II. Voter Information and Participation in Judicial Elections

Voters participate in judicial elections much as they do in other types of contests. When individuals are knowledgeable about the candidates running, they are more likely to participate. Nevertheless, these elections are not always salient to voters, and they exhibit high rates of abstention. In recent years, close to one-fifth of all Americans who turned out to the polls failed to participate in a state supreme court race (Bonneau and Hall Reference Bonneau and Gann Hall2009). Critics argue that voters are not sufficiently well informed to hold judges accountable (Geyh Reference Geyh2003). This could be because campaigns or other political intermediaries such as the press do a poor job educating voters about judicial candidates (Caufield Reference Caufield and Streb2007; Schaffner and Diascro Reference Schaffner, Segal Diascro and Streb2007).

Despite these criticisms, research shows how competitive, expensive, partisan, and even churlish campaigns can educate and animate the electorate. The party label is arguably the most important heuristic in judicial elections (Klein and Baum Reference Klein and Baum2001; Schaffner and Streb Reference Schaffner and Streb2002; Hall Reference Hall2007). Partisan ballots stimulate participation by orienting voters toward fellow partisans. Even if voters know little about a candidate’s qualifications, they can make reasonable inferences about which individual will best represent their preferences. Nonpartisan, hybrid, and retention elections do not use party labels and exhibit more abstention than partisan contests. Without party labels, voters cue off of other sources of information such as a name (whether it appears to be of a particular gender, race, or member of a prominent political family), a recent political scandal, or critical judicial performance evaluations (Johnson, Schaefer, and McKnight Reference Johnson, Schaefer and McKnight1978; Dubois Reference Dubois1984; Frederick and Streb Reference Frederick and Streb2008; Kritzer Reference Kritzer2015; Gill Reference Gill, Bonneau and Gann Hall2017).

The competitiveness of judicial campaigns can also affect voter participation. Competitive elections attract campaign spending, interest group involvement, and media coverage, all of which alert voters to where candidates stand on issues important to them (Bonneau Reference Bonneau2005; Hall and Bonneau Reference Hall and Bonneau2006, Reference Hall and Bonneau2008; Goldberg Reference Goldberg and Streb2007). Campaign messages are key to voter knowledge. Television advertising is a highly salient means of judicial campaigning. Even when advertisements are churlish, they contain information that informs voters about which candidates will represent their political values (Hall and Bonneau Reference Hall and Bonneau2013; Hall Reference Hall2015; Hughes, Reference Hughesforthcoming).

The press is yet another source of civic education. Journalists convey information to voters that enhances the policy-making process, promotes democratic accountability, and furthers representation (Snyder and Strömberg Reference Snyder and Strömberg2010). Journalistic media include newspapers, television news broadcasts, radio shows, and, more recently, social media. The press informs voters by covering and contextualizing political campaigns, policy making, and the behavior of politicians (Chong and Druckman Reference Chong and Druckman2007). In one innovative study, Mondak (Reference Mondak1995) found that residents of neighborhoods in Pittsburgh where newspaper deliverymen went on strike were less likely to participate in local elections compared to their peers in Cleveland who were not subject to a similar drought of news. Druckman (Reference Druckman2005) found that newspapers were an especially informative medium, although others argue that the quality of reporting rather than the medium itself enhances voter knowledge (Zhao and Chaffee Reference Zhao and Chaffee1995; Norris and Sanders Reference Norris and Sanders2003).

The media, however, could also abet confirmation bias. Dubbing it the “Fox News Effect,” DellaVigna and Kaplan (Reference DellaVigna and Kaplan2007) found that Fox News’ entrance into a media market energizes conservative voters, boosting turnout and Republican vote-shares in these districts. More recently, the emergence of social media has raised questions such as whether online social networks can produce more informed, engaged citizens if they share news stories with one another. These online social networks reduce the costs of information and closely parallel other forms of human communication (Bond et al. Reference Bond, Fariss, Jones, Kramer, Marlow, Settle and Fowler2012). Nevertheless, some remain skeptical that these and other forms of electronic information are little more than political amusement (Prior Reference Prior2005; Vitak et al. Reference Vitak, Zube, Smock, Carr, Ellison and Lampe2011).

Unlike the broader literature on political participation, courts scholars have yet rigorously to consider the role the press plays in educating voters and stimulating participation.Footnote ¹ This is unsurprising given how sparsely scholarly research finds local media cover state courts (Hale Reference Hale2006; Schaffner and Diascro Reference Schaffner, Segal Diascro and Streb2007; Yanus Reference Yanus2009; Vining and Wilhelm Reference Vining2011). For example, Schaffner and Diascro (Reference Schaffner, Segal Diascro and Streb2007) found that the average state court race received fewer than 10 newspaper articles per race, while Vining and Wilhelm (Reference Vining2011) found that a mere .01% of all state supreme court decisions received front-page coverage in local newspapers. If voters have few opportunities to learn about either their courts or candidates through traditional outlets such as the press, then we might not expect there to be much of a relationship between media coverage and voter participation.

Despite the scholarly literature’s skepticism, politicians and voters exist within a mutually strategic environment (e.g., Ferejohn Reference Ferejohn1986). Political agents tailor their behavior to their principals’ preferences when it is likely that an unpopular decision will result in their removal from office (e.g., Hall Reference Hall1987). Courts scholars find that incumbent, elected judges posture for reelection, particularly in high-profile cases (e.g., Brace and Boyea Reference Brace and Boyea2008), and this posturing is even more pronounced when justices’ votes are reported by local media (Cann and Wilhelm Reference Cann and Wilhelm2011). Therefore, excluding the press from models of voter participation in state court elections overlooks not only the broader scholarly literature on voter turnout but also the courts literature’s own interpretation of principal-agency. Put differently, judges already behave as if the press can generate voter engagement, so is this type of behavior rational?

III. Local Media and Voter Information in Judicial Elections

If the media inform voters and influence participation in judicial elections, then what does this process look like? Journalists investigate candidates, their positions on important issues, and their qualifications for the bench, and they report new information that will further a voter’s ability to cast rational votes. To these ends, the media fulfill their agenda-setting role as they help to filter and frame events for public consumption (Clayman and Reisner Reference Clayman and Reisner1998; Chong and Druckman Reference Chong and Druckman2007; Iyengar and Kinder Reference Iyengar and Kinder2010). Perhaps most saliently, this can involve the transmission of new, previously unknown information that might significantly sway voters who either had not planned to participate or who had intended to vote for another candidate.

For example, shortly after Wisconsin’s 2011 judicial elections, reporters from Wisconsin Public Radio and the Wisconsin Center for Investigative Journalism broke a major story relating to a physical altercation between two justices on the Wisconsin Supreme Court (Lueders Reference Lueders2011). At least three anonymous sources told reporters that David Prosser attempted to choke his colleague, Ann Walsh Bradley, over a ruling to limit collective bargaining rights for union members. Another source claimed that Bradley approached Prosser, “with fists up” (Lueders Reference Lueders2011, A9). These reports, coupled with others alleging that Prosser had verbally assaulted Chief Justice Shirley Abrahamsan, helped frame the Wisconsin Supreme Court as a combative institution (Marley Reference Marley2011). They also helped set the agenda for the next state supreme court election. In his effort to unseat incumbent Pat Roggensack, Ed Fallone aired 703 television advertisements attacking Roggensack for failing to hold Prosser accountable for “choking another justice.”Footnote ²

Local journalists break major stories like those from Wisconsin with some irregularity. More commonly, coverage of judicial candidates revolves around the day-to-day occurrences of the campaign. For example, reporters attend candidates’ speeches, appearances at community events, or debates; they ask questions, fact-check answers, record candidates’ behavior, and convey findings to their readers. Individually, these fleeting snapshots might not paint a complete picture of a candidate, her positions, or qualifications for office, but taken over the long run, they can accumulate among the electorate and help them to determine which candidate best represents their interests.

The Atlanta Journal-Constitution’s (AJC) coverage of the 2004 Georgia Supreme Court race between Leah Ward Sears and Grant Brantley illustrates how local media coverage can paint a cumulative picture of candidates and their positions and qualifications, as well as even whether the race is likely to be competitive.Footnote ³ The AJC’s coverage kicked off when it attended the January 31 “Families and Freedom” rally in Atlanta.Footnote ⁴ At that gathering, Governor Sonny Perdue (R) urged attendees to vote against incumbent Justice Sears—owing in no small part to the vote she cast in striking down the state’s antisodomy law (Baxter and Galloway Reference Baxter and Galloway2004d).Footnote ⁵ By late April, the AJC reported rumors—and later confirmed—that Brantley would oppose Sears with the help of prominent state Republican strategists (Baxter and Galloway Reference Baxter and Galloway2004e, Reference Baxter and Galloway2004f; Tharpe Reference Tharpe2004). The next month, the AJC attended Brantley’s inaugural campaign event and reported the attendance and endorsement of the state’s former attorney general—the named defendant in the 1986 US Supreme Court case originally upholding Georgia’s ban on sodomy (Baxter and Galloway Reference Baxter and Galloway2004b).Footnote ⁶

With just over a month before the election, the AJC announced that same-sex marriage had become the dominant issue in the race as gay rights groups endorsed Sears and the Christian Coalition endorsed Brantley (Baxter and Galloway Reference Baxter and Galloway2004a). But within a matter of days, the Sears campaign tried to shift the narrative to Brantley’s ethics and issues related to Sears’s race and gender.Footnote ⁷ The newspaper reported that Brantley, who claimed to have been nominated to be a US district court judge, had never actually received such a nomination (Rankin Reference Rankin2004). Sears then accused state Republicans of targeting female and minority judges in their quest for the bench (Baxter and Galloway Reference Baxter and Galloway2004c). Ten days before the election, the AJC provided its readers with extensive coverage of the only supreme court debate, which was highlighted by numerous references to the candidates’ positions on salient issues and repeated personal attacks on one another (Badertscher Reference Badertscher2004b).Footnote ⁸ By the last 3 days of the campaign, the Sears-Brantley battle was front-page material.Footnote ⁹ Between April 20 and July 20, 2004, the AJC published a total of 41 articles about the day-to-day jockeying for the court.Footnote ¹⁰ Throughout the race, the newspaper highlighted and framed the issues likely to have been of greatest importance to its readers.

Local media cannot only reflect the state of the judicial campaign, but so too can they contribute to it. Newspaper editors, for example, make endorsements and print other professional and lay opinions. These pieces can in turn become grist for political campaigns. During Alabama’s 2004 Republican primary elections, newspaper endorsements became an issue in each of the three supreme court contests. The Alabama Civil Justice Reform Committee (a special interest group aimed at limiting tort awards) aired 36 television advertisements in support of Mike Bolin and Patti Smith, touting their support from three of the state’s largest newspapers. In the third race, Tom Parker weaponized newspaper endorsements by attacking his opponent, incumbent Jean Brown, for having received the support of “every liberal newspaper in Alabama.”Footnote ¹¹ He aired 404 television advertisements at a cost of over $100,000 to this end.

Reporters convey new information to voters, but they also help to reinforce information already before them. For example, the media report on candidates’ professional experience and qualifications, how they have ruled in previous cases, their partisanship (to the extent it is known), their endorsements (or denunciations), and so on. Voters could find this information on their own, but reporters help to lower the costs of information by putting it directly in front of them.

Consider former Idaho Supreme Court justice Cathy Silak. In October 1999, Silak wrote the majority opinion in a 3–2 decision that held that the federal government owned all unappropriated water in state wilderness areas (Fadness Reference Fadness1999).Footnote ¹² Silak was up for reelection the following May. Over the next 7 months, the Idaho Falls Post Register published 33 articles and opinions discussing Silak. In two-thirds of the articles, the newspaper mentioned Silak’s role in the unpopular case, including an editorial by Silak herself justifying her opinion (Silak Reference Silak2000). During this time, the state’s Republican Party organized behind a challenger, Daniel Eismann, who made Silak’s vote in the water case the cornerstone to his opposition effort (Fadness Reference Fadness2000). Voters may not have been initially aware of Silak’s role in the water case, but the media’s repeated references to it helped to reinforce the signal that she did not support local ownership of natural resources.Footnote ¹³

Similarly, local media can help to reinforce messages issued by the campaigns. For example, West Virginia’s 2004 slugfest between Republican Brent Benjamin and Democratic incumbent Warren McGraw helps show how the media can identify and amplify extant campaign messages. Benjamin, McGraw, and their supporters flooded the airwaves with thousands of advertisements—the vast majority of which were negative attacks—and the two combined to raise over $200,000.Footnote ¹⁴ In the week preceding the election, the Charleston Daily Mail published 33 stories about the two candidates (4.7 per day). Of these, 73% referenced the expense, salience, or tone of the campaigns and their advertisements. Voters might not have personally witnessed these churlish messages, but owing to their extensive coverage, the likelihood that voters did not at least hear about them is negligible. Arguably, when the media repeat salient campaign messages, they provide candidates with in-kind contributions vis-à-vis free advertising.Footnote ¹⁵

Finally, while the media can educate voters about which candidates are best qualified or best suited to pursue their policy preferences, they can also help to generate voter interest in an election. This can be accomplished, for example, by announcing to the public when an election will take place. But perhaps more importantly, the media can help voters to decide whether a contest is competitive enough to merit their participation (i.e., Downs Reference Downs1957). This is often achieved via horse-race coverage. Horse-race coverage simply focuses on which candidate is perceived to be ahead of another at any given point in time.Footnote ¹⁶ Voters have a strong preference for horse-race coverage, which influences campaigns, fundraising, and voters’ preferences over candidates (Ansolabehere and Iyengar Reference Ansolabehere and Iyengar1994; Mutz Reference Mutz1995, Reference Mutz1997; Iyengar, Norpoth, and Hahn Reference Iyengar, Norpoth and Hahn2004).

This section’s analysis of media coverage of judicial campaigns helps to inform my expectations for the empirical analysis below. Local media coverage of judicial campaigns highlights candidates, their qualifications, and their policy preferences. It also helps to set the tone and agenda for these contests by illuminating or reinforcing salient information to the electorate. Finally, the media can give voters a sense of how competitive an election will be by identifying viable opposition efforts. For all these reasons, I anticipate that greater rates of media coverage will lead to more voter participation, ceteris paribus.

IV. Data and Methodology

I study voter participation in 317 judicial elections in 34 state courts of last resort from 2002 to 2014. These are contests in which the electorate chooses among candidates for the bench, including partisan, nonpartisan, hybrid, and retention elections.Footnote ¹⁷ In this section, I identify the variables used in the analyses, measurement strategies, data sources, and statistical methodology.

B. Media Coverage of Judicial Elections

The primary independent variable of interest is a measure of state supreme court campaign coverage by local media. In this section, I review previous measurement strategies and evaluate their suitability to this study. I then outline and evaluate new data relating to state supreme court campaign coverage.

1. Previous Measures of State Court Media Coverage

Previous scholarly research measures state supreme court salience largely with respect to coverage in local newspapers (Hale Reference Hale2006; Schaffner and Diascro Reference Schaffner, Segal Diascro and Streb2007; Yanus Reference Yanus2009; Vining and Wilhelm Reference Vining and Wilhelm2010, Reference Vining2011). For example, Vining and Wilhelm (Reference Vining2011) examine state supreme court case salience by recording whether a court decision appeared on the front page of a state’s most circulated newspaper the day after that decision was issued. Schaffner and Diascro (Reference Schaffner, Segal Diascro and Streb2007), meanwhile, examine the number of state-capital newspaper articles that cover candidates in 51 partisan and nonpartisan state supreme court elections, regardless of where they appeared in a newspaper. These previous efforts have some merits, but neither is perfectly suited for the analysis at hand.

To begin, Vining and Wilhelm’s (Reference Vining2011) case-centric approach would be inappropriate because it can only measure voter familiarity with incumbent judicial behavior and can give no impression about what voters might know about challengers or candidates to an open seat. Furthermore, their strategy of measuring front-page coverage presents its own challenges. Schaffner and Diascro (Reference Schaffner, Segal Diascro and Streb2007) show that only about 14% of courts coverage appears on the front page of a local newspaper. Their measurement strategy, then, is likely a better model for the present analysis than is Vining and Wilhelm’s (Reference Vining2011) insofar as candidates, not cases, are the unit of analysis.

An additional problem arises, though, when we decide which newspapers to study. Vining and Wilhelm (Reference Vining2011) analyze the most circulated newspapers in each state, but this means that the types of publications under analysis can differ widely from state to state. In California, the most circulated newspaper is the Los Angeles Times, but in North Dakota it is the Fargo Forum. How focused will a large paper such as the Times be on local issues compared to the Forum? How much of the North Dakota agenda can the Forum set compared to the Times? Schaffner and Diascro (Reference Schaffner, Segal Diascro and Streb2007) show that nationally circulated newspapers disfavor local courts reporting compared to local ones. But Vining et al. (Reference Vining, Wilhelm, Hiers and Marcin2010) show that smaller newspapers do not report as frequently on state courts compared to more highly circulated newspapers.

While newspaper studies dominate analyses in this field, additional problems may stem from recent declines in print media. In 2016, the Pew Research Center reported that between 2004 and 2014, 126 daily newspapers went out of business.Footnote ²¹ Over a 20-year period, the entire workforce supporting print newspapers declined by 39%. A receding industry has made it harder for smaller newspapers to survive, leading national conglomerates to assume control of smaller newspapers. This is especially problematic for civic knowledge because nationally owned papers tend not to cover local issues as much as locally owned ones (Schaffner and Diascro Reference Schaffner, Segal Diascro and Streb2007).

Even with these declines in print media, however, contestation and competitiveness in judicial elections have remained relatively steady over the years, even increasing in some areas (Kritzer Reference Kritzer2015). The same Pew report in 2016 noted that newspaper websites’ traffic increased in recent years and that “incidental” readership (clicks from social media, e.g.) was common among the adult population. Indeed, many voters receive their information today from electronic media on the internet such as news sites, blogs, or social media. Pew reports that only 51% of Americans consume their news solely in print. Consequently, measuring voter knowledge of state supreme court candidates is difficult in an environment in which information is seemingly ubiquitous and the media in which it appears are rapidly evolving.

2. New Data on State Supreme Court Campaign Coverage

To address some of the difficulties discussed above, I gather media coverage data from the Associated Press’s (AP) State and Local Wire.Footnote ²² While the AP is an international news organization, its State and Local Wire consists of 143 local bureaus located in each of the 50 American states. Stories from these bureaus come from staff writers located within them and are supplemented with coverage from affiliated newspapers and broadcasters within those regions.Footnote ²³ Because the State and Local Wire pools resources from its own bureau staff with other affiliated reporters, it not only helps to account for how voters become informed across different types of media—print versus electronic, for example—but also for the various editorial agendas present within a state. Furthermore, because its stories are reported in the newswire format, it avoids the need to determine whether stories should come from the front page or elsewhere.

I measure local media coverage of state supreme court campaigns as the number of stories the State and Local Wire publishes about the candidates in a given contest. To obtain the data, I searched the State and Local Wire using LexisNexis. I conducted this search by using every candidate’s name in a given race and the state in which they were running. I accounted for the possibility that articles might mention multiple candidates for the same race in a single story, so every story is counted only once.

I coded the number of stories published about each contest in the 3, 6, and 12 months leading up to an election date. I then content-analyzed the results of these searches to remove any duplicate or irrelevant stories. The result of this data-collection process is three event count variables of the number of stories published about a given state supreme court contest. These new data improve on previous efforts in that they derive from a single source, cover multiple time periods within a year of an election, capture salience in every type of judicial election in the United States, and are not restricted solely to print media.Footnote ²⁴

I graphically summarize some of these new coverage data in figure 1. The top-left panel shows the frequency of coverage in the 6 months leading up to a state supreme court election. The average race had 7.9 stories mentioning candidates, but this distribution is largely right-skewed. Only 34 campaigns received no coverage whatsoever (10.6%), and 29 of these were retention elections. Over three-fifths of all judicial elections (64.4%) had between one and ten stories written about the candidates in the 6 months before an election; about a quarter (24.7%) had between 11 and 20 stories; and only 6.9% had more than 20. Some of the most widely covered elections were in Alabama, Florida, and West Virginia.Footnote ²⁵

Figure 1.

Local media coverage of state supreme court campaigns within 6 months of an election (2002–14). Top-left panel, number of AP stories. Top-right panel, number of stories published by year. Bottom-left panel, number of stories published by institutional design. Bottom-right panel, number of stories published within a given window of time, by institutional design.

Figure 1 also shows some clear trends in the rates of media coverage over the years of the analysis. The top-right panel displays the average number of stories published in the 6 months before each race. Note that coverage of state supreme court elections declined after 2004. From 2002 to 2004, the average race had 12.8 stories covering its candidates, but between 2006 and 2014, that figure fell by over half to 6.1 ( $t=8.85$ ). These declines likely reflect a constricting news industry during this period.Footnote ²⁶

The bottom-left panel in figure 1 shows how media coverage differs across judicial institutions. The results show that, on average, hybrid elections receive the greatest rates of coverage compared to other types of judicial elections—15.2 stories on average to 7.5, respectively ( $t=3.52$ )—although this finding is skewed by the salient races in Ohio along with the poor rates of coverage among retention elections (e.g., Baum and Klein Reference Baum, Klein and Streb2007).Footnote ²⁷ The results also suggest that nonpartisan elections receive slightly, although not significantly, more coverage than partisan ones—11.6 stories compared to 10.1, respectively ( $t=0.83$ ). Finally, retention elections receive systematically less frequent coverage compared to competitive elections—4.6 stories in the 6 months leading up to an election compared to 11.6, respectively ( $t=\mathrm{-7.04}$ )—unsurprising given how inconspicuous these races often are.

Finally, the bottom-right panel in figure 1 depicts some temporal trends in how judicial campaigns are covered throughout the 12 months leading up to an election. First, observe that the media takes an increasing interest in candidates during the homestretch of an election, regardless of institutional design. On average, the last 3 months of a campaign have approximately 93% more media coverage than during the period 3 to 6 months before an election date ( $t=5.96$ ). This is intuitive given that many states are still engaged in judicial qualifications and nominations during the period 3 to 6 months outside of a general election, and many candidacies will not reach their zenith until the last few months of the campaign. Nevertheless, I find minimal evidence that the final 6 months of a campaign are better covered than the 6 months preceding them—7.9 stories on average, compared to 6.9, respectively ( $t=1.40$ ).

3. The Content of State Supreme Court Campaign Coverage

What kinds of stories are local media most likely to cover in state supreme court campaigns? Previous research finds that the media—of any type—are drawn to controversy in their coverage of the courts (Vining and Marcin Reference Vining and Marcin2014). For example, newspaper coverage of state supreme courts is often drawn to cases involving civil liberties or declarations of unconstitutionality or to those with minimum winning coalitions (Vining and Wilhelm Reference Vining and Wilhelm2010). Even among death penalty cases, media are apt to lend greater coverage to cases with sensational facts than to those with broader legal implications (Vining, Wilhelm, and Collens Reference Vining, Wilhelm and Collens2015).

To further examine how local media cover state supreme court elections, I took a stratified sample of 10 state courts of last resort and content analyzed every article published within 6 months of an election. My sample includes two partisan election states, three nonpartisan election states, and five retention election states.Footnote ²⁸ My sample includes 99 state supreme court contests, 144 candidates, 716 news articles, and 13 content areas.Footnote ²⁹ To perform the analysis, I read each article and coded whether a particular attribute appeared in a given story (1 if yes, 0 otherwise), and then I added up the total number of content types during the 6 months preceding an election. I present results from the content analysis in table 1.

Table 1.

Content of State Supreme Court Election Media Coverage

Note.— Average number of stories in which a content area appeared in the 6 months before an election. Values in parentheses are the average percentage of all race stories that contain one of the content areas.

Entries in table 1 show the average number of stories per race in which a specific content area appeared in the 6 months before an election. Values in parentheses represent the average percentage per race of stories that contain references to specific content areas.Footnote ³⁰ The most common type of news coverage relates to candidates’ behavior in court cases. In the average contest, over one-third of local news articles referenced a candidate’s behavior. Nevertheless, there are stark institutional differences in this type of coverage. Such content appears in approximately three-fourths of articles covering retention elections but only about one-fourth of articles covering competitive elections ( $t=6.97$ ).

States that use competitive—particularly partisan—judicial elections see greater types of coverage related to their expense or tone. For example, 29.1% of stories in partisan races and 17.9% in nonpartisan races mention the campaign activity of the candidates for judicial office.Footnote ³¹ These figures significantly outweigh those for retention elections ( $t=9.60$ and $t=4.67$ , respectively). Similarly, 27.1% of stories covering partisan races mention the involvement of special interests or political actions committees (PACs), compared to only about 10.8% of nonpartisan or retention elections ( $t=2.27$ ).

Local media also help to convey candidates’ policy positions by discussing their partisan orientations, their stances on specific issues, their ideological predispositions, or the endorsements of other individuals. Among partisan contests, local media make nearly unanimous reference to the party identification of the candidates (84.9% of all stories), which, by no surprise, dwarfs that for other types of judicial elections ( $t=13.78$ ). Other types of cues get somewhat less coverage. Only about 5.4% of news stories mention candidates’ positions on specific issues; 3.2% mention their ideology; and 2.8% mention endorsements. The media are slightly more likely to mention nonpartisan candidates’ positions on issues ( $t=3.80$ ) and their endorsements ( $t=1.70$ ) compared to other types of courts.

Local media can also help to build popular interest around judicial campaigns by drawing greater attention to them. Horse-race coverage, for example, makes up about 5.1% of local coverage of judicial elections. The bulk of this goes to nonpartisan elections, where 11.6% of all stories mention the horse race—significantly more than other types of elections ( $t=3.64$ ). Relatively few stories cover some of the more sensational types of campaign activity such as mudslinging or scandals. Only 2.6% of stories covering judicial elections were devoted to mudslinging, and less than 1% discussed a scandal. Neither attribute appeared in any of the retention elections under analysis, and the bulk of all scandal coverage is attributable to the highly salient judicial elections in Alabama.

Local media also provide voters with other types of useful, if not sensational, information in the months leading up to an election. In about 14.1% of news stories do the media make announcements that remind voters of an upcoming election.Footnote ³² In only about 11.5% of all stories are candidates’ qualifications discussed, but this figure is due to poor rates of coverage among retention contests. In 24.7% of partisan contests, and in 17.3% of nonpartisan contests, are the candidates’ qualifications discussed. In only 4% of all retention elections is an incumbent’s qualifications mentioned—significantly less than in contested races ( $t=\mathrm{-4.91}$ ). News outlets also remind their subscribers about the types of administrative duties courts perform.Footnote ³³ In only about 6.8% of stories do the media discuss courts’ administrative responsibilities.

4. A Robustness Check on the Validity of the AP Data

Does the AP’s coverage of state supreme court campaigns accurately reflect the types of information to which voters are likely to have access? It is not to be expected that voters learn about local news by literally reading articles coming across the State and Local Wire. Rather, one would expect that individuals in AP bureaus respond to the same types of stimuli as do those who work for local newspapers, for example. In this section, I present a robustness check on the validity of this assumption by comparing results from the AP to a sample of stories in local newspapers.

On their face, the new media coverage data appear to tap into concepts of salience. Some of the most highly covered races in the data set include the 2004 West Virginia race between Brent Benjamin and Warren McGraw, the 2012 Alabama race between Roy Moore and Bob Vance, and the 2002 Ohio race between Tim Black and Maureen O’Connor. Some of the least covered states are those without any challengers (i.e., retention elections). As a more rigorous check on robustness, I took a stratified sample of five state supreme courts—one partisan, two nonpartisan, and two merit selection states—and replicate the three aggregate variables from the AP using the most circulated newspapers in those states.Footnote ³⁴ In all, these total 50 state supreme court elections.

In table 2, I present Pearson’s correlation coefficients between the AP measures of campaign salience and those taken from the five additional printed newspapers. I divide the results across three periods—the last 3 months of the race, the intermediate 3 to 6 months before the race, and the beginning 6 to 12 months before the race. Correlation coefficients show how well the two sources of information covary. As can be seen from the results in table 2, all three periods have high Pearson correlation coefficients between the AP and print newspaper measures, and all three easily reject the null hypothesis that there is no covariance between the two. These results suggest that the AP’s coverage of state supreme court elections is a reasonable proxy for more traditional forms of media.

Table 2.

Comparison of AP Stories and Five Printed Newspapers

Note.— $N=50$ . Correlation coefficients relate to how well Associated Press coverage and local newspaper coverage covary.

VI. Conclusion

This research makes important new insights into the scholarly debate surrounding judicial elections, the public’s capacity to participate in them, and the media’s ability to facilitate that participation. The press is an important political intermediary in American politics, and yet virtually nothing is known about how the news media influence voter engagement in judicial elections. To address these shortcomings, I gathered new data relating to local media coverage of state supreme court campaigns. These data measure the volume of coverage in every type of judicial election across 34 states, 13 years, and 317 total state supreme court contests. Along with these new data, I also coded the content of over 700 news articles. The result is one of the most comprehensive analyses of media coverage of state supreme court elections to date.

I examined the effect of media coverage on voter participation in state supreme court elections. I found that increasing rates of coverage had a mobilizing effect on voters that was comparable to that of expensive campaigns. Additional analyses found that part of this effect was likely attributable to campaign expense and advertising rather than to the media itself. Nevertheless, robustness checks on merit selection races without any spending or advertising showed an independent effect of media coverage on voter participation. Additional tests showed that specific media content such as candidates’ voting behavior, qualifications, mudslinging, and the horse race were also likely to increase voter participation. Therefore, this research not only contributes to our understanding of how voters engage in judicial elections (e.g., Bonneau and Hall Reference Bonneau and Gann Hall2009; Hall Reference Hall2015) but also why supreme court justices might pander to popular preferences in response to increasing media coverage (Cann and Wilhelm Reference Cann and Wilhelm2011).

This research points to causal links between media coverage and voter participation, but without experimental methods, it is nearly impossible to eliminate altogether alternative explanations of voter participation. While subsequent work will need to take up this research agenda, others should consider the myriad ways in which media coverage might also affect judicial elections. For example, does greater transparency in the media hinder the incumbency advantage as campaign expense and advertising have been found to? And how has a receding news industry affected principal-agency in judicial elections? Finally, while I have briefly considered how the media choose to cover judicial campaigns, nothing approaching a full-blown theoretical or empirical model was advanced herein. Future scholarship will need to take up this task.