In a democracy, the people are supposed to get what they want. Do they?

This deceptively simple prompt conceals a host of complex normative, conceptual, and empirical questions. Who are “the people?” What does it mean for them to “want” something – as individuals, or as a group – and how can we tell what it is? What would it mean for them to “get” what they want, and does it matter if they get it because they wanted it, or for some other reason? Finally, how much of what they want can we reasonably expect them to get, within a liberal, pluralistic, constitutional democracy – and how much should we want them to get?

None of these matters admit of easy resolution. Yet any attempt to address our orienting question must grapple with them somehow. Such is the burden gracefully borne by Christopher Wlezien’s ambitious field essay, which offers four main contributions: (i) a survey of the empirical evidence scholars have collected in their efforts to assess the extent of policy responsiveness in modern democracies; (ii) a conceptual framework for organizing and interpreting this evidence; (iii) a critical argument that the most pessimistic conclusions of “realists” like Achen and Bartels (Reference Achen and Bartels2016) are overstated; and (iv) a constructive argument that the “demand” side of policy responsiveness is relatively well-understood, while the “supply” side is in greater need of further research. And he achieves a great deal on each front.

Still, certain blind spots and limitations remain. While Wlezien internalizes some of the controversies noted above – spelling out the implications of different answers rather than choosing a side – in other cases he presumes a particular approach. In doing so, he follows conventions common in this corner of the field, where certain conceptual and normative issues are set aside or treated as presumptively settled. Indeed, some such scope-narrowing is nearly always necessary to render important, big picture questions – regarding the quality and fairness of democratic governance, for instance – tractable to empirical measurement. Yet the incumbent danger here, as ever, is that we will become so invested in the answers to these narrower, more precise questions that we will mistake them for the bigger yet fuzzier ones we really care about.

In the terms of a familiar analogy, evaluating democratic governance primarily on the basis of those quantitative data that are easiest to generate is a bit like confining our search for a lost set of keys to the area under a lamppost, simply because that is where we are best able to look. If anything, in fact, the present case is actually more troubling, for the risk is not that we will fail to find an answer, but that we will be overconfident in the answers we do find. More specifically, I suggest in what follows that quantitative studies of policy responsiveness are liable to overstate the fairness and quality of democratic governance, because they neglect to account for forms of capture and distortion by powerful groups that are more difficult to operationalize and measure. Wlezien’s survey of these studies is comprehensive and generally fair: it cannot be accused of presenting an overly optimistic summary of their aggregate findings. Nevertheless, it shares the blind spots of that literature as a whole, and therefore dismisses realist skepticism (such as that of Achen and Bartels) too quickly. By properly situating this literature within broader discussions of democratic values and political equality, this response aims to recenter the big picture – and highlight what may be concealed when we give too much weight to policy responsiveness.

As noted, of course, many of this concept’s boundaries are emphasized by Wlezien himself. Instead of policy, for instance, he notes that other scholars have studied the positions taken by elected leaders, or the priorities they appear to adopt, as the relevant output variable to examine. And even within the domain of policy, there may be significant discrepancies between a law as written and as implemented (or enforced); not to mention its effects on outcomes like growth.

At the same time, he is also concerned with more than mere “congruence” between what the public “wants” and what it “gets”: the idea of responsiveness distinctively implies a causal arrow running from public preferences to outcomes. By itself, that is, policy congruence is insufficient to establish responsiveness, so long as it could plausibly result from processes other than input from independent public preferences constraining representatives’ policy decisions. In particular, responsiveness must be distinguished from advantageous selection, whereby representatives’ acts may align with public preferences, without actively responding to them, simply because voters tend to elect politicians who resemble them. Two forms of reverse causation must also be ruled out: adaptive preferences, whereby people retroactively endorse whatever policy is enacted, in order to reduce cognitive stress; and endogenous preferences, whereby people take cues from representatives about which preferences to express, rather than the other way around. Finally, we must also rule out the possibility that some omitted third cause is responsible for both the policy output and the public preferences favoring that output – such as an elite faction that shapes the agenda of all viable parties, as well as the broader media narratives that inform public opinion.

Compared to the alternatives in both respects, policy responsiveness is relatively difficult for societies to achieve – and also for social scientists to demonstrate. Indeed, one of Wlezien’s main contributions is to catalog a multitude of distinct stages in the causal chain linking public preferences to policy, all of which present distinctive (and relatively common) failure modes.

On the “input” or “demand” side, in sum, policy responsiveness requires that: (a) citizens have preferences about policy that are clear, reasonably stable, and relatively independently formulated (i.e., not primarily adaptive or endogenous); (b) these preferences are unidimensional and single-peaked, or structured in some other way that avoids cycling, such that a meaningful collective preference (e.g., median or majority) can be inferred; (c) citizens have enough accurate information about the behavior of their representatives to assess the latter’s performance in terms of realizing these preferences, under various known and unknown constraints; (d) citizens’ votes (primarily) reflect the judgments that rationally follow from their information and preferences, rather than some other process or heuristic.

On the “output” or “supply” side, meanwhile, it also requires that representatives: (e) correctly perceive “public” preferences (however defined); (f) have incentives to realize those preferences that are strong enough to outweigh other competing incentives they face (including their own personal preferences); and (g) are able to realize those preferences, despite budget and information constraints as well as legal, political, or structural barriers to implementation. As this final point suggests, indeed, many features of a country’s voting rules, federal organization, party system, media landscape, administrative structure, and degree of polarization – not to mention its place in the international order – may all create additional hurdles to responsiveness.

Wlezien characterizes all of these requirements as the “conditions of effective demand and supply” for policy responsiveness, and together they serve as an organizational schema for making sense of the ocean of research surveyed. Each of the conditions (a)–(g), as well as the various contextual factors, has been studied extensively, and Wlezien reads the accumulated results as sobering but not entirely destructive to the possibility of policy responsiveness. Voters may have few clear and independent policy preferences, and these may not always generate clear collective preferences; meanwhile, voters often lack relevant information about representative behavior, and vote on other bases even when they have it. Still, the literature shows that each of these conditions is met in some cases, meaning responsiveness is theoretically possible.

Something similar is true on the “supply” side: representatives may often perceive public preferences inaccurately, have strong competing incentives, and face large implementation challenges, but nevertheless seem to respond to public preferences sometimes. If anything, Wlezien notes, we observe more policy responsiveness than his catalog of obstacles might suggest. Our system may not be “ideal,” he concludes, but it is “pretty much as we might expect, given what we know about the public and policymakers.” Sobering as the facts may be, in other words, Wlezien rejects the strong version of Achen and Bartels’ (Reference Achen and Bartels2016) pessimistic conclusion – i.e., that “elections do not produce responsive government” – as unwarranted.

I am not so sure. There is nothing to complain about in Wlezien’s even-handed assessment of the evidence about the plausibility of each condition on its own. And indeed, he acknowledges that when we put all of these individually sobering results together, it is reasonable to wonder how responsiveness is ever achieved. Yet the proverbial glass nevertheless appears half full to Wlezien – along with many other scholars of policy responsiveness – for the simple reason that responsiveness seems to persist in spite of these formidable obstacles. Like others with a broadly bullish reading of the evidence, for one, he first points to findings of decent congruence between public opinion and policy. He then rules out certain alternative explanations for this congruence; presenting evidence that policy views react thermostatically in a negative direction – the opposite of what an “adaptive preferences” explanation would imply – and that preference changes tend to come before policy changes – the opposite of what “endogenous preferences” would seem to suggest. Yet these results rule out only the crudest versions of those “reverse causation” stories, and (relatedly) do not address omitted-variable explanations: i.e., those which hypothesize that some exogenous force causes changes in both expressed public preferences and enacted policy.

Indeed, Wlezien takes comfort in the suggestion that exogenous (rather than endogenous) causes may be responsible for relatively high degrees of observed preference-policy congruence. But this should only be comforting if we suppose a certain kind of exogenous cause. If people rationally and independently adjust their preferences in response to changes in the world, for instance – and their representatives then adjust policy in response to these preference “shocks” – it is perfectly plausible to understand the resulting policies as “responsive” to voter preferences in a democratically valuable sense. This assessment is less plausible, however, if we imagine instead that these shifts are caused by the influence of certain narrow interest groups or elite factions who simultaneously drive policy outcomes and the dominant cultural narratives from which ordinary people absorb their expressed preferences in largely passive ways.

Of course, the difference between these scenarios cannot be cashed out without examining our core assumptions about what it means for people to “want” something, and why responding to those wants is desirable in the first place. In particular, Wlezien’s glass-half-full view of the observed results depends on two key presumptions: first, that what people “want” is equivalent to “whatever they happen to say in response to a survey”; and, second, that good democratic representation is equivalent to giving people what they want in this rather superficial sense. And to be sure, if our aim is to operationalize the concept of democratic representation so that it can be quantified and measured with maximal precision, it makes sense to accept both premises. If we want a definition that tracks our broader normative intuitions about power and democracy, however, this one leaves much to be desired. And this is where it seems to me that Wlezien, like most scholars who focus on the quantitative measurement of policy responsiveness, has not grappled with the full force of skeptical concerns – as expressed by various (uncited) radical critics as well as the “democratic realists” (like Achen and Bartels) he explicitly takes on.Footnote ¹

Again, there are good reasons for scholars of responsiveness to accept these presumptions. Not only do they seem to render vague notions of democratic quality and fairness in more precise and empirically tractable terms; they also eschew dangerous claims about “false consciousness,” whereby some vanguard or elite is said to know what “the people” need or want better than the people themselves. Whether reactionary or Marxist-Leninist in origin, such claims are indeed unverifiable, insulting, and antidemocratic. Still, it strikes me as equally extreme, and equally implausible, to dismiss any form of power that acts on our preferences rather than our choice set as simply irrelevant to the quality of democracy. Wealthy oligarchs, white settlers and enslavers, upper castes, religiously dominant groups, men – all have sustained their power over dominated groups in part through control over the social norms and cultural narratives that drive preference formation (see, e.g., Wollstonecraft 1994 [1792]; Ambedkar Reference Ambedkar2014 [1936]; Césaire Reference Césaire2000 [1950]). And while these “faces” of power may be more difficult to quantify precisely, that is no excuse for ignoring them entirely (Lukes Reference Lukes1974). On the one hand, troves of qualitative and humanistic scholarship are devoted to exploring the subtle modalities and effects of such “ideological” or “productive” forms of power (Gaventa Reference Gaventa1980; Young Reference Young1990; Hayward Reference Hayward2000, Reference Hayward2013). On the other hand, mountains of experimental evidence demonstrate the critical role of social identity in shaping political beliefs, preferences, and behavior – as summarized most memorably by Achen and Bartels (Reference Achen and Bartels2016). And in doing so, they provide plausible psychological micro-foundations for the mechanisms traced in more interpretive and historical studies (see Bagg Reference Bagg2018a, Reference Bagg2021).

The key lesson of this extensive literature on social identity and motivated reasoning, in my view, is not the one that Wlezien and other scholars of responsiveness are most often at pains to refute: i.e., that policy responsiveness is impossible. Rather, its most potent implications are normative. As I have put it elsewhere, the causes that most plausibly account for the existence of responsiveness, despite the many obstacles it faces, should also lead us to doubt its democratic value (see Bagg Reference Bagg2024, pp. 42–48). After all, the core intuitive reason to value responsiveness to citizen preferences is that these preferences reflect something real and valuable about their will, or their autonomous choices, or their authentic interests. And what this research has reliably demonstrated, in study after study, is that our preferences do not generally reflect an autonomous or rational choice – even when they are decently well informed. In forming our political beliefs, preferences, and behaviors, rather, we largely follow the signals we get from prominent figures who share our identities, telling us what “people like us” think, want, and do.Footnote ² And while this process can generate preferences that reliably reflect our authentic interests, if groups themselves are organized in a healthy, democratic way (Lepoutre Reference Lepoutre2020), this is hardly guaranteed.

In many cases, rather, the profound dependance of political beliefs, preferences, and behaviors on social identity cues furnishes a potent mechanism for wealthy and powerful elites to use in shaping mass preferences. In short, they can and do wield outsized social and material resources to influence which identities people find salient; how they interpret those identities; and which beliefs, desires, and behaviors they adopt as a result. And when this project of shaping “demand” for policy is pursued alongside interventions in the policy “supply” process, the result will be a substantial degree of congruence, which may even be temporally sequenced to appear as dynamic responsiveness. Yet rather than a genuinely egalitarian distribution of power between social groups, these results will reflect the reproduction of elite power. The people may get what they “want” in some sense – but not one we should valorize as straightforwardly democratic.

Of course, none of this implies that elections are entirely meaningless, or that voting is without value (Bagg Reference Bagg2018b). Crucially, we can affirm that each person is a more reliable steward of their own interests, as compared to any particular alternative agent – and thus reject the vanguardist conclusions often thought to follow from simplistic claims about “false consciousness” – while retaining a critical perspective on the various processes of agenda construction and preference formation that translate periodic episodes of unidimensional mass voting into policy outcomes across thousands of unique dimensions. In fact, this is precisely the tenor of recent developments in the political theory of representation. Disch (Reference Disch2021), for instance, formulates her theory of representation as “constituency-making” by bringing literature on responsiveness and political psychology into dialog with humanistic and qualitative perspectives on power. And the result is a more complex and realistic account of what we can and should expect from the relationship between ordinary people and the myriad elected and unelected leaders who represent them in various contexts. Other “constructivist” theorists of representation have made similar interventions (Saward Reference Saward2010; Castiglione and Pollak Reference Castiglione and Pollak2018; Disch Reference Disch2019; Fossen Reference Fossen2019).

In one sense, it is perfectly reasonable for Wlezien to bracket these concerns. The purpose of his field essay is to survey studies of policy responsiveness, giving some structure and coherence to an already vast, unwieldy field of results. He largely succeeds in that aim. Yet as informative as these studies are, on their own strictly defined empirical terms, they are also liable to being over- or misinterpreted – and Wlezien does little to mitigate this danger. Despite efforts to state its general conclusions in relatively weak and provisional terms, his essay’s key takeaway seems to be that skeptics like Achen and Bartels are too pessimistic, and that democracy works about as well as we can expect. In my view, neither conclusion is justified by the results he presents; they get more dubious still when we consider forms of evidence he neglects.

To be sure, many criticisms of Achen and Bartels are valid – including those focusing on their now-infamous “shark attacks” result – and constructivists like Disch have their own reasons to reject the pessimism with which they are often associated. Like many of their critics, however, it seems to me that Wlezien misses the forest for the trees. As I suggested in my discussion of social identity theory above, the key insight of the democratic realist tradition (to which Achen and Bartels belong) is best understood not in strictly empirical terms – i.e., as a claim that responsiveness is impossible – but in normative terms. In short, whatever forms of “responsiveness” we do observe are of relatively limited democratic value, because many of the underlying preferences to which policies respond are themselves of dubious origin.

As Wlezien acknowledges, on the one hand, genuinely dynamic responsiveness is typically observed only on the relatively small number of issues that are highly salient for large numbers of voters. That is surely important: indeed, it is one of the reasons to favor democracy over other forms of government (Bagg Reference Bagg2018b). Yet given that only a few issues can occupy this sort of broad public attention at a time, it still leaves the vast majority of policy outcomes to the elite-dominated realm of “quiet politics” (Culpepper Reference Culpepper2010). Even on high-salience issues, on the other hand, voter preferences are profoundly shaped by social identities whose content and relative salience may be manipulated in democratically suspect ways by those with outsized social, cultural, and material power. This does not mean we should simply ignore the actual preferences of voters in favor of some vanguard’s conception of what they should want, but it is equally misguided to ignore the influence of cultural and ideological power altogether, in our efforts to assess and improve the quality of democratic representation.

Once again, Wlezien’s integrative survey of research on policy responsiveness is clearly a valuable contribution. Yet there are limitations on what that sort of research can tell us, which a field essay of this sort would do best to acknowledge and engage more explicitly. As I have suggested, doing so not only makes it more difficult to embrace the optimism implied by Wlezien’s account – however cautious and provisional it claims to be. It also, more importantly, alters both the urgency and the substantive focus of our efforts at democratic renewal.