|

What do Americans want in their state judges?

by

Spring 2021 | Volume 105 Number 1 | Download PDF Version of Article

As scholars regularly document,1 states have frequently changed their systems of judicial selection and retention. What remains unknown is whether these systems actually address the kinds of qualities citizens value in their state judges.

Since 1946, the most frequent change has been to adopt some kind of system that proponents describe as “merit selection” that constrains the governor or other selecting authority to choose from a list of candidates nominated or approved by a screening body. The most common form of merit selection is a version of Missouri’s Nonpartisan Court Plan, often referred to as the “Missouri Plan.” In this system, the governor fills all vacancies by making an appointment for an initial term from a list forwarded by a nominating commission, and incumbents stand in yes-no “retention elections” for subsequent terms. Thirteen states have adopted this system.2 But, interestingly, the Missouri Plan was last approved by voters more than three and a half decades ago — in Utah, in 1985.3 Since then, voters in several states have rejected this system for some or all courts.

The other version of merit selection, which I label more generally as “constrained appointment,” similarly limits selections to persons nominated or otherwise screened by a legally required nominating or selection body, but it does not involve retention elections.4 Six states have adopted this system for some or all major courts.5

The debate over judicial selection is often framed around judicial independence versus accountability. Those favoring independence over accountability tend to promote systems of selection that deemphasize a direct role for voters; those favoring accountability over independence tend to prefer popular elections. In a recent book, Charles Geyh, who has long favored appointment over election,6 acknowledged that there are some good reasons for preferring elective systems, especially for state supreme courts.7 Geyh goes on to note, however, that much of the debate has focused on appellate judiciaries, and argues that contested elections for trial court judges may be even more problematic. The logic is that most of the work of trial courts “consists of routine matters in which the law is clear and the policy implications of the court’s legal rulings are limited,” and that the appeals process “keeps the excesses of trial courts in check.”8 He points out the research that has found perverse effects of the election cycle on criminal sentencing in states where trial judges stand for retention in potentially contested partisan or nonpartisan elections.9 In contrast, state appellate courts, particularly courts of last resort, have a significant lawmaking function, particularly regarding common law issues; moreover, the decisions of state supreme courts are rarely subject to review by the U.S. Supreme Court.10 Given the greater policy role of appellate courts, electing the judges of those courts seems more justifiable.

Geyh’s argument that it is more difficult to justify electing trial court judges than appellate judges is counter to actual practice, assuming one counts the Missouri Plan as an appointive system. Ignoring the issue of appointments to fill interim judicial vacancies, 17 states have abandoned popular elections for the selection of some or all appellate court judges over the last 100 years.11 Sixteen of those states now use retention elections for subsequent or full terms for appellate judges; the exception is New York where appellate judges are subject to reappointment. There are no states that elect appellate judges but appoint judges of major trial courts.12

In three states, Florida, Oklahoma, and South Dakota, voters approved the adoption of a Missouri Plan system for the states’ appellate courts but later rejected proposals that would have extended the system to trial courts. In several other states that adopted the Missouri Plan for appellate courts, the legislature also considered the system for trial court — but there was never sufficient support to put it before the voters.

These divergent patterns of change suggest at least two questions: Why are voters in some states apparently unwilling to give up elections for trial courts even though they are willing to do so for appellate courts? And why are legislatures in additional states not even willing to put before voters the question of whether to forego elections for trial courts? One possible explanation is that voters and legislators see different characteristics as desirable for judges at the appellate and trial level.13 How might one describe possible characteristics that voters view as important in selecting judges?14

Surprisingly, the question of what citizens view as important in selecting judges has not been extensively explored. James Gibson, in a study of the 2006 nonpartisan election for the Kentucky Supreme Court, asked survey respondents to rate the importance of ten characteristics of a “good Kentucky Supreme Court judge.”15 Gibson omitted some obvious things such as “being fair and impartial” because he expected there would be little or no variation on those items. The two top characteristics were “protect people without power” and “strictly follow the law,” with 72.9 percent and 71.8 percent respectively rating them as “very important.” Next was “state how they stand on important legal and political issues as part of their campaigns,” with 64.2 percent rating this as “very important.” The two lowest-rated characteristics were “decide the way the majority wants” (30.1 percent, very important) and “base decisions on party affiliations” (18.5 percent, very important). To the extent these are expectations of what a judge should do, they may be more relevant when deciding whether to retain a judge — rather than elect her for the first time — since at least some of these characteristics would be difficult to assess without a history of judicial decisions made by the candidate.

Gibson also provides some data from a 2001 national survey conducted by Justice at Stake (JaS), a now-defunct organization that advocated the reform of judicial selection.16 That survey asked respondents to rate the importance of ten responsibilities of courts and judges using a 0-to-10 scale. The mean responses ranged from 8.19 (51.8 percent rating at 10) for “defending constitutional rights and freedoms” to 6.23 (18.1 percent rating at 10) for “advancing social and economic justice.” The others rated near the top were “ensuring fairness under the law,” “protecting civil liberties,” and “protecting individual rights.” Toward the bottom one finds “resisting political pressure” and “being an independent check on other branches of government.” As with Gibson’s own survey, the JaS survey focused more on expectations than on qualifications.

What, then, are the characteristics Americans want in their state judges, and do these characteristics differ depending on the type of court a judge will serve on?

Survey Design and Sampling

For answers to these questions, I conducted a short survey, distinguishing between what I label “political characteristics” and “professional characteristics.” I identified six characteristics that I believe are essentially political in nature:

  • Respected by elected political officials
  • Experience running for and/or holding political office
  • Experience as a criminal prosecutor17
  • Strong support from the leaders of my preferred political party
  • Understands community preferences
  • Active in community organizations
  • And another six that are more professional in nature:18
  • Deep legal knowledge
  • Reputation for integrity/high ethical standards
  • Excelled in law school
  • Substantial experience practicing law in the courtroom
  • Reputation as a good listener
  • Respected by leaders of the legal community

The respondents rated each of the 12 characteristics twice, once for the state supreme court and once for local trial courts, using a four-point scale with the points labeled “essential (4),” “very important (3),” “somewhat important (2),” and “not important (1).” One potential point of confusion is that in New York what is called the “supreme court” is not what most people think of as the “state supreme court”; the highest court in New York is called the Court of Appeals.19 The survey instructions alerted respondents to this issue in order to make it clear that they were to think about the state’s highest court.20 The survey also asked for the respondent’s political party affiliation, self-described ideology, gender, level of education, age, and the first three digits of the respondent’s zip code (used to determine the respondent’s state).

Due to limited time and resources, I obtained a sample using Amazon Mechanical Turk (MTurk).21 MTurk respondents are self-selected rather than randomly sampled. Consequently, one must be careful in interpreting results based on MTurk samples, because there are some known biases, including overrepresentation of males, political liberals, persons under 45, and those with at least a college education. To ameliorate over-representation of liberals and over-representation of males, I did the survey in three stages.

In the first stage, the initial sample of 500 respondents overrepresented persons describing themselves as liberal as compared to what was shown in recent random sample surveys.22 To correct for this, I collected in phase two an additional 100 responses to balance the survey using a feature of MTurk that allowed me to restrict respondents to self-identified “conservatives.” Analysis of the combined sample of 600 showed both that women were underrepresented and that the women on average rated all of the characteristics as more important than did men. Consequently, to balance on gender, I collected in phase three a second supplemental sample of 65 women, producing a final sample of 329 men and 336 women, including one transgender female (plus three respondents selecting “Gender Variant/Nonconforming” and one who preferred not to answer the gender question). The two biases I did not adjust for were age and education;23 preliminary analysis showed that there was little correlation between either age or education and respondents’ ratings of desirable judicial characteristics.24 My final sample had usable responses from 669 respondents.

Results

Professional and Political Dimensions

A first question is whether the 12 characteristics diverged along the lines of professional and political as I hypothesized. To assess this, I applied a statistical method called factor analyses that can be used to assess whether a set of questions groups along one or more dimensions.25 I applied the method — both combining the responses regarding the two levels of courts and separately for local trial courts and state supreme courts — and found that the 12 characteristics did indeed group along these two dimensions.26 With one possible exception, the specific characteristics aligned nicely with my expectations, including experience as a criminal prosecutor, reflecting a professional dimension and a political dimension. The one exception is “respected by leaders of the legal community,” which for local trial judges split across the two dimensions, but which was slightly stronger on the professional dimension.

Turning to how the characteristics were rated, Figure 1 (above) shows the distribution of responses for each characteristic, with the professional characteristics at the top and the political characteristics at the bottom. The percentages rating a characteristic as “essential” appear in orange. The figure clearly shows the greater importance assigned to professional characteristics, with the percentage of respondents rating those characteristics as “essential” higher than any of the political characteristics — with the one exception of “understands community preferences” as regards to judges of local trial courts.

Table 1 (above) shows two statistics for each court. The first column in each pair (“Mean”) is the mean rating and the second column (“% Essential”) is the percentage of respondents rating the characteristic as “essential.” The characteristics are ordered based on the percentage rating a characteristic as essential for state supreme court justices. As indicated in the table, for 10 of the 12 characteristics, respondents, on average, differentiated between the two courts when rating the characteristics.27

Regarding desired characteristics for state supreme court justices, the mean ratings for the professional characteristics all exceeded 3.4 on the 4-point scale, ranging up to 3.8, while the mean ratings for the political characteristics were all less than 3.0, ranging from 2.3 to 2.9. Over 80 percent of respondents rated two of the professional characteristics — “deep legal knowledge” and “reputation for integrity/high ethical standards” — as essential for supreme court justices. Three additional characteristics in the professional category — “excelled in law school,” “substantial experience practicing law in the courtroom,” and “reputation as a good listener” — were deemed essential for state supreme court justices by 70 to 77 percent of respondents. The highest ranked political characteristic for state supreme court justices was “respected by elected political officials,” but only 40.4 percent rated it as essential. In fact, for state supreme court justices, all the professional characteristics were rated higher than any of the political characteristics. However, it is noteworthy that three of the six political characteristics were also rated higher for state supreme courts than for local trial courts. Overall, this suggests that the public may have higher expectations for state supreme court justices than for local trial judges, particularly with regard to professional qualifications.

Although the professional characteristics also tended to be deemed the most important for trial court judges, there were notable differences in ratings compared to those for supreme court justices. The means for the professional characteristics ranged from 3.1 to 3.7, very close to but never exceeding the corresponding means for the state supreme court. The range for the political characteristics was 2.1 to 3.3, several exceeding the corresponding rating for the state supreme court. No characteristic was deemed essential for local trial judges by more than 80 percent of respondents, with the highest, “reputation for integrity/high ethical standards,” deemed essential by 76.1 percent of respondents. The top-ranked characteristic for the state supreme court, “deep legal knowledge,” was deemed essential for local trial courts by only 66.1 percent of respondents compared to 88.0 percent for the state supreme court. And although 77.0 percent rated “excelled in law school” essential for state supreme court justices, only 34.4 percent thought it was essential for local trial court judges. Interestingly, only 49.3 percent rated “substantial experience practicing law in the courtroom” as essential for trial court judges compared to 72.2 percent saying it was essential for a state supreme court justice.

While three of the professional characteristics stood at the top of the rankings for trial court judges, two of the political characteristics were rated substantially higher for trial court judges than for state supreme court justices. “Understands community preferences” was deemed essential for trial court judges by 54.4 percent of respondents; only 21.2 percent rated this as essential for state supreme court justices (means 3.3 and 2.3). Similarly, 31.4 percent rated “active in community organizations” as essential for trial court judges compared to 15.7 percent for supreme court justices (means 2.8 and 2.3). Thus, there was measurably less emphasis on professional characteristics and more emphasis on characteristics reflecting local knowledge and connections for trial court judges than for state supreme court justices.

Political Affiliation of Respondents

Given the degree of political polarization in the United States as this is written, it is interesting that “strong support from the leaders of my preferred political party” was close to the bottom for both courts. Also, one might ask whether there were systematic differences between Republicans and Democrats in preferred characteristics. The answer is largely no. Comparing Democrats and Republicans,28 only 2 of 24 comparisons met the criterion for statistically significant differences, and those differences were modest. Republicans rated “experience as a criminal prosecutor” more important for state supreme court justices than did Democrats (33.1 percent essential versus 27.2 percent). Democrats rated “active in community organizations” higher for local trial court judges than did Republicans (34.1 percent essential versus 29.8 percent). Although both met the criterion to be statistically significant, the differences are of minimal substantive significance.

Based on the two-factor analyses, I combined the responses to obtain four scales, two for each court.29 One pair of scales was associated with political characteristics and the other with professional characteristics. I adjusted all scales to have an average of five and a standard deviation of one, with higher scores indicating a higher level of importance assigned to characteristics associated with that dimension. Table 2 (below) shows the scale averages broken down for each of the political and demographic variables included in the survey.30 The table also shows the probability (the “p-value”) that the variation across the categories of a variable could be attributed to chance, which when very low is referred to as “statistical significance.”31

The only differences that meet the criteria for statistical significance (i.e., a probability of occurring by chance of .05 or less) for self-identified ideology were the professional scale for local trial courts (with conservatives rating professional qualities lower, on average, than either liberals or those labeling themselves middle-of-the-road) and the political scale for the state supreme court (with liberals rating political qualities lower, on average, than the other two groups).

In addition to asking the respondents their self-identified ideology, I asked them which political party they identified with (Democrat, Republican, Independent, or other), what political scientists refer to as “party identification.” The pattern with party identification is interesting. Those identifying as Democrats or Republicans had higher average scores on the political scale than those identifying as Independents, even Independents leaning toward one of the parties; these differences meet the criterion for statistical significance. The pattern of the relationship between the professional scale and party identification is muddled, although it tends toward the opposite direction (i.e., higher average scores for Independents than for Democrats or Republicans). Thus, the differences here reflect “partisanship” — identifying with a political party rather than which political party a respondent identified with.

Selection System in Respondents’ Home States

The respondents’ states (based on the zip code information) were recoded into the states’ initial selection system: contested election (ignoring appointments to fill interim vacancies), Missouri Plan, or appointment.32 Separate variables were created for appellate selection and for trial selection; states in which trial selection varied by county or judicial district were coded “missing” for trial selection. As Table 2 shows, there were no statistically significant differences based on the formal system of initial selection in the respondent’s state of residence, although variations in the political characteristics scale for trial courts approached statistical significance: It is unclear what to make of the fact that respondents where the Missouri Plan is used for judicial selection in trial courts rate political characteristics higher than do respondents in states using contested elections or other appointment systems.

Gender, Age, and Education of Respondents

Turning to the demographic variables, gender stands out — with women rating both scales for both courts higher than men by about one-quarter standard deviation, rising to one-third standard deviation for the professional scale for the trial court. Moreover, ratings for the professional scale appear to rise with age, while ratings for the political scale appear to decline with age. The maximum differences related to age for the political scale are greater than a full standard deviation; the maximums for the professional scale slightly exceed one-third standard deviation. Regarding education, the differences for the trial court do not reach the standard of statistical significance for either the political or professional scale. Regarding the state supreme court, both scales tend to decline as education increases, but these findings meet the criterion for statistical significance only for the professional scale.

What happens if the five predictor variables — ideology, partisanship,33 age, education, and gender — are taken together? To answer this question, I combined those predictors in a regression model, details of which are provided in the online appendix.34 The patterns shown in the regression results were generally consistent with the bivariate results in Table 2.

Summary

There are some particularly interesting observations based on my analysis. First, women tended to rate both professional and political qualities higher than did men. Second, there appears to be a kind of “partisan impact,” with those who clearly identify with a party rating the political scale higher and the professional scale lower than the two categories of Independents. Finally, the patterns for age and education raise intriguing questions: Why do younger respondents tend to view political qualities in a judge as less important, while older respondents view professional qualities as more important? And why do more highly educated respondents value professional qualities less than those with a lower level of education?

Conclusion

The major limitation of this study is the sampling source. Ideally, the survey would have been done using a true national random sample rather than a self-selected MTurk sample. Hopefully the results presented here will inspire a replication using a better sample and include additional questions that measure variables such as political knowledge and political interest. In the interim, the results presented help account for why voters (and legislatures) in some states were willing to adopt variants of the Missouri Plan for state supreme courts but rejected that system for trial courts: Members of the public view certain professional characteristics — the very characteristics that Missouri Plan advocates argue nominating commissions will emphasize — as more important for state supreme courts than for trial courts. The public apparently views an understanding of the local political situation as more important for trial court judges than for state supreme court justices. This suggests that there is at least some understanding of the differing roles played by local trial courts and the top state appellate courts.

This also partially explains a greater willingness to adopt variants of the Missouri Plan for appellate courts than for local trial courts. Specifically, popular elections are more likely to keep judges tied to the local community than is selection through appointment. However, if voters understood that most trial judges in most “election” states initially obtain their positions by appointment to fill interim vacancies, the preference for election over appointment might decrease.35 Nonetheless, using contested elections for retention arguably allows voters in a community to reject appointees of governors not from the locally dominant political party, and there is some evidence that this does sometimes happen.36

Even with the differences between the two types of courts, professional characteristics tend to be deemed more important than political characteristics for both levels of courts. This in turn raises the question as to why there has been a lack of success in recent years in adopting versions of the Missouri Plan, which tends to emphasize professional characteristics. Essentially, there is an irony here: Voters appear to want judges with strong professional characteristics but seem increasingly inclined to distrust and reject mechanisms for judicial selection designed to focus on those very characteristics.

I argue elsewhere that this partly reflects that business interests that once supported systems such as the Missouri Plan now prefer contested elections.37 Those interests have learned that, in such elections, they can get their desired candidates elected and defeat judges perceived to be hostile to interests of business. The broader conservative movement, epitomized by the Federalist Society38 and the Heritage Foundation,39 has come to see elections as preferable to systems using nominating commissions (like those central to the Missouri Plan) based on the belief that commission lawyers tend to produce liberal judges. The argument is that lawyers tend to be more liberal than the general electorate, which leads to those commissions nominating lawyers who are also more liberal than the electorate.40 Conservatives also argue, relatedly, that domination of lawyers in the nominating process of the Missouri Plan makes the judicial selection process overly elitist and lacking in democratic legitimacy.41 Opponents of the Missouri Plan have learned that they can successfully argue the plan turns selection over to lawyers and deprives the public of its right to vote on who should be selected as a judge. Moreover, the argument goes, given the very small number of judges defeated in the Missouri Plan retention elections, judges would effectively be selected to serve until they die, reach mandatory retirement, or choose to depart the bench voluntarily. In several Missouri Plan states, conservative opponents of the plan have undertaken campaigns to end the requirement that the governor appoint from a list forwarded by a nominating commission in filling vacancies on appellate courts. These efforts have been successful or partially successful in two states: Tennessee for all appellate courts42 and Kansas for the state’s intermediate appellate courts.43

The results of the survey reported here raise the interesting question of whether proponents of systems that involve formalized nominating commissions, either via a full Missouri Plan system or a system of constrained appointment for all major judicial vacancies, could educate the public about at least four things:

  • how the various judicial selection systems work in practice,
  • what characteristics the various systems effectively prioritize,
  • how the frequency of interim appointments limits the role of popular elections in most states that ostensibly use such elections for initial selection,
  • and the fact that nominating commissions are intended to yield judges with the kinds of qualities (e.g., more “professional” qualities such as deep legal knowledge, integrity, high ethical standards) citizens claim to prefer.

To be effective, such a public education program would need not be specifically tied to any event but could include a range of activities over a period of years.

A major part of this challenge is the American love affair with elections. The number of state and local offices elected, even omitting judges, is probably unique to the United States. There are even places where the dog catcher (or “animal control officer” in modern parlance) is elected.44 This love affair, baffling to people in other countries,45 is one of the major roadblocks to Americans accepting nonelective systems of state judicial selection. It is debatable whether such a public education campaign could change voters’ views on how judges should be selected, but this research does suggest how such a campaign might be framed.

However, even if the public could be convinced that the intent of systems employing a nominating commission is to produce judges with the kind of qualifications citizens view as important in judges, it is not clear that the kinds of systems now in use actually will achieve that goal. Extensive research has sought to assess whether the different selection systems used by the American states do in fact differ in the qualifications possessed by the resulting judges. Although there may be some differences in the prior background of the judges (e.g., systems of legislative election put more former legislators on the bench than do other systems), the general conclusion is that there is little or no difference in the qualifications of the judges selected.46 Perhaps if there were a way to professionalize the screening process and design that process to go beyond the kind of reputational assessment now used, nominees would have better qualifications than those produced under the current system.47

Finally, one can ask whether more should be done to further specify the nature of the political characteristics that voters are concerned about regarding who should be selected as judges. Two of the characteristics that showed statistically significant differences between the two levels of courts (“active in community organizations” and “understands community preferences”) dealt specifically with the potential judges’ connections to the local community and presumably their understandings of the local community. It is unclear whether this translates to a belief that judges will act in accordance with community preferences when confronted with difficult cases. It is important to think about the difficulties in getting at this question. Recall that the two lowest-rated characteristics in Gibson’s study of Kentuckians’ views of state supreme court candidates were “decide the way the majority wants” and “base decisions on party affiliations.” Voters may be reluctant to state a preference for judicial decisions to be based on majority preferences or party characteristics, but may still have in the back of their minds a hope or expectation that this will actually be the case.

Footnotes:
***This paper is an extended analysis of a survey reported in my recent book on judicial selection reform, Judicial Selection in the States: Politics and the Struggle for Reform (2020). Thanks to Lawrence Baum, Charles Gardner Geyh, James Gibson, and Melinda Gann Hall for helpful thoughts in the design of the survey reported below and/or on a draft of this paper. The funding for the survey was provided by the University of Minnesota Law School’s Steen Fund.***
See, e.g., Jed Handelsman Shugerman, The People’s Courts: Pursuing Judicial Independence in America (2012); Herbert M. Kritzer, Justices on the Ballot: Continuity and Change in State Supreme Court Elections (2015).
2 Alaska adopted the Missouri Plan when it became a state in 1960, and Missouri had adopted its plan in 1940.
3 The Tennessee legislature had adopted a version of the Missouri Plan for all appellate courts in Tennessee in 1971, but then repealed it for the state supreme court three years later; in 1993 the Tennessee legislature readopted a version of the plan for the state supreme court in 1993. See Herbert M. Kritzer, Judicial Selection in the States: Politics and the Struggle for Reform 104–08 (2020) [hereinafter Judicial Selection].
4 There are other states where a governor has by executive order or the legislature has by statute created a nominating/screening body, but the governor is not legally constrained to follow the recommendations of that body.
5 One of these six is South Carolina, where the selection from among the nominees is made through an election in the legislature rather than by gubernatorial appointment.
See Charles Gardner Geyh, Why Judicial Elections Stink, 64 Ohio St. L.J. 43 (2003).
7 Charles Gardner Geyh, Who Is to Judge? The Perennial Debate Over Whether to Elect or Appoint America’s Judges 124–154 (2019).
Id. at 154.
Id. at 155. See also Gregory A. Huber & Sanford C. Gordon, Accountability and Coercion: Is Justice Blind When It Runs for Office? 48 Am. J. Pol. Sci. 247 (2004). Gibson suggests that the issue is not the higher sentences as reelection nears but the lower sentences imposed outside the reelection period. James L. Gibson, Electing Judges: Future Research and the Normative Debate About Judicial Elections, 96 Judicature 223, 228 (2013).
10 Id.
11 This count does not include New Mexico, which adopted a system of initial appointment but with appointees subject to standing in a partisan election at the time of the next general election and sometimes losing that election; due to some anomalies, there are also occasional open-seat elections.
12 Some minor or special (e.g., probate) court judges, assistant judges, and magistrates are appointed in states that elect judges of appellate and major trial courts.
13 Importantly, particularly in states using nonpartisan elections for trial court judges, most incumbents standing for election do not face an opponent. This may lead trial judges to be content with the existing system because once the filing deadline has passed, they need not worry about opposition to their continuing in office. In a system using retention elections, trial judges can face a last-minute campaign opposing their continuing in office when it is too late to raise money and organize a campaign. One result of this possibility, rare as a challenge may be, is that many trial judges may prefer to continue the existing system rather than change to a system using retention elections.
14 Note that there may be a conceptual difference between characteristics evaluated for initial selection versus retention (with the former perhaps more concerned with background and experience, and the latter perhaps more concerned with past judicial decisions).
15 James L. Gibson, Electing Judges: The Surprising Effects of Campaigning on Judicial Legitimacy 93–96 (2012).
16 Id. at 92. JaS favored appointment of judges. It ceased operating in 2017. The survey is available at https://www.brennancenter.org/sites/default/files/2001%20National%20Bipartisan%20Survey.pdf, accessed January 19, 2021.
17 At first glance, one might think that “experience as a criminal prosecutor” should fall under professional characteristics. However, I chose “experience as a prosecutor” rather than the more neutral “experience as a criminal lawyer” because I viewed specifying the prosecution side as making this more political. The analysis bears this out.
18 The survey also included a seventh item in each category, “expressed commitment to protecting legal rights” in the professional category and “expressed willingness to be sensitive to community preferences” in the political category. I omitted them from the analysis presented in this paper because in retrospect I realized that they were not really characteristics but campaign pledges.
19 This is also true of Maryland, but in Maryland, there is no court named the “supreme court.” Some other states have slightly different labels for their highest courts, but those include the word “supreme”; two examples are the West Virginia Supreme Court of Appeals and the Massachusetts Supreme Judicial Court.
20 The instructions read: “The survey asks you to rate 14 characteristics [see supra note 18], both as applied to potential judges of your local trial courts and as applied to judges of your state’s highest court (called the state supreme court except in New York where it is called the Court of Appeals).”
21 I was able to exclude potential respondents outside the United States, even those using a virtual private network (VPN) to make it appear they were in the United States.
22 The ideology question in the survey included the options “progressive” and “libertarian”; for purposes of analysis, I combined progressives with liberals and libertarians with conservatives.
23 Another bias in MTurk samples that I did not try to adjust for is that the level of political knowledge of MTurk respondents tends to be higher than in other panel-based samples. However, at least in terms of basic knowledge (presidential succession, vote needed to override a veto, presidential term limit, length of U.S. Senate term, number of senators per state, and length of U.S. House term), that gap is “not large”: 71.3 percent correct for MTurk samples versus 63.5 percent for the ANES 2008-09 Panel Study. Adam J. Berinsky, et al., Evaluating Online Labor Markets for Experimental Research: Amazon.com’s Mechanical Turk, 20 Pol. Analysis 351, 359 (2012). It is possible that there is a greater gap for more obscure items of political knowledge. See Taylor C. Boas, et al., Recruiting Large Online Samples in the United States and India: Facebook, Mechanical Turk, and Qualtrics, Pol. Sci. Rsch. & Methods 1, 11 (2018).
24 The strongest single correlation with age was -.176 with “experience running for and/or holding political office” for state supreme court justices; for education, the highest correlation was -.127 with “deep legal knowledge,” also for state supreme court justices. However, one remaining problem with education was that there was only one respondent with less than a high school education.
25 The results of the factor analysis can be found in the online appendix located at http://judicature.duke.edu.
26 Factor analysis produces estimates of how each item correlates with the underlying dimensions, with each dimension having high correlations with a subset of the items; the correlations are thus used to identify groups.
27 To “test” whether there was a difference, I used the sign test. See Sidney Siegel, Nonparametric Statistics for the Behavioral Sciences 68–75 (1956). The advantage of this test is that it does not require the assumption of interval measurement or normality that is required for the matched-pairs t-test. I also did a matched-pairs t-test. I put “test” in quotes because the nonrandom nature of the MTurk sample means that the tests do not technically apply.
28 This was done by cross tabulating the four-category ratings with a two-category party variable that combined those who said they were independents but leaned toward a party with those identifying with a party; “pure” independents were omitted.
29 Technically, the scales are factor scores which are weighted sums of the responses.
30 The ideology question on the survey included the options “libertarian” and “progressive”; for purposes of analysis, I collapsed the former with “conservative” and the latter with “liberal.”
31 The p-values are the result of one of two statistical tests: a two-sample difference of means (averages) t-test or a one-way analysis of variance (ANOVA). A probability of .05 or less is typically taken as indicating “statistically significant” differences.
32 Legislative election, used in Virginia and South Carolina, was treated as a form of appointment.
33 Partisanship is a recasting of party identification into a three-point scale of Independent (1), Independent leaning Democrat or Republican (2), and Democrat or Republican (3).
34 See online appendix, supra note 25.
35 There are two states, Arkansas (see Ark. Const. amend. 29) and Louisiana (see La. Const. art. V, § 22 [adopted 1974]), where appointees to fill an interim vacancy are not permitted to run in the subsequent election to fill that position.
36 Kritzer, Judicial Selectionsupra note 3, at 35–36.
37 Id. at 120–21.
38 Michael DeBow, et al., The Case for Partisan Judicial Elections, 33 U. Tol. L. Rev. 393 (2002).
39 Deborah O’Malley, A Defense of the Elected Judiciary, 57 Legal Memorandum 1 (2010), http://thf_media.s3.amazonaws.com/2010/pdf/lm0057.pdf.
40 Brian T. Fitzpatrick, The Ideological Consequences of Selection: A Nationwide Study of the Methods of Selecting Judges, 70 Vand. L. Rev. 1729 (2017). See also Adam Bonica & Maya Sen, The Politics of Selecting the Bench from the Bar: The Legal Profession and Partisan Incentives to Introduce Ideology into Judicial Selection, 60 J.L. & Econ. 559 (2017). There is good evidence that the legal profession as a whole is more on the liberal side of the political spectrum than is the general public. See Adam Bonica, et al., The Political Ideologies of American Lawyers, 8 J. Leg. Analysis 277 (2016); see also Adam Bonica & Maya Sen, The Judicial Tug of War: HowLawyers, Politicians, and Ideological Incentives Shape the American Judiciary 124–28 (2021).
41 Stephen J. Ware, The Missouri Plan in National Perspective, 74 Mo. L. Rev. 751, 758–66 (2009).
42 Kritzer, Judicial Selectionsupra note 3, at 117–122.
43 Id. at 328. As of 2020, Republicans in Kansas have been unsuccessful in their efforts to end the role of the nominating commission for selecting state supreme court justices (id. at 325–334). There have been unsuccessful (as of 2020) efforts in Missouri to reduce the role of lawyers on the nominating commission (id. at 320–322) and Oklahoma to end the use of nominating commissions for appellate court (id. at 338–343).
44 Amy Kold Noyes, Can’t Get Elected Dogcatcher? Try Running in Duxbury, Vt., NPR Weekend Edition Saturday (March 24, 2018), https://www.npr.org/2018/03/24/595755604/cant-get-elected-dogcatcher-try-running-in-duxbury-vt; Dough Ireland, Danville Animal Control Officer Race Heats Up, North Andover(Massachusetts) Eagle Tribune (Feb. 5, 2013), http://www.eagletribune.com/latestnews/x1525010958/Danville-animal-control-officer-race-heats-up.
45 See Steven E. Schier, The Minnesota Ballot Is a Joke in Canada, Minneapolis StarTrib., Dec. 22, 2002, 4A.
46 For a recent analysis of this issue, see Greg Goelzhauser, Choosing State Supreme Court Justices: Merit Selection and the Consequences of Institutional Reform 71–83 (2016).
47 See Kritzer, supra note 1, at 255–260, for a description of how such a system might work.

About Herbert M. Kritzer

Herbert M. Kritzer is the Marvin J. Sonosky Chair of Law and Public Policy at the University of Minnesota Law School. He is former editor of Law & Society Review and author of Judicial Selection in the States: Politics and the Struggle for Reform (Cambridge University Press, 2020) and Advanced Introduction to Empirical Legal Research (Edward Elgar Publishing, 2021).