For present purposes, the most important finding is that there is a dramatic difference between politically diverse panels, with judges appointed by presidents of more than one party, and politically unified panels, with judges appointed by presidents of only one party. On divided panels in which a Republican-appointed majority of the court might be expected, on broadly speaking political grounds, to be hostile to the agency, the court nonetheless deferred to the agency 62 percent of the time. But on unified panels in which an all-Republican panel might be expected to be hostile to the agency, the court upheld the agency’s interpretation only 33 percent of the time. Note that this was the only such finding in the data. When Democratic-appointed majority courts were expected to uphold the agency’s decision on political grounds, they did so more than 70 percent of the time, whether unified (71 percent of the time) or divided (86 percent of the time). Consider the results in tabular form:46
3–0 Republican panel 2–1 Republican panel 3–0 Democratic panel 2–1 Democratic panel
Uphold agency action
33%
62%
71%
86%
Invalidate agency action
67%
38%
29%
14%
It seems reasonable to speculate that the seemingly bizarre result—a 67 percent invalidation rate when Republican appointees are unified—reflects group influences and in particular group polarization. A group of all-Republican appointees might well take the relatively unusual step of rejecting an agency’s interpretation, whereas a divided panel, with a built-in check on any tendency toward the unusual or extreme outcome, is more likely to take the conventional route. A likely reason is that the single Democratic appointee acts as a “whistleblower,” discouraging the other judges from making a decision that is in tension with the Supreme Court’s command to uphold agency interpretations of ambiguous statutes.47
Other factors are probably involved. When a court consists of a panel of judges with the same basic orientation, the median view, before deliberation begins, will be significantly different from what it would be with a panel of diverse judges. The argument pool will be very different as well. For example, a panel of three Republican appointees, tentatively inclined to invalidate the action of the Environmental Protection Agency (EPA), may offer a range of arguments in support of invalidation and relatively few in the other direction—even if the law, properly interpreted, favors validation. If the panel contains a judge who is inclined to uphold the EPA, the arguments that favor validation are far more likely to emerge and be pressed. Indeed, the very fact that the judge is a Democratic appointee increases the likelihood that this will occur, if that judge does not think of himself or herself as being part of the same “group” as the other panel members. (Recall that when people are connected by ties of solidarity, disagreement is much less likely.) And because corroboration of opinion leads to greater confidence and hence extremity, it is not surprising that a panel of three like-minded judges would lead to unusual and extreme results.
In this context, the effect is fortified by the possibility that the sole judge, being outnumbered on a three-judge panel, might produce a dissenting opinion in public. To be sure, Supreme Court review is rare, and in the general run of cases, the prospect of such review probably does not have much of a deterrent effect on courts of appeals. But judges who write majority opinions are usually not enthusiastic about having to see and respond to dissenting opinions. And if the law actually favors the dissenting view, two judges, even if they would like to reverse the Environmental Protection Agency, might be influenced to adopt the easier course of validation. The evidence so suggests.48
A glance at the table above offers some countervailing data: A panel of three Democratic appointees is not more likely than a panel with two Democratic appointees to uphold agency action in cases in which Democratic appointees might be expected to want to uphold the agency. And in the context of a challenge from an environmental group, Republican appointees are not likely to vote differently if they are accompanied by two Democratic appointees, one Republican appointee, or two Republican appointees.49 But in many important domains, at least, a panel of three like-minded judges does indeed behave differently from a panel with two.50
At this point a skeptic might note that lawyers make adversarial presentations before three-judge panels. Such a skeptic might insist that the size of the “argument pool” is determined by those presentations, not only and not even mostly by what members of the panel are inclined to say and to do. And undoubtedly the inclinations of judges are shaped, much of the time, by the contributions of advocates. But even if this is so, what matters, for purposes of the outcomes, are the inclinations of judges, whatever they are based on; and it is here that the existence of a single dissenter can make all the difference. In the punitive damage study discussed above, mock juries were presented with arguments from both sides, and polarization followed this presentation, as it has elsewhere. Notice in this regard that for the polarization hypothesis to hold, it is not necessary to know whether judges spend a great deal of time offering reasons to one another. Mere exposure to a conclusion is enough.51 A system of simple votes, unaccompanied by reasons, should incline judges to polarize. Of course reasons, if they are good ones, are likely to make those votes especially persuasive.
It remains to investigate the normative issues. If like-minded judges go to extremes, should we be troubled? Is it good if a large effect is found from a single judge from a different party? More generally, is there reason to attempt to ensure diversity on the federal courts? To promote a degree of diversity on panels? Some people think that judges appointed by presidents of different political parties are not fundamentally different and that once on the bench, judges frequently surprise those who nominated them. The view is not entirely baseless, but it is misleading. Some appointees do disappoint the presidents who nominated them, but the availability heuristic should not fool us into thinking that these examples are typical. Judges appointed by Republican presidents are quite different from judges appointed by Democratic presidents. “Partisanship clearly affects how appellate courts review agency discretion.”52
But it seems difficult to evaluate the underlying issues without taking a stand on the merits—without knowing what we want judges to do. Suppose that three Republican appointees are especially likely to uphold criminal convictions and that three Democratic appointees are especially likely to reverse those convictions. At first glance, one or the other is troubling only if we know whether we approve of one or another set of results. In the punitive damage study discussed above, the movement toward increased awards might be something to celebrate, not to deplore, if we conclude that the median of predeliberation awards is too low and that the increase, produced by group discussion, ensures more sensible awards. And if a view about what judges should do is the only possible basis for evaluation, we might conclude that those who prefer judges of a particular party should seek judges of that party and that group influences are essentially beside the point.
But the conclusion is too strong. In some cases, the law, properly interpreted, really does argue strongly for one or another view. The existence of diversity on a panel is likely to bring that fact to light and perhaps to move the panel’s decision in the direction of what the law requires. The existence of politically diverse judges, and of a potential dissent, increases the probability that the law will be followed. The Chevron study, referred to above, strongly supports this point. The presence of a potential dissenter—in the form of a judge appointed by a president from another political party—creates a possible whistleblower who can reduce the likelihood of an incorrect or lawless decision. Through an appreciation of the nature of group influences, we can see the wisdom in an old idea: a decision is more likely to be right, and less likely to be political in a pejorative sense, if it is supported by judges with different predilections.
There is a further point. Suppose that in many
areas, it is not clear, in advance, whether the appointees of Democratic or Republican presidents are correct. Suppose that we are genuinely uncertain. If so, there is reason to favor a situation in which the legal system has both, simply on the ground that through that route, more (reasonable) opinions are likely to be heard. And if we are genuinely uncertain, there may be reason to favor a mix of views merely by virtue of its moderating effect.
Consider an analogy. Modern law and policy are often made by independent regulatory commissions, such as the Federal Trade Commission, the Securities and Exchange Commission, the National Labor Relations Board, and the Federal Communications Commission. Much of the time, such agencies act through adjudication. They function in the same fashion as federal courts. And under federal statutes, Congress has attempted to ensure that these agencies are not monopolized by either Democrats or Republicans. The law requires that no more than a bare majority of agency members may be from a single party.
An understanding of group influences helps explain this requirement. An independent agency that is all Democratic or all Republican might move toward an extreme position, indeed a position that is more extreme than that of the median Democrat or Republican and possibly more extreme than that of any agency official standing alone. A requirement of bipartisan membership can operate as a check against movements of this kind. Congress was apparently aware of this general point. Closely attuned to the policy-making functions of the relevant institutions, it was careful to provide a safeguard against extreme movements.
Why do we fail to create similar safeguards for courts? Part of the answer must lie in a belief that unlike heads of independent regulatory commissions, judges are not policy makers. Their duty is to follow the law, not to make policy. An attempt to ensure bipartisan composition would seem inconsistent with the commitment to this belief. But the evidence I have discussed shows that judges are policy makers of an important kind and that their political commitments very much influence their votes. I do not mean to embrace any particular policy proposal here. But in principle, there is good reason to attempt to ensure a mix of perspectives within courts of appeals.
Of course the idea of diversity, or of a mix of perspectives, is hardly self-defining. It would not be appropriate to say that the federal judges should include people who refuse to obey the Constitution, who refuse to exercise the power of judicial review, or who think the Constitution allows suppression of political dissent and racial segregation. Here as elsewhere, the domain of appropriate diversity is limited. What is necessary is reasonable diversity, or diversity of reasonable views, and not diversity as such. People can certainly disagree about what reasonable diversity entails in this context. All that I am suggesting here is that there is such a thing as reasonable diversity and that it is important to ensure that judges, no less than anyone else, are exposed to it, and not merely through the arguments of advocates.
These points cast fresh light on a much-disputed issue: the legitimate role of the Senate in giving “advice and consent” to presidential appointments to the federal judiciary. Above all, an understanding of social influences suggests that the Senate has a responsibility to exercise its constitutional authority in order to ensure a reasonable diversity of view. The Constitution’s history fully contemplates an independent role for the Senate in the selection of Supreme Court justices.53 That independent role certainly authorizes the Senate to consider the general approach, and likely pattern of votes, of potential judges. There can be no doubt that the president considers the general approach of his or her nominees; the Senate is entitled to do so as well. Under good conditions, these simultaneous powers would bring about a healthy form of checks and balances, permitting each branch to counter the other. Indeed, that system is part and parcel of social deliberation about the direction of the federal judiciary.
Why might this view be rejected? It could be argued that there is only one legitimate approach to constitutional or statutory interpretation—that, for example, some version of originalism or textualism is the only such approach and that anyone who rejects that view is unreasonable. For true believers, it is pointless to argue for diverse views. Diversity is not necessary, or even valuable, if we already know what should be done and if competing views would simply cloud the issue. (In a scientific dispute, it is not helpful to include those who believe the earth is flat.) Or it might be argued that a deferential role for the Senate, combined with natural political competition and cycles, will produce a sensible mix over time. I do not deny this possibility. My only suggestions are that a high degree of diversity on the federal judiciary is desirable, that the Senate is entitled to pursue diversity, and that without such diversity, judicial panels will inevitably go in unjustified directions.
Diversity and Affirmative Action in Higher Education
Countless educational institutions pursue the goal of diversity. Most of America’s large private and public institutions seek a wide range of views, faculty members, and students. There are some prominent exceptions; some institutions pride themselves on a high degree of homogeneity. And here as elsewhere, the idea of diversity needs to be clarified. Colleges and universities do not make special efforts to include students who collect Taylor Swift memorabilia, hate America, smell bad, or have low SATs. Such institutions are committed to diversity, but only to a certain degree and of a certain kind. It remains possible to argue, as many do, that they give excessive attention to diversity of some kinds and insufficient attention to diversity of other kinds. The only point I am making here is that they tend to be committed to diversity of a recognizable sort.
Some people think that the pursuit of diversity is a big mistake, at least for colleges and universities. They believe that only one factor matters: merit. To be sure, the idea of merit could be defined in many different ways, but according to the view I am considering, it refers to academic potential, measured principally by reference to grades and scores on standardized tests. We could qualify that view by suggesting that if people have grown up poor, in challenging family circumstances, or with a disability, their academic potential might not be adequately measured by grades and test scores. Fair enough. But there is no question that many colleges and universities are promoting a variety of goals that the word “merit” does not capture.
For example, a preference for children of alumni is most easily defended in purely economic terms. If institutions admit such children (“legacies”), they might be more likely to get donations. If institutions seek geographical diversity, they will obtain a range of perspectives. If they seek musicians and athletes or students with unusual tastes and passions, they will have an interesting mix of people. Let us focus for the moment on the commitment to cognitive diversity—on having students who have different experiences, values, perspectives, and information.
There are many reasons for this commitment. One involves simple market pressures; a school that has different sorts of students is more likely to be able to attract good faculty members and good students. Of course people’s preferences and values vary, and some people want to go to places that are relatively homogeneous. But this seems to be the exception rather than the rule. And there is another justification for diversity, one that has received considerable attention within courts54 and that is closely related to my topic here. The idea is that education is likely to be better if the school has people with different views, perspectives, and experiences.
In principle, that idea need not focus on the question of race at all. If a university has students from New York, California, Ohio, Texas, Florida, Iowa, Mississippi, and Alabama, it is likely to have more cognitive diversity than if its students come only from New York. If an American university seeks cognitive diversity, it might well make special efforts to attract and admit students from other nations—China and France, Germany and Denmark, Japan and South Africa. Students at different ends of the income distribution are also likely to have different perspectives.
In some places, women are, as a class, st
ronger applicants than men. People can dispute whether a university might decide that it wants to have a sufficiently large percentage of men—say, at least 40 percent—even if that means it will be giving a preference to them. And of course there is a trade-off between cognitive diversity and other values. Pure academic potential may cut in one direction; the pursuit of diversity might cut in another.
It is important to see that a university can pursue cognitive diversity without discriminating against anyone on grounds that are generally agreed to be illicit. In 2018, it was alleged that some elite universities are discriminating against Asian Americans, using interviews, or other factors, to impose some kind of quota on their numbers. In the view of many people, using quotas is utterly unacceptable; it cannot be distinguished from similar ceilings on the number of Jews from decades before. Let us bracket the question whether the allegation is true. It is entirely possible to insist that universities may not discriminate against Asian Americans while also insisting that they can pursue diversity of various kinds, through geographic preferences, consideration of economic background, emphasis on extracurricular activities, and so forth. To be sure, some difficult questions might arise here about whether discrimination is being shrouded and about whether there really is a line between discrimination against one group and preferences for another.
Let us put those questions to one side and focus on the question of race-based affirmative action. In that context, an argument in favor of cognitive diversity was approved in Justice Lewis Powell’s decisive opinion in the Bakke case,55 an opinion that has the unusual distinction of having settled, for a period of decades, the constitutionality of affirmative action in higher education. My goal here is to offer a qualified defense of Justice Powell’s view. I urge that in some educational settings, racial diversity is important for ensuring a broad array of experiences and ideas and that in those settings, narrowly tailored affirmative action programs should be constitutionally permissible.
Conformity Page 12