and data that support the findings of this study are openly available at the American Political Science Review Dataverse: https://doi.org/10.7910/DVN/FW87Y1.
ACKNOWLEDGEMENTS
Several people gave comments on drafts and to them, I give my sincere thanks: Jamie Druckman, Al Tillery, Nazita Lajevardi, and Amanda D’Urso. Thanks to the Chicago Behavioral Working Group for comments on a proposal and experimental findings, including Alexandra Filindra, Mary McGrath, Dave Doherty, Stephanie Edgerly, and Meghan Condon. Thank you to SPSA and APSA Conference participants in 2019, and MPSA Conference participants in 2021. A special thank you also to the Comparative Election Pledges Workshop in 2022—the feedback was invaluable. Finally, thank you to the anonymous reviewers who each offered important contributions to clarify, extend, and strengthen the work. Of course, all shortcomings remain my own.
FUNDING STATEMENT
This work was in part funded from an APSA Small Grant, partly by the Northwestern Institute for Policy Research, and partly by the Northwestern School of Education and Social Policy.
CONFLICT OF INTEREST
The author declares no ethical issues or conflicts of interest in this research.
ETHICAL STANDARDS
The author declares the human subjects research in this article was reviewed and approved by Northwestern University IRB and the certificate numbers are provided in the appendix. The author affirms that this article adheres to the principles concerning research with human participants laid out in APSA’s Principles and Guidance on Human Subject Research (2020).
Open access
Comments
No Comments have been published for this article.