A. Cultural Theory: A Brief Primer

Ideology has been defined in a variety of ways (Gerring Reference Gerring1997; Knight Reference Knight2006), but a textbook definition we like is that “ideology is a fairly coherent and comprehensive set of ideas that explains and evaluates social conditions, helps people understand their place in society, and provides a program for social and political action” (Ball, Dagger, and O’Neill Reference Ball2016, 5). This definition shares key features with one adopted by Epstein and Segal (Reference Epstein2006, 84–85 n. 10, quoting Bawn Reference Bawn1999, 305; italics omitted): “Ideology is an enduring system of beliefs, prescribing what action to take in a variety of political circumstances.” Meanwhile, political culture—conventionally defined as the shared values and beliefs of a group or society regarding political relationships and public policy—answers the question of how human beings live together. That is, like ideology, political culture answers the question of who gets to do what with and to whom under what circumstances. Like ideology, political culture answers the question of who decides, who has authority, and who has power in a group, organization, institution, or other social unit in society (Swedlow Reference Swedlow2013, 624).

Values, for their part, have been defined as “the criteria people use to select and justify actions and to evaluate people (including the self) and events” (Schwartz Reference Schwartz1992, 1, building on Rokeach Reference Rokeach1973). “Values define what is ‘good’ and ‘bad’ in the world” (Jacoby Reference Jacoby2006, 706). Values are typically part of ideologies and political cultures alike, and they are thought to provide a bridge between them and more particular attitudes and preferences (Feldman Reference Feldman2003, 479). “Values … differ from attitudes primarily in their generality or abstractness,” which allows them to transcend specific situations, and “in their hierarchical ordering of importance,” meaning that people and organizations regard some values as more important than others (Schwartz Reference Schwartz1992, 4), leading to value rankings and value structures (Jacoby Reference Jacoby2006, 707).

Values are a central motivation for political behavior, prompting significant research programs in public opinion and political psychology on material and postmaterial values (Inglehart Reference Inglehart1990; Inglehart and Abramson Reference Inglehart1994), political culture (Wildavsky Reference Wildavsky1987; Thompson et al. Reference Thompson1990; Elazar Reference Elazar1994), political values (e.g., Feldman Reference Feldman2003; Goren Reference Goren2005; Jacoby Reference Jacoby2006, Reference Jacoby2014; Swedlow Reference Swedlow2009; Feldman and Huddy Reference Feldman2014), universal values (Schwartz Reference Schwartz1992), and moral foundations (Graham, Haidt, and Nosek Reference Graham2009; Graham et al. Reference Graham2011).

To begin to show the value added by a values-based approach to judicial ideology, this article focuses on the conceptualization of political values used by scholars relying on the cultural theory (CT) developed by Douglas, Wildavsky, Thompson, and others (Douglas and Wildavsky Reference Douglas1983; Schwarz and Thompson Reference Schwarz1990; Thompson et al. 1990). We like CT because it provides one of the most theoretically developed accounts of values-based behavior among those on offer. Moreover, CT aspires to be a general theory of social and political behavior and has undergone significant theoretical and conceptual development as well as increasingly rigorous empirical testing in recent years. The most recent bibliography of CT applications across the social sciences and humanities spans 55 pages (see Wildavsky Reference Wildavsky2006), and a symposium in PS: Political Science and Politics (Swedlow Reference Swedlow2011b), a special issue of the Policy Studies Journal (Swedlow Reference Swedlow2014), and a symposium in Public Administration (6 and Swedlow 2016) show how the theory can advance the development of political science and subfields closely related to public law.

Scholars have found CT helpful in explaining why legal institutions and practice in the United States are more adversarial than those in Europe and Japan (Kagan Reference Kagan2001); the relationships among criminal behavior, law, and penal approaches in the United States and Europe (Mars Reference Mars1982, Reference Mars2000; Mars and Weir Reference Mars2001; Sparks Reference Sparks2001; Vaughan Reference Vaughan2002, Reference Vaughan2004; Giles-Sims and Lockhart Reference Giles-Sims2005); perceptions of legally relevant facts in the United States (Kahan Reference Kahan2009; Kahan et al. Reference Kahan2009, Reference Kahan2012); the development of tort law in the United States (Polisar and Wildavsky Reference Polisar1989; Kagan Reference Kagan2006); types of administrative justice (Halliday and Scott Reference Halliday2009); and regulation in the United States, Europe, and elsewhere (Coyle Reference Coyle1994; Hood Reference Hood1995; Hood et al. Reference Hood1999, Reference Hood2004; Verweij Reference Verweij2000a, Reference Verweij2000b; Swedlow et al. Reference Swedlow2009; Lodge, Wegrich, and McElroy Reference Lodge2010; Lodge Reference Lodge2011; Lodge and Wegrich Reference Lodge2011; Heims Reference Heims2016; Nakamura Reference Nakamura2016).

This literature also includes CT applications to studying judicial behavior. Coyle (Reference Coyle1993) identifies signficant cultural differences driving California, New Jersey, and Pennsylvania state court decisions on property rights; Swedlow (Reference Swedlow2009, Reference Swedlow2011a, Reference Swedlow2012, Reference Swedlow2017) finds that CT and “cultural surprises” help explain federal court decisions in the spotted owl cases that dramatically changed federal land and wildlife management in the Pacific Northwest; and Robinson (Reference Robinson2014) uses CT and CCT to explain radical doctrinal change in US Supreme Court gender discrimination cases.Footnote ⁴

CT scholars hypothesize that different patterns of social and political relations are accompanied by specific values and beliefs—including constructs of human and physical nature—that allow people to justify these relationships to each other. In other words, different packages of social and political relations, values, and beliefs are functionally related. These mutually supportive, interdependent relations, values, and beliefs are what CT scholars call “cultures,” or ways of life, while values and beliefs taken together are what they call “cultural biases” (Schwarz and Thompson Reference Schwarz1990; Thompson et al. Reference Thompson1990; Thompson 2008).

CT aims to characterize these cultural packages in a way that is parsimonious but still comprehensive at a useful level of abstraction. CT scholars think that much of the variation in social and political relations is captured by the extent of individual autonomy and collectivism in those relations. By the “extent of collectivization” (“group” in CT terms), we mean the extent to which individuals make “us versus them” distinctions, that is, the extent to which cultural patterns are defined by an external group boundary. By the “extent of individual autonomy” (“grid”), we mean the extent to which individuals are free from regulation and can act as they please. These dimensions and the resulting patterns of social relations and their associated political values are depicted in figure 1.

Figure 1.

Dimensions of relations, resulting relational patterns, and associated political values constituting political cultures.

Hierarchical relations are highly structured, with everyone and everything having his, her, and its place, represented by the organizational pyramid in figure 1. In this cultural environment, individual identities are subsumed in organizational roles, and legitimate decisions are made by persons in the right positions to make particular types of decisions. Here, the decision rule is that the proper authority decides.Footnote ⁵ Individualistic relations, by contrast, are highly fluid and subject to individual choice, represented by a network in figure 1, where I decide. Power differences among people in individualistic relations are common and are represented here by differences in the size of the circles representing individuals. Fatalistic relations, meanwhile, are tenuous and unreliable, driven by the whim and caprice of others; these are represented by atomized individuals in figure 1, where others decide. Finally, people in egalitarian relations retain their autonomy by giving everyone an equal voice in (and thus the power to veto) collective decisions. Here, we decide. The egalitarian desire to have it all is represented here by something that looks like a chocolate chip cookie in figure 1, where the “chips” are individuals with roughly equal power, enacting the egalitarian “one person, one vote” principle.

Each of these four patterns of social and political relations is hypothesized to justify and in turn be justified by specific values and beliefs. Perhaps not surprisingly, cultural theorists hypothesize that people in hierarchical relations value order, people in individualistic relations value freedom, people in fatalistic relations value (good) luck, and people in egalitarian relations value equality (see fig. 1; Coyle Reference Coyle1993; Swedlow Reference Swedlow2008). Valuing order means valuing hierarchical order, often found in traditional institutions but also in bureaucracy, while valuing freedom means valuing the behavioral freedom to act with limited constraints in both economic and social spheres. Valuing equality means valuing not only equality of opportunity but also equality of results (Swedlow Reference Swedlow2008).

CT hypothesizes that values and relations cannot easily be mixed and matched without disrupting this functional relationship. For example, people in hierarchical relations cannot value freedom or equality more than order without undermining their relational pattern. Less obviously, beliefs regarding human nature, the environment, and economics are also associated with (again because they are functional for) each pattern of relations, providing considerable explanatory leverage and predicting much covariation (and inverse variation) among relations, beliefs, and values (Thompson et al. Reference Thompson1990).

As indicated, we use the values of liberty, equality, and order to code cases because CT hypothesizes that these values are functional, respectively, for individualistic, egalitarian, and hierarchical social relations. That is the difference between having a theory to generate values and generating an ad hoc list that seems important, as others do (see, e.g., Segal and Spaeth Reference Segal1993; Bailey and Maltzman Reference Bailey2008). Yet liberty, equality, and order are not the only values that may be expected to be associated with (because they are functional for) particular patterns of social and political relations. These core values entail a variety of related values and beliefs, which we begin to use in coding cases but do not systematically identify here.

For example, people in hierarchical relations value authority, deference to expertise, and clearly defined social roles; people in individualistic relations value limited government, individual rights, and markets; people in fatalistic relations value fortune smiling on them; and people in egalitarian relations value equality and blame unjust systems (i.e., hierarchical and individualistic systems) for social ills. To the extent that individuals are influenced by one more than another of these cultural relations, this will affect what they see as risky; whom they blame for social, economic, or political problems; and how they evaluate political issues. In attempting to explain the roots of poverty, for example, people in egalitarian relations would be attracted to arguments that referenced structural causes and discrimination, while people in individualistic relations would instead be drawn to arguments about poor individual choices or character flaws.

It is important to note here that because CT is an institutional theory of culture (6 and Swedlow 2016), it does not claim that individuals are particular cultural “types,” although it does hypothesize a “strain to consistency” for individuals (Thompson et al. Reference Thompson1990), meaning there will be dominant cultural tendencies in individuals. Empirical studies validate this hypothesis of cultural consistency for some people—particularly among the more politically active (Ellis and Thompson Reference Ellis1997) and knowledgeable (Gastil et al. Reference Gastil2011; Ripberger et al. Reference Ripberger2012), among whom Supreme Court justices should be counted—but not for others (Olli Reference Olli2012).

A related point that we do not develop here is that in CT culture is dynamic rather than static. CT’s relational dimensions of culture generate pressures that create this dynamism (Thompson Reference Thompson2008), as do encounters with situations that do not conform to cultural expectations. These “cultural surprises” are sources of cultural change for both institutions and individuals (Thompson et al. Reference Thompson1990; Swedlow Reference Swedlow2011a, Reference Swedlow2012; Robinson Reference Robinson2014, Reference Robinson2016).

Finally, we should mention that we emphasize CT’s values-based contribution to the study of judicial behavior primarily in order to introduce law and courts scholars to CT. However, CT offers significantly more than this to the study of judicial behavior. As indicated, CT is a theory of how relational patterns, values, and beliefs arise and relate to each other, forming four cultures. Thus, as culture is conceived in CT, it has indicators beyond values. The ideational aspect of CT and its cultural biases (which are most comparable to ideology as conventionally conceived) consist not just of values but of a variety of beliefs. Our CT-based coding of judicial decisions takes advantage of these various markers of culture to code not just values but relational and belief indicators of culture where these are available, as we will discuss.Footnote ⁶

B. Cultural Theory and Judicial Ideology

As a way of conceptualizing judicial ideology, CT addresses our prior critiques better than the AM. First, its dimensions (at least the three that would most often likely apply to political elites such as US Supreme Court justices) are grounded in specific values: order, freedom, and equality. Building on past findings, individuals should have preferences regarding not only which values they support in specific contexts but also how these values should be ranked (Jacoby Reference Jacoby2006, Reference Jacoby2014). Justices supportive of hierarchy should value order over freedom and equality, justices supportive of individualism should value freedom over order and equality, and justices supportive of egalitarianism should value equality over freedom and order (Swedlow Reference Swedlow2008, Reference Swedlow2009). Coding Supreme Court decisions as advancing order, liberty, or equality, as opposed to being liberal or conservative, should make coding more consistent and intelligible, avoiding the sort of ad hoc adjustments to coding criteria that the AM requires.

Similarly, CT’s values should be valid across a longer timespan than the AM’s loose liberal-conservative conception of attitudes. The specific packages of ideas and preferences that constitute legal liberalism and conservatism have continued to change since the 1940s (the starting point for most systematic, quantitative examinations of the Court’s decisions). CT’s values, by contrast, should allow for more consistent comparisons over time, showing which values predominate in particular case areas or for individual justices at particular points in time, rather than presuming consistency in ideological labels.Footnote ⁷ CT can also shed light on lay usages of “liberal” and “conservative” with respect to a specific issue at a specific time, providing systematic evidence, for example, on whether egalitarianism has become more common among legal liberals in regard to disputes over rights or whether legal conservatives have become less hierarchical and more individualistic since the 1970s. Given that liberal and conservative labeling is likely to remain robust in political discourse and political science, such specifications would be helpful.

Finally, given empirical work that suggests the superiority of multidimensional characterization of opinions generally (see the review in Carmines and D’Amico [Reference Carmines2015]) and court decisions specifically (Anderson and Tahk Reference Anderson2007; Peress Reference Peress2009; Fischman Reference Fischman2015), CT benefits from having multiple dimensions. While fatalism is rare among US political elites (Swedlow Reference Swedlow2001), CT’s differentiation of equality, liberty, and order still allows a more valid and nuanced discussion of legal phenomena than the status quo. A liberal-conservative framework struggles, for example, to incorporate the distinctions between social and economic conservatives (Maddox and Lilie Reference Maddox1984; Knight Reference Knight2006; Carmines and D’Amico Reference Carmines2015). CT, by contrast, posits both hierarchical and individualistic values, which may at times both be considered “conservative” but are, in fact, quite different from one another (Swedlow Reference Swedlow2008; Swedlow and Wyckoff Reference Swedlow2009; Jackson Reference Jackson2015).Footnote ⁸ Moreover, while moving beyond a single dimension, CT remains parsimonious and abstract enough to explain phenomena in general terms.

A. A Cultural Theory Coding of First Amendment Votes and Case Outcomes

To establish CT’s plausibility as a theory-based explanation for judicial decision-making, we classify both case outcomes and justice votes in our First Amendment data set along CT’s cultural dimensions.Footnote ¹² Our coding framework assesses the latent CT cultural dimensions advanced by the parties in question, as determined by the nature of the rights claim and the parties involved in the case. Importantly, CT provides not only a theory of ideology in which values figure prominently but also a theory of how relational patterns, values, and beliefs arise and relate to each other, forming four cultures. Given this, our CT-based coding of judicial decisions takes advantage of various markers of these cultures to code value as well as relational and belief indicators of culture where these are available.

As further explained in appendix A, our coding practice is functionally similar to that in the AM. We code a justice vote or case outcome as individualist when it supports the party making a rights claim (in almost all cases, this means supporting the rights-seeking plaintiff), as well as when it supports individualistic values such as choice and competition. We code a vote or outcome as hierarchical if it advances government interests in order, aesthetics, traditional norms, or other state interests, as well as when it supports traditional racial or religious hierarchies. In almost all cases, a vote to reject the civil liberties claim and support the government contains a hierarchical dimension. We code a vote or outcome as egalitarian when it supports the poor or racial, religious, or sexual minorities, as well as specific aims (antiwar speech, pro-environmental speech, regulation of money in politics) that CT posits and public opinion research has shown are valued in egalitarian cultures (e.g., Rothman and Lichter Reference Rothman1987). Justices valuing equality often serve as a swing vote in First Amendment cases, siding with justices valuing liberty to protect the speech rights of civil rights protestors, for example, while siding with justices valuing order to support campaign finance or antidiscrimination regulation. Not all cases involve an egalitarian dimension; that is, in some cases, neither potential outcome advances or threatens equality in any meaningful way.

Both case outcomes and justice votes are first classified using three dummy variables, coded 1 if they support a specific CT culture and 0 if they do not. We use these dummies to create a categorical variable that lists each possible outcome from an array of CT combinations. For example, a decision or vote to strike down government attempts to censor African American civil rights protestors (e.g., Edwards v. South Carolina [1963]) would be coded as 1 for individualist, 1 for egalitarian, and 0 for hierarchy, resulting in a categorical coding of individualist-egalitarian for both the case outcome and the justice votes in the majority coalition. Importantly, because justices often vote without writing separate concurrences or dissents, we lack the information needed to assess whether a justice backs these protestors because he or she supports racial equality, free speech, or both. As such, we assume that because the justices ultimately supported the case outcome, they are comfortable advancing all CT dimensions present in that vote or position, since even if they do not explicitly support a CT political culture, they still prefer it to the alternative. We employ this procedure for all our cases.

Generally, in First Amendment cases, votes to support government regulation of speech or religion are coded as hierarchy since such votes serve hierarchical values like maintaining order. However, if a government victory also supports egalitarian or individualistic values of equality and freedom, then we additionally code the case to reflect that, such as a vote for a campaign finance regime that purports to advance equality while limiting freedom. In the AM scheme, by contrast, government wins are coded as conservative outcomes and government losses are coded as liberal outcomes, with no possibility of additional nuance.

Similarly, unlike the AM, in which a liberal decision by definition means a conservative dissent, our coding allows for the possibility that both sides in a case may support the same culture. Hudgens v. NLRB (1976), for example, pitted the speech and assembly rights of striking workers against the property rights of their employers, leading us to code both sides of the case as supporting individualism. Such cases seek to resolve within-culture disputes. About 97% of our cases, however, fall into one of four alignments. First, a case might lack any substantive egalitarian dimension, pitting the individualistic rights-seeking plaintiff against the hierarchical progovernment position. Boos v. Barry (1988), involving the government’s desire to regulate protests outside foreign embassies to avoid diplomatic embarrassment, is such a case, as generally are cases involving obscenity, loyalty oaths, or libel. Second, our data contained many cases presenting coalitions of individualist and egalitarian interests united against the government’s hierarchical position, as seen in Edwards above. Third, we saw evidence of individualist positions pitted against an egalitarian-hierarchical alliance, as in recent campaign finance or commercial speech cases, such as Sorrell v. IMS Health (2011).Footnote ¹³ Finally, we have cases in which the civil liberties of hierarchical groups—such as the free-exercise claims of the Green family in Hobby Lobby (2014)—come into conflict with egalitarian-hierarchical government rules. Table 1 provides examples of how we applied our CT coding criteria, while appendix B provides a complete list of our coded cases.Footnote ¹⁴

Table 1.

Cultural Theory Political Cultural Coding Criteria: Sample Cases and Coding Rationales

B. Trends in Support for Individualism, Egalitarianism, and Hierarchy in First Amendment Case Outcomes and Judicial Votes

In figure 2, we show decision outcomes for First Amendment cases over time according to our categorical CT variable, which shows how frequently the three CT political cultures are supported by case outcomes. We first see that the mostly hierarchical pattern of outcomes in the early years of the data set gives way to increasingly individualist and individualist-egalitarian decision patterns in the 1960s and 1970s, in line with what one would expect with the doctrinal revolutions of the Warren Court. Beginning in the 1970s, however, there are fewer individualist-egalitarian outcomes, whether because Warren Court liberals are replaced by more conservative justices or because individualist-egalitarian challenges become less frequent or less tenable in light of existing Warren Court gains. Also noteworthy is that after the 1970s, there are a small but stable number of egalitarian-hierarchical decisions, signaling that in areas such as campaign finance or antidiscrimination, egalitarians gained the political power to pass rules that elicited First Amendment challenges.

Figure 2.

Decision outcomes by cultural theory political cultures, First Amendment cases, 1946–2014 terms, plotted over 5-year periods. “Other” refers to either rare alignments of cultural theory political cultures or decision outcomes that were unclassifiable.

For comparative purposes, we also track pairwise correlations between the USSC Database’s decision direction variable—specifically whether a decision is coded as liberal—and each of our three CT cultures (see fig. 3).Footnote ¹⁵ Our results here provide a theoretically grounded and more refined explanation for analyses done by Volokh (Reference Volokh2000), Epstein and Segal (Reference Epstein2006), and others. In the 1960s and 1970s, there is a high degree of correlation between conservatism and hierarchy on the one hand and liberalism, individualism, and egalitarianism on the other. Beginning in the mid-1970s, however, these correlations weaken. While these results are likely driven in part by changes in the Court’s agenda, they also suggest a conflict between a USSC Database coding scheme that defines liberalism as support for free-speech plaintiffs and an increasingly mixed reality in which there are weaker relationships between hierarchy and conservatism on the one hand and individualism and liberalism on the other. Moreover, our analysis finds these patterns despite recent attempts to improve the facial validity of the USSC Database by recoding outcomes on issues such as campaign finance.

Figure 3.

Pairwise correlations of cultural theory political cultures with US Supreme Court Database liberal outcomes, First Amendment cases, 1946–2014 terms, aggregated over 5-year periods. Egalitarian correlations are calculated using only cases in which egalitarianism was advanced by at least one party to the case.

To examine trends in justice votes, we use our three CT dummy variables to conduct an analysis of how often, over his or her tenure, a justice made a vote that aligned with each of the three CT cultures, similar to a “percentage liberal” analysis that one might conduct using the AM and database. Figure 4 provides a scatterplot illustrating to what degree a justice’s votes supported one of our three cultures. To contrast our codings with the AM, we color-code justices by their mean Martin-Quinn scores across all their terms (Martin and Quinn Reference Martin2002); justices with mean scores under −1 (liberals) are labeled light gray, justices with mean scores between −1 and 1 (moderates) are labeled gray, and justices with mean scores over 1 (conservative) are labeled black. To aid in interpretation, we also present a rank ordering of all justices along each of our three dimensions in table 2.Footnote ¹⁶

Figure 4.

Supreme Court justices by the percentage of their votes that support cultural theory political cultures, First Amendment cases, 1946–2014 terms. Justices are color-coded as liberals (light gray), moderates (gray), or conservatives (black), according to whether their average Martin-Quinn score (over all terms) is below −1, between −1 and 1, or above 1. Egalitarian figures are scored as a percentage of cases in which the egalitarian culture was advanced by at least one party to the case.

Table 2.

Justice Votes for Cultural Theory Political Cultures: Percentage of Votes, First Amendment Cases, 1946–2014 Terms

Note.—Supreme Court justices ranked by the percentage of their votes that support cultural theory political cultures. Justices are labeled as liberals (standard text), moderates (italics), or conservatives (boldface), according to whether their average Martin-Quinn score (over all terms) is below −1, between −1 and 1, or above 1. Egalitarian figures are scored as a percentage of cases in which an egalitarian culture was present; egalitarian votes as a percentage of all decisions are listed in parentheses.

Our results again provide a theory-based, further refined explanation for Volokh’s (Reference Volokh2000) and Epstein and Segal’s (Reference Epstein2006) claims that the ideological dynamics of First Amendment decisions have changed in recent decades. Take, for example, our opening discussion of Justices Clark and Kennedy. As seen in table 2, Clark scores low on individualism (38%), in the middle on egalitarianism (43%), and relatively high on hierarchy (63%). By contrast, Kennedy is higher on individualism (57%), lower on egalitarianism (27%), and lower on hierarchy (51%). Contrary to an AM coding scheme that simply compares whether Kennedy or Clark is more liberal, our values-based coding allows us to tease out more nuances, finding differences between the two that fit our preexisting understanding of their values and behavior and explain why it is misleading to think of them as interchangeable ideological moderates.

As with the pairwise correlations in figure 3, we use our justice vote results to make broader inferences about ideological change over time. First, we show that more recently appointed conservative justices are more individualistic than previous conservatives, being comparable in this dimension to sitting liberal justices. Roberts and Alito, for example, take individualist positions in higher percentages of decisions than Sotomayor, Breyer, Kagan, and Ginsburg.Footnote ¹⁷ By contrast, the lowest individualist scores come from earlier conservative and moderate justices, such as Rehnquist, Minton, Reed, and Clark. Second and similarly, support for or opposition to hierarchy—generally viewed in our coding as support for or opposition to the government’s regulation of speech or religious activity—does not break down along liberal-conservative lines as cleanly as it once did. The bottom of the ranks of support for hierarchy is filled by earlier liberal justices such as Douglas, Black, Brennan, and Marshall. The most recent liberal appointees, however—Kagan, Breyer, Sotomayor, and Ginsburg—are far more willing to vote to support government regulation of speech, while Thomas, Alito, and Roberts have slightly lower scores than prior conservatives.

It is the egalitarian analysis, however, that is most illuminating. The most recent liberal justices support egalitarian positions at high rates, while recent conservative appointees rank at the bottom for this dimension. In today’s cases, then, the strongest cleavage on First Amendment issues is not over support for speech rights per se but over support or opposition to egalitarian positions, particularly regulations of speech and religion that ostensibly promote egalitarian ends. Rather than a contest between order and freedom, then, contemporary First Amendment cases increasingly divide along disputes over equality. In these cases, contemporary liberal justices are increasingly likely to vote in support of egalitarian and hierarchical positions; contemporary conservatives are increasingly likely to vote in support of litigants whose actions violate laws promoting egalitarianism. Of course, the extent to which these changes are occurring because of a substantive change in judicial views or because of changes in the Court’s agenda requires further study.

Importantly, these results do not emerge from the AM’s coding scheme. In figure 5, we present the same data used in table 2 and figure 4 but this time using only the USSC Database’s direction-of-decision variable, ordering the justices along the percentage of First Amendment decisions they make in a liberal direction. Although the AM’s coding does show a relationship between overall justice ideology and support for free speech, the AM does little to tell us why. More importantly, the AM has little to say about the declining correlation between how the AM says liberal judges should vote and what liberal judges actually do. At best, we could say that more recently appointed Supreme Court liberals are “less liberal” than those who served in the 1960s—a framing that only makes sense if one views free-speech liberalism as concerned with freedom but not equality. CT, by contrast, makes it clear that contemporary liberal justices tend to support speech that advances egalitarian ends but oppose speech that hinders them, while contemporary conservatives oppose restrictions on speech that advance egalitarian ends. Coding votes and case outcomes using values, then, provides crucial information about judicial behavior that the AM’s coding fails to supply.

Figure 5.

Supreme Court justices by the percentage of their First Amendment votes that the US Supreme Court Database codes as liberal, 1946–2014 terms. Justices are color-coded as liberals (light gray), moderates (gray), or conservatives (black), according to whether their average Martin-Quinn score (over all terms) is below −1, between −1 and 1, or above 1.

C. Trends in Support for Individualism, Egalitarianism, and Hierarchy in First Amendment Cases—Four Issue Areas