Does the Public Care How the Supreme Court Reasons? Empirical Evidence from a National Experiment and Normative Concerns in the Case of Same-Sex Marriage

BY Courtney Megan Cahill & Geoffrey Christopher Rapp

Click here for PDF*

Courtney Megan Cahill** & Geoffrey Christopher Rapp***

Can the Supreme Court influence the public’s reception of decisions vindicating rights in high-salience contexts, like same-sex marriage, by reasoning in one way over another? Will the people’s disagreement with those decisions—and, by extension, societal backlash against them—be dampened if the Court deploys universalizing liberty rationales rather than essentializing equality rationales? Finally, even if Supreme Court reasoning does resonate with the people as a descriptive matter, should the Court minimize anxiety-producing characteristics in decisions vindicating civil rights—such as homosexuality in the marriage-equality context—simply in order to assuage the people?

This Article combines constitutional theory and empirical legal analysis to ask and answer each of these questions. It uses the Supreme Court’s disposition of a marriage-equality issue in United States v. Windsor as an opportunity to test empirically a theoretical claim made most recently by Professor Kenji Yoshino, namely that the Court’s reasoning in high-salience contexts resonates with the people. Yoshino was not the first to argue that Supreme Court reasoning matters to the public or that the Court ought to decide cases in a certain way in order to influence the public’s reception of its decisions. He was, however, the first to argue that the Court ought to lead with liberty (rather than with equality) in a variety of contexts involving group-based civil rights because of liberty’s putative power to satisfy the people.

Skeptical of Yoshino’s positive claim and troubled by his strategic advice to courts, we subjected his theory to empirical analysis by way of a national experiment. We conducted our experiment in May 2013—two months after oral argument in Windsor and Hollingsworth v. Perry, when the issue of same-sex marriage was fresh in the public’s mind—to determine whether the Court’s reasoning in a fictional same-sex marriage case affects the public. Consistent with our intuition, we did not find broad support for the notion that reasoning resonates with the people. Based in part on our results and in part on our normative reservations with Yoshino’s tactical advice, we believe that judges and commentators ought to approach Yoshino’s suggested strategy with restraint.

Introduction.. 103

  1. The Theory: Minimizing Pluralism Anxiety with Liberty.. 113
  2. Does Judicial Reasoning Really Resonate with the People?.. 120
  3. Three Reasons for Skepticism…. 120
  4. Evidence Supporting Our Skepticism…. 123
  5. Empirical Scholarship on the Public’s Reception of Supreme Court Decisions 123
  6. Empirical Evidence on the Relationship Between Court Decisions and Public Support for Marriage Equality.. 127

III.        Methodology.. 129

  1. Results. 136
  2. Initial Results. 136
  3. Extension: Subject Comprehension.. 142
  4. Extension: Limiting Analysis to Those with Strong Positive Reactions to the Court’s Decision 145
  5. Discussion.. 146
  6. Differences Between Liberty and Equality Do Not (Really) Resonate with the People 146
  7. Our Study’s Limitations. 152
  8. Normative Critique.. 154
  9. Liberty and Sexual Minorities’ Peculiar Rhetorical History.. 155
  10. Liberty As Bias Avoidance.. 162

Conclusion.. 166

Appendix I. Survey Vignettes 167

Introduction

On June 26, 2013, the United States Supreme Court struck down section 3 of the Defense of Marriage Act (“DOMA”)[1] and declined to rule on the merits in California’s “Proposition 8” case,[2] leaving in place a district court ruling striking down California’s same-sex marriage prohibition on federal constitutional grounds.[3] These two decisions are considered victories for marriage equality.[4] They also virtually guarantee that in the not-too-distant future the Supreme Court of the United States will be asked to elaborate its view of the constitutionality of state laws restricting marriage to opposite-sex couples.[5]

Let us suppose that the not-too-distant future is today and that the Supreme Court finds that same-sex marriage prohibitions are unconstitutional under the U.S. Constitution.[6] Would the constitutional basis or bases on which such a decision rests matter to most people? Would people care whether the Court rendered a pro-marriage equality decision on equal protection grounds or on fundamental rights grounds? Does how the Court reasons in cases dealing with culturally contested issues like same-sex marriage resonate with the public? Or, will the public either embrace or oppose a Supreme Court decision in such cases regardless of the reasons that the Court offers in support of it?

This paper combines empirical legal analysis and constitutional theory to begin to probe these (and related) questions. It presents an original empirical study on the relationship between judicial reasoning and public reception of a court decision.[7] Its objective is part descriptive and part normative. Descriptively, it argues that judicial reasoning likely does not matter to the people, contrary to the claims of some legal commentators.[8] Normatively, it argues that courts ought not to reason in one way over another in a certain subset of decisions vindicating group-based civil rights simply to alleviate the anxiety those decisions might provoke in the general public, as one prominent constitutional theorist, Kenji Yoshino, has recently suggested.[9]

In the past, scholars of constitutional law have been concerned with what the Supreme Court should have said, but did not, in landmark constitutional decisions like Brown v. Board of Education[10] and Roe v. Wade.[11] Increasingly, scholars of constitutional law are interested in elaborating on how the Supreme Court should reason in potential future landmark cases, including a marriage equality decision.[12] Why is this?

Scholars have offered several reasons for why the Court’s reasoning matters, particularly in cases involving deeply divisive issues like desegregation, abortion, and same-sex marriage. For some, the Court’s reasoning matters because it could impact a decision’s precedential value.[13] For others, the Court’s reasoning matters because it could strengthen the constitutional basis for a particular right.[14] For still others, the Court’s reasoning matters because it could help to contain the backlash that particular decisions produce in the court of public opinion.[15] It is the conjectured relationship between judicial reasoning and public reception of a decision (and backlash against it) that concerns us here. In particular, we are interested in the recent argument advanced by Kenji Yoshino––namely, that courts can minimize the people’s pluralism anxiety, defined as “apprehension of and about . . . demographic diversity,”[16] by vindicating civil rights on liberty rather than on equality grounds.[17] To be sure, Professor Yoshino is not the first to argue that courts can influence the public’s reception of (and backlash against) judicial decisions by reasoning in one way over another in their opinions.[18] Most notably, Justice Ginsburg has argued that “[t]he Roe decision might have been less of a storm center had it . . . homed in more precisely on the women’s equality dimension of the issue.”[19] Under this view, justifying the abortion right on equality rather than on privacy grounds might have muted Roe’s ostensible social and political backlash[20]—at least among academics but possibly among the public as well—because people would have been more willing to accept (or at least tolerate) abortion had they understood the deep relationship between a woman’s reproductive autonomy and her “ability to stand in relation to man, society, and the state as an independent, self-sustaining, equal citizen.”[21]

Yoshino is, however, the first to argue that courts should foreground liberty and minimize equality in a whole swath of cases involving group-based civil rights. In his view, liberty justifications for civil rights—of the sort relied on by the Supreme Court in Lawrence v. Texas[22]—can dampen the angst that equality justifications ostensibly provoke (or at least do nothing to alleviate) in the general public.[23] We describe Yoshino’s theory in greater detail in Part I. Suffice it to say here that Yoshino contends that pluralism anxiety is a good reason why the Court should continue to vindicate civil rights on liberty rather than on equality grounds, as it did in Lawrence.[24] While “liberty claims [might not] quash [pluralism] anxiety altogether,” Yoshino admits, they “do so more than equality claims.”[25] Assuming “courts . . . believe something is at stake in how [civil rights] claims are framed,”[26] he argues, they ought to follow Lawrence’s lead of vindicating rights on the basis of liberty rather than equality.[27] In this sense, Yoshino advances the conversation initiated by other scholars about the Court’s ability to influence the public’s reception of its decisions. Cass Sunstein has argued that the Court can limit public backlash by deciding “minimally.”[28] Other commentators have argued that the Court can control public backlash by not deciding at all.[29] Yoshino appears to suggest that the Court can alleviate the people’s pluralism anxiety—and, by extension, control public backlash[30]—by deciding in a certain way. Indeed, we interpret Yoshino’s pluralism-anxiety theory as a backlash-minimizing device, as pluralism anxiety is a necessary condition for public backlash to occur. We describe the relationship between pluralism anxiety and backlash in greater detail below.[31]

But do the people—the group whose pluralism anxiety Yoshino appears to care about the most—believe that something is at stake in the way civil rights claims are framed? Will Americans be more willing to accept a Supreme Court decision vindicating marriage equality for same-sex couples if that decision is grounded in liberty, “emphasiz[ing] what all Americans . . . have in common,” rather than in equality?[32] Our study empirically tests the claim that how the Court reasons in decisions vindicating civil rights actually matters to the public. We chose to test Yoshino’s particular theory of “pluralism anxiety” because he is the first to argue that the “how” matters in any case involving group-based civil rights—as opposed to, say, Justice Ginsburg, who has more modestly suggested that the “how” might have mattered in the abortion context specifically.[33] Moreover, we thought it was important to subject Yoshino’s theory to empirical study because Yoshino provides no evidence to substantiate his claim that the Supreme Court’s reasoning actually resonates with the people. This oversight is curious, especially given that reasoning’s ostensible relevance constitutes the very basis for Yoshino’s principal normative argument (the one with which we have strong normative reservations), namely that courts ought to foreground liberty in decisions vindicating group-based civil rights.[34]

For reasons explained below, our initial intuition was that the Court’s reasoning in constitutional cases involving divisive issues like same-sex marriage and abortion would not quiet any pluralism anxiety or contain any backlash[35] that those cases produce. If the Supreme Court as an institution does not cause a distinctive backlash in the public—as scholars David Fontana and Donald Braman have recently found[36]—then our intuition was that the Supreme Court could not contain backlash through the reasoning that it employs.

In 2012, Professors Fontana and Braman published the findings of an empirical study on institutional backlash.[37] Their findings arguably cast doubt on Yoshino’s thesis, but they do not actually test it.[38] In brief, Fontana and Braman examined whether the public reacted more strongly when the Court decided an ideologically divisive issue like same-sex marriage than when Congress decided that issue.[39] Conventional wisdom on backlash posits that the Court, because of its intrinsic institutional features and the so-called “counter-majoritarian difficulty,”[40] causes a distinctive backlash in the court of public opinion.[41] Fontana and Braman found that this was not the case, and that people’s “cultural priors,” rather than any pre-existing institutional preference, best predicted how they would react to the resolution of a high-salience issue by either the Court or Congress.[42]

Our study, like Fontana and Braman’s, experimentally tests a common constitutional law claim, albeit on a more granular level. Whereas Fontana and Braman focus on whether the Supreme Court causes a distinctive public backlash,[43] we focus here on whether the Supreme Court’s reasoning in cases dealing with divisive issues like same-sex marriage contains public backlash. Whereas Fontana and Braman focus on whether the public cares if the Supreme Court decides important constitutional law issues,[44] we focus here on whether the public cares if the Supreme Court employs particular justifications over others when deciding those issues—and, more narrowly, when deciding those issues in a way that ultimately vindicates a constitutional right.

We chose to focus on the issue of same-sex marriage for three reasons. First, Yoshino contends that leading with liberty in a marriage case would be more appealing to the public.[45] Second, the issue of same-sex marriage, considered in June 2013 by the Supreme Court in two high-profile cases and currently under review in a number of state and federal courts,[46] is very much in the public eye. Moreover, Michael Klarman recently argued that a Supreme Court decision establishing a national marriage equality precedent—the so-called “ ‘fifty state’ solution”—is well within the realm of possibility for the near future.[47] Thus, we believe that the time is ripe to test the public’s response to such a precedent. Third, our normative reservations with Yoshino’s theory center on that theory’s application to the class of people most directly affected by a same-sex marriage decision: sexual minorities.

Our results, explained at greater length in the Parts that follow, did not overcome our intuition that the precise reasoning employed by the Court (when vindicating a civil right like same-sex marriage) has little impact on the public. While we found some evidence that the public is less likely to accept a same-sex marriage decision justified on equality grounds than one justified on liberty grounds or which contains no justification at all, we did not find robust evidence in support of that proposition—certainly not robust enough to overcome our initial skepticism of Yoshino’s theory. Importantly, even if we had found a robust relationship between equality-based reasoning and decreased support for a Supreme Court decision vindicating marriage for same-sex couples, we would still reject Yoshino’s tactical advice to courts for normative reasons.

More specifically, we believe that Yoshino’s tactical advice constitutes a form of judicial covering, defined as “ton[ing] down a disfavored identity to fit into the mainstream.”[48] We find it curious that Yoshino, who has written eloquently on the subject of covering and who has criticized the covering demands that the law places on sexual minorities in particular,[49] would advocate what we construe to be a form of judicial covering. The Supreme Court’s most recent gay-rights decision, United States v. Windsor, engages in a similar sort of judicial covering that we find objectionable.[50] We elaborate on this normative critique and on Windsor in Part VI.

It is our position that courts ought to reject Yoshino’s theory on normative grounds alone. That said, we are aware that courts likely care about how the public receives their decisions[51] and might therefore be tempted by Yoshino’s suggestion that liberty alleviates the anxiety that equality provokes. With that possibility in mind, it is our hope that the results of our study at the very least prompt courts to approach Yoshino’s tactical advice with restraint. Our study found that there is probably little to lose as a positive matter should courts foreground equality over liberty in a same-sex marriage decision. It is our contention, however, that there is quite a bit to lose as a normative matter should courts go that route.

Part I sets forth in greater detail Yoshino’s pluralism-anxiety theory, and Part II discusses our positive reservations with it. Part III describes our methodology, Part IV reports our results, and Part V offers discussion. Part VI closes with the normative reservations that prompted us to undertake this study in the first place.

  1. The Theory: Minimizing Pluralism Anxiety with Liberty

Kenji Yoshino is not the first person to suggest that Supreme Court reasoning in any particular decision can affect the way in which that decision is received by various constituencies outside the Court.[52] He is, however, the first person to argue that the Court ought to foreground liberty rationales in any decision vindicating civil rights on the theory that liberty is more likely than equality to quell the “pluralism anxiety” that any such decision will provoke in the “people.”[53] Our study is limited to testing an important—and until now untested—aspect of this more specific claim. Before turning to it, a more complete description of Yoshino’s theory and of our positive reservations with it is necessary.

Yoshino’s theory has both a descriptive and a normative dimension, both of which center on what he calls “pluralism anxiety.”[54] Defined as the country’s “apprehension of and about its demographic diversity,”[55] pluralism anxiety is weighing all of us down. Citing the work of political scientist and Bowling Alone author, Robert Putnam, Yoshino argues that increased racial, ethnic, and sexual diversity in the United States has led to decreased “social solidarity and . . . social capital.”[56] While we might “celebrate diversity” in theory,[57] we are exhausted by it in fact—so much so that we are experiencing “less happiness and lower perceived quality of life”[58] and desiring “a recommitment to the ideals of assimilation and integration.”[59]

No less than it is affecting “we, the people,” pluralism anxiety is affecting the Supreme Court. “Just as the War on Terror has transformed our separation of powers jurisprudence and the internet has transformed our First Amendment obscenity jurisprudence,” Yoshino argues, “pluralism anxiety has pressed the Court away from traditional group-based identity politics in its equal protection and free exercise jurisprudence.”[60] Pluralism anxiety explains why the Court has placed limitations on heightened scrutiny classifications,[61] why the Court has foreclosed pure disparate impact claims under the Constitution’s equality and free exercise guarantees,[62] and why the Court has reined in Congress’s exercise of its enforcement power under section 5 of the Fourteenth Amendment.[63] In “each line of jurisprudence,” Yoshino observes, the Court “has alluded to pluralism anxiety,” thereby suggesting that its exhaustion with identity politics is driving it to close “three separate doors through which it had permitted the advancement of group-based civil rights under the Equal Protection Clause.”[64] As a result, and at first blush, “[t]he future of constitutional civil rights . . . seems grim.”[65]

Not so grim, however, as to suggest that “we are witnessing the end of constitutional civil rights in this country.”[66] Yoshino argues that to compensate for the closure of various equality doors, the Court has turned to liberty—or, more precisely, to liberty-based dignity claims—when vindicating civil rights. While the Court “seems to understand pluralism as a challenge to a progressive agenda,” Yoshino remarks, “it has seen that challenge as one that can be overcome by using liberty analysis, which draws on a broader, more inclusive form of ‘we.’ ”[67] Unlike essentializing equality claims, liberty claims can bridge the differences between us, sounding as they do in a “universal register.”[68] He says:

The new equal protection paradigm stresses the interests we have in common as human beings rather than the demographic differences that drive us apart. In this sense, the shift from the “old” to the “new” equal protection could be seen as a movement from group-based civil rights to universal human rights.[69]

To exemplify this shift from the “old” to the “new” equal protection, Yoshino offers Lawrence v. Texas,[70] in which the Supreme Court struck down Texas’s sodomy statute on liberty rather than on equality grounds.[71] There, Yoshino reminds us, the Court reasoned that “the statute violated the fundamental right of all persons—straight, gay, or otherwise—to control their intimate sexual relations.”[72] In Yoshino’s view, the Court decided Lawrence on a liberty theory because of its exhaustion with the “old” equality “paradigm.”[73] He claims that “by deciding Lawrence on liberty grounds, the Court quieted pluralism anxiety.”[74]

When Yoshino says that the Lawrence Court “quieted pluralism anxiety,” presumably he means that it quieted the public’s anxiety,[75] given that he immediately follows that assertion by saying: “The Court evaded the charge that it was picking and choosing among groups by highlighting that the right in question belonged to all persons within the United States.”[76] In other words, Yoshino suggests that liberty not only minimizes the Court’s pluralism anxiety, but also inoculates the Court from the “charge” that it is playing favorites among identity groups.[77]

Even more, Yoshino suggests that liberty is the judicial antidote to the people’s pluralism angst, and for that reason might help to contain any backlash that Court decisions vindicating the rights of minority groups are thought to precipitate.[78] Although Yoshino does not explicitly say so, his theory suggests that courts might play a role in containing the public’s backlash in culturally contested cases—like same-sex marriage, the focus of our study—by reasoning one way over another. To be sure, pluralism anxiety and backlash are distinct concepts; the former denotes an uncomfortable feeling, whereas the latter denotes a strong reaction against something (such as a court decision or a court itself) that can have negative consequences.[79] Yoshino’s pluralism anxiety is a feeling provoked by the increasing diversity in society, whereas backlash is an antagonistic reaction provoked by judicial decisions.[80] At least on its face, then, Yoshino’s pluralism-anxiety theory does not directly tell us anything about backlash or about the Court’s ability to contain backlash through its reasoning.

On close inspection, however, Yoshino’s theory indirectly posits a connection among pluralism anxiety, backlash, and judicial reasoning. An uncomfortable feeling, such as one emanating from pluralism anxiety, might be thought of as a necessary, but perhaps not sufficient, condition for backlash. The public, feeling uncomfortable, can be targeted for mobilization by elites, and that mobilization, by focusing reaction on a Court decision, ripens into backlash. The more the public is persuaded by an opinion, however, the less anxious it feels about the opinion—and the less likely it is to backlash against it.

Liberty, in Yoshino’s view, performs that persuasive function. He says: “The universality of [liberty] claims will make them more persuasive [than equality claims] to many.”[81] If backlash can occur when people are not persuaded by judicial decisions, then liberty presumably minimizes backlash (at least vis-à-vis equality) because it makes decisions “more persuasive” to the polity than equality can.[82] Indeed, for Yoshino, there is something about liberty that makes people more likely to agree with a judicial decision or to accept it as law. A decision justified using liberty (rather than equality) is less likely to trigger backlash, either against the decision or the Court that rendered it.

It is for this reason that we interpret Yoshino’s pluralism-anxiety theory as a backlash-limiting mechanism. Where other constitutional theorists have argued that the Court can limit backlash either by deciding “minimally”[83] or by not deciding at all (at least sometimes),[84] Yoshino appears to be suggesting that the Court can contain backlash by deciding in a certain way.

Importantly, we do not read Yoshino as offering a complete explanation for the complicated process associated with backlash against judicial opinions.[85] Instead, the “anxiety” he describes represents a form of immediate or “gut” response to a Court decision.[86] Members of the public may not fully understand a Court decision yet react negatively to it based on the inclusion of particular words or concepts in the Court’s explanation for its holding.[87] In this sense, the reaction that Yoshino anticipates to an equality-based decision resembles the kind of “implicit bias” that has attracted the serious attention of behavioral psychologists and legal scholars.[88] The reaction Yoshino describes may be an intuitive, yet powerful, one, and we posit that a fully developed backlash is unlikely to occur where the public has a positive intuitive response but may occur when the public’s reaction is intuitively negative.[89] We revisit the issue of implicit bias and its relationship to Yoshino’s pluralism-anxiety theory in Parts V and VI.

Moreover, it is important to keep in mind that Yoshino does not consider liberty and equality in a vacuum; rather, his claim is that the liberty rationale suppresses pluralism anxiety (and possibly backlash) more than equality. In other words, his theory poses the comparative question of which mode of reasoning courts ought to choose––liberty or equality. Under this account, equality does not necessarily provoke pluralism anxiety (and therefore backlash). Rather, it could simply fail to alleviate it.[90] Nor does liberty completely eliminate pluralism anxiety, as Yoshino admits.[91] As explained in greater detail below, our results provide limited support for the proposition that equality is less effective than liberty in garnering support for Supreme Court decisions vindicating civil rights—and therefore, by implication, less effective than liberty in curtailing pluralism anxiety.[92]

If that understanding of Yoshino’s pluralism-anxiety theory is correct, then his theory is much more than just an explanation for why the Court has abandoned equality as a basis for vindicating civil rights. It is also a tactical recommendation for how the Court ought to decide group-based civil rights cases in the future. Yoshino frames the article where he presents his theory of pluralism anxiety, The New Equal Protection, in a way that throws this normative objective into relief. At the end of Part I of that article, where Yoshino describes pluralism anxiety, he states: “This Article considers how the United States Supreme Court might help create [a] ‘new, broader sense of “we” ’ through its constitutional jurisprudence.”[93] In the final substantive part of his piece, Yoshino gives courts (or really the Supreme Court) the following advice: “[W]here a claim can be validly characterized as either a liberty-based or an equality-based dignity claim, it should be characterized as the former.”[94] In other writings articulating his theory, Yoshino has been clear that his intention is to convince courts—rather than legislatures or litigators—to lead with the liberty argument when vindicating civil rights because of its anxiety-minimizing potential.[95] His theory is exclusively juriscentric, and deliberately so.

By his own admission, Yoshino is much more “confident” and less “tentative” about the descriptive dimension of his piece than he is about his strategic or normative recommendations.[96] We share his restrained optimism about the normative “desirability of this shift” from equality to liberty,[97] and articulate our normative reservations at greater length in Part VI. In fact, it was our normative reservations with Yoshino’s theory that prompted us to undertake our empirical inquiry—an inquiry whose results convince us that Yoshino’s lack of confidence “about the inevitability or desirability of [the] shift” he advocates is warranted.[98]

  1. Does Judicial Reasoning Really Resonate with the People?

In addition to our significant normative reservations with Yoshino’s theory, discussed below, we have a straightforward positive reservation: We doubt that Supreme Court reasoning actually resonates with the people enough for it to curtail the people’s pluralism anxiety. Subpart A sets forth the three reasons why we are skeptical of Yoshino’s normative claim. Subpart B supports our skepticism with empirical evidence on the relationship between judicial decisions and the public’s acceptance of them, both in general and in the particular context of marriage equality.

  1. Three Reasons for Skepticism

To recall, Yoshino argues that “the United States Supreme Court might help create “ ‘[a] new, broader sense of “we” ’ ”[99] by leading with liberty over equality in decisions vindicating civil rights. We have doubts about this claim for three interrelated reasons.

First, we are skeptical that the public is sufficiently attuned to the reasoning of any single Supreme Court opinion to be persuaded by that reasoning. In other words, Yoshino’s claim conflicts with our intuitive belief that the general public is unfamiliar with the discrete aspects of a Supreme Court opinion, including whether or not the Court vindicates a right on the basis of liberty or of equality.[100] While Yoshino might be right that some Americans respect and have confidence in the Supreme Court—thirty six percent according to a July 2010 Gallup poll[101]—it is unlikely that Americans would have more respect for the Court’s decisions in ideologically salient contexts like abortion and same-sex marriage because it reasons in one way over another. We suspect that how the Court reasons is too granular a concern for most people, even people who are interested in what the Supreme Court has to say.[102] Moreover, even if some attributes of a Supreme Court decision have been experimentally found to influence the public’s acceptance of that decision, those attributes—whether the decision came from a divided Court and whether the decision follows or overrules precedent[103]—are highly visible to the public.[104] Whether the Court vindicates the abortion or same-sex marriage right on the basis of privacy or of equality is not.

Second, even if members of the public were sufficiently attuned to the reasoning of a Supreme Court opinion, we are skeptical that they could tell the difference between liberty-based reasoning and equality-based reasoning, given the close interrelationship between liberty and equality. Laurence Tribe has described liberty and equality as two strands in a single “double helix,” a hybrid right which he conceptualizes as “dignity.”[105] Several Supreme Court opinions—including, notably, Lawrence v. Texas—recognize the deep interconnectedness between constitutional equality and liberty guarantees.[106] Even Yoshino admits that these two guarantees are deeply interwoven.[107] We suspect that these guarantees are so closely connected that it is difficult to determine when the Court is “doing” civil rights primarily on a liberty basis or primarily on an equality basis. For reasons explained in greater detail below, our study does not rule out the possibility that equality and liberty have become so intertwined that it is unlikely that one will ultimately have more purchase than the other in the court of public opinion. Because we tested the public’s receptivity to liberty reasoning and to equality reasoning separately, it remains uncertain how the public would receive a decision that incorporates both rationales nearly simultaneously.[108]

Third and most relevant to our study, even if the public were attuned to Supreme Court reasoning and could differentiate between liberty and equality rationales in a single decision, we are skeptical that judicial reasoning would ultimately sway or persuade the people in the way that Yoshino envisions. Other than briefly citing two scholars who have suggested that casting group-based civil rights in more universal (and less identity-based) ways might yield positive results for those groups,[109] Yoshino provides no empirical support for the proposition that Supreme Court reasoning can influence the public’s general response to Court decisions in high-salience contexts. This is not surprising, given that there is a dearth of research on the link between reasoning and public reactions, something our study attempts to address.[110]

  1. Evidence Supporting Our Skepticism
  2. Empirical Scholarship on the Public’s Reception of Supreme Court Decisions

Recent empirical evidence on the relationship between Supreme Court decisions and the public’s reception of those decisions supports our skepticism of Yoshino’s claim that the Supreme Court can quell the public’s pluralism anxiety through the reasoning that it uses in high-salience contexts like same-sex marriage. For instance, empirical evidence on the link between decision attributes and public opinion[111] suggests that individuals’ ideological predispositions with respect to issues that the Court decides are the most significant predictor of how those individuals react to a Court decision.[112] This is particularly the case for high-salience issues, like abortion, where the prior views of citizens will likely dominate the nuance of a Court’s decision in shaping public response.[113]

Scholars empirically testing the relationship between decision attributes and public opinion have found an inverse relationship between visible decisional attributes, like majority size, and the ideological salience of the issue considered in a decision.[114] As the authors of a recent study observe, “[T]he effects of [certain] decision attributes grow stronger as the ideological salience of the legal issue declined, so that we see the least consistent effects for the high-salience issue of abortion and the most consistent effects for the low-salience issue of bankruptcy.”[115]

Others have similarly found that “it is only in ‘cases with low salience’ that large majorities have an effect on public attitudes.”[116] In an unpublished article cited this August by The New York Times,[117] Cass Sunstein summarizes recent empirical evidence suggesting that unanimous decisions in culturally contested settings like same-sex marriage do not guarantee increased public acceptance of them,[118] despite some justices’ professed belief that judicial consensus promotes stability and enhances the Court’s legitimacy in the court of public opinion.[119] Sunstein discusses the empirical work of political scientist Michael Salamone,[120] who has found that “reaction to judicial consensus is dependent on the ideological salience of the issue involved”[121] and that “the public is unmoved by the majority size in highly salient decisions.”[122]

If visible decisional attributes (e.g., majority size, adherence to precedent) matter less for decisions dealing with high-salience issues like abortion, then we suspect that less prominent decisional attributes like judicial reasoning will matter significantly less for decisions dealing with those issues. The decisional attributes tested in the literature are not just more visible than judicial reasoning, they are also more comprehensible. An opinion is likely to be described in the media as “unanimous” or “divided,” or as “overruling” a well-known prior case or leaving that case the settled law; the public also likely has a firm understanding of what those descriptions mean. If highly visible and easily understood characteristics do not affect public reaction in high-salience cases, it is hard to imagine how more submerged and complex characteristics could.

Other recent empirical scholarship supports our skepticism that the Supreme Court can contain backlash through the reasoning that it employs.[123] As mentioned earlier,[124] David Fontana and Donald Braman have subjected the judicial backlash theory—the claim that the Supreme Court causes a distinctive backlash in the court of public opinion[125]—to empirical analysis, and with interesting results.[126] Fontana and Braman experimentally examined the following question: “Does it matter to members of the public whether it is the Supreme Court or Congress deciding important constitutional issues of the day?”[127] Conventional wisdom suggests that the public reacts more strongly when the Court decides certain issues than when Congress decides those issues—that is, that judicial resolution of an issue (and, in particular, of an ideologically divisive issue) causes a distinctive public backlash that legislative resolution of that issue ostensibly avoids.

Contrary to that wisdom, Fontana and Braman found that people generally do not care whether Congress or the Court decides important constitutional law issues.[128] If people disagree with same-sex marriage, then they will react quite negatively to either the Court or Congress as an institution were either to extend marriage to same-sex couples. If people agree with same-sex marriage, then they will react quite positively to either the Court or Congress as an institution were either to extend marriage to same-sex couples.[129] In other words, Fontana and Braman found that there is no distinctive public reaction when the Supreme Court decides constitutional issues, leading them to theorize that backlash against Congress can be just as strong as backlash against the Court.[130] People’s institutional preferences do not exist in a vacuum, Fontana and Braman found.[131] Rather, those preferences are determined by whether those institutions resolve constitutional issues in a way that vindicates people’s cultural worldviews or “cultural priors.”[132] This central holding from Fontana and Braman’s study has been reinforced by other empirical research.[133]

We hypothesize that if the Supreme Court does not cause a distinctive backlash in the court of public opinion because of institutional features intrinsic to the Court, then it is highly unlikely that the Supreme Court’s reasoning in ideologically charged cases will contain public backlash. Fontana and Braman’s study suggests that cultural worldviews with respect to the issues that the Court decides drive people’s views of the Court more than anything else. If that is right, then it is unlikely that the Court can do much to change those worldviews through its reasoning.[134] As Fontana and Braman themselves assert, “[I]n high-salience cases there is little chance the Court can convince opponents.”[135] If people disagree with same-sex marriage, then it is unlikely that they will be more supportive if the Court rests a marriage-equality decision on liberty grounds rather than on equality grounds. In short, if the people do not care about who decides a constitutional issue, then why would they care about how a constitutional issue is decided?

An unpublished paper by political scientists Patrick Egan and Jack Citrin provides support for our hypothesis.[136] Egan and Citrin conducted a two-wave experiment to evaluate whether the reasoning used by the Supreme Court affects the public’s attitude towards the decisions themselves and the public’s belief in the legitimacy of the Court.[137] Egan and Citrin began by assessing respondent attitudes toward abortion, flag burning, the decriminalization of same-sex relations, limits on campaign finance, assignment of public school students by race, and handgun ownership bans.[138] One half of respondents were provided with a one-sentence description of the reasoning used in a case and the other half were provided with no such description.[139]

In five of the six issues studied, Egan and Citrin found that learning of the Court’s reasoning had no effect on the public’s attitude toward the decision.[140] Learning the Court’s reasoning had “relatively paltry effects,”[141] they write. The authors conclude that the Court’s “power to shift mass opinions towards accepting its rulings was essentially nil.”[142] The sole exception was Lawrence v. Texas, where the Court decriminalized same-sex sex.[143] Learning that the decision was based on the right to privacy moved opinion towards the Court’s decision.[144] The authors dismiss that result, however, as inconsistent with the actual reaction observed in the public; in the aftermath of Lawrence, they note, support for criminalization of same-sex relations actually increased (albeit somewhat temporarily).[145] Egan and Citrin’s finding that the Court’s reasoning has no effect on public opinion appears to undercut Yoshino’s argument that the mode of reasoning employed by the Court can constrain backlash.

  1. Empirical Evidence on the Relationship Between Court Decisions and Public Support for Marriage Equality

If Yoshino were correct in his prediction that equality incites (or fails to contain) pluralism anxiety, then what would we make of the fact that we have witnessed increased public support for gay rights over the past four years despite a consistent trend of courts deciding gay rights cases not just on an equality basis but also on the assumption that gays deserve special judicial scrutiny as a class under equality provisions in state constitutions? Arguably, this trend started with a few state marriage equality cases where state supreme courts—Connecticut, Iowa—decided a same-sex marriage case on equality grounds, finding that gays and lesbians constituted a quasi-suspect (rather than a non-suspect) class for the purpose of state constitutional equality guarantees.[146]

Several federal district and appeals courts have since followed their lead, deciding to apply not rational basis review to laws that discriminate against gays and lesbians but rather heightened judicial scrutiny—also on the basis that gays as a class deserve that level of review.[147] Importantly, since this trend started in 2009, public support for gay rights legislation in myriad domains—employment, marriage, the military—has only increased, and in some cases dramatically so.[148] This is all just to say that if Yoshino’s prediction were correct, then we would at least expect to see decreased support for gay rights following decisions grounded in equality. If anything, the data suggest just the opposite.

Maybe the answer is that the public’s growing support for gay rights simply has nothing to do with judicial decisions; rather, that support results from changing demographics, greater exposure to gays and lesbians in popular culture, or advocacy efforts. But if that is the case, and the judicial decisions embracing gay rights on equality grounds have been simply irrelevant, that would seem to undermine Yoshino’s worry that decisions expanding rights on the basis of equality would reduce the public’s receptiveness to them.

Given what we believe is a strong normative downside to Yoshino’s tactical advice—one considered at length in Part VI—we wanted to empirically test the positive foundation of his theory. Evidence of a robust link between Supreme Court reasoning and public opinion might outweigh the normative reservation we identify (for some of Yoshino’s audience, though not for us). However, evidence of a weak connection between the two variables that Yoshino assumes to be linked—Supreme Court reasoning and public opinion—would be even greater reason to approach his tactical advice with restraint. It is to that empirical inquiry that we now turn.

III. Methodology

To test the relationship between the expressed rationale for rights-expanding Supreme Court decisions and the public’s reaction to those decisions, we employed a methodology adapted from several previous studies. We used the experimental vignette approach of Professor Zink and his colleagues.[149] In addition, we used the survey sampling approach of Fontana and Braman, attempting to “replicate actual and potential real-world constitutional conflicts.”[150] The experimental nature of our study is designed to “isolate the impact of the Court relative to other possible sources of influence” by minimizing the “noise of the outside world.”[151]

We retained a private survey firm, Survey Sampling International, Inc. (“SSI”), to conduct the study.[152] SSI maintains a standing panel reportedly numbering in the millions from which survey respondents can be drawn.[153] The panel for our study was based on a Census-representative random selection of SSI panelists. That is, SSI attempted to match our sample’s demographic characteristics to those of the broader population. We requested a sample consisting of 1500 subjects. We divided this pool into three groups, each of which was exposed to a vignette describing a Supreme Court decision.

 

Table 1: Sample Demographics

 

  Treatment 1 – Equality Vignette Treatment 2 – Liberty Vignette Treatment 3 – Neutral Vignette Overall
Number 500 500 502 1502
Gender (percentage female) 48.20% 48.80% 51.00% 49.33%
Age (mean years) 42.35 41.57 42.19 42.03
Education (percent having completed college) 40.22% 48.21% 42.33% 43.59%

 

Table 1 displays selected demographic data for our overall sample and each of the three treatment groups. Our panel had a median level of education of “completed some college” but not “college degree.”[154] Forty-nine percent of our panel was female.[155] The mean age for our sample was forty-two.[156] The median annual household income was between $30,000 and $39,000.[157]

Our panel had a high level of support for same-sex marriage. Fifty-eight percent of our panel either “agreed” or “strongly agreed” that same-sex couples should be permitted to marry in the same manner as opposite-sex couples. This is slightly higher than the fifty-one to fifty-five percent of Americans found to support the legalization of same-sex marriage in various June 2013 public opinion polls.[158] This difference likely reflects the fact that in order to participate in the study, our panelists, at a minimum, must have had regular access to a computer and the Internet. Younger Americans have a much higher level of support for same-sex marriage than older Americans[159] and are also presumably more likely to have access to a computer and the Internet,[160] and to consider enrolling as members of SSI’s standing panel.

Respondents were surveyed using SSI’s Dynamic platform.[161] By participating in the survey, they received points which could (when accumulated with points earned for participation in other surveys) lead to the reward of a gift card. No other reward was provided to panelists for participation. Panelists were asked to verify their consent to the study and were told that they could exit the survey at any time if they were uncomfortable continuing.[162]

We limited the focus of our study to subjects’ reactions to vignettes relating to same-sex marriage. The survey was administered in the second week of May 2013. This was approximately two months after the Supreme Court heard oral arguments in the DOMA[163] and “Proposition 8” cases[164] and was timed to precede the publication of the Court’s decisions in those cases. Our goal was to administer the survey at a point in which it might be plausible to a member of the public that the Court had reached (and was prepared to release the results of) its decision but before any media coverage concerning an actual, real-world release of the Court’s opinions.

We exposed panelists to one of three vignettes (dividing our total survey sample into three separate groups). Each vignette was written in the style of a newspaper story reporting on a Supreme Court decision legalizing same-sex marriage. The vignettes were presented in a font resembling newspaper formatting.[165]

The first vignette provided information suggesting that the Court’s decision had been based upon equality concerns. The second provided information suggesting that the Court’s decision had been based upon liberty concerns. The third vignette was meant to provide a “neutral” stimulus—reporting a same-sex marriage endorsement from the Supreme Court without providing a description of the reasoning behind the Court’s decision.

The headlines of the “Liberty” and “Equality” vignettes explicitly referenced the rationale for the Supreme Court’s reported decision. In addition, all three vignettes included a block or pull quote, located in a separate box, which reported to excerpt the Supreme Court’s decision.

Prior to reading the vignette, we asked respondents to answer three sets of questions. The first set of questions, adapted and updated from the survey used by Fontana and Braman, involved knowledge of current political events and individuals. The second set of questions asked respondents about their views on same-sex marriage.[166] The third set of questions asked respondents how important the issue of same-sex marriage was to them personally and how closely they followed the issue. After completing these preliminary questions, respondents were presented with one of the three vignettes. The survey vignettes appear in Appendix One.

After reading the vignettes, we asked panelists to answer two sets of questions. The first set of post-stimulus questions asked them to provide information about the story that they had read, and the second set of post-stimulus questions asked them to provide their reaction to the stories. We included the first set of questions in an attempt to assess panelist comprehension of the stimulus they had been provided. One possible concern in survey research is that any statistical results obtained might reflect panelists being confused or not understanding what they had just read. By asking them to answer questions about the vignette, we aimed to be able to analyze whether the effects of each stimulus differed depending on reader comprehension.

In this comprehension module, we asked five True/False questions. The first two questions dealt with the Court’s ruling, and the second two dealt with its reasoning. In the first two questions, we asked if the Court’s decision requires governments to allow same-sex marriage or leaves the question to the states. In the third and fourth questions, we asked if the reason for the Court’s decision was equality-based or liberty-based. In the fifth question, we asked whether the Court had provided any reason for its decision.[167]

The second set of post-stimulus questions represents our primary dependent variables. We asked panelists, following Zink and his colleagues, to provide us with an opinion, first, on whether they agree with the Court’s decision and, second, on whether they accept the Court’s decision.[168] We included a third question on the panelists’ support for an amendment to the U.S. Constitution which would prohibit same-sex marriage.

In both our prestimulus questions on panelist attitudes towards same-sex marriage and our post-stimulus questions seeking reaction to the reported Court decision, we use a four-point Likert response scale: strongly disagree, disagree, agree, and strongly agree.[169] While Zink and his colleagues use a five-point response scale for their “agreement” dependent variables and a four-point response scale for “acceptance,”[170] we utilize the four-point scale throughout. We declined to use the neutral response for two reasons: (1) The neutral response, if coded as “3” on a five-point scale, has the “effect of pulling the item means to the middle of the scale” and “causes items on a survey to look more similar;”[171] and (2) not including the neutral response forces respondents to make a choice, even if it involves expressing a weak opinion.

Other questions involved true/false responses, or, in the case of the political awareness module, choices reflecting the specific items. We included a free-form space for panelists to explain why they reacted to the Court decision in the way they did to facilitate subsequent qualitative analysis. After providing their free-form response, panelists were informed that the vignettes they had just read were fictional in nature (though drawn from issues currently being litigated in the courts). Finally, we included in matrix form ten other questions on various issues.

We subjected the results of our study to an ordered logit regression, utilizing the various reaction questions as dependent variables and utilizing the prestimulus attitude questions and the selected stimulus as our primary independent variables. The ordered logit model is appropriate where ordinal responses are analyzed, but there is no reason to suppose that the distance between two responses remains constant along a scale.[172] In other words, if one cannot say for certain that the difference in preference between “strongly disagree” and “disagree” is the equivalent of the difference between “strongly agree” and “agree,” then an ordered logit approach is proper.

We created dummy variables for each stimulus, coded as “1” if the person was presented with a particular stimulus (LIBERTY, EQUALITY, or the NEUTRAL option). Since the independent prestimulus attitude variables were also ordinal in nature, we followed the accepted practice of creating dummy variables for each response option, coded as “1” if the respondent selected that option and 0 otherwise. Our regression models omit the “disagree” dummy variable, which serves as a reference for interpretation of the “strongly disagree,” “agree,” and “strongly agree” coefficients.

The regression model for the first cut, where the dependent variable, Y, would reflect agreement with the Court’s decision, willingness to accept the Court’s decision, or support for a constitutional amendment to prohibit same-sex marriage, would be:

 

Y                                 =               α + β1EQUALITY STIMULUS

+ β2LIBERTY STIMULUS

+ [βmSSM POLICY PREFERENCE DUMMY VARIABLES]

+ [βnSSM SALIENCE DUMMY VARIABLES]

+ AGE

 

In this model, by omitting the dummy variable for the NEUTRAL stimulus, the coefficients on the LIBERTY and EQUALITY stimulus dummies become measures of the impact of reading each of those stories, with the NEUTRAL stimulus serving as referent. That is, the coefficient we obtain on the LIBERTY STIMULUS dummy would tell us, by comparison to those receiving the NEUTRAL STIMULUS, what impact the LIBERTY STIMULUS had on agreement, acceptance, and support for a constitutional amendment to undo the supposed Supreme Court decision. By including both the LIBERTY STIMULUS and EQUALITY STIMULUS dummy variables in our regression specification we measure the effect of each stimulus, in comparison to the other, and by reference to the NEUTRAL STIMULUS.

The null hypothesis, H0, would be

β1 = β2 = 0.

That is, the null hypothesis would be that it should not matter in determining reactions which stimulus was presented to a subject; instead, a person’s reaction should be entirely determined by her prior views about the issue.

  1. Results
  2. Initial Results

The first model specification uses the question of agreement as the dependent variable:

 

Overall, I agree with the Supreme Court’s decision in this case.[173]

 

Subjects responded along a four-point Likert scale (strongly disagree, disagree, agree, strongly agree). Numerically, these were coded with the numbers 1–4, so a higher number would reflect a more positive level of “agreement.”

Table 2 reveals that there was no statistically significant difference among the three subject groups in regard to agreement with the Court’s decision based on which article they read. The coefficients on both the LIBERTY and the EQUALITY variables did not differ from zero at the .05-level of significance. Instead, the respondents’ level of agreement with the Court’s decision was largely shaped by their prior views. With a pseudo-R-squared value of .27, a fairly high level of the variation in respondents’ answers to the “agreement” question was explained by their prior views and age.[174]

We obtain statistically significant and predictable results in regard to the three “prior view” questions included in this regression.[175] The questions asking respondents to indicate how closely they follow the issue of same-sex marriage and how important it is to them personally did not yield statistically significant variables.[176]

 

Table 2: Agreement with Court’s Decision

 

VARIABLE                            OLOGIT

 

Coefficient   P-value

Stimulus

Equality Dummy               -.02          .86

Liberty Dummy                       -.12          .35

 

SSM Prior View

SSM=loving and committed

Strongly Disagree Dummy      -.64*         .00

Agree Dummy                .51*         .00

Strongly Agree Dummy  1.1*         .00

SSM=undermine traditional

Strongly Disagree Dummy      .13          .48

Agree Dummy                  -.49*         .00

Strongly Agree Dummy  -.92*         .00

SSM should be permitted

Strongly Disagree Dummy      -1.2*         .00

Agree Dummy                .97*         .00

Strongly Agree Dummy  1.9*         .00

 

SSM Salience

Important to me personally

Strongly Disagree Dummy      -.02          .89

Agree Dummy                .05          .71

Strongly Agree Dummy  .13          .53

Closely follow SSM issue

Strongly Disagree Dummy      .04          .82

Agree Dummy                -.01          .93

Strongly Agree Dummy  .14          .50

 

Age

Age                                 -.00          .15

 

Number of Observations              1502

Log Likelihood               -1414

Pseudo R-squared                    .27

LR chi2                             1038

 

* p < .05; two-tailed test

 

Our second model specification uses the question of acceptance as the dependent variable:

 

I accept the Supreme Court’s decision. That is, I think that the decision ought to be accepted and considered to be the final word on the matter. I do not think that there ought to be an effort to challenge the decision and get it changed.[177]

 

Here, as indicated in Table 3, we obtain what at first blush appear to be the most striking results associated with our study. We find a statistically significant negative coefficient on the dummy variable for the EQUALITY STIMULUS. That is, reading the equality stimulus (unlike the liberty stimulus) with the “neutral” stimulus as a reference point makes respondents less likely to indicate acceptance of the Court’s decision.[178]

Although the coefficient on this variable is statistically significant, its magnitude arguably is not.[179] When we calculate the “Odds Ratio” for this variable, we get a ratio of .74, suggesting that a change in whether or not a party read the EQUALITY STIMULUS has only approximately a one in four chance of changing the reader’s response in terms of acceptance.[180] Nevertheless, the results on acceptance, while perhaps anomalous, merit further discussion, in which we engage below.

The coefficient on the LIBERTY STIMULUS, by comparison, is not statistically significant at the .05 level.[181]

As reflected in Table 3, we continue to observe statistically significant coefficients (with predictable signs) on the “prior views” questions concerning whether to permit same-sex marriage and whether permitting same-sex marriage would expand the number of persons in loving and committed relationships.[182]

 

Table 3: Accept Court’s Decision

 

VARIABLE                            OLOGIT

 

Coefficient   P-value

Stimulus

Equality Dummy               -.29*         .01

Liberty Dummy                       -.19          .12

 

SSM Prior View

SSM=loving and committed

Strongly Disagree Dummy      -.44*         .02

Agree Dummy                .39*         .02

Strongly Agree Dummy  .87*         .00

SSM=undermine traditional

Strongly Disagree Dummy      .28          .12

Agree Dummy                  -.12          .44

Strongly Agree Dummy  -.65*         .00

SSM should be permitted

Strongly Disagree Dummy      -1.1*         .00

Agree Dummy                .83*         .00

Strongly Agree Dummy  1.4*         .00

 

SSM Salience

Important to me personally

Strongly Disagree Dummy      .21          .16

Agree Dummy                -.02         .89

Strongly Agree Dummy  .32          .11

Closely follow SSM issue

Strongly Disagree Dummy      -.05          .76

Agree Dummy                -.08          .55

Strongly Agree Dummy  .48*         .02

 

Age

Age                                 -.00          .22

 

Number of Observations              1502

Log Likelihood               -1575

Pseudo R-squared                    .19

LR chi2                             742

 

* p < .05; two-tailed test

 

Our third model specification uses the question of support for a constitutional amendment as the dependent variable. Respondents were asked for their level of agreement with the following statement:

 

I support an amendment to the United States Constitution that would prohibit same-sex couples from marrying.

 

The results of the regressions for amendment support are reported in Table 4. Neither the EQUALITY nor the LIBERTY treatments had a statistically significant effect on respondents’ support for a constitutional amendment.[183]

 

 

Table 4: Support for Constitutional Amendment Banning SSM

 

VARIABLE                            OLOGIT

 

Coefficient   P-value

Stimulus

Equality Dummy               -.00          .96

Liberty Dummy                       -.11          .39

 

SSM Prior View

SSM=loving and committed

Strongly Disagree Dummy      .48*         .02

Agree Dummy                -.18         .29

Strongly Agree Dummy  -.43         .07

SSM=undermine traditional

Strongly Disagree Dummy      -1.6*         .00

Agree Dummy                  .85*         .00

Strongly Agree Dummy  1.44*         .00

SSM should be permitted

Strongly Disagree Dummy      1.2*         .00

Agree Dummy                -.49*        .00

Strongly Agree Dummy  -1.71.9*     .00

 

SSM Salience

Important to me personally

Strongly Disagree Dummy      -.02          .87

Agree Dummy                .95          .52

Strongly Agree Dummy  .54*         .01

Closely follow SSM issue

Strongly Disagree Dummy      .23          .18

Agree Dummy                .15          .26

Strongly Agree Dummy  .71*         .00

 

Age

Age                                 -.01*         .00

 

Number of Observations              1502

Log Likelihood               -1443

Pseudo R-squared                    .29

LR chi2                             1208

 

* p < .05; two-tailed test

 

  1. Extension: Subject Comprehension

After obtaining the preliminary results discussed above, we next attempted to determine whether panelists’ comprehension of the vignettes they encountered affected their reactions to those decisions. Unfortunately, we found that our respondents, at a very high rate, answered “True” both when asked if the Court used a liberty rationale and when asked if the Court used an equality rationale.[184] For instance, of our 500 subjects reading the EQUALITY STIMULUS, 378 answered “True” both when asked whether the reason for the Court’s decision was equality-based and liberty-based. For the 500 subjects reading the LIBERTY STIMULUS, 409 answered “True” both when asked if the reason for the Court’s decision was equality-based and liberty-based. Among respondents exposed to the NEUTRAL STIMULUS (with no mention of the Court’s reasoning), 371 answered “True” when asked whether the Court had used both equality-based and liberty-based rationales.

Given these results, we constructed a new set of variables to reflect panelist comprehension without requiring an “entirely correct” response.[185] This allows the reapplication of the regression models to a smaller sample with only those answering “correctly” being included. With N=322 for that population, we end up with coefficients on the EQUALITY and LIBERTY STIMULUS dummy variables in an ordered logit regression that are not statistically significant.[186]

Instead, then, we constructed a new regression model in which we exploited the information generated regarding whether panelists answered correctly but did not excise 75% of our study subjects.[187] This model attempts to take into account whether subjects demonstrated a relatively higher level of understanding and comprehension of the vignettes with which they were presented. We created a new dummy variable coded as “1” for a “correct” response (as described above) and zero otherwise. We then added that dummy variable to ordered logit regressions. Here, for the specification using the most interesting of our dependent variables, “acceptance,” we found that having a “correct” understanding of the vignettes presented was associated with a more positive level of “acceptance.” The coefficient on the new “correct” variable is .28, with a p-value of .022.[188]

On the other hand, we found a statistically significant negative coefficient on the EQUALITY STIMULUS, and a negative and nearly significant coefficient on the LIBERTY STIMULUS dummy variable.[189] In other words, we can say that, by controlling for subject understanding of the vignettes, both the EQUALITY STIMULUS and the LIBERTY STIMULUS have negative effects on panelist acceptance when compared to the NEUTRAL STIMULUS. To be clear, what these results do not suggest is that those who best understood our instrument had more negative responses when seeing one of the two stimuli that contained an explanation of the Court’s reasoning. This specification tests the effect of the LIBERTY and EQUALITY STIMULI taking into account the level of panelist comprehension. In fact, the positive and statistically significant coefficient on the “correctness” measure, read in light of the negative coefficients on both reasoning stimulus variables, suggests the following: (1) understanding was associated with a higher level of acceptance; and (2) the negative effects of being presented with “reasoning” explanations may be concentrated among those who demonstrated a relatively low level of understanding of the reasoning in the Court’s supposed decision.

To extend this inquiry, we developed an additional regression model that utilized interaction terms connecting the “correctness” and “stimulus” variables. The coefficient on the term interacting the LIBERTY or EQUALITY STIMULUS dummy with the comprehension dummy would isolate the effect of reading either stimulus (as opposed to the NEUTRAL STIMULUS) for those with a high level of comprehension from the effects of reading each stimulus generally. For the coefficients on both interaction terms, we get positive signs, just outside of the range of statistical significance at the .05-level.[190] The coefficients on the EQUALITY and LIBERTY dummies remain negative and statistically significant. This suggests that for those who understood the vignettes, explanations of either liberty or equality grounds enhanced acceptance when compared to the no-explanation rationale. On the other hand, for those with a lower level of understanding of the vignettes, either rationale continued to exert a negative effect on acceptance.

  1. Extension: Limiting Analysis to Those with Strong Positive Reactions to the Court’s Decision

Our study aimed to explore backlash, but our survey only seeks to obtain information about subjects’ immediate reactions and responses to the vignettes to which they were exposed. In the real world, backlash is a more complicated phenomenon. For backlash to a Supreme Court decision to arise, there would likely need to be some mobilization of popular objections by elites and other policy entrepreneurs.[191] Our study cannot answer the question of when a Court decision is likely to trigger backlash in that it does not include any other mediating influences besides a simulated newspaper report discussing the Court’s decision.

That said, it is logical to posit that elites can only mobilize a backlash against a decision where the public does not strongly support or accept that decision. Therefore, we construct an additional regression model in which we limit our analysis to those respondents who “strongly agreed” or “strongly accepted” the Court’s decision. Here, we posit that such individuals could not be mobilized by elites to form part of a backlash. To the extent that the Court’s reasoning affects the proportion of respondents “strongly agreeing” with the decision or who “strongly agree” that they accept the decision, it would be far more difficult for elites to mobilize a backlash. Arguably, even respondents who “agreed” with a decision could, under the right circumstances, be convinced by elites that the decision was objectionable and could thus become part of a backlash.

To construct this next set of regressions, we create dummy variables coded as “1” if a respondent strongly agrees with the Court’s decision or strongly agrees with the statement that she accepts the Court’s decision. Since this limited dependent variable is not an ordinal one, we use a logit rather than ordered logit approach.

The results confirmed the earlier regression model.[192] There is no statistically significant effect in terms of the level of strong agreement with the Court’s decision based on which stimulus respondents viewed, controlling for age, prior views on the issue and salience.[193] However, there is a statistically significant negative effect on acceptance based on viewing the equality-based stimulus. This could be read to suggest that a Court decision using an equality rationale will produce a smaller share of the public that strongly accepts the decision and which would be predisposed in a manner inconsistent with mobilization for backlash. But with little effect on strong agreement with the decision, it is hard to conclude that overall an equality rationale does not do as good a job as a liberty rationale or no rationale at all in containing backlash by minimizing the pool of people receptive to mobilization by elites against the Court’s decision.

  1. Discussion

Our study tested Yoshino’s theory that Supreme Court reasoning in decisions vindicating civil rights in high-salience contexts—like same-sex marriage—can affect the public’s reception of those decisions. For reasons set forth more fully below, we remain unconvinced that the Court’s reasoning matters to the people. While our results do intriguingly suggest that any reasoning might negatively affect the public’s reception of certain decisions, they do not overwhelmingly support the proposition that a particular kind of reasoning will dampen any pluralism anxiety that the public is likely to exhibit in response to those decisions.

Before our study, our intuition was that reasoning was not as significant a determinant of public reception of certain decisions as were other factors, including cultural priors with respect to high-salience issues like same-sex marriage. After our study—with some important limitations which are discussed here—our intuition remains the same.

  1. Differences Between Liberty and Equality Do Not (Really) Resonate with the People

Our hypothesis was that judicial reasoning is simply too granular and nuanced to affect public reaction to judicial decisions in culturally contested cases such as same-sex marriage. The results of our study overwhelmingly confirm our preliminary intuition. Whether subjects agree or disagree with a Court opinion, and whether they support a constitutional amendment to prohibit same-sex marriage, appear to be determined not by the reasoning employed by the Court but instead by the subjects’ prior views of the issue.

To be sure, we did find limited evidence that where the Supreme Court employs equality-based reasoning in legalizing same-sex marriage, the public’s “acceptance” of the Court’s decision is reduced.[194] These results conflict with those of past experiments connecting public acceptance and Court decisions. In those experiments, researchers found little relationship between the basis for a Court decision and public acquiescence.[195] That said, the magnitude of the effect we observe is relatively small, and thus, even though we are somewhat intrigued by the results of our study, we are unable to conclude that, broadly speaking, an equality-based rationale produces a different response in terms of public support for an opinion.

Figures 1, 2, and 3 display our results visually, by focusing on respondents’ prestimulus response to the question of whether gay and lesbian couples should be permitted to marry in the same fashion as heterosexual couples (the most robustly predictive of our “prior views” dependent variables). Dividing survey respondents according to their level of agreement with that statement (on a four-point scale) and the stimulus to which they were exposed produces a 4 X 3 between-subjects design. Figure 2 may overstate the effect of reasoning on acceptance. The effects we observed in the regressions may have been driven by shifts only at the margins—between “acceptance” and “strong acceptance,” for instance.

 


Figure 1: Agreement with Court’s Decision Based on Prior Views
and Stimulus Presented

 
Figure 2: Acceptance of Court’s Decision Based on Prior Views and Stimulus Presented


Figure 3: Support for Constitutional Amendment Banning Same-Sex
Marriage Based on Prior Views and Stimulus Presented

 

 

Consistent with our belief that Court decisions may be too granular for the public to understand, relatively few of our subjects demonstrated a full understanding of the vignettes presented. The subjects, generally, were unable to respond correctly to questions, which asked them about the reasoning behind a Court decision, even though they were given explanations for the Court’s decision in our two “treatment” vignettes.[196]

For two of our three dependent variables—agreement with the Court’s decision and support for a constitutional amendment banning same-sex marriage (and thus overturning the Court’s decision)—we observed no statistically significant differences among respondents based on the vignette to which they were exposed.

In spite of this, one reading of our data would be that something appears to be going on for the subjects exposed to the equality rationale stimulus. Subjects exposed to equality were less likely to accept a Supreme Court decision vindicating marriage equality for same-sex couples.[197] Still, the magnitude of the effect appears to be small, and weighed alongside the lack of any observable difference in “agreement” with a decision, or support for an amendment to overturn a decision, we are unwilling to conclude at this point that our study suggests a clear influence of judicial reasoning on public reception. Given that reading the equality stimulus had only around a one in four chance of changing a respondent’s answer on “acceptance” and no effect on a respondent’s “agreement” or support for a constitutional amendment overturning the Court’s decision, we are unable to conclude that judicial reasoning, as a positive matter, is decisive in shaping public reactions.

Interestingly, while our results do not provide strong support for the proposition that equality is that much worse than liberty in terms of losing public acceptance of a Supreme Court opinion vindicating marriage for same-sex couples, they do provide support for the proposition that any reasoning tends to be worse than no reasoning at all. That is, when controlling for subjects’ comprehension, we see a negative effect of both the liberty rationale and the equality rationale when compared to the neutral rationale.[198] This is intriguing. It could be that the public does not understand that justifying decisions is commonplace—and required, of course, by principles of sound jurisprudence.[199] Perhaps the public presumes that a decision that does not need to be justified is most in line with governing law and authority. This may be especially pronounced for those subjects that are confused about the questions/stimuli. When low-comprehension subjects—the vast majority of subjects in our study—encounter any rationale, they may be less likely to accept the decision.[200]

Alternatively, it could be that the negative effects of equality and liberty vis-à-vis no rationale in our study was the result of implicit bias against same-sex marriage, sexual minorities, or both. That is, it could be that subjects were less likely to accept a Supreme Court decision that vindicated marriage equality on the basis of a rationale rather than no rationale because the vignettes presenting a rationale (whether liberty or equality) contained more reasoning than the neutral stimulus. More reasoning, in turn, involves more language about things like “same-sex marriage” and “gays/lesbians”—language that could conceivably trigger subjects’ implicit bias against same-sex marriage and/or sexual minorities.

More specifically, hundreds of validated studies have shown that implicit bias against gay people not only exists, but also influences biased individuals’ treatment of them.[201] Even people who explicitly support gay rights exhibit implicit bias against gay people.[202] Moreover, one of the tests used by social scientists to measure implicit bias against sexual minorities, the Implicit Association Test, uses words like “gay” and “lesbian” to trigger unconscious bias.[203]

It could be, then, that the more the Court in our vignettes reasoned about same-sex marriage—and, in particular, about the rights of “gays” and “lesbians” to marriage—the more our subjects’ implicit biases were triggered. The negative effects of both liberty and equality relative to no rationale—as well as the slightly more negative effect of equality, whose vignette drew attention to “gays” and “lesbians”—might therefore be capturing that implicit bias.[204] While subjects may not have understood what they were reading, they clearly reacted to it; their reactions, perhaps unconscious, were pronounced.[205]

In sum, our primary positive reservation with Yoshino’s theory was not overcome by our results. We remain unconvinced that it matters to the people (in terms of alleviating or curtailing their pluralism anxiety) whether the Court rests a marriage equality decision on liberty or on equality. Our findings with respect to the equality rationale were simply not strong enough to suggest to us that liberty does that much better a job at containing pluralism anxiety—and perhaps, by extension, backlash—than does equality. Importantly, we did find that reasoning generally matters in terms of public reception of a hypothetical marriage equality decision, as our subjects exhibited a more negative reaction to vignettes that contained some reasoning as opposed to no reasoning at all. An explanation for this finding is hard to pinpoint, although one possibility is that more reasoning contains more language and that there is something about language—and particularly the language used in our vignettes—that triggers an intuitive response against same-sex marriage generally, and sexual minorities specifically.

  1. Our Study’s Limitations

Our study, like most empirical and experimental work,[206] has some limitations. The first and most obvious limitation of our study is one that is also reflected in prior work such as that of Professors Zink, Braman, and Fontana, namely that backlash involves a complex process that cannot be captured through the administration of a single survey instrument.[207] We accept that our method does not capture the nuance that shapes, over a period of months or years, how a Court decision may be received by the public.[208] Perhaps it is more appropriate to say that what we are really capturing—as suggested in the previous subpart[209]—is the implicit or unconscious bias that the public has against particular terms or concepts that appear in descriptions of a Court decision.

Second, our model presumes that the content of Court decisions reaches the members of the public, which may not always happen in the real world. Our research question was whether the reasoning articulated by the Supreme Court could serve to alleviate the people’s pluralism anxiety and, by extension, constrain public backlash. For the Court’s reasoning to matter in the real world, the following must occur: (1) The Court articulates reasons; (2) the public learns of the Court’s articulation; and (3) the public understands the Court’s articulation. Our survey skips step two in this process, or, to be precise, assumes that step two occurs. Our findings cannot measure whether the public would be likely to encounter Court reasoning if the public does not seek it out; instead, all we can measure is how the public responds when it is presented with judicial reasoning.

Third, our study isolates the Supreme Court’s role from the role of other cultural actors that will affect public disposition in the aftermath of Court decisions in controversial cases. Whatever the Supreme Court might say could, in the real world, be “drowned out by the polarizing re-biasing being performed by interested parties in these cases.”[210] Even the best efforts by the Court to constrain backlash through messaging can thus be undermined.[211] Our study, of course, subjects panelists only to news about a decision and does not expose them to the full panoply of voices they would likely encounter in the real world.

Fourth and related, our study is limited in that all subjects received word of the Court’s decision in a uniform fashion. In the real world, “citizens learn about different Court decisions based on information available and salient to them,” which means it may be “misguided” to assume “uniform national effects.”[212]

Fifth, we tested equality and liberty rationales in isolation, whereas Yoshino’s precise argument is not that courts should ignore equality in favor of liberty, but instead that courts ought to lead with liberty.[213] Further work could present vignettes embracing both arguments but alter the order of presentation to see if the ordering of arguments leads to different reactions. Given low levels of subject comprehension in our study, we are skeptical that such an approach would produce observable differences. Subjects in our study attributed rationales to the Court not included in the vignettes they read. There is no reason to believe that members of the public would be discerning enough to grasp differences in the order in which arguments were presented.

Sixth and last, our study was limited to the issue of same-sex marriage. Yoshino’s theory is not so limited; instead, he argues that the “leading with liberty” tactic should be applied in a larger subset of cases dealing with the vindication of “civil rights.”[214] Future work could adopt our methodology to test the impact of judicial reasoning on public reactions in a wider variety of scenarios.

  1. Normative Critique

As noted in this Article’s introduction, our primary objection to Yoshino’s theory is a normative one, and it is to this objection that we now turn. Even if Yoshino is descriptively correct that pluralism anxiety is alleviated better by liberty than by equality—which we seriously doubt for all of the reasons set forth above—we would still reject his suggestion that courts downplay the equality demands of identity groups in decisions vindicating their civil rights. The results of our study, which found scant support for the proposition that equality is significantly worse than liberty in terms of cultivating public acceptance, simply make a stronger case for why rejecting Yoshino’s theory makes sense not just from a normative standpoint (as we argue) but from a positive standpoint as well.

More specifically, we argue that Yoshino’s suggested strategy of “leading with liberty” in gay rights cases amounts to judicial “covering” of what is really “at stake” in them.[215] Liberty, under this view, is less antidote and more anodyne. Our critique is limited to an examination of the “leading with liberty” idea in the context of gay rights specifically—a context that seems especially important to Yoshino, given his reliance on gay rights and on key gay rights cases to articulate his theory.[216]

Should the Supreme Court establish a national marriage equality precedent at some point in the near future—as some scholars have predicted[217]—we believe that it ought to resist Yoshino’s suggested strategy for two reasons relating to the “covering” idea. The first reason, discussed in Subpart A, concerns the relationship between liberty rationales and the peculiar rhetorical history that has attended the law’s treatment of sexual minorities for decades. The second reason, addressed in Subpart B, concerns the relationship between liberty rationales and implicit bias against gay people.

  1. Liberty and Sexual Minorities’ Peculiar Rhetorical History

To borrow an image with which Yoshino is well familiar, we believe that leading with liberty in gay rights cases constitutes undesirable judicial “covering”[218] of what those cases are really about: discrimination against gays and lesbians. Consider, for instance, Yoshino’s recommendation that courts frame the question of gay marriage in liberty terms rather than in equality terms in order to make gay marriage more palatable to the general public.[219] He asks readers to consider how different the following two claims sound “to the American ear:

(1) ‘Gays should have the right to marry because straights have the right to marry and gays are equal to straights;’ or

(2) ‘All adults should have the right to marry the person they love.’ ”[220]

Echoing a similar argument made by Martha Nussbaum,[221] Yoshino argues that proposition two is more “persuasive” than proposition one because

it performs the empathy it seeks. It frames the right at a high enough level of generality that opposite-sex couples are urged to imagine a world in which they were denied the right. In contrast, equal protection claims tend to stress distinctions among us, even as they ask us to overcome those distinctions.[222]

Under this view, proposition two is more persuasive than proposition one because it downplays an essentializing equality claim in favor of a more universalizing liberty claim.

But proposition two downplays something else that might make it more appealing to the public than proposition one: gay and lesbian identity. In this particular example, it is not just equality that is elided from the liberty statement, it is equality for gays and lesbians specifically that is absent. The “gays” of proposition one are collapsed into the “all adults” of proposition two. The gay discrimination objected to in proposition one becomes the people-who-want-to-marry discrimination objected to in proposition two. If “covering,” as Yoshino elsewhere defines it, means “to tone down a disfavored identity to fit into the main stream,”[223] then this particular liberty frame exemplifies the covering idea.

Admittedly, there is evidence to support the notion that covering or toning down the “gay” aspect of the marriage question might make gay marriage an easier sell to the public. At least one study has shown that the very words “gay” and “lesbian” can bring implicit bias—the sort of bias that people are either unwilling or unable to acknowledge, or both—against sexual minorities to the surface.[224] If that is right, then framing gay marriage in a way that omits the “gay” might make people less anxious and more supportive of it. As mentioned earlier, our study provides some support for this proposition, as we found that any reasoning—and, to some extent, equality-based reasoning in particular—makes people less receptive to Supreme Court decisions vindicating same-sex marriage. Thus understood, framing marriage for “gays” as marriage for “all adults” might serve the strategic ends of winning marriage for gays and alleviating pluralism anxiety for all of us.

However, framing same-sex marriage a certain way simply in order to appease the public is no less problematic than framing sexual orientation discrimination as sex discrimination or as conceptualizing marriage discrimination against gays and lesbians as the contemporary incarnation of Jim Crow.[225] Each of these arguments substitutes one kind of discrimination (liberty, sex, race) for another (gay/lesbian). Each has a “transvestic quality”[226] that camouflages gay and lesbian discrimination in the “garb” of something else. To quote Edward Stein, each “closets, rather than confronts, homophobia.”[227]

Casting exclusionary marriage laws as a liberty violation rather than an equality violation of gays and lesbians also “closets rather than confronts, homophobia”[228]—as well as its targets. In so doing, the liberty argument reflects a centuries-old legal tradition of not “naming” gays and lesbians[229]—a legal tradition embraced in Bowers v. Hardwick[230] and only moderately rejected in Lawrence v. Texas,[231] one of the Court’s most pro-gay-rights decisions to date. To be sure, suppressing the equality demands of all identity groups, not just those of sexual minorities, has the effect of covering those groups. However, suppressing the equality demands of gays and lesbians in particular—and thereby covering their identity—is arguably worse given gays’ unique history of covering (and of being covered).

A reading of the Supreme Court’s most recent gay rights case, United States v. Windsor,[232] suggests that the Court has continued this long tradition of covering gay identity. It is worth briefly looking at Windsor because it at once departs from Yoshino’s theory and reinforces it. Windsor strikes down section 3 of DOMA primarily on equality, not liberty, grounds—thus suggesting that Yoshino erred in his prediction that the Supreme Court would continue to lead with liberty in future civil rights cases, as it did in Lawrence.[233] By mentioning the word “equality” and its variants more than a dozen times throughout its opinion, the majority strongly suggests that it was deciding Windsor on the basis of equality, or of what it calls “equal dignity.”[234]

At the same time, though, Windsor could also be read as the Court’s attempt to quell pluralism anxiety by covering gay identity—in the same way that universalizing liberty arguments, which omit any reference to “gays” and “lesbians,” do. Aside from three moments where the majority quotes from DOMA’s legislative record,[235] it nowhere mentions the words “gay, “lesbian,” “homosexuality,” or their variants; as Noa Ben-Asher has recently observed, Windsor is “striking for . . . the conspicuous absence of the words ‘homosexual,’ ‘lesbian,’ or ‘bisexual,’ ”[236] particularly striking in light of the fact that its named plaintiff, Edith Windsor, argued that DOMA discriminated against her on the basis of sexual orientation.[237] The majority opinion does not argue, as the Second Circuit did,[238] that section 3 harms gays and lesbians, notwithstanding the fact that “moral disapproval of homosexuality” was very much on Congress’s mind when it passed section 3—language from the legislative record that not only came up during Windsor’s oral argument but also was cited directly by the Windsor majority.[239] Nor does it recognize that marriage prohibitions like section 3 warrant heightened scrutiny because they harm a class which deserves that level of review—a finding made by several state and federal courts around the country in myriad gay rights contexts.[240]

Rather, the majority argues that section 3 harms married same-sex couples and interferes with the “equal dignity of same-sex marriages.” Gays and lesbians are alluded to only obliquely in dyadic terms as composing the “couple, whose moral and sexual choices the Constitution protects, see Lawrence.”[241] The targeted class, in other words, is not gays and lesbians, but rather the relationships into which they presumptively enter.[242] Nowhere does the Court acknowledge that marriage discrimination of this kind is relational discrimination, and that relational discrimination of this kind is gay and lesbian discrimination.[243] This omission is curious in light of the fact that both the Supreme Court and lower federal courts have found that discrimination against same-sex relationships is synonymous with discrimination against gays and lesbians.[244] In so doing, the Court (and lower courts) has suggested that relational discrimination and status discrimination is a distinction without a difference.

Windsor’s peculiar reasoning “covers” sexual orientation identity no less than universalizing liberty arguments do and therefore illustrates what we believe is problematic about judicial arguments that effectively closet the very class that seeks constitutional redress. The only time the Court comes close to recognizing that DOMA harms gay and lesbian persons rather than just “same-sex marriages” is when it says that “the principal purpose and necessary effect of this law [DOMA] are to demean those persons who are in a lawful same-sex marriage.”[245] But even here one wonders why Justice Kennedy did not just say who, precisely, “those persons”[246] are. Moreover, it seems rather uncontroversial to say that another “necessary effect” of DOMA is to “demean” even “those persons” who are not in a “lawful same-sex marriage”[247] but whom DOMA nevertheless symbolically touches: all gays and lesbians, an entire class of persons rendered presumptively unfit for marriage under DOMA.

Covering over sexual orientation identity in this way—and in the way that Yoshino’s theory appears to require—elides the history of discrimination that underwrites and motivates both DOMA and other laws that discriminate against gays and lesbians. We are wary of judicial covering of this sort, failing as it does “to make significant moral claims”[248] about the status of gays and lesbians in American society as well as about why laws that exclude them from various spheres of American public life warrant condemnation. We agree with Yoshino that the language in which courts vindicate the rights of the subjects before them matters,[249] but we disagree with him as to why that is so. In his view, language ought to serve the goal of gaining a right (say, marriage) in a way that is unobjectionable to the people—in a way that assuages, rather than incites, pluralism anxiety (and possibly backlash as well). In our view, language ought to serve the goal of gaining a right (say, marriage) in a way that vindicates the dignity of those individuals who seek it. Indeed, “even if advocates may feel compelled to offer every argument that might persuade a court, judges, in selecting among those arguments, have a choice about the language and the light in which they cast the subjects who appear before them.”[250] We strongly reject a choice that hides subjects behind the universalizing rhetoric of liberty.

Thus, even if we did find strong support for the idea that the rhetoric of liberty has a purchase that the rhetoric of equality lacks, we believe that there is still good reason to reject Yoshino’s tactical advice.[251] Rather than submerge equality in favor of liberty, we favor the doctrinal strategy relied on by the Executive Branch—which repudiated DOMA for reasons relating to equality, not liberty, in its 2011 announcement[252]—as well as by most federal courts that have struck down exclusionary marriage laws, including DOMA.[253]

  1. Liberty As Bias Avoidance

Our second normative objection concerns the issue of implicit bias and how best to manage it. As already mentioned, implicit bias might explain why we found a negative effect of both the liberty rationale and the equality rationale when compared to the neutral rationale. In addition, implicit bias might explain why equality-based rationales had more negative effects than did liberty-based rationales (albeit only at the margins). If people do react negatively to equality-based reasoning because such reasoning triggers unconscious bias against same-sex marriage, sexual minorities, or both, then Yoshino’s suggestion that courts downplay the equality claims of discrete identity groups when vindicating those groups’ civil rights[254] might be read as a strategy of bias-avoidance or bias-containment. Thus understood, Yoshino’s suggested strategy is unappealing because it urges courts to do the very thing—avoid confrontation with anxiety-provoking identity categories—that implicit bias scholarship cautions against.

If first generation implicit bias scholarship aimed to show that implicit bias not only exists but also influences behavior, then second generation implicit bias scholarship—where we are today—aims to devise the best way to manage it.[255] Scholars from a number of fields have recently argued that the optimal way to deal with implicit bias is to confront, rather than evade, the identity categories that precipitate it.[256] For instance, in their work on race and juries, Samuel Sommers and Phoebe Ellsworth have found that white jurors who attempt to avoid seeing race in cases involving white and black parties—that is, jurors who champion the ideal of colorblindness—are actually more biased in those cases.[257] Indeed, Sommers and Ellsworth have found that there is something about trying to avoid confrontation with the category of race that makes people more racially biased.[258] Conversely, when white jurors suspend the colorblindness ideal and confront race, they are less racially biased.[259]

Other scholars have similarly found that “actively contemplating others’ psychological experiences weakens the automatic expression of racial biases.”[260] For instance, white individuals who adopt the perspective of black individuals—who place themselves in the proverbial shoes of the latter—are less likely to exhibit automatic bias against them.[261] Based on these and other studies, legal commentators have argued that judges ought to encourage juries “to be conscious of race, gender, and other social categories.”[262] They point to “evidence [that] suggests that it is precisely this greater degree of discussion, and even confrontation [with social categories], that can potentially decrease the amount of biased decisionmaking.”[263] These kinds of “perspective-taking interventions,” scholars observe, “substantially decreased implicit bias in the form of negative attitudes.”[264] Their recommendations are consistent with those of law-and-mind scholars like Terry Maroney, who has argued that judges are their most effective when they consciously and deliberately engage, rather than evade, uncomfortable emotions and the stimuli that provoke them.[265] They are also consistent with empirical data that show that “consciously held beliefs and values about equality . . . can override the effect of automatically activated prejudice . . . toward outgroups.”[266]

Yoshino’s proposal deviates from the debiasing interventions suggested above. Rather than engage with social categories like race, gender, and sexual orientation, Yoshino’s tactical advice encourages judges to avoid those categories in order to serve the larger goal of managing the public’s pluralism anxiety—or, as we suspect, its implicit bias, given the likely relationship between those two things. Where implicit bias experts advocate direct engagement with the social categories that precipitate implicit bias in order to decrease it, Yoshino advocates the muting of those same categories in order to avoid it. In this sense, Yoshino’s suggested strategy does nothing to address implicit bias (or, in his words, “anxiety”)—and, ironically, might even perpetuate it.[267]

To be sure, we are not suggesting that Supreme Court opinions ought to function as debiasing devices. Indeed, the primary purpose of Court decisions is to “say what the law is,”[268] not to cure the public’s implicit bias against social outgroups like sexual minorities.

That said, we object to Yoshino’s recommendation not because we think that it will ultimately fail to remedy Americans’ implicit bias. Rather, we object to his recommendation because we do not think that judges, among other legal actors, ought to be encouraged to engage in bias-minimizing strategies. Perhaps it is very well true that few people read Supreme Court opinions and even fewer understand them. Our study provides at least some support for that proposition (given the extremely low level of subject comprehension that we found); in that sense, our study coheres with a wealth of social science data that paint a rather dim picture of the relationship between Supreme Court opinions and the public’s engagement with them.[269] That does not mean, however, that the nation’s most respected branch of government[270] ought to be embracing strategies that perpetuate implicit prejudice against outgroups—or, at the very least, strategies that do very little to eliminate that prejudice. Given the judiciary’s perceived stature in American public life, we believe that it ought to lead by example—and that minimizing the salient characteristics of groups when vindicating their civil rights is less than exemplary.

Conclusion

Our study tested empirically a theoretical claim commonly made by scholars of constitutional law: that Supreme Court reasoning matters in the court of public opinion. The most recent version of this claim comes from Professor Kenji Yoshino, who argues that courts can alleviate pluralism anxiety, and presumably contain public backlash, in decisions vindicating civil rights by leading with liberty and downplaying equality in those decisions. Our intuition was that Yoshino’s belief in the power of judicial reasoning was wrong as a positive matter. If, as Professors Braman and Fontana recently found, the public does not really care about which institutional actor resolves a high salience issue, then why would it care about something even more nuanced and granular, namely how an issue is decided? If that intuition is correct, then Yoshino’s tactical advice to courts is not only normatively undesirable (as we believe and argue) but also descriptively inaccurate.

Our results do not alter our intuition. Yoshino’s tactical advice to courts—that they lead with liberty in order to quell the public’s pluralism anxiety—is not robustly endorsed by our results, since we do not find that liberty increases acceptance or agreement when compared to a no-rationale decision. Nor do we find the negative effects of equality significantly pervasive to convince us that equality is that much worse than liberty (indeed, if anything, our results suggest that both are bad in terms of decreasing public support). Perhaps most important, even if Yoshino is correct that liberty does a better job than equality in containing backlash, his advice must be weighed against the significant normative concerns that follow from it—particularly when it comes to gay rights.

 

APPENDIX I. SURVEY VIGNETTES[271]

(Complete copies of the survey instrument text are available upon request from the authors.)

 

Major Victory for Same-Sex Marriage Advocates:

Supreme Court Issues Decision Protecting Same-Sex Marriage, Citing Liberty Interests

 

“The Constitution requires the government to respect everyone’s fundamental right to marry the person of his or her choice. Marriage is a fundamental freedom that the Constitution protects. The government acts unconstitutionally when it denies people, like same-sex couples, that freedom.” – Supreme Court of the United States

WASHINGTON, DC – In recent months, advocates of marriage between same-sex couples have been arguing to a federal court that the federal government and state governments must permit same-sex couples to marry, and now have taken their case to the Supreme Court. Advocates of marriage between same-sex couples have argued that the United States Constitution protects the right of all adults to marry the person of their choice, without regard to that person’s gender. In response, opponents have argued that the Constitution only protects traditional marriages and families, and that same-sex marriage is not mentioned in the Constitution. These opponents have argued that a Supreme Court decision protecting marriage between same-sex couples would be unwise.

The Supreme Court held oral arguments on marriage between same-sex couples, which were presented by the lawyers for same-sex marriage advocacy groups and by the lawyers for groups opposing same-sex marriage. After these arguments were presented to the Court, the members of the Court engaged in private discussions about the case.

Just a few days ago, the Supreme Court decided the case—and advocates of marriage for same-sex couples are very pleased. The new decision requires the federal government and state governments to permit same-sex couples to marry. The Supreme Court’s decision stated that everyone should be able to marry the person of their choice under the Constitution. Marriage, the Court declared, is a fundamental right for all adults and laws that prohibit same-sex marriage violate that right.

Advocates of marriage for same-sex couples praised the new decision as a “major step forwards in the history of American civil rights and a vindication of the Constitution” while opponents criticized the decision as a “blow to the people and traditions of the United States and to the importance of the institution of traditional marriage.”

 

Major Victory for Same-Sex Marriage Advocates:

Supreme Court Issues Decision Protecting Same-Sex Marriage, Citing Equality

“The Constitution requires the government to treat everyone equally, including gays and lesbians. When the government withholds marriage from gay and lesbian people, it treats them unequally, and therefore unconstitutionally. Gays and lesbians deserve to be treated equally to straight persons.” – Supreme Court of the United States

Concerns

 

WASHINGTON, DC – In recent months, advocates of marriage between same-sex couples have been arguing to a federal court that the federal government and state governments must permit same-sex couples to marry, and now have taken their case to the Supreme Court. Advocates of marriage between same-sex couples have argued that the Constitution prohibits governments from denying to gay people what they give to straight people—namely, the ability to marry—because gay people are the same as, and equal to, straight people. In response, opponents have argued that the Constitution only protects traditional marriages and families, and that same-sex marriage is not mentioned in the Constitution. These opponents have argued that a Supreme Court decision protecting marriage between same-sex couples would be unwise.

The Supreme Court held oral arguments on marriage between same-sex couples, which were presented by the lawyers for same-sex marriage advocacy groups and by the lawyers for groups opposing same-sex marriage. After these arguments were presented to the Court, the members of the Court engaged in private discussions about the case.

Just a few days ago, the Supreme Court decided the case—and advocates of marriage for same-sex couples are very pleased. The new decision requires the federal government and state governments to permit same-sex couples to marry. The Supreme Court’s decision stated that the Constitution prohibits the government from discriminating against people by singling them out for differential treatment on the basis of a particular characteristic—which is exactly what the government does when it prohibits gay people from marrying the person of their choice because they are gay.

Advocates of marriage for same-sex couples praised the new decision as a “major step forward in the history of American civil rights and a vindication of the Constitution” while opponents criticized the decision as a “blow to the people and traditions of the United States and to the importance of the institution of traditional marriage.”

 

Major Victory for Same-Sex Marriage Advocates

Supreme Court Issues Decision Protecting Same-Sex

“The Constitution requires the government to allow same-sex couples to marry.” – Supreme Court of the United States

Marriage

 

WASHINGTON, DC – In recent months, advocates of marriage for same-sex couples have been arguing to a federal court that the federal government and state governments must permit same-sex couples to marry, and now have taken their case to the Supreme Court. Advocates of marriage for same-sex couples have argued that the Constitution prohibits governments from enacting laws that ban same-sex marriage. In response, opponents have argued that the Constitution only protects traditional marriages and families, and that same-sex marriage is not mentioned in the Constitution. These opponents have argued that a Supreme Court decision protecting marriage between same-sex couples would be unwise.

The Supreme Court held oral arguments on marriage between same-sex couples, which were presented by the lawyers for same-sex marriage advocacy groups and by the lawyers for groups opposing same-sex marriage. After these arguments were presented to the Court, the members of the Court engaged in private discussions about the case.

Just a few days ago, the Supreme Court decided the case—and advocates of marriage for same-sex couples are very pleased. The new decision requires the federal government and state governments to permit marriage between same-sex couples.

Advocates of marriage for same-sex couples praised the new decision as a “major step forwards in the history of American civil rights and a vindication of the Constitution” while opponents criticized the decision as a “blow to the people and traditions of the United States and to the importance of the institution of traditional marriage.”

                     *   © 2014 Courtney Megan Cahill & Geoffrey Christopher Rapp.

                  **   Donald Hinkle Professor, Florida State University College of Law. I am grateful to David Landau for his illuminating thoughts and suggestions on this piece, and thank the Florida State University College of Law for the generous financial assistance that it provided for this study.

               ***   Harold A. Anderson Professor of Law & Values, University of Toledo College of Law. This study was supported by research grant from the University of Toledo College of Law. Greg Stone, Ph.D., of the University of Toledo, provided useful insight in connection with the design of the project. Christine Fox, Ph.D., of the University of Toledo, was exceptionally generous with her time and insights. Professor Quin Shao and her graduate students also provided outstanding feedback.

         [1].   See generally United States v. Windsor, 133 S. Ct. 2675 (2013) (striking down section 3 of DOMA as a violation of the Fifth Amendment’s Due Process Clause). Section 3 of DOMA defined marriage in cross-sex terms for the purpose of federal benefits. Defense of Marriage Act, Pub. L. No. 104-199, § 3, 110 Stat. 2419, 2419 (1996) (codified at 1 U.S.C. § 7, 28 U.S.C. § 1738C (2012)).

         [2].   See generally Hollingsworth v. Perry, 133 S. Ct. 2652 (2013) (dismissing petitioners’ appeal for lack of standing under Article III, thereby reinstating the lower court’s decision striking down California’s exclusionary marriage initiative on federal constitutional grounds).

         [3].   Perry v. Schwarzenegger, 704 F. Supp. 2d 921, 991 (N.D. Cal. 2010) (striking down a state constitutional marriage exclusion as a violation of the U.S. Constitution’s Due Process and Equal Protection Clauses).

         [4].   See, e.g., Elizabeth Weise, San Francisco Celebrates Gay Marriage Rulings, USA Today (July 5, 2013, 12:51 PM), http://www.usatoday.com/story/news/nation/2013/06/26/san-francisco-celebrates-gay-marriage-ruling/2461259/ (noting that “[a]fter both [Windsor and Hollingsworth] were announced, a line of politicians and gay activists . . . let out a roar of approval”).

         [5].   Dan Levine, U.S. High Court Ruling Sets Up New Wave of Gay Marriage Battles, Chi. Trib. (June 26, 2013), http://articles.chicagotribune.com/2013-06-26/news/sns-rt-usa-gayrightscampaignsl2n0f21jd-20130626_1_marriage-ban-marriage-rights-gay-marriage; see also Michael J. Klarman, Windsor and Brown: Marriage Equality and Racial Equality, 127 Harv. L. Rev. 127, 158 (2013) (“Within a few years of Windsor, as public support for gay marriage continues to increase and as more states enact it into law, one can imagine some Justices being tempted to extend that ruling to forbid the states from excluding same-sex couples from marriage. Indeed, the Windsor dissents of both Chief Justice Roberts and Justice Scalia seem mostly addressed to that eventuality . . . .”).

         [6].   A number of cases have been filed in the aftermath of Windsor and Hollingsworth challenging state same-sex marriage bans in both state and federal court; it is at the very least possible that one of them will find its way to the Supreme Court of the United States. See, e.g., Bishop v. U.S. ex rel. Holder, 962 F. Supp. 2d 1252, 1296 (N.D. Okla. 2014) (striking down Oklahoma’s constitutional marriage ban on federal equal protection grounds), aff’d, Nos. 14-5003, 14-5006, 2014 WL 3537847 (10th Cir. July 18, 2014); Kitchen v. Herbert, 961 F. Supp. 2d 1181, 1216 (D. Utah 2013) (striking down Utah’s constitutional and statutory marriage ban on federal due process and equal protection grounds), aff’d, 755 F.3d 1193 (10th Cir. 2014). During its October 2014 Term, the Supreme Court denied certiorari for all cases arising from challenges to state laws banning same-sex marriage, which left a split among the states regarding the issue. See, e.g., Baskin v. Bogan, 766 F.3d 648, 653 (7th Cir. 2014), cert. denied, 135 S. Ct. 316 (2014); Kitchen v. Herbert, 755 F.3d 1193, 1198–99 (10th Cir. 2014), cert. denied, 135 S. Ct. 265 (2014); Bostic v. Schaefer, 760 F.3d 352, 367–68 (4th Cir. 2014), cert. denied, 135 S. Ct. 308 (2014); Bishop v. Smith, 760 F.3d 1070, 1074 (10th Cir. 2014), cert. denied, 135 S. Ct. 271 (2014).

         [7].   Our study contributes to a growing body of empirical research undertaken by political scientists on the relationship between decision attributes and public opinion. See, e.g., James R. Zink, James F. Spriggs II & John T. Scott, Courting the Public: The Influence of Decision Attributes on Individuals’ Views of Court Opinions, 71 J. Pol. 909, 910 (2009) (testing whether decisions with larger majorities and that follow precedent positively influence public opinion).

         [8].   See infra Part II.

         [9].   See infra Part VI.

       [10].   See generally 347 U.S. 483 (1954) (striking down laws separating the races in public education as a violation of the Fourteenth Amendment’s Equal Protection Clause).

       [11].   See generally 410 U.S. 113 (1973) (striking down Texas’s criminal abortion law and finding that the abortion right is protected as an aspect of privacy under the Fourteenth Amendment’s Due Process Clause). For a sampling of this scholarly commentary, see, for example, Anita L. Allen et al., What Roe v. Wade Should Have Said 18–22 (Jack M. Balkin ed. 2005) (summarizing the authors’ revised opinions of Roe and noting that they “took a variety of different approaches to answer [the book’s central] question of what Roe v. Wade should have said”), and Bruce Ackerman et al., What Brown v. Board of Education Should Have Said (Jack M. Balkin ed. 2001) (noting that a group of constitutional scholars was asked how they would “have written the Brown opinion”).

       [12].   See generally Kenji Yoshino, The New Equal Protection, 124 Harv. L. Rev. 747 (2011) [hereinafter Yoshino, The New Equal Protection] (theorizing how the Supreme Court should reason); Kenji Yoshino, Tribe, 42 Tulsa L. Rev. 961 (2007) [hereinafter Yoshino, Tribe] (theorizing how the Supreme Court should reason). Professor Yoshino’s theory is elaborated more fully below. See infra Part I.

       [13].   William M. Richman & William L. Reynolds, Elitism, Expediency, and the New Certiorari: Requiem for the Learned Hand Tradition, 81 Cornell L. Rev. 273, 282–85 (1996). Consider here Brown v. Board of Education, which held that state-mandated separation of the races in public education was unconstitutional because it made African American schoolchildren feel inferior to whites. 347 U.S. at 494. The Brown Court has been criticized not for the result that it reached but rather for the reasoning on which that result rests—reasoning that putatively lacks “neutral principles” with clear applicability outside the public education context. See Herbert Wechsler, Toward Neutral Principles of Constitutional Law, 73 Harv. L. Rev. 1, 22 (1959) (arguing that the Supreme Court’s decision to extend its Brown “ruling to other public facilities, such as public transportation, parks, golf courses, bath houses, and beaches, which no one is obliged to use––all by per curiam decisions” is not necessarily a “principled” one given Brown’s context-specific reasoning). If separate-but-equal was unconstitutional in public education because of how it made African American schoolchildren feel, then what of separate-but-equal on public golf courses—or in any other context in which that repudiated legal doctrine was the law of the land? See Michael J. Klarman, From Jim Crow to Civil Rights 321 (2004) (discussing the cases extending Brown’s holding to other contexts as well as contemporary criticism of them).

       [14].   Consider here Roe v. Wade, which struck down Texas’s criminal abortion law on due process grounds and found that the federal Constitution protects the abortion right under the larger rubric of privacy. 410 U.S. at 164. Many commentators have argued that the Roe Court should have supported its holding on equality grounds because equality is a better constitutional basis for abortion rights than is privacy, given that privacy is mentioned nowhere in the Constitution whereas equality is and given that abortion laws affect women in both sex-specific and sex-stereotypical ways. See, e.g., Sylvia A. Law, Rethinking Sex and the Constitution, 132 U. Pa. L. Rev. 955, 981 (1984) (arguing that abortion laws violate the Equal Protection Clause because they create a sex classification—only women can have an abortion because only women can get pregnant); see also Neil S. Siegel & Reva B. Siegel, Equality Arguments for Abortion Rights, 60 UCLA L. Rev. Disc. 160, 170 (2013) (arguing for a “synthetic understanding of the constitutional basis of the abortion right—as grounded in both liberty and equality values”); Reva B. Siegel, Sex Equality Arguments for Reproductive Rights: Their Critical Basis and Evolving Constitutional Expression, 56 Emory L.J. 815, 816 (2007) (outlining the “constitutional arguments that have been advanced in a variety of doctrinal frameworks” in favor of a “sex equality approach to reproductive rights”); Reva B. Siegel, The New Politics of Abortion: An Equality Analysis of Woman-Protective Abortion Restrictions, 2007 U. Ill. L. Rev. 991, 994 (arguing that “[a]n abortion ban reflecting and enforcing [gender stereotypes] violates constitutional guarantees of equal citizenship”); Cass R. Sunstein, Neutrality in Constitutional Law (with Special Reference to Pornography, Abortion, and Surrogacy), 92 Colum. L. Rev. 1, 36 (1992) (making the sex stereotype argument against restrictive abortion laws). Moreover, had the Court justified the abortion right on equality grounds, the abortion right would be secure even if Roe were reversed. See Reva B. Siegel, Concurring, in What Roe v. Wade Should Have Said 63, 63 (Jack M. Balkin ed. 2005). Roe also illustrates how the Court’s reasoning in a particular decision can affect that decision’s precedential scope, as discussed earlier in the context of Brown. See supra note 13 and accompanying text. Had Roe been decided on equality grounds, some commentators maintain, it could have been strong authority for why the government’s refusal to cover even medically necessary abortions for indigent women was unconstitutional, even though the Court held otherwise in cases after Roe. See, e.g., Harris v. McRae, 448 U.S. 297, 318 (1980) (holding that the abortion right does not require the state to fund non-therapeutic abortions of Medicaid recipients).

       [15].   See Yoshino, The New Equal Protection, supra note 12, at 796–97 (arguing that the Supreme Court can “fashion a new, more inclusive sense of ‘we’ ” by reasoning in one way over another in decisions addressing issues that might otherwise provoke the people’s “pluralism anxiety”).

       [16].   Id. at 751.

       [17].   Id. at 750 (arguing that the “liberty-based dignity claim” is a “way for the Court to ‘do’ equality in an era of increasing pluralism anxiety”).

       [18].   Justice Ginsburg has argued that by avoiding “giant strides” in constitutional adjudication, a court can reduce the risk of producing a “backlash too forceful to contain.” Ruth Bader Ginsburg, Speaking in a Judicial Voice, 67 N.Y.U. L. Rev. 1185, 1208 (1992) [hereinafter Ginsburg, Speaking in a Judicial Voice].

       [19].   Id. at 1200; see also Ruth Bader Ginsburg, Some Thoughts on Autonomy and Equality in Relation to Roe v. Wade, 63 N.C. L. Rev. 375, 382 (1985) [hereinafter Ginsburg, Some Thoughts on Autonomy and Equality in Relation to Roe v. Wade] (arguing that “[a]cademic criticism of Roe . . . might have been less pointed had the Court placed the woman alone, rather than the woman tied to her physician, at the center of its attention”). But see id. at 382 (“I do not pretend that, if the Court had added a distinct sex discrimination theme to its medically oriented opinion, the storm Roe generated would have been less furious.”).

       [20].   We say “ostensible” because the Roe backlash thesis—the claim that Roe precipitated backlash against both the abortion right and the Court—has recently undergone serious critique. See Linda Greenhouse & Reva Siegel, Before Roe v. Wade: Voices That Shaped the Abortion Debate Before the Supreme Court’s Ruling 253 (2010) (questioning the Roe backlash thesis); Linda Greenhouse & Reva B. Siegel, Backlash to the Future? From Roe to Perry, 60 UCLA L. Rev. Disc. 240, 242 (2013) [hereinafter Greenhouse & Siegel, Backlash to the Future?] (questioning “the one-dimensional story of court-centered backlash so often attributed to Roe”); Linda Greenhouse & Reva B. Siegel, Before (and After) Roe v. Wade: New Questions About Backlash, 120 Yale L.J. 2028, 2030–31 (2011) (raising “questions about the conventional assumption that the Court’s decision in Roe is responsible for political polarization over abortion”).

       [21].   Ginsburg, Some Thoughts on Autonomy and Equality in Relation to Roe v. Wade, supra note 19, at 383 (citing Kenneth L. Karst, Foreword: Equal Citizenship Under the Fourteenth Amendment, 91 Harv. L. Rev. 1, 57–59 (1977)).

       [22].   See generally 539 U.S. 558 (2003) (striking down Texas’s criminal sodomy law on due process grounds and finding that the federal Constitution protects a limited right to sexual autonomy).

       [23].   See infra Part I.

       [24].   Yoshino made this argument before its appearance in the Harvard Law Review. See, e.g., Yoshino, Tribe, supra note 12, passim (setting forth his pluralism anxiety theory in a symposium tribute to Laurence Tribe); Kenji Yoshino, Gerken-Yoshino Debate on Liberty and Equality, Round 2, Slate (May 22, 2008, 10:17 AM), http://www.slate.com/blogs/convictions/2008/05/22/gerken_yoshino_debate_on_liberty_and_equality_round_2.html [hereinafter Yoshino Debate].

       [25].   Yoshino, The New Equal Protection, supra note 12, at 794.

       [26].   Id.

       [27].   See id. at 778 (praising Lawrence’s liberty rationale as one that “quieted pluralism anxiety”).

       [28].   See Cass R. Sunstein, One Case at a Time: Judicial Minimalism on the Supreme Court 4–6 (1999) [hereinafter Sunstein, One Case at a Time]; Cass R. Sunstein, Foreword: Leaving Things Undecided, 110 Harv. L. Rev. 4, 6–8 (1996) [hereinafter Sunstein, Leaving Things Undecided].

       [29].   See, e.g., Klarman, supra note 13, at 6–7 (describing how courts can avoid public backlash); William N. Eskridge, Jr., Backlash Politics: How Constitutional Litigation Has Advanced Marriage Equality in the United States, 93 B.U. L. Rev. 275, 278 (2013) (using pluralism reinforcing theory to argue when courts should and should not incite backlash); Michael J. Klarman, Civil Rights Law: Who Made It and How Much Did It Matter?, 83 Geo. L.J. 433, 433 n.4 (1994) [hereinafter Klarman, Civil Rights Law] (discussing the Brown “backlash thesis”); Michael Klarman, How Brown Changed Race Relations: The Backlash Thesis, 81 J. Am. Hist. 81, 117–18 (1994) [hereinafter Klarman, How Brown Changed Race Relations]; Michael Klarman, Harvard Law School, Hart Lecture at Georgetown Law Center: Courts, Social Change, and Political Backlash (Mar. 31, 2011) [hereinafter Klarman Lecture] (discussing the disruptive backlash that resulted from judicial decisions in high-salience contexts like desegregation and marriage equality).

       [30].   For a description of the relationship between pluralism anxiety and backlash, see infra notes 78–91 and accompanying text.

       [31].   See infra note 78–91 and accompanying text.

       [32].   Yoshino, The New Equal Protection, supra note 12, at 796.

       [33].   In fact, Yoshino and Ginsburg differ in an even more fundamental way than the relative scope and applicability of their theories. Whereas Yoshino argues that liberty-based rationales are more appealing to the public, Ginsburg argues that an equality-based rationale for the abortion right specifically would have been more appealing to the public. Compare Yoshino, The New Equal Protection, supra note 12, at 792–94 (arguing that liberty-based rationales “may appeal to the libertarian streak in some conservatives” and generally have a greater “effect on quieting pluralism anxiety” than equality-based rationales), with Ginsburg, Speaking in a Judicial Voice, supra note 18, at 1200 (“The Roe decision might have been less of a storm center had it [] homed in more precisely on the women’s equality dimension of the issue . . . .”). She roots this belief in the fact that starting in the 1970s, the public was very supportive of Supreme Court decisions striking down sex-specific laws on equality grounds. See Ginsburg, Speaking in a Judicial Voice, supra note 18, at 1200.

       [34].   This oversight is also curious in light of existing empirical accounts of Supreme Court opinions and the American public—accounts which suggest that “large portions of the American public are unaware even of major decisions like Roe v. Wade.” Gerald N. Rosenberg, Romancing the Court, 89 B.U. L. Rev. 563, 567 (2009); see also Nathanial Persily, Jack Citrin & Patrick J. Egan, Public Opinion and Constitutional Controversy 8 (2008) (surveying empirical data that overwhelmingly suggest that “Supreme Court decisions had no effect on the overall distribution of [the American] public opinion”).

       [35].   We discuss the relationship between pluralism anxiety and backlash below. See infra notes 76–91 and accompanying text.

       [36].   David Fontana & Donald Braman, Judicial Backlash or Just Backlash?, 112 Colum. L. Rev. 731, 735 (2012) [hereinafter Fontana & Braman, Judicial Backlash] (reporting the results of a national survey that tested whether the public was less supportive of rights-enhancing decisions if they were issued by the Supreme Court rather than by Congress); see also Donald Braman & David Fontana, The New Republic: Supreme Anxiety, NPR (Jan. 19, 2012, 9:02 AM), http://www.npr.org/2012/01/19/145445550/the-new-republic-supreme-anxiety [hereinafter Braman & Fontana, Supreme Anxiety] (summarizing the Columbia Law Review piece for a more wide-spread audience).

       [37].   See Fontana & Braman, Judicial Backlash, supra note 36, at 735.

       [38].   Professors Fontana and Braman empirically tested whether it matters to the public if Congress or the Supreme Court decides a high salience issue like same-sex marriage, not whether it matters to the public if the Court decides a high salience issue in a particular way. See id. at 734–35 (summarizing their study and its results).

       [39].   Id. at 734.

       [40].   Alexander M. Bickel, The Least Dangerous Branch: The Supreme Court at the Bar of Politics 16 (1962) (“The root difficulty is that judicial review is a counter-majoritarian force in our system.”).

       [41].   Fontana & Braman, Judicial Backlash, supra note 36, at 740–41.

       [42].   Id. at 746 (“[W]e find in our study that cultural priors are much stronger predictors of behavior than other attitudinal priors.”).

       [43].   See id. at 742.

       [44].   Id. at 734.

       [45].   See Yoshino, The New Equal Protection, supra note 12, at 793–94.

       [46].   See supra notes 2–3, 6 and accompanying text.

       [47].   See Klarman, supra note 5, at 128 (referring to the “ ‘fifty-state’ solution” of identifying “a federal constitutional right to same-sex marriage”).

       [48].   Kenji Yoshino, Covering: The Hidden Assault on Our Civil Rights ix (2006) [hereinafter Yoshino, Covering: The Hidden Assault].

       [49].   See generally Yoshino, Covering: The Hidden Assault, supra note 48 (criticizing the covering demands that the law places on sexual minorities); Kenji Yoshino, Covering, 111 Yale L.J. 769 (2002) [hereinafter Yoshino, Covering] (explaining the problem of judicial covering). Yoshino has also written about the law’s erasure of sexual minorities. See generally Kenji Yoshino, Assimilationist Bias in Equal Protection: The Visibility Presumption and the Case of “Don’t Ask, Don’t Tell, 108 Yale L.J. 487 (1998) (arguing that the assimilationist bias contained in immutability and visibility factors of equal protection jurisprudence should be retired); Kenji Yoshino, The Epistemic Contract of Bisexual Erasure, 52 Stan. L. Rev. 353 (2000) (seeking to explain why bisexuals are erased in society).

       [50].   United States v. Windsor, 133 S. Ct. 2675, 2709 (2013); see infra notes 233–39 and accompanying text.

       [51].   For opposing views on this question, compare Lawrence Baum & Neal Devins, Why the Supreme Court Cares About Elites, Not the American People, 98 Geo. L.J. 1515, 1516 (2010) (“Supreme Court Justices care more about the views of academics, journalists, and other elites than they do about public opinion.”), with Barry Friedman, The Will of the People: How Public Opinion Has Influenced the Supreme Court and Shaped the Meaning of the Constitution 367–68 (2009) (arguing that the Supreme Court is responsive to public opinion in its decisionmaking).

       [52].   See supra notes 13–21 and accompanying text.

       [53].   Yoshino, The New Equal Protection, supra note 12, at 794. Like Yoshino, other scholars have argued against using equality-based and identity-based reasoning. See, e.g., Jessica Knouse, From Identity Politics to Ideology Politics, 2009 Utah L. Rev. 749, 785.

       [54].   See Yoshino, The New Equal Protection, supra note 12, at 751.

       [55].   Id.

       [56].   Id. at 752 n.37 (citing Robert D. Putnam, E Pluribus Unum: Diversity and Community in the Twenty-First Century, 30 Scandinavian Pol. Stud. 137, 138 (2007) [hereinafter Putnam, E Pluribus Unum]). Putnam is the author of the landmark book Bowling Alone, which posited a link between increasing diversity in the United States and decreasing “social capital.” Robert Putnam, Bowling Alone 352 (2000). Putnam’s studies are limited to ethnic and racial diversity, see Putnam, E Pluribus Unum, supra note 56, at 138–39, whereas Yoshino posits that diversity generally—including gender diversity and sexual diversity—might be contributing to pluralism anxiety, see Yoshino, The New Equal Protection, supra note 12, at 747 (discussing diversity broadly in terms of religious, racial, ethnic, gender, and sexual diversity).

       [57].   Yoshino, The New Equal Protection, supra note 12, at 754.

       [58].   Id. at 753 (quoting Putnam, E Pluribus Unum, supra note 56, at 150).

       [59].   Id. at 752.

       [60].   Id. at 755 (citations omitted).

       [61].   Id. at 755–63.

       [62].   Id. at 763–68.

       [63].   Yoshino, The New Equal Protection, supra note 12, at 768–73.

       [64].   Id. at 774.

       [65].   Id. at 776.

       [66].   See id.

       [67].   Id.

       [68].   Id. at 793.

       [69].   Yoshino, The New Equal Protection, supra note 12, at 793.

       [70].   Id. at 776 (citing 539 U.S. 558 (2003)).

       [71].   Lawrence v. Texas, 539 U.S. 558, 578 (2003) (striking down Texas’s sodomy statute on federal due process grounds); id. at 575 (explaining the inadequacies of striking down Texas’s sodomy statute on federal equal protection grounds).

       [72].   Yoshino, The New Equal Protection, supra note 12, at 777.

       [73].   See id. at 777–78. For a more tempered reading of the Lawrence Court’s turn toward liberty, see Russell K. Robinson, Uncovering Covering, 101 Nw. U. L. Rev. 1809, 1827 (2007) (reviewing Yoshino, Covering: The Hidden Assault, supra note 48, and stating that “[t]he cases [Yoshino] cites as leading the way toward his liberty-based model are also revealing. To the extent that Lawrence v. Texas . . . discussed liberty or universality, [its] actual holding[] clearly benefited outsiders, with hardly any impact on insiders.” (citations omitted)).

       [74].   Yoshino, The New Equal Protection, supra note 12, at 778. Some commentators have argued that Lawrence was more polarizing than Yoshino suggests. See, e.g., Patrick J. Egan & Nathaniel Persily, Court Decisions and Trends in Support for Same-Sex Marriage, The Polling Report, Aug. 2009, at 1, 6 (discussing the backlash that Lawrence initially precipitated).

       [75].   See Yoshino, The New Equal Protection, supra note 12, at 778. A plausible but less compelling interpretation would be that Yoshino meant to describe the effect of the Court’s opinion on pluralism anxiety among elites (such as political and media elites). Even if that is the case (perhaps suggested by who would be making a “charge” against the Court), the reason why elite reaction might matter is that it can shape public reaction.

       [76].   Id. at 777–78.

       [77].   See id.

       [78].   Id. at 754 (arguing that “the United States Supreme Court might help create  . . . ‘[a] new, broader sense of “we” ’ ” through its constitutional jurisprudence (quoting Putnam, E Pluribus Unum, supra note 56, at 139)).

       [79].   The term “backlash” is susceptible to multiple interpretations. We argue that backlash involves mobilized political action to reverse and/or erode a judicial decision. While a lack of “acceptance” or “agreement” with a decision is not the same as backlash, acceptance of a decision (or agreement with it) would seem to deny the field of battle to elites seeking to create a public backlash (or at least to make it harder for them to achieve victory on it). For a more complete definition of backlash and a history of its use in law, see Robert Post & Reva Siegel, Roe Rage: Democratic Constitutionalism and Backlash, 42 Harv. C.R.-C.L. L. Rev. 373, 388–91 (2007).

       [80].   To be sure, it does not have to be judicial decisions, see Greenhouse & Siegel, Backlash to the Future?, supra note 20, at 243 n.2 (“Backlash . . . is not limited to courts.”), although conventional wisdom says that courts provoke backlash to a much greater degree than the two politically accountable branches, see id. (“The premise of the Roe backlash narrative is that there is something about the judicial declaration of minority rights that produces an especially virulent and polarizing reaction among losers who would not respond in a similar fashion to legislative defeat.”). Recent empirical findings suggest that this conventional wisdom may be wrong. See Fontana & Braman, Judicial Backlash, supra note 36, passim.

       [81].   Yoshino, The New Equal Protection, supra note 12, at 794.

       [82].   See id. at 794.

       [83].   See, e.g., Sunstein, One Case at a Time, supra note 28, at 5; Sunstein, Leaving Things Undecided, supra note 28, at 8.

       [84].   See, e.g., Eskridge, supra note 29, at 281–91 (using pluralism reinforcing theory to argue when courts should and should not incite backlash); see generally Klarman, supra note 13 (arguing that the Court’s narrow decision in Brown did more to encourage southern white opposition); Klarman, Civil Rights Law, supra note 29 (reviewing Mark Tushnet, Civil Rights Law: Thurgood Marshall and the Supreme Court, 1936–1961 (1994), and discussing the Brown “backlash thesis”); Klarman, How Brown Changed Race Relations, supra note 29 (questioning Brown’s impact and arguing that Brown encouraged southern resistance to racial change); Klarman Lecture, supra note 29 (discussing the disruptive backlash that resulted from judicial decisions in high-salience contexts like desegregation and marriage equality).

       [85].   See Jeffrey Rosen, The Supreme Court: Judicial Temperament and the Democratic Ideal, 47 Washburn L.J. 1, 8 (2007) (describing backlash as a “complicated phenomenon”).

       [86].   See Robert A. Garda, The White Interest in School Integration, 63 Fla. L. Rev. 599, 625 (2011) (describing how anxiety is a manifestation of implicit or unconscious bias).

       [87].   See Deana A. Pollard, Unconscious Bias and Self-Critical Analysis: The Case for a Qualified Evidentiary Equal Employment Opportunity Privilege, 74 Wash. U. Rev. 913, 918 (1999) (explaining that even those with no conscious bias may react negatively as a result of subconscious bias).

       [88].   See, e.g., Jerry Kang et al., Implicit Bias in the Courtroom, 59 UCLA L. Rev. 1124, 1126–28 (2012) [hereinafter Kang, Implicit Bias]; Jerry Kang & Kristin Lane, Seeing Through Colorblindness: Implicit Bias and the Law, 58 UCLA L. Rev. 465, 492–93 (2010); Jerry Kang, Trojan Horses of Race, 118 Harv. L. Rev. 1489, 1506–12 (2005); Linda Hamilton Krieger & Susan T. Fiske, Behavioral Realism in Employment Discrimination Law: Implicit Bias and Disparate Treatment, 94 Calif. L. Rev. 997, 1061 (2006); Linda Hamilton Krieger, The Content of Our Categories: A Cognitive Bias Approach to Discrimination and Equal Employment Opportunity, 47 Stan. L. Rev. 1161, 1199–1200, 1209–12 (1995) (discussing anti-discrimination law’s failure to capture implicit bias); Kristin Lane, Jerry Kang & Mahzarin R. Banaji, Implicit Social Cognition and the Law, 3 Ann. Rev. L. And Soc. Sci. 427, 428 (2007); Charles R. Lawrence III, The Id, the Ego, and Equal Protection: Reckoning With Unconscious Racism, 39 Stan. L. Rev. 317, 318–23 (1987) (discussing anti-discrimination law’s failure to capture unconscious racism); Justin D. Levinson, Superbias: The Collision of Behavioral Economics and Implicit Social Cognition, 45 Akron L. Rev. 591, 592–96 (2012); Philip E. Tetlock, Gregory Mitchell & L. Jason Anastasopoulos, Detecting and Punishing Unconscious Bias, 42 J. Legal Stud. 83, 83–84 (2013).

       [89].   See Fontana & Braman, Judicial Backlash, supra note 36, at 766 (“[A] base of support may insulate a Court decision from immediate backlash.”).

       [90].   Such an interpretation, however, is not implausible. If pluralism anxiety causes people to be less persuaded than they would otherwise be by a judicial decision, if being less persuaded by a judicial decision could eventually precipitate backlash against both it and the court that rendered it, and if equality at the very least does nothing to minimize pluralism anxiety (and might even exacerbate it), then it would be fair to say that equality rationales, on Yoshino’s view, precipitate backlash more than liberty rationales do.

       [91].   Yoshino, The New Equal Protection, supra note 12, at 794 (acknowledging that “liberty claims [might not] quash [pluralism] anxiety altogether but . . . they do so more than equality claims”).

       [92].   See infra Parts IV–VI.

       [93].   Yoshino, The New Equal Protection, supra note 12, at 754 (quoting Putnam, E Pluribus Unum, supra note 56, at 139).

       [94].   Id. at 792.

       [95].   See e.g., Yoshino Debate, supra note 24 (arguing that “the fact that I want the courts to lead with liberty does not mean that other bodies must do so as well”).

       [96].   Yoshino, The New Equal Protection, supra note 12, at 750 (stating that he is “confident in [his] descriptive claim that the Court shut doors in its equality jurisprudence in the name of pluralism anxiety and has opened doors in its liberty jurisprudence to compensate” but recognizing also that he is “less confident about the inevitability or desirability of this shift”).

       [97].   Id.

       [98].   Id.

       [99].   Id. at 754 (quoting Putnam, E Pluribus Unum, supra note 56, at 139).

     [100].   See Fontana & Braman, Judicial Backlash, supra note 36, at 774 (arguing with respect to the Lawrence opinion specifically that “[i]t would have been hard to focus on discrete elements of the opinion by the Court and tune out the pre-decision and post-decision framing of the case”).

     [101].   Government, Gallup, http://www.gallup.com/poll/27286/Government.aspx (last visited Sept. 7, 2014); see Yoshino, The New Equal Protection, supra note 12, at 754 n.55.

     [102].   Indeed, after canvassing the social science literature on the relationship between Supreme Court decisions and the American public, political scientist Gerald Rosenberg confidently asserts: “I could devote a considerable number of pages to reviewing the literature that uniformly finds most Americans do not have a clue as to what the Court is doing or has done.” Rosenberg, supra note 34, at 566.

     [103].   See, e.g., Zink et al., supra note 7, at 923 (finding that the public is generally more receptive of Supreme Court decisions with larger majorities and that follow precedent).

     [104].   See id. at 911 (stating that the two opinion attributes that the authors test “are among the most visible features of Court opinions both in Court decisions themselves and in the reporting of these decisions in the popular press and therefore among the most visible to the public”).

     [105].   Laurence H. Tribe, Lawrence v. Texas: The “Fundamental Right” That Dare Not Speak Its Name, 117 Harv. L. Rev. 1893, 1898 (2004).

     [106].   As Yoshino observes, Lawrence recognized the “hybrid structure” of equality and liberty when it asserted: “Equality of treatment and the due process right to demand respect for conduct protected by the substantive guarantee of liberty are linked in important respects, and a decision on the latter point advances both interests.” Lawrence v. Texas, 539 U.S. 558, 575 (2003); Yoshino, Tribe, supra note 12, at 967. It is also worth mentioning that the fundamental right to marry has always, nearly since its inception, been considered alongside equality, or at least “equal access,” concerns. See, e.g., Martha Nussbaum, A Right to Marry?, 98 Calif. L. Rev. 667, 688 (2010); Nelson Tebbe & Deborah A. Widiss, Equal Access and the Right to Marry, 158 Penn. L. Rev. 1375, 1413 (2010).

     [107].   Yoshino, Tribe, supra note 12, at 970 (“We should . . . recognize that liberty and equality are intertwined values.”).

     [108].   The majority opinion in United States v. Windsor, 133 S. Ct. 2675 (2013), which struck down section 3 of DOMA as a violation of the Fifth Amendment and which we discuss at relative length in Part VI, might be one of these opinions. See Douglas NeJaime, Windsor’s Right to Marry, 123 Yale L.J. Online 219, 220 (2013), available at http://www.yalelawjournal.org/pdf/1205_3mchpr78.pdf (arguing that while the Windsor majority technically struck down section 3 on equal protection grounds, it is “conceptually, if not doctrinally, a right-to-marry case”).

     [109].   See Yoshino, The New Equal Protection, supra note 12, at 795 (citing Theda Skocpol, Targeting Within Universalism: Politically Viable Policies to Combat Poverty in the United States, in The Urban Underclass 411, 414 (Christopher Jencks & Paul E. Peterson eds., 1991), and Richard H. Pildes, The Future of Voting Rights Policy: From Anti-Discrimination to the Right to Vote, 49 How. L.J. 741 (2006)).

     [110].   While there is a rich discourse on the relationship between Court decisions and the American public, see, e.g., Klarman Lecture, supra note 29 (discussing the disruptive backlash that resulted from judicial decisions in high-salience contexts like desegregation and marriage equality), there is little empirical data on the relationship between Court reasoning and the American public.

     [111].   Although several commentators have theorized that a connection between decisional attributes and public acceptance of a decision exists, very few have subjected that theory to empirical analysis. See Zink et al., supra note 7, at 910 (stating that prior to the authors’ 2008 study there was “no convincing empirical evidence for the claim that the attributes of Court decisions have any influence on individuals’ perceptions of those decisions”).

     [112].   See id. at 923 (observing that there is little empirical evidence that any particular attribute of a Court decision alters the public’s reaction to it); Patrick Egan & Jack Citrin, The Limits of Judicial Persuasion and the Fragility of Judicial Legitimacy 7 (July 2011) (unpublished manuscript), available at http://politics.as.nyu.edu/docs/IO/4819/egancitrin.pdf.

     [113].   Some scholars have found that the Court is more likely to affect opinion in regard to less controversial issues. See, e.g., Valerie J. Hoekstra, Public Reaction to Supreme Court Decisions 9 (2003).

     [114].   See Michael F. Salamone, Judicial Consensus and Public Opinion: Conditional Response to Supreme Court Majority Size, 67 Pol. Res. Q. 320, 320–21 (2014); Zink et al., supra note 7, at 910; Egan & Citrin, supra note 112, at 17–19.

     [115].   Zink et al., supra note 7, at 919.

     [116].   Cass R. Sunstein, Unanimity and Disagreement on the Supreme Court 33, (Working Paper 2014) (quoting Salamone, supra note 114, at 321), available at http://ssrn.com/abstract=2466057.

     [117].   Adam Liptak, On Supreme Court, Does 9-0 Add Up to More Than 5-4?, N.Y. Times, Aug. 11, 2014, at A13 (citing Sunstein, supra note 116, at 33, and Salamone, supra note 114).

     [118].   See Sunstein, supra note 116, at 32–34 (summarizing this empirical evidence).

     [119].   See Liptak, supra note 117, at A13 (discussing Justice Ginsburg’s recent endorsement of the position that “closely divided rulings may be perceived to be less legitimate than united ones”).

     [120].   See Sunstein, supra note 116, at 32–33.

     [121].   Salamone, supra note 114, at 320.

     [122].   Id.

     [123].   We should mention that scholars have also contested the thesis that the Supreme Court caused a distinctive backlash in the court of public opinion in discrete cases by looking to history rather than to empirical analysis. Most notable here are Reva Siegel and Robert Post, who have used history to seriously undermine the Roe v. Wade backlash thesis, that is, the claim that Roe caused a social and political backlash, one that not only seriously undermined the public’s respect for courts but has also placed Roe and the abortion right that it upheld in perpetual peril. Post & Siegel, supra note 79, at 377 (observing that for Roe’s progressive critics, “Roe illustrates the terrible consequences of judicial decisionmaking that provokes intense opposition”); see also id. at 406–07 nn.180–82 (summarizing those critics’ arguments).

     [124].   See supra notes 37–42 and accompanying text.

     [125].   Fontana & Braman, Judicial Backlash, supra note 36, at 768.

     [126].   Id. at 735.

     [127].   Id. at 734.

     [128].   See id. at 766–67.

     [129].   See id. at 759.

     [130].   Fontana and Braman did find some slight differences with respect to voter turnout following a decision by the Court versus one by Congress: “Regardless of whether the Court reached a conservative or liberal outcome, when the Court issued a major decision conservatives were slightly more likely to vote and liberals were slightly less likely to vote than if Congress had acted. However, these differences in turnout were not major.” Braman & Fontana, Supreme Anxiety, supra note 36.

     [131].   See Fontana & Braman, Judicial Backlash, supra note 36, at 765–66 (discussing “contingent institutional preferences”).

     [132].   Id. at 771. Fontana and Braman define “cultural priors” as “prior worldviews (‘priors’) that might be motivating one’s institutional preferences.” Id. at 734–35.

     [133].   See Benjamin G. Bishin et al., Testing Backlash: The Influence of Political Institutions on Public Attitudes Toward Gay Rights 23–24 (May 5, 2013) (unpublished manuscript), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2279675.

     [134].   Furthermore, even if one posits that Supreme Court Justices act rationally to try to promote their own policy preferences (and, in so doing, may try to constrain backlash against their policy views), see Lee Epstein & Jack Knight, The Choices Justices Make 11 (1998), Justices face considerable complexity in acting strategically. They must balance the degree to which a particular decision promotes their policy goals as well as its effects on institutional legitimacy and the likelihood that it will trigger a reaction by Congress. See id. at 11–14.

     [135].   Fontana & Braman, Judicial Backlash, supra note 36, at 782.

     [136].   See generally Egan & Citrin, supra note 112 (finding that the “persuasive powers” of the Supreme Court are limited).

     [137].   Id. at 10–11.

     [138].   Id. at 10.

     [139].   Id. at 11.

     [140].   See id. at 14.

     [141].   Id. at 22.

     [142].   See id. at 15.

     [143].   Id. at 23.

     [144].   Id. at 15.

     [145].   Id. at 2, 15–16; see also Nathaniel Persily, Patrick Egan & Kevin Wallsten, Gay Marriage, Public Opinion and the Courts 17 (Apr. 29, 2006) (unpublished manuscript), available at http:///www.law.upenn.edu/fac/npersily/gaymarriagessrn.pdf (“Before Lawrence only 36 percent [of people surveyed] thought homosexual relations should be illegal; afterwards, that figure rose to 41 percent.”). Based on public polling, Persily and his colleagues show that the upsurge in hostility against same-sex sexual relations that occurred immediately after Lawrence was decided in 2003 eventually subsided in 2005. See id. at 1 (stating that the post-Lawrence backlash “appears to have leveled off and even returned to pre-Lawrence levels by the summer of 2005”).

     [146].   See generally Kerrigan v. Comm’r of Pub. Health, 957 A.2d 407 (Conn. 2008) (applying heightened scrutiny to sexual orientation classifications); Varnum v. Brien, 763 N.W.2d 862 (Iowa 2009) (applying heightened scrutiny to sexual orientation classifications). Alternatively, one could assert that the rise in public support emanated from the liberty rationale espoused in Lawrence, but that would require some explanation for the evident lag between that decision and rising support for gay rights.

     [147].   See, e.g., Windsor v. United States, 699 F.3d 169, 179 (2d Cir. 2012) (applying heightened scrutiny to sexual orientation classifications), aff’d, 133 S. Ct. 2675 (2013); Pedersen v. Office of Pers. Mgmt., 881 F. Supp. 2d 294, 326 (D. Conn. 2012) (applying heightened scrutiny to sexual orientation classifications); Golinski v. U.S. Office of Pers. Mgmt., 824 F. Supp. 2d 968, 985–90 (N.D. Cal. 2012) (applying heightened scrutiny to sexual orientation classifications).

     [148].   With respect to employment, a recent poll by the Human Rights Campaign (“HRC”) found that 73% of 2012 likely voters supported a federal law banning sexual orientation discrimination in employment; 60% of self-identified conservative voters supported such a ban. See Americans Overwhelmingly Support Executive Action to Ban Anti-LGBT Workplace Discrimination, Hum. Rts. Campaign, http://www.hrc.org/

resources/entry/americans-overwhelmingly-support-executive-action-to-ban-anti-lgbt-workplac (last visited Sept. 14, 2014). With respect to marriage, in 2009, 40% of Americans supported same-sex marriage. See Jeffrey M. Jones, Majority of Americans Continue to Oppose Gay Marriage, Gallup (May 27, 2009), http://www.gallup.com/poll/

118378/Majority-Americans-Continue-Oppose-Gay-Marriage.aspx. As of July 2013, that number jumped twelve points. See Lydia Saad, In U.S., 52% Back Law to Legalize Gay Marriage in 50 States, Gallup (July 29, 2013), http://www.gallup.com/poll/163730/back-law-legalize-gay-marriage-states.aspx. With respect to the military, 75% of Americans supported the repeal of “Don’t Ask, Don’t Tell” before its demise in 2011. See Brian J. McCabe, Public Opinion on ‘Don’t Ask, Don’t Tell, N.Y. Times (Nov. 30, 2010), http://fivethirtyeight.blogs.nytimes.com/2010/11/30/public-opinion-on-dont-ask-dont-tell/?_r=0. Professor Katie Eyer also summarizes some of these developments. See Katie R. Eyer, Marriage This Term: On Liberty and the “New Equal Protection, 60 UCLA L. Rev. Disc. 2, 9 (2012). She argues that “these developments suggest a major transformation in the minds of a majority of Americans regarding the social meaning of LGB status.” Id.

     [149].   See Zink et al., supra note 7, at 913. Professor Zink and his co-authors explore whether survey respondents, controlling for ideology, responded to decisional attributes such as the size of a Supreme Court’s majority coalition and whether the decision follows or overrules precedent. Id. The study exposed respondents to a brief newspaper article describing a fictional Supreme Court decision with the characteristics in question and then measured respondents’ agreement with and willingness to accept a decision. See id.

     [150].   See Fontana & Braman, Judicial Backlash, supra note 36, at 754.

     [151].   Hoekstra, supra note 113, at 25. Hoekstra uses panel data to evaluate whether Court decisions affect public support for the court. Id. at 3–4, 26–28.

     [152].   Survey Sampling Int’l, Inc., http://www.surveysampling.com (last visited Sept. 10, 2014).

     [153].   See N.J. Schweitzer et al., Neuroimages as Evidence in a Mens Rea Defense: No Impact, 17 Psychol. Pub. Pol’y & L. 357, 366–67 (2011).

     [154].   This was also the median level of education for Fontana & Braman. See Fontana & Braman, Judicial Backlash, supra note 36, at 751.

     [155].   Our sample had a slightly higher level of male participation than the Fontana & Braman study. See id. at 751.

     [156].   This is several years younger than the Fontana & Braman study. See id.

     [157].   This is one step lower than the level of income in the Fontana & Braman study. See id.

     [158].   Sarah Dutton et al., Poll: Slim Majority Backs Same-sex Marriage, CBS News (June 6, 2013, 11:46 PM), http://www.cbsnews.com/8301-250_162-57588131/poll-slim-majority-backs-same-sex-marriage/ (51% support legalization of same-sex marriage); Bill Mears & Paul Steinhauser, CNN Poll: Judging the Supreme Court, CNN News (June 20, 2013), http://politicalticker.blogs.cnn.com/2013/06/20/cnn-poll-judging-the-supreme-court/ (55% of the public supports same-sex marriage).

     [159].   Marjorie Connelly, Support for Gay Marriage Growing, but U.S. Remains Divided, N.Y. Times (Dec. 7, 2012), http://www.nytimes.com/2012/12/08/us/justices-consider-same-sex-marriage-cases-for-docket.html?_r=0.

     [160].   Updated: Change in Internet Access by Age Group, 2000–2010, Pew Internet & Am. Life Project (Sept. 10, 2010), http://www.pewinternet.org/Infographics/2010/Internet-acess-by-age-group-over-time-Update.aspx.

     [161].   One can join SSI’s panel by visiting the site https://www.opinionworld.com/. A person without internet access, obviously, would not be able to do so.

     [162].   None chose to do so.

     [163].   United States v. Windsor, 133 S. Ct. 2675 (2013).

     [164].   Hollingsworth v. Perry, 133 S. Ct. 2652 (2013).

     [165].   See Zink et al., supra note 7, at 913.

     [166].   Although Fontana & Braman included questions asking respondents to provide their views on what the Constitution says about same-sex marriage, Fontana & Braman, Judicial Backlash, supra note 36, at 795, we omitted those questions to avoid mixing questions involving panelist opinions about same-sex marriage and questions about panelist views on the meaning of the Constitution.

     [167].   An entirely “correct” set of answers for a panelist receiving the equality stimulus would be true/false/true/false/false.

     [168].   Zink et al., supra note 7, at 913.

     [169].   Fontana & Braman use a six-point scale in the “Policy Preference Module,” which includes “slight agree” and “moderately agree,” see Fontana & Braman, Judicial Backlash, supra note 36, at 796, a four-factor scale in their “Salience Module,” id. at 795, and a six-point measure for their “Institutional Preference” dependent variable module, see id. at 796. To facilitate consistency, we employ the four-response scale throughout.

     [170].   Zink et al., supra note 7, at 913.

     [171].   What Is the Right Scale for Agree-Disagree Items on a Survey?, Managing with Measures Blog (Feb. 28, 2011, 4:01 PM), http://www.managingwithmeasures.com/what-is-the-right-scale-for-agree-disagree-items-on-a-survey/.

     [172].   Robert K. Fleck & F. Andrew Hanssen, When Voice Fails: Potential Exit As a Constraint on Government Quality, 35 Int’l Rev. L. & Econ. 1, 32 n.42 (2013).

     [173].   This represents a slight rephrasing of a dependent variable used in the Zink study. See Zink et al., supra note 7, at 913.

     [174].   A pseudo-R-squared of between .2 and .4 is considered “highly satisfactory.” Andrew Ainsworth, Logistic Regression, Presentation at California State University at Northridge, available at https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=

web&cd=1&cad=rja&ved=0CCwQFjAA&url=http%3A%2F%2Fwww.csun.edu%2F~ata20315%2Fpsy524%2Fdocs%2FPsy524%2520lecture%252019%2520logistic_cont.ppt&ei=r7vFUuK9N4mMyQG-w4HYDg&usg=AFQjCNEYCv3Basg6ywl0R9CBv7MRhIEMiA&

sig2=gkBMwAo72Gs-JF1rwrNlfQ&bvm=bv.58187178,d.aWc.

     [175].   First, subjects were asked to give their level of agreement (again, using a four-point scale) to the following statement: Permitting gays and lesbians to marry will allow more Americans to enter into and benefit from loving and committed relationships. This variable was used in the Fontana & Braman study. Fontana & Braman, Judicial Backlash, supra note 36, at 70. A subject with a pro-same-sex marriage view would answer “agree” or “strongly agree,” while a subject with an anti-same-sex marriage view would answer “strongly disagree” or “disagree.” One would expect the sign on the coefficient on “disagree” dummies to be negative (disagreeing that same-sex marriage is beneficial would likely be correlated with a more negative reaction to the Court’s decision), and the coefficient on “agree” dummies to be positive.

As Table 2 reveals, the coefficients on all three dummy variables for the “loving and committed relationships” statement included in this specification were statistically significant at the .05-level (the “disagree” response becomes the reference point through exclusion from the specification).

The second prior-view question asked respondents to give their level of agreement (again, using a four-point scale) to the following statement: Allowing gays and lesbians to marry will undermine traditional marriage and American families. Again, this variable is derived from the Fontana & Braman study. Fontana & Braman, Judicial Backlash, supra note 36, at 70.

A pro-same-sex marriage respondent would be expected to disagree with this statement, while an anti-same-sex marriage respondent would be expected to agree with this statement. Here, we obtain statistical significance on the two “agree” dummy variables and the sign is, as one would expect, negative (agreeing that same-sex marriage would undermine traditional marriage would be expected to correlate with disagreement with a Court decision legalizing same-sex marriage).

By contrast, the coefficient on the dummy for a “strongly disagree” response is not statistically significant. This could be interpreted to mean that it does not matter whether a respondent “disagrees” or “strongly disagrees” on the “undermine marriage” statement, since the omitted “disagree” response serves here as the referent.

Our third prior-view question asked for respondents to indicate their level of agreement with the following statement: Marriage should be defined as the union of one man and one woman. We omitted this question from our regression models to avoid multi-collinearity. Tests for multi-collinearity are available upon request.

Our fourth prior-view question asked respondents to give their level of agreement (again, using a four-point scale) to the following statement: Gay and lesbian couples should be allowed to marry in the same manner as heterosexual couples.

Again, since we use “disagree” as the referent, we would expect positive coefficients on the “agree” and “strongly agree” response dummy variables, and a negative coefficient on the “strongly disagree” dummy variable. As expected, the signs on agree and “strongly agree” dummies are positive (and the coefficients increase), and the sign on “strongly disagree” is negative; the coefficients are also statistically significant at the .05 level. The more a person agrees that gay couples should be permitted to marry, the more positive her response is to a Court decision requiring states and the federal government to allow same-sex marriage.

     [176].   Similarly, in this model we did not find that age had a statistically significant effect on agreement with the Court’s decision, controlling for the “prior view” variables.

     [177].   This is a slight rephrasing of a variable described as “the most commonly used measure in the literature.” See Zink et al., supra note 7, at 913 (citing James L. Gibson, Gregory A. Caldeira & Lester Kenyatta Spence, Why Do People Accept Public Policies They Oppose? Testing Legitimacy Theory with a Survey-Based Experiment, 58 Pol. Res. Q. 187, 190–91 (2005)).

     [178].   Dropping the subjects viewing the “neutral” stimulus from the pool, we do not find a statistically significant difference in regard to either acceptance or agreement with the Court’s decision based on viewing the equality, rather than the liberty stimulus. That is, while there is a negative effect of EQUALITY on acceptance in the primary regressions, and no effect of LIBERTY, a comparison of only LIBERTY and EQUALITY does not produce a statistically significant difference. Full regressions are available from authors upon request.

     [179].   The -.30 coefficient suggests that, for subjects reading the EQUALITY STIMULUS, controlling for other variables, they would move less than a third of the way towards a “lower” level of agreement (that is, less than a third of a way from, say, “strongly agree” to “agree”). The 95% confidence interval for this coefficient is between -.54 and -.06, suggesting that it is possible that the actual effect of reading the EQUALITY STIMULUS may be even lower.

     [180].   Odds-ratio calculations are available from the authors upon request.

     [181].   Similarly, if we drop the subjects receiving the EQUALITY STIMULUS from our model and compare only the subjects receiving the LIBERTY STIMULUS to those receiving the NEUTRAL STIMULUS, we get a positive but not statistically significant coefficient on the LIBERTY STIMULUS dummy. Full regressions are available from authors upon request. In other words, compared to the NEUTRAL STIMULUS, the LIBERTY STIMULUS dummy appears to have no discernible impact on subject acceptance.

     [182].   Similarly, we continue to find mixed results for the question relating to whether permitting same-sex marriage would undermine traditional families.

     [183].   Interestingly, in this specification, for the first time, all of the response dummies for the question relating to whether legalizing same-sex marriage would undermine traditional marriage were statistically significant. Similarly, for the first time we find statistically significant effects for those who “strongly agree” that they closely follow the issue and that it is important to them personally—with strong followers and those feeling the issue is personally important statistically more likely to support a constitutional amendment. We also, for the first time, find that age has a statistically significant (though low magnitude) effect on our dependent variable—the older the person, the lower his support for a constitutional amendment banning same-sex marriage (again, though, the coefficient is quite small).

     [184].   In retrospect, it would have been better to present the Comprehension module as a single question, from which respondents selected a single “right” answer, rather than as three separate “True/False” questions. It is also possible that our efforts to describe in understandable yet specific terms whether the reasoning was based on equality or liberty remained confusing for panelists. While we labored to construct clear questions in this area, we may not have gotten to the point we had hoped.

     [185].   For those receiving the LIBERTY STIMULUS, we treated their comprehension response as “correct” if they: (1) correctly responded that the Court’s decision requires governments to allow same-sex marriage and does not leave the question open to the states; (2) answered “True” when asked whether the Court used a liberty-based rationale; and (3) answered “False” to the question of whether the story omitted any discussion of the Court’s explanation. Similarly, for those receiving the EQUALITY STIMULUS, we treated as “correct” accurate answers to the two questions relating to the impact of the Court’s decision, and a “True” response to the question regarding whether the rationale was equality-based accompanied by a “False” response to the no-rationale question. In other words, we counted as “correct” a portion of responses that answered “True” to both our LIBERTY and EQUALITY comprehension question, so long as they did not also answer “True” to our no-rationale question.

     [186].   The full results from this regression are available from the authors upon request.

     [187].   Our subjects demonstrated what appears to be a low level of comprehension, or at least they were unable to evidence that comprehension in their survey responses; yet their responses did appear to be influenced by the nature of the stimulus they had been administered.

     [188].   Full regression results are available from the authors upon request.

     [189].   We find statistically significant negative coefficients on both terms in the ordered logit specification.

     [190].   Regression results are available from the authors upon request.

     [191].   Egan & Citrin, supra note 112, at 5 (“[I]n reality a Supreme Court ruling is just the opening salvo of a debate among the nation’s elites that can quickly overwhelm any of the persuasive power of the ruling in a case.”).

     [192].   Full results are available from the authors upon request.

     [193].   The coefficient on the EQUALITY STIMULUS here was negative, with a p-value of .052, which is just outside of the range of statistical significance.

     [194].   It bears mentioning that while we find that equality rationales have a slight negative effect on acceptance, we do not observe a greater level of support for a constitutional amendment based on exposure to an equality rationale. The respondents are simultaneously reporting that they are less likely to accept the Court’s equality-based decision as the final word on the matter, but are no more likely to want to amend the Constitution. It may be that the public views constitutional amendments as a particularly harsh response. Instead of an amendment overturning a Court decision, those less inclined to accept the decision may prefer that legislation or a subsequent Court decision weaken a same-sex marriage precedent at the margins.

     [195].   Gibson et al., supra note 177, at 192.

     [196].   See infra app. I.

     [197].   In this sense, our results stand apart from previous work, which has found a stronger relationship between signals from the Court and public agreement than between such signals and public acceptance. See Zink et al., supra note 7, at 922. Our study finds the opposite in the case of Court signals relating to the reasons for decisions––those articulated reasons have a greater effect in our study on acceptance than on agreement. See supra tbls. 2–4. Perhaps there is something special about same-sex marriage. Or perhaps there is something distinguishing our treatment (the Court’s reasoning) from the treatment employed in prior research (such as whether the decision overrules precedent and is unanimous).

     [198].   The negative effect of the liberty rationale, while not statistically significant at the .05-level, was statistically significant at the more generous .10-level.

     [199].   See Liza Vertinsky, Comparing Alternative Institutional Paths to Patent Reform, 61 Ala. L. Rev. 501, 532 (2010) (“Courts must give reasons for their decisions, and their decisions must reasonably relate to the specific case they are deciding.”). That the public does not believe that this is so is supported by public opinion research, which has found that a near majority (47%) of the public either believes that the Court does not regularly give reasons for its rulings (18%) or does not know (29%). See Michael Serota, Popular Constitutional Interpretation, 44 Conn. L. Rev. 1637, 1659 (2012).

     [200].   Future work could explore this possibility by inserting an additional stimulus involving an implausible line of reasoning (e.g., justifying a same-sex marriage decision by reference to the bankruptcy code).

     [201].   See Kang, Implicit Bias, supra note 88, at 1170.

     [202].   See Yoel Inbar et al., Disgust Sensitivity Predicts Intuitive Disapproval of Gays, 9 Emotion 435, 438 (2009).

     [203].   For a description of the Implicit Association Test, see Anthony G. Greenwald et al., Measuring Individual Differences in Implicit Cognition: The Implicit Association Test, 74 J. Personality & Soc. Psychol. 1464, 1464–66 (1998) (introducing and describing the Implicit Association Test (“IAT”)); see also Brian A. Nosek, Anthony G. Greenwald & Mahzarin R. Banaji, The Implicit Association Test at Age 7: A Methodological and Conceptual Review, in Social Psychology and the Unconscious: The Automaticity of Higher Mental Processes 265, 267–71 (John A. Bargh ed., 2007).

     [204].   If that is right—that is, if it is true that our findings on the slightly negative effects of equality on acceptance are capturing the well-documented phenomenon of implicit bias against homosexuality—then what Yoshino calls “pluralism anxiety” might actually be the product of an implicit bias against equality-related concepts.

     [205].   See supra note 190 and accompanying text.

     [206].   Fontana & Braman, Judicial Backlash, supra note 36, at 756.

     [207].   See Ilya Somin, The Limits of Backlash: Assessing the Political Response to Kelo, 93 Minn. L. Rev. 2100, 2159–63 (2012) (describing failure of survey data to accurately capture backlash as a phenomenon).

     [208].   Among the many factors—aside from Supreme Court reasoning—that might generate backlash are: “public opinion on the underlying issue; the relative intensity of preference on the two sides of the issue; the degree to which public opinion is divided along geographic or regional lines; and the ease with which a particular Court ruling can be circumvented or defied.” Klarman, supra note 5, at 148.

     [209].   See supra notes 178–81 and accompanying text.

     [210].   Fontana & Braman, Judicial Backlash, supra note 36, at 773.

     [211].   Id.

     [212].   Hoekstra, supra note 113, at 3.

     [213].   See Yoshino, The New Equal Protection, supra note 12, at 747, 802 (Yoshino notes that “constitutional equality and liberty claims are often intertwined” as so-called “dignity claims” but advocates for “liberty-based dignity jurisprudence [because it] leads with [liberty] to quiet pluralism anxiety in an increasingly diverse society”).

     [214].   Id. at 792 (“[W]here a claim can be validly characterized as either a liberty-based or an equality-based dignity claim, it should be characterized as the former.”); id. at 793–97 (discussing his “new equal protection” theory in the context of “civil rights” more generally).

     [215].   See Heather K. Gerken, Larry and Lawrence, 42 Tulsa L. Rev. 843, 851 (2007) (“[W]hat is really at stake in these debates [over sexual orientation discrimination] is not whether all humans should enjoy a right, but whether gays and lesbians, in particular, should do so, and that idea is better captured by the equal protection paradigm.”).

     [216].   Yoshino anticipates four objections to his theory in his Harvard Law Review article. One is that we ought to encourage people to overcome their anxieties rather than capitulate to them. Yoshino, The New Equal Protection, supra note 12, at 797. A second is that “with respect to certain groups, we should be careful about jumping to a higher level of generality” through, say, liberty-based rationales. Id. at 798. A third is that “liberty-based dignity claims allow subordinated groups to contest their subordination only in a piecemeal fashion.” Id. at 799. A fourth is that the “liberty-based dignity paradigm . . . is a false rescue because it substitutes one slippery-slope claim for another.” Id. More specifically, the liberty claim replaces the question of which groups ought to receive protection with the question of which rights ought to receive protection. Id. at 800. Yoshino responds to each of these objections in turn. Professor Katie Eyer has recently articulated a different normative objection to Yoshino’s theory. See Eyer, supra note 148, at 6. She argues that “there are strong signs that the [LGB] movement has finally ‘arrived,’ rendering it plausible for the first time that the Court may extend full equal protection coverage.” Id. Professor Eyer further contends that “there are reasons to believe that Yoshino’s suggested focus [on liberty over equality] might actually be counterproductive for groups that have reached this equal protection tipping point.” Id.

     [217].   Klarman, supra note 5, at 158 (“Within a few years of Windsor, as public support for gay marriage continues to increase and as more states enact it into law, one can imagine some Justices being tempted to extend that ruling to forbid the states from excluding same-sex couples from marriage. Indeed, the Windsor dissents of both Chief Justice Roberts and Justice Scalia seem mostly addressed to that eventuality . . . .”).

     [218].   Yoshino, Covering: The Hidden Assault, supra note 48, at ix; Yoshino, Covering, supra note 49, at 794.

     [219].   Yoshino, The New Equal Protection, supra note 12, at 793.

     [220].   Id.

     [221].   Martha C. Nussbaum, From Disgust to Humanity 205–06 (2010) (suggesting that cultivating empathy—the ability to place yourself in another’s shoes—is a promising way to move past disgust and toward humanity for gays and lesbians).

     [222].   Yoshino, The New Equal Protection, supra note 12, at 794.

     [223].   Yoshino, Covering: The Hidden Assault, supra note 48, at ix.

     [224].   Implicit Association Test, Project Implicit, https://implicit.harvard.edu/implicit/ (last visited Sept. 3, 2014). The word “homosexual” apparently triggers even more implicit bias than the more politically correct “gay” and “lesbian.” According to a New York Times poll conducted before the military’s exclusionary policy was lifted, 42% of Americans opposed allowing “homosexuals” to serve openly in the military, whereas only 28% opposed allowing “gay men and lesbians” to serve openly. See Geoffrey R. Stone, Gays and Lesbians or Homosexuals?, Huff. Post (Apr. 15, 2010, 5:12 AM), http://www.huffingtonpost.com/geoffrey-r-stone/gays-and-lesbians-or-homo_b_

461668.html. Indeed, the results of the New York Times poll helps explain why the only opinion to mention the word “homosexuals” and its variants in the Supreme Court’s most recent landmark gay rights case, United States v. Windsor, is that of Justice Scalia—in dissent. See, e.g., United States v. Windsor, 133 S. Ct. 2675, 2709 (2013) (Scalia, J., dissenting).

     [225].   With respect to the sex-discrimination argument, Edward Stein contends that it “mischaracterizes the core wrong of these laws [that discriminate against gays and lesbians],” which are motivated, in his view, not by sexism but rather by a particular moral view of gays and lesbians. Edward Stein, Evaluating the Sex Discrimination Argument for Lesbian and Gay Rights, 49 UCLA L. Rev. 471, 503–04 (2001). “[B]y failing to address arguments about the morality of same-sex sexual acts and the moral character of lesbians, gay men, and bisexuals,” he argues, “the sex-discrimination argument ‘closets,’ rather than confronts, homophobia.” Id. With respect to the race-discrimination argument, one of us contends that it ignores the centuries-old history of discrimination that such laws embody—discrimination that is unique to gays and lesbians. See Courtney Megan Cahill, (Still) Not Fit to Be Named: Moving Beyond Race to Explain Why ‘Separate’ Nomenclature for Gay and Straight Relationships Will Never Be ‘Equal, 97 Geo. L.J. 1155, 1159–62 (2009).

     [226].   William N. Eskridge, Jr., Gaylaw: Challenging the Apartheid of the Closet 223–24 (2009).

     [227].   Stein, supra note 225, at 504.

     [228].   Id.

     [229].   Sodomy—the behavioral metonym of gay identity—was long known as the “crime that shall not be named.” 4 William Blackstone, Commentaries *215. For commentary on sodomy as the metonym of gay identity, see Janet E. Halley, Reasoning About Sodomy: Act and Identity in and After Bowers v. Hardwick, 79 Va. L. Rev. 1721, 1722 (1993). As recently as 1986, the Supreme Court invoked that trope to refer to what it understood to be gay sex, see Bowers v. Hardwick, 478 U.S. 186, 197 (1986) (Burger, C.J., concurring), a description of which the Court relegated to a place “named after the foot, that lowly organ which spends its life near the ground, in the dirt,” J.M. Balkin, The Footnote, 83 Nw. U. L. Rev. 273, 275 (1989). And even though Lawrence v. Texas overruled Bowers and brought the definition of sodomy from footnote to text, see Lawrence v. Texas, 539 U.S. 558, 578 (2003), it, too, participated in that tradition of un-naming by discussing the “conduct,” but even more so the “right” at issue in that case, in such an oblique and abstract way that it is often difficult to tell what, exactly, the Lawrence majority is saying, see Heather Gerken, supra note 215, at 847 (stating that the opening of the majority opinion in Lawrence is “lovely” but then asking: “But what on earth does it mean?”); Mary Anne Case, Of “This” and “That” in Lawrence v. Texas, 2003 Sup. Ct. Rev. 75, 75–77; Tribe, supra note 105, at 1898–99.

     [230].   476 U.S. at 196–97.

     [231].   539 U.S. at 578–79; see also Eyer, supra note 148, at 8 (arguing that Lawrence at most represents a jurisprudence of “tolerance”—not one of full “recognition”).

     [232].   133 S. Ct. 2675 (2013).

     [233].   But see NeJaime, supra note 108, at 220 (arguing that while the Windsor majority technically struck down section 3 on equal protection grounds, it is “conceptually, if not doctrinally, a right-to-marry case”).

     [234].   To be sure, in invoking the “equal dignity of same-sex marriages,” the Windsor Court adverted to a concept with close associations to liberty: dignity. According to Laurence Tribe, “dignity” is the concept that unites the two interlocking strands of the Fourteenth Amendment: equality and liberty. See Tribe, supra note 105, at 1898. Yoshino refers to hybrid liberty and equality claims as “dignity” claims, thus suggesting that he would interpret Windsor as resting its holding, at least in part, on liberty. See Yoshino, The New Equal Protection, supra note 12, at 749. That said, Yoshino makes a distinction between “liberty-based” and “equality-based” dignity claims, depending “on whether the liberty or the equality dimension of the hybrid claim is ascendant.” Id. Given the Windsor majority’s repeated references to equality and “equal dignity,” as well as its reliance on canonical equal protection precedents like Moreno and Romer, it seems safe to say that it is one such case where equality was in the ascendant. See Windsor, 133 S. Ct. at 2681–93.

     [235].   Windsor, 133 S. Ct. at 2683.

     [236].   See Noa Ben-Asher, Conferring Dignity: The Metamorphosis of the Legal Homosexual, 37 Harv. J. L. & Gender 243, 245 (2014). Ben-Asher further argues that “Windsor introduces us to the new legal homosexual: the ‘same-sex couple.’ ” Id. She later refers to this rhetorical strategy as an erasure of “the terms ‘homosexual’ [and] ‘lesbian’ ” and ultimately homosexual identity. Id. at 263.

     [237].   Brief on the Merits for Respondent Edith Schlain Windsor at 17, Windsor, 133 S. Ct. 2675 (No. 12-307).

     [238].   Windsor v. United States, 699 F.3d 169, 185–88 (2d Cir. 2012) (finding that sexual minorities constituted a quasi-suspect class under the federal Equal Protection Clause and striking down section 3 under heightened judicial scrutiny), aff’d, 133 S. Ct. 2675 (2013).

     [239].   Windsor, 133 S. Ct. at 2693 (quoting H.R. Rep. No. 104-664, at 16 (1996)).

     [240].   See, e.g., Pedersen v. Office of Pers. Mgmt., 881 F. Supp. 2d 294, 334 (D. Conn. 2012) (finding that “homosexuals warrant judicial recognition as a suspect classification” although striking down DOMA under rational basis review only); Golinski v. U.S. Office of Pers. Mgmt., 824 F. Supp. 2d 968, 985–90 (N.D. Cal. 2012); Varnum v. Brien, 763 N.W.2d 862, 885–96 (Iowa 2009).

     [241].   133 S. Ct. at 2694.

     [242].   In this sense, the Windsor Court embraces an “entity” rather than “associational” view of marriage, similar to the approach currently taken in the law of partnership. See Unif. P’ship Act § 201 (amended 1997), 6 U.L.A. 91 (2001).

     [243].   See, e.g., Douglas NeJaime, Marriage Inequality: Same-Sex Relationships, Religious Exemptions, and the Production of Sexual Orientation Discrimination, 100 Calif. L. Rev. 1169, 1196 (2012) (making this argument).

     [244].   See, e.g., Christian Legal Soc’y Chapter of Univ. of Cal., Hastings Coll. of the Law v. Martinez, 561 U.S. 661, 689 (2010) (“declin[ing] to distinguish between status and conduct in this context [of sexual orientation]”); Lawrence v. Texas, 539 U.S. 558, 575 (2003) (finding that discrimination on the basis of “homosexual conduct” constitutes discrimination against gays and lesbians as a class); Perry v. Brown, 671 F.3d 1052, 1081 (9th Cir. 2012) (finding that “marriage” discrimination constitutes discrimination against gays and lesbians as a class), vacated sub nom. Hollingsworth v. Perry, 133 S. Ct. 2652, 2659 (2013); Bishop v. United States ex. rel. Holder, 962 F. Supp. 2d 1252, 1287 (N.D. Okla. 2014) (stating that discrimination on the basis of a “same-sex marriage” amounts to discrimination on the basis of sexual orientation because the conduct, same-sex marriage, “is so closely correlated with being homosexual that sexual orientation provides the best descriptor for the class-based distinction being drawn”).

     [245].   133 S. Ct. at 2695.

     [246].   Id.

     [247].   Id.

     [248].   Stein, supra note 225, at 505.

     [249].   See Yoshino, The New Equal Protection, supra note 12, at 793–94 (discussing the ramifications of framing the marriage equality claim as an equality right for gays and lesbians or as a liberty right that is universally enjoyed).

     [250].   Ben-Asher, supra note 236, at 284.

     [251].   We are reminded of an observation made by Michael Sandel nearly twenty-five years ago, also in the context of gay rights. Sandel argued then that the “neutral case for toleration [of homosexuality]” is undesirable because it “leaves wholly unchallenged the adverse views of homosexuality itself.” Michael Sandel, Moral Argument and Liberal Toleration: Abortion and Homosexuality, 77 Calif. L. Rev. 521, 537 (1989). “Unless those views can be plausibly addressed,” he maintained,

even a Court ruling in their favor is unlikely to win for homosexuals more than a thin and fragile toleration. A fuller respect would require, if not admiration, at least some appreciation of the lives homosexuals live. Such appreciation, however, is unlikely to be cultivated by a legal and political discourse conducted in terms of autonomy rights alone.

Id. In this sense, liberty arguments reflect what Eskridge has called a politics of “toleration” rather than one of “recognition.” See William Eskridge, Jr., Lawrence’s Jurisprudence of Tolerance: Judicial Review to Lower the Stakes of Identity Politics, 88 Minn. L. Rev. 1021, 1086 (2004).

     [252].   See Letter from Eric H. Holder, Jr., Att’y Gen., to Hon. John A. Boehner, Speaker, U.S. House of Representatives (Feb. 23, 2011), available at http://www.justice.gov/opa/pr/2011/February/11-ag-223.html.

     [253].   See Windsor v. United States, 699 F.3d 169, 188 (2d Cir. 2012) (striking down section 3 of DOMA under the Equal Protection Clause of the U.S. Constitution), aff’d, 133 S. Ct. 2675 (2013); Pedersen v. Office of Pers. Mgmt., 881 F. Supp. 2d 294, 347 (D. Conn. 2012) (striking down section 3 of DOMA under the Equal Protection Clause); Golinski v. U.S. Office of Pers. Mgmt., 824 F. Supp. 2d 968, 1002 (N.D. Cal. 2012) (striking down section 3 of DOMA under the Equal Protection Clause).

     [254].   See Yoshino, The New Equal Protection, supra note 12, at 796 (arguing that “[e]quality claims inevitably involve the Court in picking favorites among groups, a practice attended by pluralism anxiety” whereas “[l]iberty claims, in contrast, emphasize what all Americans (or, more precisely, all persons within the jurisdiction of the United States) have in common”).

     [255].   See Kang, Implicit Bias, supra note 88, at 1126 (stating in 2012 that “[g]iven the substantial and growing scientific literature on implicit bias, the time has now come to confront a critical question: What, if anything, should we do about implicit bias in the courtroom?”); John Powell & Rachel Godsil, Implicit Bias Insights As Preconditions to Structural Change, Poverty & Race, Sept.–Oct. 2011, at 3.

     [256].   See infra notes 259–68 and accompanying text.

     [257].   See Samuel R. Sommers & Phoebe C. Ellsworth, Race in the Courtroom: Perceptions of Guilt and Dispositional Attributions, 26 Personality & Soc. Psychol. Bull. 1367, 1374 (2000); Samuel R. Sommers & Phoebe C. Ellsworth, “Race Salience” in Juror Decision-Making: Misconceptions, Clarifications, and Unanswered Questions, 27 Behav. Sci. & L. 599, 608 (2009); Samuel R. Sommers & Phoebe C. Ellsworth, White Juror Bias: An Investigation of Prejudice Against Black Defendants in the American Courtroom, 7 Psychol. Pub. Pol’y & L. 201, 218 (2001) [hereinafter Sommers & Ellsworth, White Juror Bias].

     [258].   See Sommers & Ellsworth, White Juror Bias, supra note 257, at 225.

     [259].   Id.

     [260].   Kang, Implicit Bias, supra note 88, at 1185 (summarizing this research).

     [261].   Id. (observing that “perspective-taking interventions substantially decreased implicit bias in the form of negative attitudes, as measured by both a variant of the standard IAT (the personalized IAT) and the standard race attitude IAT”).

     [262].   Id. at 1184.

     [263].   Id.

     [264].   Id. at 1185.

     [265].   Terry A. Maroney, Emotional Regulation and Judicial Behavior, 99 Calif. L. Rev. 1485, 1492 (2011).

     [266].   Nilanjana Dasgupta & Luis M. Rivera, From Automatic Antigay Prejudice to Behavior: The Moderating Role of Conscious Beliefs About Gender and Behavioral Control, 91 J. Personality & Soc. Psychol. 268, 277 (2006).

     [267].   If colorblindness can perpetuate (rather than alleviate) racial bias, as scholars have found, then it is at the very least conceivable that decisions committed to “social category blindness” do not just fail to alleviate bias but even perpetuate it on some level. We understand that the results of our study do not strongly support this claim. If, as we found, Supreme Court reasoning does not really affect public acceptance of a decision, then it is unlikely that Supreme Court reasoning would affect implicit bias. If people are not affected by reasons in terms of acceptance, then it is unlikely that they will be affected by reasons in terms of implicit bias. That said, we are mostly concerned here with Yoshino’s belief that the Court cares about pluralism anxiety (and, by implication, implicit bias)—both its own and that of the people. If that is right, then we believe that the Court at the very least ought to engage the best strategy to manage that bias. If implicit bias scholars are right, then the best strategy is one that engages, rather than deflects, the social categories that purportedly trigger bias.

     [268].   Marbury v. Madison, 5 U.S. 137, 177 (1803).

     [269].   See supra notes 110–44 and accompanying text.

     [270].   See Yoshino, The New Equal Protection, supra note 12, at 754 n.55.

     [271].   The formatting of the vignettes has been altered here from that presented to survey subjects, so as to facilitate publication.

DOWNLOAD PDF | 93 N.C. L. Rev.303 (2015)