Facing the Public: Voter Assessments of Judicial Scandal

Abstract

A decades-long debate has sought to determine whether judicial elections are detrimental to the impartiality of judges and public support for state court systems. We contribute to that discussion by assessing whether elections work to hold errant judges accountable. We use a novel dataset of judicial scandal, including newspaper reports of misconduct against elected state supreme court justices. Our data examines the effect of scandal from 2000–2023 to determine whether controversy affects voter support for incumbents. We find voters pay attention to the harmful effects of scandal on judiciaries and that impacted incumbents experience diminished support, relative to other incumbents.

In 2005, Nathan Hecht, then an associate justice of the Texas Supreme Court, gave roughly 120 interviews supporting the nomination of Harriet Miers to the US Supreme Court.¹ Critics responded that Hecht’s support for Miers violated the Texas Code of Judicial Conduct, which prohibits judges from endorsing candidates for public office. The State Commission on Judicial Conduct reprimanded Justice Hecht; however, he was exonerated in 2006 after an appeal of his sanctions. Hecht experienced additional problems after the attorney who represented him during his appeal discounted his legal fees. Hecht was fined $29,000 for that infraction, though he paid only $1,000 after years of appeals.² Those controversial actions were reported throughout state press outlets, including Texas newspapers, exposing Hecht’s scandals to the voting public. Though Justice Hecht was reelected in 2006 and appointed by Governor Rick Perry in 2013 to serve as chief justice of the Texas Supreme Court, reports of political and legal controversy nearly derailed his judicial career.³

Scholars have debated the normative implications of judicial elections for decades. Critics claim that judicial elections threaten the independence of the judiciary and are detrimental to judicial legitimacy (Geyh 2003; O’Connor and McGregor 2011). Others claim that concerns about electing judges are overstated, with critics ignoring the beneficial attributes of elections and preferences of state citizens for judicial accountability (Bonneau and Hall 2009; Gibson 2012; Hall 2015). While this debate continues, questions about the standing of state judiciaries remain a central focus for understanding the administration of the law. Justice Hecht’s story illustrates that scandal does not always lead to electoral defeat; however, we consider whether media attention to judicial controversy affects voter support for judges tainted by scandal.

Research confirms that instances of scandal and misconduct erode trust in the performance of accused judges (Boston et al. 2023; Baker and Canelo 2024), but the electoral consequences of judicial scandal are less understood. We believe an examination of elections for the states’ highest courts (which we refer to as state supreme courts) provides a unique opportunity to understand how judicial (mis)behavior shapes public responses to scandal. Specifically, we address whether scandal decreases the vote share of incumbent judges when they seek reelection after being associated with scandal.⁴

We examine judicial scandal by collecting and examining newspaper reports of judicial scandal from 2000–2023 in the twenty-two states with competitive partisan or nonpartisan elections. We find that both lingering (since a judge joined a state supreme court) and recent (since a judge’s last election or interim appointment) controversies decrease an incumbent’s vote share. Further, as the number of articles covering a scandal increases, voter support decreases. The results suggest the public is highly attentive to judicial controversy and less willing to support judges where they are connected to scandal.

Changing judicial elections

Though debate continues over the use of elections for selecting judges, research on state courts notes that state supreme court elections have nationalized (Weinschenk et al. 2020). Voters in judicial elections are responsive to the same stimuli as those that determine the winners of presidential, congressional, or gubernatorial elections (Hall 2007, 2015; Hall and Bonneau 2008; Bonneau and Hall 2009). Though critics of judicial elections counter that the same forces that mobilize voters reduce the legitimacy of elective courts by making them “noisier, nastier, and costlier” (Schotland 1985), other scholars see active and engaged groups of voters in judicial elections. As a result, state supreme court justices tend to resemble the preferences of their constituents due to electoral accountability (Brace and Boyea 2008; Canes-Wrone, Clark, and Kelly 2014).

Studies of state courts find the transformation began during the late twentieth century. Moreover, recent decisions by the US Supreme Court, including Republican Party of Minnesota v. White (2002),⁵ have encouraged contemporary judicial elections to be highly competitive (Hall 2015; Hall and Bonneau 2008) with higher levels of voter participation (Hall 2015) and larger sums of money spent (Hall and Bonneau 2008). This change partially relates to the increased use of television campaign advertising and evolving rules for campaigns that have changed the tenor of what candidates can signal to the public (Hall 2015). Kritzer (2015) reasons, however, that changes in judicial election competition were driven by elections in the South due to the end of that region’s Democratic one-party domination. Throughout judicial elections, studies show that while incumbents generally win reelection (Hall 2001; Kritzer 2015), salient contests mobilize voters with important differences where elections allow or do not allow partisan affiliation (Hall 2007, 2015; Hall and Bonneau 2008; Bonneau and Hall 2009; Hughes 2019).

Empirical scholarship on state court elections finds media coverage has a positive effect on voter engagement (Hughes 2020). Local media coverage provides information relating to the qualifications of judicial candidates, which is meaningful beyond the essential role of party labels (Hall 2007). Media coverage also has an independent effect on voter participation that can reinforce the messages of campaigns, leading to less or more voter participation (Hughes 2020). Whether the substantive effects of media coverage extend past voter participation is evaluated in this paper, as we look to the electoral consequences of media coverage of judicial scandal.

The implications of legislative scandal

The electoral consequences of legislative scandal at the state and national levels have received much scholarly attention. Scandals can erode trust in institutions (Chanley, Rudolph, and Rahn 2000; Bowler and Karp 2004) and increase voter turnout (Praino, Stockemer, and Moscardelli 2013). While scandals may lead to decreased incumbent vote shares in subsequent elections, it does not mean imminent electoral defeat (Banducci and Karp 1994; Dimock and Jacobson 1995; Basinger 2012; Hamel and Miller 2019). Incumbents are shown to regain voter support in later elections (Praino, Stockemer, and Moscardelli 2013), though Basinger (2012) found that over 40% of scandal-tainted incumbents did not win reelection.

The nature of a scandal and candidate attributes matter for voter assessment. Research on political scandals finds the public differentiates between financial and moral controversies (Funk 1996; Doherty, Dowling, and Miller 2011; Pereira and Waterbury 2018; Gulati and Brown 2021), citizens are more forgiving of scandalous co-partisans (Dimock and Jacobson 1995; Walter and Redlawsk 2019; Funck and McCabe 2022), and men, more than women, are more likely to emerge from scandal successfully (Newark, Vaughan, and Pleites-Hernandez 2019). In sum, evaluations of scandal are context specific – varying based on the type of scandal and attributes of the accused.

Expectations about the electoral consequences of judicial scandal

The implications of scandal differ across institutions (Rottinghaus 2014, 2023). By theorizing about the electoral consequences of judicial scandal, it is important to identify distinctions between judges and other elected officials. While legislators, executives, and judges are guided by various codes of conduct, expectations for judges are unique. In the contemporary polarized environment, there are strong expectations for legislators and governors to work for their party and advocate for their position. Judges, on the other hand, are expected to be impartial, or at least maintain the appearance of impartiality, to preserve public confidence in the judiciary (Baum 2006; Vickrey, Denton, Jefferson 2012). Judges in all states, regardless of whether party labels are on election ballots, are expected to adhere to their state’s code of conduct and maintain a degree of impartiality.

State reliance on the American Bar Association’s “Model Code of Judicial Conduct,” which emphasizes impartiality, independence, and the avoidance of “impropriety and the appearance of impropriety,” reinforces this unique judicial structure.⁶ Thus, actions that might not be scandalous or unacceptable behavior for legislators might be perceived as inappropriate for judges. For example, while legislators and governors are free to endorse political candidates or claim membership in a political organization, endorsements like Justice Hecht’s backing of Harriet Miers are deemed inappropriate for state judges due to perceptions of bias. While overlap exists between what is considered inappropriate for legislators and judges (e.g., sexual misconduct, accepting bribes, etc.), we assume that judges are held to higher standards than legislators and are expected to conduct themselves in a manner that preserves the appearance of judicial impartiality.

Courts recognize the importance of maintaining public trust and are attuned to the public’s expectations. The public is more supportive of judges (Benesh 2006), courts (Tyler and Rasinski 1991; Benesh 2006), and decisions (Scheb and Lyons 2001; Zink, Spriggs, and Scott 2009) where the appearance of judicial behavior appears grounded in the law and procedural fairness. This support aligns with the strong emphasis on impartiality and independence in state codes of judicial conduct. Scandal can be damaging to a court’s image (Casey 1988) and decreases specific support for a judge accused of wrongdoing (Boston et al. 2023). Further, the public views judges accused of discriminatory misconduct as a threat to the rights of the group they target, as well as other minority groups (e.g., women, racial minorities, or ethnic minorities) (Baker and Canelo 2024). This line of research demonstrates the ways in which scandal can be detrimental to public perceptions of individual judges and courts as institutions.

For the public to hold errant judges accountable, we believe the public pays some level of attention to media reports of scandal. We expect the public will be attentive to localized instances of judicial scandal for several reasons. Judicial elections have become increasingly salient. Campaign spending has increased over time (Bonneau 2005) and candidates boost their visibility through television advertisements (Hall 2015), both of which increase participation in judicial elections (Hall and Bonneau 2008; Hall 2015). Vining, Wilhelm, and Collens (2015) find that “drama, novelty, and sensationalism” drive newspaper coverage of state supreme courts’ death penalty cases over legal salience. The most circulated state newspapers are more likely to dedicate front page coverage to state supreme court cases that address controversial issues (Vining and Wilhelm 2010). We assume the media is attuned to scandal, including judicial misconduct, given it is frequently salient in nature. We hypothesize that newspaper articles involving scandal (as measured by whether or not an incumbent has been connected to scandal) will lead to decreased vote shares for incumbent judicial candidates (H1).

Not all reports of judicial scandal are the same. Scandal varies from one report to those that occupy the media for an extended duration. We expect the effect of scandal on voter support for incumbent candidates will be more pronounced when the media dedicates added attention to incidents of scandal. When reports of scandal are more frequent, it is likely the public has more exposure to the issue. We hypothesize that increased coverage of scandals (as measured by the number of articles) will be tied to a decrease in vote share for an incumbent judge (H2).

We expect that more recent newspaper descriptions of scandal (those during an incumbent’s current term) should weigh heavily on the evaluations of voters. We hypothesize that more than reports of scandal during an incumbent’s entire career, recent scandal will be associated with a larger reduction in incumbent vote share (H3).

Models of electoral performance for incumbent candidates

We examine the effects of scandal on the electoral performance of incumbents in state supreme court elections. We pay careful attention to how public reports of scandal affect an incumbent’s vote totals. The research design employed in our study benefits from the advantages of comparative state analysis to examine various race-specific, contextual, and institutional factors that affect the outcomes of elections.

Dependent variables

To explore electoral competition, we follow the approach used by Hall (2001, 2015) and Bonneau (2007). Our outcome variable is the percentage of the vote received by an incumbent state supreme court justice (incumbent vote share).⁷ Data for incumbent votes were collected for all competitive partisan and nonpartisan elections from 2000–2023. Retention elections are not included due to their non-competitive format. Our judicial elections data originates from Chris Bonneau from 2000 to 2008, and we collected election data from 2009 to 2023.

Independent variables

Our models of electoral competition seek to explain how scandal affects the performance of judicial incumbents. We apply four primary measures of scandal. To identify where incumbents were involved in scandal, we collected instances where incumbent state supreme court justices were subject to events reported in a state newspaper. We follow the coding strategy used by Vining and Wilhelm (2010) and Vining, Wilhelm, and Hendricks-Benton (2025; also, Vining, Wilhelm, and Collens 2015) to collect media coverage of scandal events in a state’s primary newspaper. We direct our focus to state newspapers due to their assigned state court reporters and specific attention to political and legal events within a state.⁸ We apply Vining and Wilhelm’s (2010) list of state newspapers for the twenty-two states with partisan or nonpartisan elections. Though local newspapers in recent years have closed (Darr, Hitt, and Dunaway 2018), access to the newspapers listed by Vining and Wilhelm was not interrupted. We searched newspapers in Nexis Uni and other databases (e.g., Factiva and NewsBank), as well as direct access to several state newspapers. Our search terms included the name of the court, the individual justice, and a variety of issues connected to professional misconduct.⁹ We apply our focus to newspaper articles, but do not include editorial opinions among articles collected. A list of the newspapers searched and methods for access can be found in Appendix B.

The first primary independent variable (scandal) measures whether an incumbent had associated newspaper articles involving scandal prior to her election. This measures the lingering effects of scandal and whether allegations continue to affect incumbent performance throughout a judge’s career. We apply the year 1999, the year before our first incumbent election, as our first year searched. Our second independent variable (scandal before election) takes a similar approach to scandal yet considers only scandal reports since an incumbent’s last election or interim appointment. This alternative approach seeks to identify whether voters are prone to consider recent events more strongly than a candidate’s total history of scandal.

Two additional variables explore an alternative approach to scandal reports. We consider whether the number of articles covering scandal affect the performance of incumbents (scandal articles). To measure the effect of article quantity, we collected the total number of scandal articles involving an incumbent. The measure includes all newspaper articles from the point that an incumbent joined the court. The final scandal measure is a count of scandal articles since an incumbent’s most recent election or interim appointment (scandal articles before election). That measure, like the scandal before election variable, is designed to examine the effect of more recent scandal.

Control variables

We consider other fundamental attributes connected to judicial elections, including candidate-related and state context-connected attributes. Following studies by Hall (2001, 2015), Bonneau (2007), and Kritzer (2015), we consider how attributes of judges affect election performance. To control for candidate features, we include variables connected to whether an incumbent was appointed by a governor (interim appointment), is a woman (female), is seeking an elected chief justice position (chief justice race), was challenged by a candidate with lower court experience (quality challenger), participated in a contested primary election (primary contested), and had a competitive race (60% or less) in their most recent election (competitive position).¹⁰ To identify the attributes of candidates, we searched state election records, newspapers, campaign websites, and contribution records.

With the structure of state supreme courts, we consider election designs (partisan election), district versus at-large election formats (district election), number of years per term (term length), and professional incentives for service in state supreme courts (professionalization).¹¹ We also consider three attributes of elections – whether an election decided more than one position (multimember design), whether a race in the prior election year was won by 60% or less (competitive court), and the number of positions decided in an election year (number of seats). We control for two-year election cycles, with the 2000–2001 election cycle our base category.

Beyond our primary scandal-related independent variables, we must be alert to potential confounders. Specifically, we consider vote shares for judicial incumbents where we control for the electoral environment of states with competitive judicial elections. Our statewide election variable represents the winning vote percentage for statewide elections involving seven statewide executive offices and races for the US Senate. Information about how the statewide election variable was operationalized, as well as results for the expanded model are in Appendix C.

Appendix Table D1 provides a detailed description of the variables applied to the models in our study.

Modeling strategy

To test the effect of scandal where incumbent judges seek reelection in partisan or nonpartisan judicial elections, we focus on the 388 state supreme court elections involving an incumbent that took place between 2000–2023. Forty-five of the 388 incumbent campaigns (11.6%) involved reports of scandal in the state’s primary newspaper.¹² We explore all incumbent outcomes during the twenty-first century, a period connected to the era of new-style judicial elections and changes influenced by decisions of the US Supreme Court (e.g., Republican Party of Minnesota v. White (2002) and Citizens United v. Federal Election Commission (2010)).¹³

Our study of electoral competition involves candidate attributes which are nested in states. To explore how scandal impacts performance, we estimate multilevel models with random intercepts with two levels of analysis. Levels incorporated in our analysis include general features tied to states and specific attributes connected to candidates and elections.

Though studies have traditionally used ordinary least squares to explore judicial election outcomes, a benefit of multilevel models is recognition of data hierarchies by allowing for residual components at each level of hierarchy. Our two-level analysis allows for grouping of election outcomes within states that includes residuals at the candidate/election and state levels. The residual variance is partitioned into a between-state component (the variance of the state-level residuals) and a within-state component (the variance of the candidate and election-level residuals). State residuals represent unobserved state characteristics that we anticipate affect election outcomes.

Results

Descriptive overview of scandal

Figure 1 depicts the total number of incumbent candidates involved in scandal across the twenty-two states with competitive judicial elections. Figure 2 displays the total number of articles by state reporting incumbent scandal. A darker intensity of blue represents states with higher levels of scandal, while light blue are states with few or no reports of scandal. States that are excluded from the study are colored gray. Twelve judicial election states did not have a scandal, while Minnesota had the most implicated judges with six scandal-affected incumbents.¹⁴ Figure 2 reports that Texas had the most newspaper reports involving scandal with 119 articles. Wisconsin ranked second with 70 news articles.

Figure 1. Number of Incumbent Candidates Involved in Scandal by State.

Figure 2. Number of Newspaper Articles on Incumbent Scandal by State.

Incumbent history of scandal

In this section we report the results for our multilevel models of incumbent vote share. Table 1 provides results where we consider scandal over time (Model 1), scandal since an incumbent’s last election or appointment (Model 2), articles covering incumbent scandal over time (Model 3), and articles covering incumbent scandal since her last election or appointment (Model 4). Appendix Table E1 presents the full results for our models, including those for specific election cycles.¹⁵ Turning our attention first to an incumbent’s history of scandal, Models 1 and 2 perform exceedingly well toward explaining incumbent performance. The r-squared goodness-of-fit statistic indicates that both fitted models explain more than half of the variance connected to incumbent performance. Model 1 reports that scandal has a discernibly strong effect on how incumbent judges perform.¹⁶

Table 1. Multilevel Models Predicting Impact of Scandal on Incumbent Vote Share

Note: Dependent variable is incumbent vote share. Standard errors are in parentheses.

*** p < .01, ** p < .05, * p < .1 two tailed test of significance.

The substantive effects of our primary explanatory variables connected to scandal are presented in Figures 3–4. Figure 3 shows that where scandal was reported by a state’s primary newspaper, a judge’s vote percentage decreased by 7.3%. With Model 2’s analysis of recent scandal’s effect on incumbent performance (i.e., scandal since a judge’s last election or interim appointment), the results for scandal are slightly stronger and statistically significant.¹⁷ Figure 4 demonstrates that incumbents with scandal events during their most recent term saw their vote percentage decrease by 7.36. While voters react more intensely to recent scandal, the effect of scandal lingers throughout an incumbent’s career. The histograms embedded within Figures 3–4 display the overall distribution (by percentage) of scandal-affected incumbents. The histograms show that scandal was involved in 11.6% (over an incumbent’s career) and 10.1% (since an incumbent’s last election or appointment) of the contests involving incumbents from 2000–2023.

Figure 3. Predicted Effect of Incumbent Scandal Since Joining Court on Vote Share. Note: Histogram on x-axis displays distribution of incumbent scandal by percentage of elections.

Figure 4. Predicted Effect of Recent Incumbent Scandal on Vote Share. Note: Histogram on x-axis displays distribution of incumbent scandal by percentage of elections.

Returning to the control variables for Model 1, two candidate-related variables have meaningful relationships with incumbent vote percentage. Incumbents who ran against challengers with lower court experience saw their general election total decrease by more than 20%.¹⁸ Incumbents with a primary election challenger saw their vote decrease by almost 4.6%. Of the state level variables, multimember elections were associated with a 27% decrease in incumbent vote share. Two state characteristics were marginally associated with incumbent vote shares – a pattern that exists for each of our four models. States with an additional two years of term decreased incumbent votes by approximately 4%, while competitive races in the prior cycle were tied to a 3.5% decrease in incumbent vote shares.

From Model 2, the same control variables affected incumbent vote percentage. Challengers with judicial experience were tied to 19.9% smaller incumbent vote shares than where opposed by challengers without lower court experience. Incumbents with primary election challengers saw their general election vote drop by 4.5%. Among our state-level variables, multimember elections were associated with 27.2% smaller incumbent shares than where elections decided just one office. Length of term (4% decrease for two additional years of term) and competitive judicial races in the prior cycle (3.6% decrease) were marginally tied to the incumbent vote percentage.

Incumbent scandal articles

Models 3 and 4 in Table 1 provide evidence that the number of scandal articles is statistically associated with voter support for incumbents. Once again, a majority of the variance is explained by the features of the models. We evaluate in Model 3 the impact of the total number of scandal articles on an incumbent justice’s vote.¹⁹ As expected, a larger number of newspaper articles detailing scandal contributed to diminished voter support for incumbents. Figure 5 shows that each article detailing scandal decreased an incumbent’s vote percentage share by 0.44. With five articles, the incumbent’s share decreased by 2.2% or, considering the full range of scandal articles (0 to 34 articles), voter support decreased by 15.1%. The histogram on the x-axis of Figure 5 confirms that scandal articles are not the norm (88.4% of the observations were not tied to scandal). Yet, where we consider scandal articles throughout an incumbent’s career, we observe a wide range in the number of articles from one (2.8%) to thirty-four articles (0.3%) where we consider an incumbent’s entire career.

Figure 5. Predicted Effect of Incumbent Scandal Articles on Vote Share. Note: Histogram on x-axis displays distribution of scandal articles by percentage of elections.

Model 4 provides another examination of scandal articles, yet here articles published since a judge’s most recent election or interim appointment.²⁰ The model supports our expectation that recent events weigh on voter decisions to support scandal-affected incumbents. Each scandal article diminished support for incumbents by 0.77%. Five scandal articles lowered an incumbent’s vote share by 3.8%. Figure 6 displays the predicted effect of scandal articles on vote share across the range of articles (0 to 33). Voters lowered their support for incumbents by 25% where recent scandal was most intense. With the histogram in Figure 6, we again see a wide range in scandal articles from one (2.8%) to thirty-three articles (0.3%).

Figure 6. Predicted Effect of Scandal Articles Since Last Election. Note: Histogram on x-axis displays distribution of scandal articles by percentage of elections.

The control variables for Models 3 and 4 operate in a similar manner to Models 1 and 2. Support for incumbents in Model 3 decreased by 20.4% when opposed by candidates with lower court experience. Incumbents with primary election challengers saw their general election shares decrease by 4.5%. Of our state-level variables, incumbent vote share decreased by 27.6% in states with multimember elections compared to conventional designs that determine one office. Two state-level variables register a marginally significant effect. A two-year increase in term length was tied to a 3.7% decrease in incumbent vote total and competitive elections in a prior election cycle saw incumbent vote shares decrease by 3.8%. With the control variables in Model 4, incumbents received 20.5% smaller vote shares where opposed by challengers with lower court experience. Incumbents challenged in a primary election saw their general election support drop by 4.6%. Multimember elections diminished incumbent vote shares by almost 28%. Term length and electorally competitive courts were marginally connected to incumbent vote shares. A two-year increase in term length was tied to a 3.4% decrease in incumbent vote, and states with recent competitive judicial elections were connected to 3.8% smaller incumbent totals.

Several conventional measures failed to reach statistical significance across our four models, including variables tied to interim appointments, district elections, and state high court professionalization. We specifically note the performance of the partisan election variable. While the research on state court elections notes a prominent role for partisan and nonpartisan ballots on the performance of elections (Bonneau and Hall 2009; Hall 2015), we find no observable effect for election designs that do or do not allow partisan information.²¹

In our expanded study of scandal’s effect on incumbent vote shares where we control for the statewide election environment in Appendix C, we find the average winning vote for statewide elections had a statistically discernible and positive relationship with incumbent vote. In each model in Appendix Table C1, a one-percentage-point increase in the size of the average winning percentage for non-judicial races was tied to about half a percentage point increase for judicial incumbents. States with environments associated with larger shares for winners of statewide elections also have higher vote totals for judicial incumbents. The statistical relationship between scandal and incumbent vote share is nearly the same as those reported in Table 1.

Conclusion

Whether it is misconduct relating to the endorsement of a political actor, failure to perform the duties of office, misuse of state funds, or the physical attack of another justice, there have been many scandals involving state supreme court justices since the beginning of the twenty-first century. Research shows that scandalous behavior by judges may erode the public’s trust (Boston et al. 2023; Baker and Canelo 2024). However, prior to this study, we lacked a thorough understanding of how scandal translates into political action and whether judicial elections work to hold errant judges accountable.

Our primary goal has been to examine the effects of scandal on the electoral performance of incumbents in state supreme court elections. To do so, we examined data on media accounts of scandal from the primary newspapers of the twenty-two states with partisan or nonpartisan elections. We examined both the impact of incumbent scandal and the quantity of media reports on scandal events on incumbent vote share. We find that both lingering and recent scandals have a direct and strong negative impact on an incumbent’s performance at the polls. Further, as the number of articles covering the details of judicial scandal increases, electoral support for the incumbent again declines.

These findings make important contributions to our understanding of judicial election performance and whether public responses to scandal translate to political action. First, our findings reveal there are electoral consequences for scandal-affected judges. Incumbent candidates who engaged in or were accused of controversial behavior saw diminished support, relative to other incumbents, when seeking reelection. As such, the findings suggest judicial elections do “work” in that voters receive election signals from press reports and shift support away from tainted incumbents. Thus, on a normative dimension, our results should reassure those concerned about the performance of judicial elections. Second, our findings shed light on the ways in which the incumbency advantage can be eroded. While incumbent judges may win reelection, the public is not blind to their occasional misdeeds.

As we look toward future research and a more complete understanding of scandal’s role in judicial elections, we must further untangle how forms of scandal affect incumbent performance. For types of scandal (e.g., corruption, campaign violations, courtroom behavior), knowledge about what the public perceives to be more or less important is essential for evaluating the limits of judicial misbehavior. Our results reveal the electoral consequences for incumbents who decide to seek reelection. Investigating an incumbent judge’s choice to retire or face voters and what role the type of scandal plays in that decision is an important next step toward understanding the implications of judicial scandal.