This paper is about two cultures of democratic theory. An empirical culture, characteristic above all of Americanists who do “large-N,” statistical work, assumes that the normative standard for a well-functioning democracy is whether policy outcomes track public preferences. A political-theory culture assumes that it is something—as it might seem, almost anything—else. Footnote 1 This divide is all the more striking given that it involves what should be the “best case” of scholars on each side. It concerns not the kind of empirical research sometimes faulted for being abstruse and driven by exclusively technical concerns, but scholarship that aims to treat questions of great and obvious normative interest to both scholars and citizens. And it concerns not the kind of political theorists accused of being obsessed with antiquarian or exclusively conceptual questions but, on the contrary, those who regard themselves as socially engaged and publicly involved: who aim to criticize contemporary democracy and spur improvements in it. Footnote 2 My hope is to document this divide, do a bit to explain it, suggest ways of bridging it, and give some reasons why fully bridging it is neither possible nor desirable.
At the least, I hope to enable a better class of intellectual quarrel: one that is aware of disagreements, their causes, and the reasons, good and bad, for acknowledging or ignoring them. At risk are order and comity in the discipline, and the fate of recurrent calls for empirical researchers to take on board (and incorporate into their research programs) the conceptual sharpness, political imagination, and normative richness that political theorists can offer, and for theorists to do work that both draws on what empirical researchers know and aims to contribute usefully to the projects they work on. Footnote 3
But the stakes are higher than that. Recent empirical work on responsiveness has drawn much of its edge and inherent interest from the claim made by many scholars—and hotly denied by others—that the U.S. political system has become tilted towards oligarchy, towards a system where the preferences of the affluent are often reflected in policy change whereas those of the poor and middle class, unless they coincide with those of the rich, are not. Footnote 4 As I shall argue, this work forfeits much of its critical edge through its penchant for comparing the status quo, explicitly or implicitly, to a simplistic normative model of democracy whereby democratic majorities are to get whatever they want, on every issue, and in short order. Students of American oligarchy risk losing the attention of those whose scholarship alerts them to the multiple, complex dimensions on which democratic quality may be measured. This is a distinct shame, since what might be called oligarchic responsiveness is hard to defend on any serious account of how democracy should work. On the other hand, democratic theorists, due to their relative inattention to mass public opinion and its link to policy outcomes, often write as if the best objections to contemporary democracy remain essentially what they were from the 1950s through the 1970s: that democracy was too institutionalized and bureaucratic, insufficiently participatory. Footnote 5 They risk focusing on small-scale experiments in participation and deliberation while missing what amounts—if scholars who claim oligarchic trends are right—to a sea change in the mass polity from a past that in comparison seems quite benign.
The Doctrine of Responsiveness
In his 2012 American Political Science Association Presidential address, G. Bingham Powell wrote, “Democratic theory assumes that successful democracy will create close ideological congruence,” defined as a “fit between the preferences of the citizens and the committed policy positions of their representatives.” Footnote 6 Assertions that democratic quality can be measured by such “fit” are ubiquitous, though this criterion goes by many names: “policy responsiveness” or plain “responsiveness;” Footnote 7 “policy congruence;” Footnote 8 “issue congruence;” Footnote 9 “democratic responsiveness;” Footnote 10 “political responsiveness” or “government responsiveness;” Footnote 11 “substantive democracy;” Footnote 12 and “correspondence.” Footnote 13 (I shall use “responsiveness” for brevity and consistency; attempts to impose on the field other terms, or more precise distinctions, have not succeeded. Footnote 14 ) However they name it, empiricists who study the functioning of democratic systems consistently regard responsiveness as a key marker of democratic quality, either “a central consideration in evaluating the quality of democratic governance” or “the key indicator,” “the explicit standard.” Footnote 15 Often fit or responsiveness is deliberately adopted to the exclusion of other criteria: “The quality of representation should be indicated … by the extent to which representatives’ actions are related to the preferences of those being represented.” Footnote 16 Whether public opinion influences government decisions has been called “the central question of public opinion research.” Footnote 17 Lisa Disch aptly cites Kuklinski and Segura’s summary of the empirical literature on representation and public opinion: “to most scholars, responsiveness is what representative government is all about.” Footnote 18
Empiricists typically claim, or assume, that “democratic theory” tells us how important responsiveness is. However, to the extent that democratic theory means the reflections of political theorists who study democracy, this claim is essentially false. Political theorists do not, and never have, regarded responsiveness as the central measure of democratic quality. When they have imagined a perfectly responsive regime, they have judged that this would be a bad thing. Democratic theorists, put simply, are not playing the role that empiricists have written for them. And empiricists are not interested in the play that theorists are actually starring in.
Both sides could benefit from mutual learning and adaptation. Rather than latching on to responsiveness in the hope of avoiding complexity and normative commitment, empiricists can find in democratic theory ways of managing complexity and commitment that would render their arguments more persuasive. Normative theorists, for their own part, stand to learn from empirical researchers many sorts of lessons: methodological lessons regarding the virtues of adequate (rather than perfect) criteria, and moral lessons regarding the need to sympathize with ordinary citizens who may not meet ideal standards of democratic reflection but still yearn to exert a rough control over government outcomes. A substantial division of labor remains appropriate. But better mutual understanding will help us better determine what kind, and why.
What Counts as “Democratic Theory”?
Empiricists who study the health of democracy almost universally cite “democratic theory” in support of the responsiveness criterion. When they do this, however, they may not mean the thought of normative theorists Footnote 19 who study democracy. They may intend democratic theory in a looser or folk sense. Some digging into the literature reveals that this is sometimes the case, but often not. And to the extent that empiricists do intend a loose or folk theory of democracy, there is reason to think that they should not: that only a deeper, more substantive kind of normative theory can serve the function intended.
Sometimes empirical scholars who cite democratic theories or democratic ideals clearly intend nothing very definite or precise. The reference to “democratic theory,” citing no particular literature, is essentially for effect, expressing the authors’ commonsense belief that responsiveness is good. Footnote 20 In these cases, democratic theory is a rhetorical place-holder for a criterion whose importance the authors are really asserting in their own voice.
More often, empirical researchers asserting that democratic theory implies congruence are citing a folk or popular theory of democracy. They mean the common lay belief that politics can and should give that people what they want, and should change in response to changes in what they want. Footnote 21 If they mean to puncture a popular misconception, empiricists can hardly be faulted for quoting it in popular form. But more commonly, researchers apparently intend to puncture a scholarly position—without being fully explicit as to their scholarly enemies. Footnote 22 The result can be singular. Larry Bartels, for instance, at the close of an argument that citizens lack stable “preferences,” suggests the need for “a style of theorizing unfamiliar to political philosophers and political scientists alike,” namely one that does not “evaluate institutions by reference to their success or failure in reflecting citizens’ preferences.” Footnote 23 In fact, a theory that did evaluate institutions primaily with reference to their success or failure in reflecting citizens’ preferences would be deeply unfamiliar to political philosophers and normative theorists—though very familiar to popular rhetoricians. Footnote 24
Finally, some empirical scholars (though in fact fairly few) mean, by “democratic theory,” theory in the scientific sense: a body of causal explanations built up by testing empirical hypotheses, or a set of deductive hypotheses deriving from formal assumptions. Footnote 25 If their purpose is to establish that their findings are relevant to existing research and represent a contribution to it, this use of theory is natural. However, because theory taken in this sense names a set of value-neutral propositions—causal in the first sense, deductively true on certain formal assumptions in the second—citing “theory” cannot further the purpose discussed here: to establish a project’s substantive worth, relevance, or importance. An observed departure from Downs’ median voter theorem, for instance, can establish that Downs’ model is too simple but not that democracy faces some sort of problem. To draw the latter conclusion from a formal theory would get things backwards. The median voter theorem was only interesting because it seemed to show that party competition would be good for democracy on common normative assumptions of what is good. Footnote 26 Awareness of this in fact seems widespread. Empiricists who cite scientific political theory in support of their project’s field-relevance and originality also usually cite normative theory in support of its significance, though they do not always separate the two.
In other words, if an allusion to democratic theory is to establish the substantive interest or moral urgency of a research program, it had better cite a theory that is both normative and serious, widely held in systematic form. And this is often the clear intention: empiricists cite a normative account of what democracy ought to be, in order to justify a study of how fully it in fact meets this standard (or doesn’t). In doing so, however, they practice radical selection bias. Only those normative theorists are cited who say—or can be quoted as saying—that policy outcomes should closely track public preferences. This selection takes a few different forms.
Repeated Citation of a Misleading Authority
Robert Dahl once wrote, “A key characteristic of a democracy is the continuing responsiveness of the government to the preferences of its citizens, considered as political equals.” Footnote 27 Roberts and Kim, in citing this quotation, call Dahl’s Polyarchy “perhaps the most cited work in democratic theory”—and indeed, the same quotation is cited frequently in the responsiveness literature. Footnote 28 Yet the quotation’s context reveals the pitfalls of citing it. First of all, it is a throwaway line in a work not of democratic theory but of comparative politics: to cite it is to appeal to naked authority, not to an argument. It is not surprising that normative theorists almost never cite this line—and when they do, it is to complicate a responsiveness theory of democratic quality, not to adopt it. Footnote 29 Where Dahl himself gives an extended account of democratic quality, he finds great fault with a simple responsiveness criterion (which he calls “populistic”). He doubts that anyone would be so foolish as to demand “instantaneous translation of majority preferences into government policies,” and says that only a fanatic would make political equality and popular sovereignty the only standards for assessing democracy. Footnote 30 Dahl’s other work, sometimes cited globally in support of a responsiveness test for democratic quality, is in the same vein. So-called “polyarchy” is defined in terms of near-universal suffrage, free electoral competition, and other process criteria—but nowhere is it said that policy outcomes should closely track voter preferences. Footnote 31 Given that the Dahl quotation is ubiquitous, often stands alone, and is glossed as advocating simple responsiveness, its unsuitability in this role is a problem. But the bigger problem is that no suitable normative substitutes for Dahl could be found. For no normative theorist—except, arguably, Bentham, whose arguments on this subject are notoriously crude and inconsistent—supports a simple responsiveness criterion as the main, or even a central, hallmark of democratic quality.
Representation Theory as False Proxy for Democratic Theory
Contemporary democratic theory, and on some accounts more classic varieties as well, typically starts not by examining democracy as it is but by exploring the values that it is supposed to further. It then proposes as desirable institutions that would embody or further those values. Sometimes this has led democratic theorists to reject mass democracy in favor of direct, participatory forms, Footnote 32 or to seek democracy in constructed minipublics, e.g., so-called deliberative polls or citizen assemblies selected by lot. Footnote 33 Empirical students of mass democracy somewhat understandably neglect such theories as tangential to their concerns. But empiricists have continued to neglect democratic theory in spite of its recent, strong turn back to engagement with mass democracy. Footnote 34 In this recent work, minipublics are to be used to inform and improve mass electoral institutions Footnote 35 or else studied as part of a larger democratic order in which both mass and mini- or deliberative institutions find a place. Footnote 36 Empiricists might engage with democratic theorists who attribute some value to electoral responsiveness but who seek to improve or educate public preferences; Footnote 37 to ensure that underrepresented groups achieve enough democratic voice that their perspectives are heard in policy debates and their social status as equal citizens maintained; Footnote 38 or to radically reform campaign finance, e.g. through progressive donation vouchers, so that democracy can better fulfill its promise of effectively aggregating political information. Footnote 39 Even reforms such as compulsory voting, which exists in many countries and which would not require constitutional amendment, Footnote 40 are not usually canvassed—even to reject them—by Americanists interested in democratic quality (though the comparative literature considers them extensively). The result is again to shrink radically the range of democratic theory that is taken to be useful or applicable. The temptation is great to exalt as state-as the art those who seem to stick closest to existing practices and institutions.
In particular, classic and recent treatments of representation, which often take such an imminent approach, Footnote 41 are taken to be typical of contemporary democratic theory, even though neither their authors nor political theorists as a whole would think of describing them as such. And in any case, none of these representation theorists in the end endorses the responsiveness criterion in anything like a simple or pure form. In particular, Hanna Pitkin’s oft-cited work does not do so. Her occasional and unfortunate language regarding “responsiveness” is, as Lisa Disch has shown, in profound tension with a work that as a whole radically questions both the coherence and the relevance of a simple responsiveness criterion. Footnote 42
The Slide from Minimal to Perfect Responsiveness
Third, and perhaps most crucially, empiricists often slide between a criterion of some responsiveness and one that posits tight or perfect responsiveness. “Through the overwhelming mass of historical and contemporary argument about democracy,” writes Powell, “runs the belief that the preferences, or wishes, or (treacherous term) interests of the people should to some degree prevail.” Footnote 43 Other empirical scholars summarize the consensus of political analysts in language that is very similar—and, like the previous quotation, hedged. Footnote 44 Powell rightly adduces Pitkin in support of this claim, and indeed few democratic theorists would fail to share it. Footnote 45 But the italicized phrase—“to some degree”—does a great deal of work here. Democratic theorists, as well as folk opinions about democracy, differ sharply regarding how important such responsiveness is; how close the correspondence between public opinion and policy outcomes should be; whether lags and delays are permissible or desirable; and whether certain policy areas (e.g., those involving personal or minority rights) should rightly be unresponsive. To prize a certain amount of responsiveness—call it minimal or baseline responsiveness—is not to say that the more responsiveness, the better.
Empirical scholars recognize this distinction. But their research often proceeds as if they did not. In the rest of his book Powell often slips into much more uncompromising statements that no theorist of democracy or representation would endorse (e.g., “democratic policymakers should do what their citizens want them to do”). Footnote 46 Departing from the possibility of a baseline standard, he repeatedly assesses the degree to which a democratic outcome is “successful” according to the smallness of its departure from zero distance between the median voter’s preferences and policy outcomes. Footnote 47 Similarly, while Erikson, Wright, and McIver take a baseline statement from Key as the epigraph to their book, Footnote 48 their book proceeds to measure the ideological distance between policy outcomes and public opinion in the several states with the implication that the best distance would be none. This slip sometimes works through a type of quarantining. One or more insights from democratic theory that qualify the crude zero-distance view of democratic responsiveness are noted in introductions or conclusions, but the body of the analysis proceeds as if these insights did not exist. Footnote 49
Refuting One-Dimensional Critiques of Responsiveness
Finally, one common trope of the responsiveness literature is to take seriously only one imagined objection to the responsiveness criterion, one based on doubts that the people have the capacity to choose good outcomes (being ignorant of policy, or fickle, or one because of the other). Page and Shapiro call these concerns “the classic justifications for ignoring public opinion.” Other authors make similar assumptions. Footnote 50 The empirical claim that the public as a whole (in accord with Converse’s “miracle of aggregation” Footnote 51 ) has stable, rational preferences and that government is responsive to them is said by the authors of The Macro Polity to establish “strong” versions of democratic theory—an evaluative term, at least implicitly normative. Footnote 52 These authors in effect consider the empirical hypothesis that public opinion does not exist, but not the normative claim that it should not be omnipotent. Footnote 53
The Rejection of Perfect Responsiveness in All Democratic Theory
That few (if any) respectable democratic theorists endorse a simple responsiveness criterion for democratic quality is, as a negative claim, hard to prove. But canvassing the variety of things that democratic theorists do say may give us some idea of how far the field is from considering policy responsiveness a central—much less the central—consideration.
That the traditional “canon” of political theory rejects the idea that the best democracy is the most responsive is in a sense uninteresting and overdetermined. For the vast majority of canonical theorists, before the twentieth century, rejected democracy—if that means having rulers chosen through competitive elections involving universal and equal adult suffrage—in the first place. (For one thing, almost none even contemplated enfranchising women, or racially subordinate groups where they existed.) Footnote 54 But even contemporary thinkers who endorse democracy, who fully accept universal suffrage and in general the equal political rights of all citizens, have not thought that a responsiveness theory exhausts democratic values—or, in most cases, is even particularly relevant to them. So-called “deliberative” democrats, for their part, reject the whole idea that public preferences ought to be taken as given. John Dryzek has noted that while rational choice theorists regard preferences as exogenous, and talk as cheap or strategic, “deliberative theorists … have a very different expectation: that the reflection central to deliberation enables (though it does not always require) preference change.” Footnote 55 Even leaving this aside, democratic theory contains a variety of measures of democratic quality. None of them is limited to the responsiveness criterion, and few even include it. On a civic republican view, the test of democracy—or, better, of a “republic” based on civic equality—is widespread devotion to the common good; on a more sober republican view, it is non-domination, the prevention of arbitrary power. Footnote 56 On an “epistemic” view, building on a tradition that begins with Condorcet’s jury theorem and more recently emphasizes instead the merits of epistemic diversity, the purpose of democracy is to achieve through aggregation not the enactment of popular wishes but maximally accurate judgments regarding public problems that admit, more or less, of objectively better and worse answers. Footnote 57 On an Emersonian view, democracy is valuable to the extent that it fosters independent and egalitarian mores, disinclined to docility. Footnote 58 On a mitigated-elitist view, democracy allows for a salutary mix of democratic and aristocratic elements—a select few govern, responsible to the many Footnote 59 —and perhaps some elitist ones as well, with the unusually intelligent or educated rightly having disproportionate influence. Footnote 60 Conversely, on a radical view, full democracy would entail a universal “ability to use and develop … presently unusable or denied human capacities.” Footnote 61 On a different and more power-based radical view, democracies should be assessed by how well they “undermine and otherwise mitigate the corrosive effects of social hierarchies.” Footnote 62 On a social-democratic account of democracy, notably uncommon among normative theorists in America but very common among some comparative politics scholars, a key—perhaps the key—indicator of democratic quality is the level of economic equality. Footnote 63 All these viewpoints seem plausible expressions of values historically considered important in the American tradition (or a portion thereof) and that of other advanced, electoral democracies. And this does not even consider a range of democratic models put forth by serious comparative scholars but less common among normative theorists. Footnote 64 None of these models of democracy, involving diverse indicators of its quality or good functioning, is captured by the examination of democratic responsiveness and policy-preference congruence.
A research program that took any one of these normative positions as its starting point might, while assuming and endorsing some level of responsiveness, adopt as its measures of democratic quality a range of criteria not particularly related to it. To take just a few examples: on the anti-hierarchy view, a well-functioning democracy might be one in which the poor, and those otherwise disadvantaged in society, are more likely to participate in politics than the wealthy and powerful, since the former have a greater need than the latter for the democratic power that alone can challenge social inequality. Footnote 65 On the mitigated-elitist view, one might test the quality of democracy by seeing whether representatives in fact had the qualities of “distinction” that electors claimed to prize (such as education). Policy outcomes would be somewhat less important because, on this view, it is normatively desirable for representatives to make policy in relative independence from popular wishes. (On the other hand, if representatives were disproportionately rich, white, male, or educated without voters actually prizing these qualities, this too might be faulted, again regardless of policy outcomes.) An adherent of the Emersonian view might reverse a common measure of democratic legitimacy, namely trust in government: in a healthy democracy, most people would not trust “the government in Washington to do what is right.” Finally, a small-r republican view might have milder but still crucial implications. Republicans typically endorse constitutional mechanisms for ensuring that majorities cannot oppress certain minorities, and would assess democratic quality as higher when these are present, lower when they are absent. Footnote 66 Thus normative ideas have consequences for research programs. Taking them seriously might suggest agendas that are both practical and very different from those that prevail.
Strikingly, even political-theory defenders of responsiveness in some form almost never endorse it in the pure form favored by empiricists, which takes as the implicit null hypothesis a gap of zero, and a time lag of zero, between changes in preferences and their reflection in policy. (This null hypothesis is “positive” but threatens, for lack of explicit discussion regarding its limits, to become normative as well—as if departures from it would require explanation or justification.) Even those democratic theorists who agree with Madison that policy should reflect the “cool and deliberate sense of the community” Footnote 67 take the qualifier seriously and believe that there should be a substantial lag between preference change and policy shifts, so that fleeting or zigzagging opinion change may be, rightly, ignored. Footnote 68 Some theorists defend administrative expertise at the cost of some responsiveness, a position that hardly seems implausible on its face. Footnote 69 Many democratic theorists with populist sympathies, skeptical of judicial review, make an exception for “discrete and insular minorities” who will predictably be victimized by majoritarian politics. Footnote 70 Even those who reject Burke’s view that legislators should vote their own judgment as a rule (the “trustee” conception famously described, though not endorsed, by Pitkin Footnote 71 ) typically grant that representatives should occasionally follow their own judgment, in defiance of constituent opinion, giving reasons in the hope that constituents will eventually see the merits of what they did.
Very likely, most empirical scholars of policy responsiveness would, if asked, endorse one or more of these departures from perfect and immediate responsiveness. Footnote 72 Indeed, no empirical scholar can endorse perfect responsiveness without rejecting, as a grave offense to democracy, the U.S. Bill of Rights and its enforcement by the judiciary. Even those empiricists who do explicitly question the U.S. regime of judicial review and supermajority requirements for constitutional amendment Footnote 73 rarely oppose two other persistent checks on perfect responsiveness: judicial review of executive action, as a central component of the rule of law; Footnote 74 and requirements of deliberation and delay embodied in parliamentary practice (second reading requirements, majority quorums, and so on). Footnote 75 Finally, that stability of policy is a good in itself, allowing citizens and economic actors to form stable plans, Footnote 76 is a claim that few empiricists would likely deny but that none builds into his or her responsiveness theory. The cumulative result is for empirical researchers to give the impression, through omission, that they would regard perfect responsiveness as at least prima facie a good thing.
Behind the Disconnect
The leading researchers into democratic responsiveness and the leading normative theorists of democracy are, on both sides, intelligent, curious, public-spirited, and careful. So why do they argue at such cross-purposes? It is partly a matter of cultural style, with the mental and cultural habits of empiricists and theorists still mapping remarkably well onto the “two cultures” that C.P. Snow attributed to scientists and traditional humanists in his day. Footnote 77 But since divergent intellectual cultures often stem from real and serious disagreements, it is worth looking more substantively at the reasons for the two cultures and the prospects for bringing them closer.
The Tyranny of Zero
First, empiricists are more comfortable with numbers; theorists, with words. While that much is obvious, it has non-obvious implications: empiricists and theorists seek rigor in very different places. Empiricists regard a rigorous claim as one that can be expressed through counting. Theorists, contrary to many empiricists’ belief, have their own standards of rigor, but they involve classification—does claim X involve a right or not; is coercive government action Y legitimate or not—or assessments of quality, involving careful judgments backed by argument and capable of being rebutted. Consider the study of “focal points” in coordination problems (that is, choices or concepts that will reliably occur to people if each agent among many is trying to do as others do but has trouble settling on what that will be). Thomas Schelling, the contemporary discoverer of focal points, thought that while some aspects of focal points could be described, the points themselves could not be calculated. Reaching them was an intuitive enterprise more like “puns and anagrams” than like chess. Footnote 78 When an economist tackled the issue, he was determined to reduce quality to quantity: a focal point was defined, provisionally (and not ridiculously), as the thing that became most obvious to a wide range of actors by virtue of being most often mentioned. Footnote 79
This particular example actually goes a long way towards explaining the puzzling result mentioned above: that empiricists posit as a fallback or default a zero divergence, and zero time lag, between preferences and policies, while having no particularly strong reason to think this would be a good thing. If we are asking how large the divergence should be, or how long the typical time lag, the only “natural stopping point” or focal point once we are seeking a quantitative answer is zero. (In verbal terms, the natural question is whether we have a “match” between preference and policy, which presumably means a zero gap and a zero lag. Footnote 80 ) The commonsense answer, “a departure that is not too great, or too systematically biased, and a lag that allows for some reflection and debate,” fails to be adopted by empiricists not because it seems normatively wrong but because adjectival terms like “not too great,” “not too systematically biased,” “time enough for debate” are too hard to operationalize, to express as numbers. Footnote 81
The Game of Concepts: Empirical Coordination and Theoretical Crowding
Second, there is an inherent disconnect between how empiricists and normative theorists use concepts. Simply put, empiricists are playing a coordination game and theorists, a crowding game. When empiricsts define something, whether as complex as “democracy” or as apparently simple as “the more conservative political party,” they seek a definition that is good enough to enable rough-and-ready agreement (again, they believe true precision can only come from quantitative measures and tests). And they have a strong interest in sticking to more or less the same definition of a phenomenon as other researchers have used, so that empirical research on that phenomenon can move forward. This is a classic case of a coordination problem, in which finding one among a number of plausible solutions, and sticking to it so that common action is possible, is more important than choosing the solution that is substantively best. Footnote 82 One might say it is because—as an empiricist—she sees concept definition as a coordination game that Diana Mutz, frustrated at theorists’ multiple efforts at defining what deliberation is and why it might be a good thing, protests that “in empirical social science, conceptual definitions are often arbitrary” and that it is pointless to “haggle” over them. “If we cannot agree on what the independent variable is,” Mutz writes, “we cannot hope to systematically evaluate its impact”; in Mutz' view “openness” in conceptual definitions only “delays progress” in empirical research. Footnote 83 To be sure, empiricists aspire to, and do, innovate radically and brilliantly in how they operationalize or measure concepts. But they can only build on one another’s research to the extent that they see themselves as operationalizing or measuring the same concept, or close enough.
Theorists, for their part, face not a coordination problem but its less famous cousin, a crowding problem. In a crowding problem it is to each actor’s advantage, once one has reached a certain number of participants, not to do as others do. It is hard to do worthwhile political theory without innovating in the use of political concepts. Many political theorists adopt “conceptual change” as the whole subject of their research. To stipulate a single definition of something like democracy would be, on their view, to abstract from everything that is in fact important about how political events and processes work, i.e. though changing how we think and talk about politics. Footnote 84 But even theorists who do not study conceptual change directly engage in it implicitly. In order to write an important book or article about justice, or equality, or democracy, a theorist must try to change readers’ basic ideas about what these words mean, or at least about what they amount to in practice. Here again, the empiricist/theorist divide results from different ideas of what counts as research. Empiricists want simple and solid conceptual definitions of what X and Y are so that they can get on with their real research, which involves claims about to what extent X (along with other things) causes Y. Research “progress” for theorists, on the other hand, involves precisely ever-better arguments regarding how we should regard X and Y. Good research involves either a clarification or classification of previous arguments, or a new way of understanding a political concept that points out a new normative direction or illuminates, as has not been done before, a folk usage.
Falsification vs. Hope
Third, there is a direct tension between the empiricists’ scientific drive to enable (and respect) falsification and normative theorists’ civic drive to preserve democratic hope. Consider the debate between Mutz and Dennis Thompson on how to regard empirical findings that certain kinds of deliberation do not have the beneficial results one might hope for. Mutz regards these results as evidence that the theory of deliberative democracy—which out of attempted interpretive charity she glosses as aspiring to put forth falsifiable causal hypotheses—has in fact been, provisionally, falsified. On her view, these findings suggest that we should spend more resources investigating things other than deliberation that might make democracy better. Footnote 85 Thompson regards matters very differently:
The most promising approach for empirical research would therefore seem to be to continue trying to discover the conditions in which deliberative democracy does and does not work well, while paying more attention to the question of to what extent the unfavorable conditions could change. Footnote 86
In earlier work Thompson argued that no amount of empirical research could bring him and those who thought like him to give up on a wide-ranging “reconstructive” agenda of making both citizens and the political system more fully democratic. In fact, such ideals, which provided visions for long-term, fundamental reform of the polity and the larger society, were, on this early view, all the more useful the further they were from anything that seemed possible in “existing political reality.” Footnote 87 Similarly, in this more recent article Thompson does not—to Mutz’s dismay—appear to think that any data would be sufficient to make him give up on deliberative democracy. Nor is Thompson’s an unusual view.
Political theorists who reject falsifiability may seem to empirical scholars to have placed themselves outside the bounds of reasonable discussion: to have entered the realm of faith, or at least to have left that of social science. Footnote 88 Even those who argue in less Popperian mode sometimes follow Dewey in insisting that ideals as well as empirical generalizations must adapt in the face of experience. Footnote 89 But in fact the disagreement is less methodological than normative, as a comparison may illustrate.
Consider the career of democratization theory. In the 1930s, it could have been argued on empirical evidence that the causes or preconditions of liberal democracy—even defined in fairly minimal terms as governments chosen through periodic free and fair elections with universal adult suffrage and rough rights to free political speech—was ceasing to be a useful research question. Given that democracy was rare and retreating, it could have been argued that the best course for political scientists was to find the least-bad way of accommodating ourselves to the authoritarian and totalitarian forms of government that alone seemed able to contain (or channel) mass ideological movements and to enable a necessary level of economic planning. In spite of this, however, most social scientists in the remaining democracies did not regard democratic setbacks as reasons to give up on democracy. On the contrary, they regarded them as ever more reasons to understand threats to democracy, so as better to parry them. In doing so, they were resting their research agenda not on falsifiable causal hypotheses but on a normative commitment driving empirical research. Giving up on democracy was taken to be poor social-science practice because doing so would be ethically unacceptable. In the 1950s, when most of the world’s peoples, by far, still lived in fragile democracies or non-democracies, matters were qualitatively similar (though less critical). Those who continued to study democratization did so out of hope, not in response to clear evidence.
The fact that modern empirical scholars typically reject (or ignore) some of the aspirational or transformative ideals of democratic theory—when they would not in an earlier age have rejected ideals of electoral competition and free speech—does not show that they are “more empirical” than their predecessors. It shows that they do not believe the normative claims that many normative theorists insist are true: that democracy in its current state, without far-reaching, indeed transformative changes that might deeply alter the political and social structure, is normatively unacceptable, not sufficiently legitimate to merit our indefinite allegiance. One difference between normative political scientists and empirical researchers thus involves a latent but crucial normative disagreement: whether familiar democratic institutions and practices—though perhaps not their specific, skewed operation over recent decades—are, or are not, good enough for government work. One might speculate that empiricists (or at least Americanists) and theorists self-select into their respective subfields, on average, according to their different orientations towards that question.
Quality, Quantity, and Value Neutrality
A fourth difference between empiricists and theorists is that the former, obviously unlike the latter, are loath to make value judgments. This is partly because many empiricists doubt that such judgments can be justified through reasons, as opposed to being expressions of subjective opinion or preference. (Critics of social-science positivism, on the other hand, claim that value-neutrality is either impossible, since all facts are value-laden, or undesirable, since observing past regularities distracts from the goal of changing society in ways that will render that past irrelevant.) Footnote 90 Again this distinction is obvious but carries non-obvious consequences: responsiveness can be seen as an attempt to render a value judgment that does not appear to be such.
How one assesses the quality of democracy, or of anything else, seems very clearly to depend on what one values. In fact, what counts as democracy going well was precisely W.B. Gallie’s central example of an “essentially contested” concept: the kind of thing on which reasonable people disagree and none can be judged demonstrably wrong. Footnote 91 The problem is not a lack of good criteria for democratic performance but too many. There are many plausible and incompatible opinions of what one thinks democracy is for, which of many possibilities regarding its practices and institutions is most appropriate, which actions or policies best embody democratic aspirations. Footnote 92 One criterion of quality, however, seems to stand out among the rest in involving a kind of neutrality or self-effacement: surely there can be nothing partial or biased in saying that democracy should give voters—not political scientists—whatever they want. And to the extent that this would cause outcomes that a particular researcher considers bad, surely the right solution is to persuade a majority of our fellow citizens that they are bad, not to hope that these outcomes persistently fail to occur even though the people favor them. It seems, in short, that the scholar who adopts the responsiveness criterion is not making value judgments but simply leaving them to the people.
Normative critiques of this position are legion, and have been canvassed here. Footnote 93 But there is a more straightforward problem: the position is at war with itself to the extent that the majority of a given public itself values something other than responsiveness. If the American public, or that of another country, actually favors institutions designed to frustrate maximum responsiveness (like the US Constitution in general and judicial review in particular), or thinks it wants leaders who sometimes stand up for their beliefs when doing so is not popular, a choice must be made as to whether we should empower the median voter on the one hand, or institutions favored by the median voter on the other. The quest to avoid value judgments via a responsiveness criterion is therefore either indeterminate or incoherent.
I have focused so far on diagnosis. It is time to consider treatments. I believe that the divide between empiricists and theorists on questions of democracy can to some extent be bridged, with potentially great benefit for both sides. There follow some suggestions for both sides, stated bluntly for clarity but meant as tentative proposals rather than final words.
Bridging the Two Cultures: Empiricists
Damped Responsiveness
Andrew Rehfeld has suggested (crediting Michael Minta for the idea) that empiricists may favor something like responsiveness as a criterion because it is easy to measure. Footnote 94 Things might not be quite so simple. For some normative theories that downplay responsiveness imply alternative criteria that could also be measured quite easily. If compulsory voting is a good thing, or a bad thing, its presence or absence is an easy dummy variable. If the radical conception of democracy is adopted, or else the mitigated-elitist one, these things too are easily measured (as noted earlier). Operationalizing these and other measures of democratic quality would involve judgment and labor—but probably no more labor than is currently devoted to measuring public opinion and policy outcomes. In a comparative context, where the norms that drive American research programs are absent, a great variety of measures—including the extent of civil liberties; the kind and nature of constraints on executive power; workers’ rights; women’s rights; and the proportion of adults who vote—are in fact used as indicators of democratic quality. Comparative empirical scholars acknowledge that they must name the normative basis of their chosen criteria, and argue (controversially) for those criteria in the face of serious alternatives, before proceeding to their data and tests. Footnote 95
However, even stipulating that responsiveness were particularly easy to measure, this pragmatic reason for measuring it would not entail regarding perfect responsiveness as a value. Scholars already measure lags in responsiveness and gaps between popular wishes and policy outcomes. Measurement problems provide no excuse for rejecting alternative ways of judging what is already found, and already done; the gaps and lags might be, up to a point, good things, not bad ones. Empiricists could embrace, in accord with Madisonian principles, a mitigated or “damped” responsiveness standard: in a maximally healthy democracy, policy typically adapts, to a large degree, to public preferences—but not always, and usually with a lag. Footnote 96 This adjustment would respect the empiricists’ aspiration towards non-elitism. A system with some friction does not empower particular elites, nor even the status quo, but merely ensures that shifts in the status quo reflect durable shifts in public opinion rather than ones that the median voter will soon regret. As for what lag would be appropriate, empiricists need not make this judgment on their own. They could merely report the lags in particular cases or on particular policy-issues, and, in best value-neutral fashion, let each reader decide whether those lags are too long or too short. Once again, it is not as if zero is a normatively natural baseline, as opposed to a quantitatively unique one, in the first place.
Side Constraints, Benign Non-Decisions, and Defensive Democracy
A paradox of responsiveness and constitutionalism was mentioned above: it seems odd to take perfect responsiveness to public opinion as a normative baseline when one of the things that public opinion endorses is a constitutional system designed to frustrate perfect responsiveness. A damped responsiveness criterion is one possible response to this. But there are others. One would be to render explicit, and justify, Martin Gilens’ methodological decision to leave out of his data set public preferences whose satisfaction would require a constitutional amendment. Footnote 97 Through this exclusion, Gilens effectively accedes to a substantial and pervasive side constraint on the responsiveness criterion. But by studying only questions of government spending rather than (say) LGBT rights or affirmative action, many other researchers make a similar concession by omission. They thereby omit from their studies the issues on which respect for the median voter seems most normatively fraught. There seems no good reason to reach such normative judgments by default or omission. Exceptions to a general assumption that responsiveness is good should be either defended, on pain of admitting that the ideal political system is not maximally responsive, or abandoned, on pain of also abandoning the populist conceit that the responsiveness criterion unambiguously gives the people what they want (since many of them value the exceptions).
One could go much further in this direction, recognizing that what people want might not always be captured by policy that quickly responds to changes in what they want. Constraints involving constitutional rights can be considered a benign form of “non-decisions,” matters kept off the political agenda (at least in the short term). Footnote 98 But a focus on non-decisions might suggest a more fundamental adjustment in responsiveness theorists’ research agenda. In a famous paper, Footnote 99 Achen and Bartels have decried the fact that retrospective voting rewards or punishes incumbents for random and uncontrollable events, or trivial and manufactured ones, rather than reflecting a reasoned assessment of how deliberate policy has brought about substantial outcomes. Buried in that paper, however, is an alternative account of respect for popular wishes and one that implies more optimism. The authors note that what they call “American oligarchy” is limited by the “bounds on what any administration can do.” Footnote 100 Successive political battles, working through the party system, have “settled” some questions in ways that neither party dare depart from. These settled questions include national independence; civil rights and liberties at the federal and state level; at least minimal welfare provision in the face of accidents, unemployment, sickness, and natural disasters; and internationalism as opposed to isolationism.
But one might add much more to this list: Social Security, Medicare, free public primary and secondary education, lots of infrastructure spending, a notably large and powerful military by international standards, and federal and state regulation of health, safety, and the environment. While lamenting the fact that politicians can do as they please “within those limits,” Achen and Bartels might have looked for, and found, a great deal of democratic responsiveness contained within those limits, within the limits of what might be called “valence policies.” Footnote 101 The list of what politicians dare not do on pain of political retribution is actually quite impressive. From one perspective, the scope of what they can get away with, given the theoretical range of policy options, is quite narrow. Empiricists might do more to note, and celebrate, the bright side of non-decisions. We might call this bright side “pre-electoral responsiveness,” or perhaps “defensive democracy.” A focus on dynamic responsiveness, though highly tempting (because easier to measure than respect for static preferences) and in many ways very informative, once again becomes a handicap if it becomes too one-dimensional and obscures indicators of democracy’s durable effects.
Against Conceptual Blowback: Anti-Oligarchy as an Overlapping Consensus
If one reason that empirical researchers adopt zero-gap, zero-lag responsiveness as a criterion for democratic quality is that they are at a loss to name a number other than zero, another might be called conceptual blowback, traceable to a trap empiricists sometimes set for themselves. Researchers present a stylized, oversimplified, normative version of democracy in order to contrast that model—crude, perfect responsiveness—with the pathologies they mean to investigate and highlight, whether elite manipulation Footnote 102 or unequal influence, skewed towards the wealthy and towards organized business interests. Footnote 103 But the stylized model of democracy comes back to haunt the account because empiricists then end up defending, or seeming to defend, a model of democracy that they do not in fact support.
Empirical researchers may fall into this trap because they assume that one cannot critique a current practice without comparing it to a clear alternative (used as a baseline). But this assumption should be rejected. Both normative theorists and ordinary citizens may be united in rejecting crude responsiveness. But they also unite in rejecting the states of affairs that empirical critics of current outcomes mean to highlight and criticize: either a patterned oligarchy whereby the preferences of the wealthy are regularly reflected in policy outcomes while those of the lower two-thirds of the population are more or less ignored Footnote 104 or an unconstrained elitism under which governing parties avoid accountability for policy choices Footnote 105 or presidents can start wars based on manufactured information. Footnote 106 Empirical researchers can reasonably assert that such practices, if proven to exist, are bad on any reasonable account of how mass electoral democracy should operate—including accounts that envision delay, deliberation, side-constraints, a polity that gives priority to dismantling social hierarchy, a robust welfare state, or a role for expert (but not wealth-based) opinion.
Works of applied ethics often argue that particular normative conclusions can be reached, that certain things count as wrongs or abuses (and others as unproblematic), regardless of disagreement on ethical frameworks. That is, some things are bad, or impermissible, according to all respectable systems of ethics. Footnote 107 Or when this sort of practical agreement among moral frameworks is not possible, one can still draw practical conclusions by analyzing a problem, acknowledging disagreements regarding value, and explicitly noting the different conclusions that will follow from the analysis depending on the weight one attaches to different values. Footnote 108 Again, patterned oligarchy and unconstrained elitism are likely to warrant condemnation under any credible weighting.
In short, once empiricists realize that they do not need to posit a crude theory of democracy—or indeed, any specific theory—in order to condemn abuses, they will no longer need to put forth, or imply, theories that they do not really believe in. In a context of political and scholarly disagreement, empirical findings are most compelling when they can be shown to be troubling on a variety of accounts of how democracy should go.
Avoiding Straw Guardians
Finally, there is one particularly common mode of argument that is both unfortunate in itself and prevents communication with theorists: the argument that the only alternative to perfect responsiveness is deference to a selected group of guardians. This claim, presumably at its root a reaction to Plato (or perhaps to Mill’s Representative Government) Footnote 109 was popularized, once again, by Dahl. Footnote 110 Versions of it continue to be popular among empiricists. Footnote 111 Whatever its provenance, it makes no sense. As noted, the typical alternative to perfect responsiveness is a constitutional system that purposely builds in lags and side constraints, or institutional assurances of the time and opportunity for deliberation. What these constitutional or deliberative mechanisms have in common is that they do not name a particular elite that is to be given superior political rights. These mechanisms are persistently and deliberately consistent with full political equality. They follow Madison in rejecting the model of a regime that builds in aristocratic privilege and preferring instead to improve “unmixed” republics. Footnote 112
This is not to deny that particular constitutional mechanisms may, compared to simpler forms of majority rule, amount to minority vetoes, thereby passively favoring elites who benefit from the status quo. Footnote 113 But the mechanism through which this occurs is precisely impersonal: it does not name guardians. Minority vetoes do not displace voter judgments in favor of those of intellectual elites. Their effect is negative: they diminish the ability of democratic assemblies to control outcomes, with the effect of leaving private hierarchies undisturbed. We do no one any favors by claiming a danger of paternalism when the real problem is, to use a Walter Lippman term favored by many contemporary scholars, “drift.” Footnote 114 Putting it this way avoids the false implication that political analysts who dissent from pure responsiveness must have a secret agenda of putting themselves in charge. In fact, the oligarchic distortions of democracy that many empiricsts note (and that other empiricsts contest) are overwhelmingly likely to empower economic elites, not intellectual ones.
Bridging the Two Cultures: Normative Theorists
The normative side of the cultural divide might also make adjustments, so as to take on board the crucial insights of empirical research and make proper use of its findings.
The Dignity of Counting, Intellectual and Moral
Behaviorists used to say to theorists and historical institutionalists, “learn how to count.” Whether that admonition is still common or not, the sentiment is—and theorists often respond to it with complete and angry rejection. Where an argument is at issue, theorists tend to believe, numbers simply do not matter. Theorists very commonly gloss simple majoritarianism as “a numerical version of might makes right.” Footnote 115 They suspect that empiricists have forgotten the difference between rebutting a case and shouting it down.
But there are both intellectual and moral reasons for reassessing this blanket rejection. Intellectually, knowing something about public opinion allows theorists to break out of ideological ruts by confronting opinions that are common among people different from themselves. As with other social science results, findings regarding public preferences embody “submerged biographies of the silent majority of humanity,” a form of “knowledge for its own sake about everyone, because everyone is interesting as a social being.” Footnote 116 Public opinion provides a remedy for the Millian problem that “the world, to each individual, means the part of it with which he comes in contact; his party, his sect, his church, his class of society.” Footnote 117 To be interested in public opinion, especially in its details and the reasons that it changes, is to broaden the scope of opinions one takes seriously. This is, of course, bound up with a moral claim: what ordinary people value is worth a provisional respect. Theorists who spend most of their time opining about what people ought to value should spend at least a little time exploring what they do value. To the extent that they do not do this, they may be accused of placing their own sense of moral superiority over what should be a near-automatic civic sympathy for ordinary people and their aspirations. More uncomfortably, theorists should recognize that empiricists often ignore their insights out of a suspicion that they lack such sympathy.
Essential Contestability vs. Common Games
Because democracy is essentially contested, there will never be agreement about what it means for it to function maximally well. At the same time, however, electoral democracy could not roll along at all, enjoying a certain rough legitimacy, if basic rules did not obtain: governments that radically fail to give people what they want, and might reasonably hope to get, will be turned out. And if state policy (under a succession of governments) radically departs from public wishes for long enough, electoral democracy will lose its ability to command citizens’ allegiance. The current trend in democratic theory, stepping back from its utopian mood, has been to propose democratic innovations, often involving enhanced deliberation, within a mass democracy that enjoys well-functioning electoral institutions. Footnote 118 To the extent that they endorse these trends, those who specialize in detailed critiques and evaluations of democratic quality ought to take an interest in empiricists’ assessments of whether the system is performing its basic functions as it should. “The worse, the better”—indifference to the malfunctioning of mass democracy on the grounds that only radical transformation is worthwhile anyway—is neither a responsible position nor one likely to result in more perfect democratic institutions. Refinements of democracy thrive when public belief in democracy is at least sufficient to allow for confident experimentation, not when citizens lose hope that democracy (in the only form they have ever known it) can do its basic job.
Critique without Snobbery
One of the biggest divides between empiricists and normative theorists concerns their respective senses of moral superiority. Theorists often regard themselves as the only true guardians of democracy in the face of empiricists’ complacency towards the true dangers it faces. They should acknowledge that empiricists regard it as precisely the other way around. Empiricists see themselves as sounding the alarm about what ails democracy while theorists sit back and enjoy their tenured right to read abstruse books, engage in lengthy political arguments over value, and demand that all citizens more fully share their taste for both (rather than their actual tastes for things like decent schools, job opportunities, and low casualties in foreign wars).
Theorists should be aware of the latter accusation without accepting it. There is every reason for a division of labor between those who study how things stand, and those who think about what those findings imply and what should be done about them. And there is no reason to doubt the social benefit of political theory as an activity that explores and evaluates, to the highest and most stringent possible standards, the quality of civic discourse, the ethical limitations of our prevailing opinions, and the potential of improving our institutions’ design. The endemic problem is one of role slippage. Empiricists may be faulted for slipping from a proposition of democratic authority—that the majority of voters should often get what they want—into to the odd position that those who study politics are not permitted to criticize what most voters want, not permitted to argue, with reasons, in favor of their coming to want something different. Footnote 119 But theorists frequently, and wrongly, slip the other way. That theory “challenges political reality,” is rightly “critical, not acquiescent,” Footnote 120 in no way implies that theorists should stop caring whether public opinion—in all its imperfections—is regularly flouted by those with wealth and power. The point of normative critique should be to inform the public, not to ignore it. Footnote 121
Practicing critique without snobbery requires, fundamentally, that theorists accept a society that is mostly civil, not civic. Ben Berger, adopting an eighteenth-century distinction, has distinguished between civic engagement, explicitly political, and civil engagement in the associations of everyday moral and social life. Theorists should acknowledge that most people care more about the latter than the former: “We who have dedicated our careers to studying or practicing politics should not project our tastes onto the general population.” Footnote 122 Even those who aspire to make citizens more civic must acknowledge that even if any given citizen of ordinary capacity might become, with study, fairly competent to judge (almost) any subject, the complexity of modern society prevents any citizen, or even any small group of citizens, from becoming competent on every subject. Footnote 123
Critique without snobbery also requires that theorists acknowledge public opinion as it currently stands as imperfect but presumptively legitimate. If it is a staple of normative theory (and deliberative theory in particular) not to take preferences as “given” but to critique both their content and possible distortions in their origin, it does not follow that such preferences deserve no respect at all. They deserve the respect owed to what our fellow citizens currently believe. Even accepting for argument that “if the public thinks X should be done, but has not thought about the issue much, has not tested its views in comparison to alternative policies and the reasons for them, then there is an issue about how seriously, from the standpoint of normative legitimacy, one should take those views,” Footnote 124 we should not accept, along with this principle, the hidden premise that ordinary life in a democratic society provides no occasions for thought and testing. Such in fact regularly take place under the status quo, informally and episodically but with substantial effect over time. Footnote 125 A substantial tradition in normative theory, deriving from Mill and to some extent Dewey, endorses deliberation and reflection but looks to the ordinary processes of partisan disagreement and citizen judgment, rather than to constructed deliberative forums, to provide them. Footnote 126 No one supposes that this process is perfect. Proposing reforms—even very substantial reforms—in mass politics, so as to improve the way opinions are formed, is fully consistent with an essentially democratic attitude of respect for fellow citizens. But a form of democratic respect that takes place exclusively in the subjunctive, that respects only the attitudes that citizens would form if theorists’ preferred reforms were adopted, is not. It amounts to a desire, paraphrasing Brecht, to dissolve the mass public and elect another. Footnote 127 Theorists might do well to adopt a language common among empiricists (comparativists even more than Americanists): the presence or absence of our preferred democratic practices bears on democracy’s quality, not its legitimacy.
Conclusion: Conversation, Not Identity
Separate vocations exist for a reason. Empirical and normative students of democracy will never think exactly the same way, nor should they. But each can learn from the other without becoming the other.
Empiricists can turn to theory for useful reminders of things they already know but sometimes slight: that the full health of a democracy involves institutional and civic qualities beyond mere responsiveness; that constitutions can depart from perfect responsiveness without betraying democratic values; that a critical attitude towards our fellow citizens’ preferences, within a broader attitude of respect for democratic outcomes, is not in itself undemocratic but a normal part of democratic debate. They should also learn from theorists, on my view, a permanent change in how they frame their results. They should portray inequalities in democratic representation, corruptions in the political process, or opportunities for manipulation and deception by elites as bad not because they depart from a crude, perfect responsiveness that very few citizens would actually welcome, but because they depart from a variety of democratic ideals. None of these ideals endorses the outcomes that various schools of empirical scholarship claim to have discovered, namely, electoral and policy outcomes that are greatly affected by random events; that result from manipulation; or, probably worst, that reflect a strong durable bias in favor of wealthy citizens over the rest.
This change in framing would strengthen, not weaken, the theoretical basis of empirical research. A concern with normative credibility might also lead empiricists whose exclusive focus has been on large-N studies to combine such work with more historical, case-based work. For such work would help make an essential normative case: that observed departures from responsiveness are not the kind of departures that ordinary citizens and democratic theorists might consider ethically neutral, or desirable, but embody on the contrary a systematic and indefensible skew towards oligarchy. Many empiricists have already used case studies to provide strong evidence for that. Footnote 128
With all that, however, progressive empirical research will still necessarily require much more simplification of democratic theory, for the purpose of testing causal hypotheses, than normative theorists would recommend for the purpose of critique. Frankly, empirical research would not benefit from constant reflectiveness or infinite theoretical sophistication. Researchers can only perfect ever better ways of measuring a phenomenon, and of tracing its causal relationships to other phenomena, if they eventually cut short their reflections regarding how to define that phenomenon or whether their chosen measure fully captures it. No empirical researcher who consistently applied a Socratic, aporetic, attitude towards a category like “democracy” could get anything done. It is not even clear that when theorists attempt the task they often take to be most useful for empirical research—that of distinguishing one category from a series of closely related ones so that boundaries are sharp—the resulting work will be taken up by empiricists. The problem is that such examinations yield a multiplicity of specialized definitions of a phenomenon, each appropriate for different purposes or to the extent that we care to stress a certain value or set of values. Footnote 129 What empiricists really want from theorists, however, is that they pick one among this Babel so that researchers can get on with studying how a thing is working rather than arguing about what it is. To the extent that normative theorists do affect empirical research, they will do so in a way resembling the “tournament” markets known to economists. A few accounts of what students of democracy should pay attention to will become the bases of extensive research programs; the rest, regardless of argumentative quality, will necessarily be ignored.
In turn, to the extent that normative theorists want to draw insight from empirical political science, they had best do so by studying a variety of specialized debates, each employing a different conceptual toolkit (e.g. debates about democratic peace theory, comparative democratization, democratic responsiveness, and the relationship between democracy and the welfare state), rather than expecting any one of these research programs to acknowledge, much less grapple with, a political concept such as democracy in all its diversity. The reasons for this are partly practical, partly theoretical. To evaluate democracy in all its conceivable dimensions, across all time, we would have to expand colossally the gathering of political information worldwide. This goal, while laudable, is for now very distant. We would also have to accept a model, to a great extent already present in comparative politics (as opposed to empirical American research), in which different teams of researchers each adopt an essentially partisan model of democracy that places greater weight on some indicators than on others. Each such model, or school, or whatever one calls it, would suggest research programs that tracked a favored, essentially ideological specification of the values more essential to democracy. And each would necessarily neglect research questions that partisans of other models would favor (not to mention the contributions of innovative or idiosyncratic theories that would call all existing models into question). Footnote 130
For their part, normative theorists can learn from empiricists a discipline against over-sophistication: One must not derive from the imperfections of public opinion formation a perverse indifference towards evidence that governing elites ignore many voters’ opinions altogether. Finally, theorists might learn from empirical research a sense of which of democracy’s problems are most pressing and tractable and which must remain more distant ideals. Empiricists have found that in many countries, policy outcomes track voter preferences more closely than in the US. But they have not yet discovered a country in which, for instance, typical voters have extremely high interest in politics and judge political questions without reference to partisan cues. Given that, it might make sense to focus immediate attention on bringing the United States closer to the former state, devoting less concern to distant prospects of the latter. Footnote 131 On a more purely intellectual level: an urge for normative insight should lead theorists not to look down on those who trace causal regularities but rather to consume their research with a sense of gratitude. For theorists’ normative and ontological concepts regarding politics ultimately derive not from disembodied reflection on other concepts but from systematic observation of how politics works. Footnote 132
But theorists will not, and should not, abandon the desire to explore democracy in its many meanings. Some of those meanings will enrich large-scale research programs or serve as the basis for new ones. But others will not. Much, perhaps most, democratic theory will never be designed for that use but rather for the sake of critiquing existing conditions and imagining new political possibilities. The primary purpose of normative theory is putting forth, and defending with clarity and rigor, assessments of existing institutions and practices, and persuasive reasons for changing them—not manufacturing tools to help empiricists study their current causal workings (tools that would in any case go mostly unused). Political theory operates most usefully when it turns empirical findings into resources for evaluation and critique. An enterprise devoted to making democracy more perfect, and to raising new issues for debate, will not always be able to give empiricists the kind of conceptual definitions they really want: accessible rather than conceptually precise, easy to operationalize, widely acceptable (at least within a given school or ideological tendency), and above all, unchanging.
