In 1992, Leah Wards Sears became the first Black woman to serve on the Georgia Supreme Court. While Sears was frequently mentioned on President Barack Obama’s shortlists for US Supreme Court vacancies, she faced serious electoral challenges in a state where nonpartisan state supreme court races are generally sleepy affairs. This is perhaps best illustrated by the Georgia Republican Party and the Georgia Christian Coalition’s campaign to unseat Sears in 2004. This campaign was prompted by Sears’s frequent dissents in criminal procedure cases, especially death penalty cases.^[1] Sears won reelection handily with 62 percent of the vote and was soon elevated to the chief justiceship by her colleagues. But the concerted effort against her underscores the electoral consequences of decisions that do not align with voter preferences (e.g., Brace and Boyea 2008; Curry and Romano 2018). Since justices on almost all state supreme courts must face voters in some fashion periodically,^[2] they engage in strategic behavior to ensure voters give them another term on the court (e.g., Baum 1994; Hall 1992). We argue this strategic behavior is shaped, in particular, by justices’ overlapping identities.

Rooted in the idea of judicial accountability (e.g., Bonneau and Hall 2009; Brace and Boyea 2008; Devins and Mansker 2010; Fernandez and Husser 2020), judicial elections prompt justices to strategically alter their votes to be more in line with voter preferences (Hall 1992; Romano and Curry 2018). But not all justices are equally strategic (e.g., Baum 1994). While any number of things might predict strategic behavior (e.g., Hall 2018; Murphy 1964; Peltason 1971), recent studies find it often turns on immutable characteristics such as race or gender (e.g., Collins and Moyer 2008; Nguyen 2019). There is an extensive literature finding diverse justices decide differently from their white and male counterparts in some issue areas and under certain conditions (Collins et al. 2010; Haire and Moyer 2015; Kastellec 2013; Scheurer 2014). However, most studies focus on race or gender; rarely do they explore the intersection of minority statuses.

When existing studies examine female judges they typically mean white women; when they look at Black judges, they usually explore Black men. This is problematic, as a number of studies note overlapping identities create altogether different patterns of behavior (e.g., Crenshaw 1989, 1991; Dowe 2020; Szmer et al. 2015). Ultimately, one of the most fundamental questions in judicial behavior is how a justice’s characteristics shape her behavior. If scholars are to answer that question, it is imperative to account for how intersecting identities contribute to voting on state supreme courts.

In this chapter, we explore whether intersectional justices strategically alter their decision-making calculus in election years to account for voter preferences. In doing so, we draw on classic studies of state supreme court justice decision-making which find strategic behavior increases as elections approach (Hall 1992) and literature stressing minority candidates are more strategic and sophisticated than their less diverse counterparts (e.g., Milyo and Schosberg 2000; Nguyen 2019) because of their lived experiences (Anzia and Berry 2011; Fox and Lawless 2005). We test our expectations by analyzing all nonunanimous criminal procedure cases from twelve state supreme courts from 1995 to 2005. We draw on a number of datasets (Fairbanks 2020; Windett and Hall 2013) supplemented with original data collection on election proximity for all 109 justices who served in this time period. We find support for our expectations. Our chapter proceeds in several parts. First, we detail our theoretical expectations. We then overview our data and operationalization. We subsequently present our results. We then discuss our findings in the broader context of the judicial politics literature and acknowledge the limitations of this study. We close with suggestions for future scholars who might build on our study.

Strategic Behavior, Election Proximity, and Justice Identity

Many factors go into judicial decision-making, not the least of which is public preferences. Even federal judges, shielded by life tenure, consider public preferences for the purposes of implementation (Fix et al. 2017; Johnson and Canon 1984), institutional legitimacy (Badas 2019; Christenson and Glick 2019), or the quality of their daily lives (Peltason 1971). Importantly, while the public can critique or ignore the decisions of federal judges (e.g., Dolbeare and Hammond 1971), they cannot actually harm the judge’s continued tenure. This is not the case for state supreme court justices who must routinely face voters to secure another term. The precise electoral mechanism varies from state to state. Some justices face voters in competitive (non)partisan elections. In those using the Missouri Plan, voters answer a yes/no question on the ballot about whether the justice should get another term. In a handful of states, the state senate votes on whether the justice should remain on the bench (Goelzhauser 2018). If we accept that elected officials would like to remain in office (Mayhew 1974), justices must pay attention to voters and their preferences (e.g., Fernandez and Husser 2020; Curry and Romano 2018).

Work on state supreme court decision-making repeatedly finds justices strategically alter their decision-making calculus in high profile cases as elections approach (see for instance: Brace and Boyea 2008; Hall 1992). This work finds much of its genesis in a 1992 study by Melinda Gann Hall; there Hall notes justices are more likely to vote to uphold death sentences in election years. While this may deviate from a justice’s sincere preference, she must take care to ensure her decisions are in line with the voters, even if this deviates from personal preferences (Brace and Boyea 2008). The role of an impending election on strategic voting behavior is further underscored by another study examining judicial behavior in death penalty cases when all electoral restraints are removed.

Many states have mandatory retirement ages for their judges. Thus a subset of judges is (re)elected but face no future electoral pressures to consider voter preferences because their age precludes them from seeking another term. Hall (2014) finds these terminal-term justices vote more liberally in death penalty cases than similarly situated justices who are eligible for reelection. In tandem with Hall (1992), this provides strong evidence that justices strategically consider voter preferences as elections approach. But it is important to recall that justices vary in the extent of their strategic behavior (Baum 1994). Research on candidate emergence and campaigning, in judicial races and beyond, finds diverse candidates are often more strategic than their white and male colleagues (Fox and Lawless 2005; Milyo and Schosberg 2000; Nguyen 2019).

In order to face reelection, one must first ascend to the bench. Despite evidence voters prefer diverse justices (Arrington 2018) and do not electorally punish members of minority groups in some instances (Gill and Eugenis 2019), women are more risk-averse when deciding to run for office (Fox and Lawless 2005; Lazarus and Steigerwalt 2018). Indeed, women typically only commit to a race when they are well qualified (Milyo and Schosberg 2000; Nguyen 2019). This hesitancy to run helps explain why many diverse jurists initially take the bench through appointment (Bratton and Spill 2002; Goelzhauser 2020). But even if a justice is initially appointed, she must soon face the voters to continue her tenure. In anticipation of that and subsequent elections, she will likely engage in strategic behavior.

It is, of course, important for us to note that not all members of a given group have the exact same experience; nor is any group monolithic. Ruth Bader Ginsburg and Amy Coney Barrett are both women but have very different views on the law. Likewise, Thurgood Marshall and Clarence Thomas, the only two African Americans to serve on the US Supreme Court, are ideologically distant from each other. However, the fact remains, immutable characteristics contribute to unique experiences that may prompt a different way of looking at the world than one who has not lived that life (e.g., Boyd 2016; Norgren 2018).

Female state supreme court candidates are, on average, more strategic than their male counterparts (Nguyen 2019). This can be partially attributed to the Jackie (Jill) Robinson Effect (Anzia and Berry 2011). Briefly, diverse officeholders, much like Jackie Robinson when he joined the Brooklyn Dodgers in 1947, have to be twice as good to be considered half as good. To this end, female state supreme court justices are more likely to receive discretionary citations from their peers on other courts (Gleason et al. 2020). In the current context, women and racial minorities are more strategic and should be more inclined to think about how their decisions could impact their fortunes on election day. But questions remain about women who are also racial minorities. Existing research suggests overlapping minority statuses should make these intersectional justices even more strategic.

Intersectionality and Strategic Decision-Making

The scholarship on diversity in judicial decision-making usually looks at women or racial minorities (see, for instance, Boyd et al. 2010; Haire and Moyer 2015; Kastellec 2013; Walker and Barrow 1985). However, what of jurists like Justice Sears who are both female and a racial minority? To this end, Crenshaw (1989) introduces the concept of intersectionality, which occurs when an individual has multiple intersecting minority statuses.^[3] Intersectional identities create experiences that cannot be fully captured by just one minority status. Crenshaw (1991) examines the experiences of a Black woman pursuing employment discrimination litigation in the 1980s. At the time, the law allowed for discrimination claims on the basis of gender or race, but not both simultaneously. While there have been gains in recent years, intersectional discrimination claims remain “complex” and complicated by both structural and ideological factors (Mayeri 2015). This makes it difficult to impossible for a Black woman to fully articulate her experience at the intersection of two minority identities. Importantly for our purposes, Crenshaw’s (1991) example plays out repeatedly in a growing body of work on intersectional judges.

In the last fifteen years, the number of intersectional judges on state supreme courts has grown rapidly, but they still make up a small percentage of the bench (Goelzhauser 2020). Against this backdrop, there are a small but growing number of studies looking at intersectionality in the legal profession broadly and the judiciary specifically. These studies generally find intersectional identities lead to distinct patterns of strategic behavior. Intersectional attorneys have higher levels of perceived discrimination (Collins et al. 2017), which necessarily results in a more difficult path to the bench. However, after intersectional lawyers ascend to the state trial court bench, they are more ambitious than their counterparts (Jensen and Martinek 2009). Should this ambition lead them to the state supreme court, intersectional justices are more likely to support criminal defendants than their nonintersectional counterparts (Collins and Moyer 2008).

Other research indicates intersectional justices exhibit a distinct judicial style sooner than their unidimensional minority colleagues. Early studies of judicial diversity find women and racial minorities decided similarly to their white and female counterparts under most conditions (e.g., Songer et al. 1994; Songer and Crews-Meyer 2000; Walker and Barrow 1985). However, once women obtain a considerable number of seats on a given court, they exhibit a distinct decision-making style (e.g., Collins et al. 2010; Scheurer 2014). While intersectional justices are few in number (Golezhauser 2020; Szmer et al. 2015) note that intersectional justices take on a distinct decision-making style even when occupying a token number of seats on a court. This underscores that intersectional identities prompt a different strategic calculation than that employed by women or racial minorities. Thus, much as Crenshaw (1991) argues race or gender cannot fully capture the experiences of Black women filing suit, we contend existing accounts of strategic behavior by justices must likewise account for intersectional identities.

For over half a century, scholars have repeatedly found judicial behavior is partially based on a judge’s experiences (see, for instance, Boyd et al. 2010; Glynn and Sen 2015; Pelatson 1971). Importantly, experiences are shaped by identities and their intersection (Crenshaw 1989, 1991; Haire and Moyer 2015; Norgren 2018). To this end, accounts of strategic behavior on state supreme courts must account for justice identities and their intersection. We explore this in the context of vote direction as elections approach. We expect to find that while women and minority justices will more strategically consider voter preferences in election years than their less diverse counterparts (e.g., Nguyen 2019), intersectional judges will be the most strategic and exhibit the most attention to voter preferences (Crenshaw 1989, 1991).

Data and Methods

We evaluate our argument with a dataset of all nonunanimous criminal procedure cases from 1995 to 2005 for the states of Alaska, California, Georgia, Illinois, Louisiana, Maine, Michigan, Oregon, Pennsylvania, South Dakota, Vermont, and West Virginia (Windett and Hall 2013). These states are advantageous as they represent a cross-section of state supreme courts with respect to selection mechanism. Additionally, Fairbanks (2020) compiles demographic information on all justices who sat on these courts over the period of our study. This allows us to account for justice race, gender, and length of tenure. Using Kritzer’s (2015) data on state supreme court elections, we note the years in which each jurist faces voters. We ultimately are able to analyze 4,518 justice votes across 776 cases.^[4] We now turn to a more in-depth account of how we construct our data.

Hall (1992) finds justices usually vote to uphold death sentences in election years. Thus we use the direction of the justice vote to serve as our dependent variable. This means we need to know whether a justice’s vote in a given case is liberal or conservative. Unfortunately, Windett and Hall (2013) do not include a measure of ideological direction in their dataset of state supreme court decisions. However, we are able to create one by utilizing a unique feature of criminal procedure cases; the accused is always liberal and the state actor is always conservative (Gleason and Provost 2016; Segal 1984). Accordingly, we write a text crawler to search the petitioner and respondent variables for words typically associated with the state or law enforcement. Sample words include state, probation, sheriff, and warden. Using this technique, we are able to determine which of the two parties represents the state and is therefore the conservative party.^[5] We then use the disposition, or summary, variable to determine whether the Court ruled for the petitioner or the respondent. If the justice supports the criminally accused, she has cast a liberal vote and therefore the dependent variable takes on a value of 1. Should she support the state, she has cast a conservative vote and the dependent variable is set to 0.

Our primary independent variables exist on the justice level. We employ two of Fairbanks’s (2020) dichotomous identity markers. We set a gender variable to 1 if a justice is female, 0 otherwise. In a similar fashion, we set a race variable to 1 if a justice is Black; 0 otherwise. Since we posit that intersectional judges approach judging differently, we multiply our race and gender variables together. This takes on a value of 1 for Black female justices and 0 for all other justices. We also note if each vote is cast in a year in which the justice faces reelection. We do so via Kritzer’s (2015) data on state supreme court elections. This variable is set to 1 if it is an election year for a given justice, 0 otherwise.^[6]

We utilize Berry et al.’s (1998) ideology scores to note voter preferences. This measure, updated continuously since its initial release in the late 1990s (e.g., Berry et al. 2010), has two scores for state political preferences: one for citizens and another for state institutions. For each justice, we use the score that corresponds to the voters that will decide continued tenure in office. For justices reelected by (non)partisan or retention elections, we utilize the citizen score. For justices retained by the state senate, we employ the state institutions score. The scores range from 0 to 100 with higher values being more liberal. Since we argue strategic behavior is the product of the interplay between identity, election proximity, and voter preferences, we interact with all three variables. That is to say, we multiply our three primary independent variables together. Since we are interested in three separate identities, Black justices, female justices, and Black female justices, we create three interaction terms. The first is (Black justice X election year X voter ideology). The second is (Female justice X election year X voter ideology). The final one is (Black female justice X election year X voter ideology). These are our primary independent variables.

We also include a number of control variables suggested by the prior literature. Since different selection mechanisms tend to attract different kinds of jurists (e.g., Brace and Boyea 2008; Bratton and Spill 2002; Goelzhauser 2020; Hurwitz and Lanier 2003) and can even change a justice’s strategic calculation as she heads into reelection (Curry and Romano 2018), we include two variables. The first is set to 1 for justices who are retained via (non)partisan elections and 0 for all others. The second variable denotes justices retained via retention elections. This is set to 1 for justices in Missouri Plan states; 0 for all others.^[7] Since justices who are in their terminal term because of age restrictions are less likely to consider voter preferences (Hall 2014), we include a dichotomous marker set to 1 if the justice is ineligible to run for another term because of age (Fairbanks 2020).

A justice sitting on a less professionalized court, which is to say one with a heavy docket, less support, more mandatory cases, or less pay, may be inclined to deal with cases in the most routine manner. Accordingly, we include a measure of court professionalism (Squire 2008). Since a longer tenure likely results in more efficient writing, we include a measure of the total number of years a justice is on the bench (e.g., Savchak and Bowie 2016). Finally, ideology is one of the most powerful predictors of judicial decision-making (e.g., Brace et al. 2000; Segal and Spaeth 2002). As such, we include a measure of judicial ideology (Bonica and Woodruff 2015).^[8]

Since our dependent variable is binary, we employ a logit model. Moreover, since it is possible that idiosyncratic characteristics of each court may shape the decision-making process (e.g., Kritzer 2015), we cluster our standard errors on the state. We now turn to our results.

Results

We find some limited evidence that Black justices are more strategic in election years. We find no evidence female justices engage in this calculus. However, we find strong evidence that intersectional justices strategically consider voter preferences as elections approach. Our results underscore that while identity is a powerful predictor of behavior on state supreme courts, not all identities operate in the same way. Indeed, intersectional identities shape judicial decision-making differently than a given identity in isolation. We now turn to a detailed discussion of our results.

Our results are presented in table 1. An initial glance at the results indicates that Black justices are more likely to vote liberally. Female justices are less likely to vote liberally. Black female justices, for their part, are more likely to vote liberally. It also appears that voter preferences and election-year decisions do not have an impact on the probability of a liberal vote. However, it is important to remember that we argue that identity, voter preferences, and election proximity operate in tandem. That is to say, we cannot evaluate our expectations without considering all three independent variables at the same time. For this we must look to our interaction terms.

Typically, we assess a hypothesis by “star gazing” at the table and looking for stars next to our variable of interest. If we do that for the interaction terms, we might conclude Black justices and Black female justices are more mindful of voter preferences in election years, but female justices overall are not. This approach to quickly assessing statistical significance does not work for interaction terms because interaction terms have to consider all possible combinations of all the variables included in the interaction. Fortunately, interaction terms can be easily assessed via a graphical plot (Brambor et al. 2006).

We create three separate three-way interaction plots, one for each of the interaction terms. In each, the first panel displays the interaction in nonelection years, the second panel displays election years. The y-axis displays voter preferences, with higher scores representing more liberal electorates. The x-axis shows the impact of justice identity on the probability of casting a liberal vote. That is to say, higher values reflect a probability that the justice’s identity will shape the propensity to vote liberally. The solid sloped line represents the effect of the justice race on the probability of casting a liberal vote at that particular value of voter preferences. A positive slope means that as the liberalism of the voters goes up, the more likely justice identity is to shape the decision to vote liberally. A negative slope means that as voter liberalism goes down, identity has less of an impact on the justice’s decision to vote liberally. Importantly, at y = 0, there is a reference line. Should the dashed lines, which represent the 95 percent confidence intervals, include the reference line, the interaction is not significant at that particular value of voter ideology.

We now turn to interpreting the interaction figures. Figure 1 shows the interaction for Black justices. In the first panel, which represents nonelection years, we find as voters become more liberal, the justice race has a diminishing effect on predicting Black justice liberal votes. Importantly, once voter preferences pass approximately 50 (which is a pure moderate), voter preferences cease to predict Black justice liberal voting (as the dashed lines include the reference line). We note much the same in election years in the second panel. This suggests that in both nonelection and election years Black justices’ liberal votes are shaped by voter preferences; but only when the voters are more conservative.

In figure 2 we evaluate the interaction term for female justices. The layout and interpretation of the figure are the same as in figure 1. In both nonelection years and election years, we note the dashed lines always include the reference line at y = 0.^[9] This indicates identity does not shape female justice votes at any level of voter liberalism.

Figure 3 displays the three-way interaction for intersectional justices. Looking first at nonelection years, we note Black women are shaped by voter preferences at the extremes. That is to say, Black women consider voter preferences when a state is quite conservative. As a state becomes more liberal, identity exerts less influence over Black female justices’ votes in nonelection years. When a state is moderate there is no statistically significant effect. In election years, however, we find intersectional justices are highly strategic. When voters are conservative, Black women are less likely to vote liberally. As voters become more liberal, Black female justices become more likely to cast liberal votes. This demonstrates that intersectional justices are attentive to voter preferences in election years and provides support for our expectations that intersectional judges are more strategic than single-dimension minorities and all justices writ large. We also note one of our control variables reaches statistical significance. Justices retained by contested elections have a predicted probability of casting a liberal vote of 0.22 lower than their peers retained by other means.^[10]

Discussion and Future Directions

Judicial decision-making is a strategic enterprise in which justices weigh a number of competing factors (e.g., Baum 1994; Maltzman et al. 2002; Murphy 1964; Spriggs and Wahlbeck 2002). Of course, strategic considerations are shaped by the context at hand (e.g., Campbell and Wilcox 2020; Peltason 1971). At state supreme courts, these considerations include the justice’s continued tenure on the court, particularly as election time approaches (Hall 1992; Curry and Romano 2018). While retaining office is a powerful motivator (e.g., Mayhew 1974), not all justices are equally strategic. We demonstrate that strategic behavior is tied to justice identity.

Importantly, we note that not all identities are equally associated with strategic consideration of voter preferences. The most strategic justices are intersectional. Whereas Black judges appear to be strategic only when a state is quite conservative and women do not appear to be strategic in election years, we note that Black women are highly strategic. When the electorate is conservative she is less likely to vote liberally. As the voters become more liberal, she becomes more likely to vote liberally. This provides support for our expectations and underscores the importance of lived experiences; identities can often overlap and intersect in ways that create unique experiences and alter the strategic calculus (e.g., Crenshaw 1989, 1991). While our findings add to our overall knowledge of strategic behavior, they also pose a number of new questions brought on by both limits to our design and our results. In both instances, we encourage future scholars to more fully explore this topic.

Perhaps one of the primary limits to our study is the scope of our data. A future study, inclusive of the other 39 states and more recent terms, will allow scholars to produce a more nuanced understanding of how identity shapes judicial decision-making. This is particularly true as a number of scholars (Brace and Boyea 2008; Brace and Hall 1998; Curry and Romano 2018) stress that changing institutional context shapes the strategic considerations a justice enters into. In recent years elections have gotten more competitive (Hughes 2019, 2020; Kritzer 2015) and the ideological map has changed. To this end, we find Alaska in 2000 is the most conservative state (ideology score of 21.4). This is not surprising and is consistent with 2021 popular understandings of ideology. However, the most liberal state in our data is West Virginia in 1998 (ideology score of 70.7). This is surprising in the 2020s, an era where West Virginia is one of the most reliably “red” states. But in the late 1990s, it was still a Democratic stronghold. While we suspect the main patterns of our results would hold in the more contemporaneous and diverse landscape today, the contours of our results would change slightly.

Beyond the changing institutional and ideological map, one of the primary benefits to expanding the temporal scope of our data is the growing diversification of state supreme courts. While thirty-six Black women have served on state supreme courts from 1960 to 2016, twenty-two of them took the bench after 2005 (Goelzhauser 2020). Since our data stops in 2005, we are only able to analyze three intersectional justices. It is possible that our findings are the product of a Leah Ward Sears effect rather than an intersectionality effect. The addition of almost twenty additional intersectional justices would lend greater confidence to our findings and perhaps allow for further nuanced understanding of other overlapping identities.

Future studies should also look beyond criminal procedure cases. We adopt this restriction because Windett and Hall’s (2013) data does not contain a direction variable. Given the rapid progress in state supreme court case-level data (e.g., Brace et al. 2000; Hall and Windett 2013), we are hopeful future scholars will expand analysis to other issue areas. This would be particularly valuable as the impact of identity on judicial decision-making is often issue area specific (e.g., Boyd et al. 2010; Songer et al. 1994). Relatedly, legal considerations often restrict the range of options available to justices in a given case (Epstein and George 1996). To this end, perhaps one of the largest limitations of our study is the omission of case-level legal factors from analysis, such as the presence of dissent on the intermediate court or whether the defendant alleges unconstitutionality (e.g., Segal 1984). We omit these factors from our model because our data source does not include them. We are hopeful that future work will be able to take this into account.

Future studies might also look beyond the justice vote. While the justice vote is central to the study of judicial behavior, it is hardly the only activity of interest. Justices ultimately justify their view in written opinions. Importantly, opinions can be written to avoid retribution during retention (Curry and Romano and 2018). Scholars already know intersectional justices are more likely to pen dissenting opinions than their nonintersectional peers (Szmer et al. 2015). Might, however, electoral proximity alter the extent to which justices dissent? Drawing on recent work analyzing the content of opinions (e.g., Corley 2008; Fix and Fairbanks 2020; Fix and Kassow 2020), we suspect justices may temper the language in their opinions as elections approach (e.g., Bryan and Ringsmuth 2016). Particularly as data on state supreme court decisions are becoming increasingly available, we encourage future scholars to more fully explore this topic.

Opinions are further complicated by the back-and-forth that occurs between coalition members (Maltzman et al. 2002). While dissents can be solo affairs, in closely split cases dissent authors often hope to seize the majority. A robust literature explores how justices accommodate or reject requests from their colleagues in crafting opinions. While the literature indicates women are more conciliatory and collaborative (Bear and Woolley 2011; Boyd 2016), there is evidence that intersectional jurists are more likely to break from the group and stand alone in dissent (Szmer et al. 2015). Given these dynamics, we encourage future scholars to more fully explore what we suspect is the highly nuanced relationship between identity and coalition formation.

Building upon expansive work on race and gender, we contribute to the small but growing scholarship on intersectional state supreme court justices. We find intersectional justices are strategic in ways different from Black justices and women justices. While previous work explores this in the context of deciding to dissent, we have explored it here in the context of strategic consideration of electoral factors when casting votes in criminal procedure cases. Our results indicate intersectional justices are highly strategic. While our study has limitations, we encourage future scholars to revisit our results with more expansive data. We also encourage future studies to explore the new questions posed by our results here.