Introduction
Concerns about citizen competence are a frequent basis for objections to democracy. From Ancient Athens to contemporary debates, the worry that average citizens are ignorant, prone to false forms of reasoning, and easily swayed by demagogues presents challenges to defenders of democracy (Mill Reference Mill2011; Plato Reference Bloom1968). The emergence of mass opinion research in the mid-twentieth century reinforced these worries (Converse Reference Converse1964; Lippmann Reference Lippmann1993; Schumpeter Reference Schumpeter1976). Contemporary empirical research (for example, Achen and Bartels Reference Achen and Bartels2016; Brennan Reference Brennan2016) is often taken to confirm this pessimistic portrayal of the reasoning capacities of ordinary citizens. To these historical worries about the ability of ordinary people to participate in politics, empirical political science has added concerns about the incoherence of mass opinion (for example, Kinder and Kalmoe Reference Kinder and Kalmoe2017), motivated reasoning (for example, Lodge and Taber Reference Lodge and Taber2013), misinformation (for example, Nyhan Reference Nyhan2021), and echo chambers (for example, Sunstein Reference Sunstein2017). Yet while we now have an abundance of evidence about how ordinary citizens reason about politics, scholars’ evaluation of the implications of this evidence remains unsatisfying. Both competence and democracy are complex, multidimensional concepts, making it challenging to draw normative implications from empirical evidence. Furthermore, recent research has undercut many of the most dramatic negative claims about voter ignorance and political rationality.
This paper has two goals. First, we synthesize recent evidence on citizen competence, drawn largely from the United States but with some replication elsewhere, which undermines some of the more pessimistic accounts. Much established research on citizen competence is based on smaller and older survey results regarding public opinion. More recently, public opinion and political behaviour research has taken a causal turn, often but not always using experiments, while incorporating larger data sets than before. These shifts in empirical standards have made for a more nuanced and ultimately a more optimistic portrayal of citizen competence.
We are not the first to wonder about the strength of empirical evidence for citizen incompetence. Within political theory, scholars have pushed back against sceptical views by emphasizing the virtue of group-based cognition (Chambers Reference Chambers2018; Farrell et al. Reference Farrell, Mercier and Schwartzberg2023; Landemore Reference Chambers2012; Lepoutre Reference Lepoutre2021). Simone Chambers (Reference Chambers2024, 95–96) notes the inconclusive nature of the individual-level evidence for citizen incompetence. Our argument against the incompetence position joins together a range of recent empirical findings. For example, while psychological research on motivated reasoning (Lodge and Taber Reference Lodge and Taber2013) is taken to support some of the more dire concerns about democratic citizens, more recent experimental evidence indicates that citizens can become substantially more knowledgeable about politics, even when the issues before them are complex and cut against their partisan commitments (Prior et al. Reference Prior, Sood and Khanna2015; Bullock et al. Reference Bullock, Gerber, Hill and Huber2015; Wood and Porter Reference Wood and Porter2019). When misinformation is followed by corrective information, the effects of misinformation on belief accuracy are virtually eliminated (Porter and Wood Reference Porter, Velez and Wood2022; Nyhan, Porter, Reifler and Wood Reference Nyhan, Porter, Reifler and Wood2019).
In this paper, we synthesize new evidence which shows that various putative threats to citizen competence, including motivated reasoning, echo chambers, affective polarization, a propensity for violence, and elite manipulation, do not reflect permanent characteristics of citizens. Insofar as these cognitive issues arise, they can be mitigated through widely available interventions. Citizens, in short, are not immutably incompetent. But are they competent? That is, are they able to behave as democracy demands? We argue that, by the standards of several theoretical perspectives, they are.
This brings us to the paper’s other contribution, which is to provide a new conceptual framework within which to think about citizen competence in electoral democracy.Footnote 1 Institutions realize multiple political values, and so our evaluation of citizen competence depends, in large part, on how scholars consider the relationship between those different moral goods that democracy produces. We make no claim to settle the debate about the value of democracy; rather, our hope is that drawing these ideas into current research on citizen competence can enable a more refined evaluation of the evidence as well as a more productive conversation between democratic theorists and empirical political scientists. In particular, we distinguish between the individual epistemic capacities and virtues required for reliability, accountability, and inclusive equality.
Reliability requires that individuals in democratic systems do not make decisions that fail to track certain scientifically knowable features of external reality and so are objectively harmful to the interests of citizens. Reliability in a democratic regime would be undermined if citizens lacked the capacity to at least somewhat accurately understand the contested factual issues of the day. Accountability requires citizens to connect political causes to consequences such that they can correctly ascribe responsibility for political outcomes. Finally, relating to fellow citizens as equals – inclusive equality – requires accurately understanding the beliefs and motivations of their fellow citizens. Each value, however, requires different sorts of competence and is vulnerable to different sorts of pathologies, and each requires different institutional contexts and systemic supports to function well.
Ultimately, this paper aims to both advance a more complex account of competence and the relationship between competence and the value of democracy and then to show that this can inform a renewed research agenda into citizen competence. We define citizens as competent if they have the necessary cognitive abilities to ensure that democratic institutions generally meet the criteria of reliability, accountability, and inclusive equality. Our review of new empirical evidence largely suggests that they meet this criterion. More often than not, citizens are capable of forming beliefs that mirror empirical reality, they can incrementally hold politicians accountable for their decisions, and they can abide by democratic procedural norms, very rarely willing to embrace violence to resolve political conflict.
Yet these findings do not mean we are blind to the challenges facing democracies in an era of new forms of media and increasing polarization around scientific debates such as climate change. All of these point to strains on the epistemic infrastructure of democratic decision-making (Herzog Reference Herzog2023). But we contend that much research has become deadlocked because it is inordinately attentive to various individual-level measures of competence, thereby neglecting consideration of competence at the system level (Chater and Loewenstein, Reference Chater and Loewenstein2023).
Our focus is primarily on the competence required for institutional mechanisms that produce binding outcomes. We focus mainly on elections within representative democracy, although we also examine evidence drawn from direct democracies in which referendums play a significant role. Within democratic theory, there has been a long-standing interest in various forms of small-scale deliberation through minipublics, citizens’ assemblies, or other devices (for example, Fishkin Reference Fishkin2009). While we are sympathetic with these projects, we set them aside to focus on the competence of citizens for more basic elements of democracy that have also attracted scepticism. We want to know whether citizens can realize democratic values even in the absence of demanding institutional reforms that are difficult to scale. But we hope our findings can also provide succour to advocates of deliberative democracy and institutional innovations.
We begin with our account of the relationship between citizen competence and the value of democracy. We next turn to a revisionary evaluation of the latest evidence on citizen competence within democracy, showing that the most recent evidence reveals surprisingly optimistic conclusions about their competence. Finally, we tease out a multi-disciplinary research programme focused on competence at the system level. We believe that such a programme would be able to better reflect recent evidence and account for the valuable perspectives of both normative theorists and empirical scholars.
Citizen Competence and the Value of Democracy
What should we expect of citizens in a democracy? Our answer depends on what we expect democracy to accomplish. Most mainstream research on citizen competence tends to accept what Andy Sabl calls the ‘folk or popular theory of democracy’ where the central value of democracy is responsiveness, ‘the common lay belief that politics can and should give the people what they want, and should change in response to changes in what they want’ (Sable Reference Sabl2015, 347). For example, Druckman describes the basic problem of citizen competence as follows: ‘A basic conception of democratic competence, then, requires that citizens be well qualified or capable to meet their assigned role. Their assigned role, in turn, concerns the expression of their preferences to which governors can and should respond’ (Druckman Reference Druckman2001, 232).
Without a doubt, responsiveness is an important aspect of democracy. But as many democratic theorists have argued (Sable Reference Sabl2015; Disch Reference Disch2021), the mainstream view tends to reduce democracy to responsiveness, crowding out other values. Our goal is not to replace the mainstream view but to situate it within a larger framework. For us, responsiveness helps spell out the normative value of accountability, which we situate alongside reliability and inclusive equality. In this section, we turn to different views of the normative value of democracy and examine the different sorts of knowledge they assume voters have. In particular, we distinguish between the epistemic assumptions of the so-called instrumental view of democracy – which we divide into strong and weak instrumentalism – and of the alternative, intrinsic view of democracy. According to instrumental views of democracy, democratic institutions are valuable insofar as they produce certain good outcomes. By contrast, defenders of the intrinsic value of democracy argue that there are qualities inherent to democratic procedures that generate value. While often pitched as rival views, there is no reason to think they are mutually exclusive (Anderson Reference Anderson, Christiano and Christman2009). We are also not saying that the authors we identify with each camp only support the single value that we identify. Rather, our goal is to produce a stylized typology that can clarify the different dimensions along which we may worry about citizen competence. Democracy is a multi-faceted, complex concept that points to institutions that realize multiple values, and the different values, in turn, require different sorts of citizen competence. Drawing these distinctions enables us to better evaluate the empirical evidence regarding citizen competence, as well as to identify the epistemic requirements of citizenship more precisely in the context of an epistemic division of labour. We also do not claim that our typology of democratic values is exhaustive. Rather, we select three values – reliability, accountability, and inclusive equality – that we take to be most relevant for thinking about citizen competence.
Voter competence-based critiques of democracy tend to conflate several different types of relevant knowledge and political capacities. If the purpose of voting is just for citizens to express their interests or values on equal terms with others, it may be irrelevant if they can identify specific political officeholders or policies. Conversely, if the purpose of voting is to hold politicians accountable, we would be justifiably worried if citizens proved incapable of connecting political decisions to consequences. Before we can fully evaluate the citizen competency objection, we need a more precise, differentiated account of the nature of citizen competency itself. This section pursues this goal, considering the latest accounts of the value of democracy.
Broadly speaking, we can distinguish three broad camps when it comes to thinking about the normative basis of democracy: strong instrumentalism, weak instrumentalism, and intrinsic value, each of which accords with a different cluster of values we may want democratic institutions to achieve. Strong instrumentalists contend that democratic institutions have to track some independent standards, whether of truthfulness or justice or good outcomes (Arneson Reference Arneson, Dowding, Goodin and Pateman2004). Thus, Jason Brennan argues that individuals have a right to be ruled by the competent, insofar as political power is only justified insofar as it leads to good outcomes for those subject to it. From this perspective, we can compare democracy to various systems of ‘epistocracy’ that delimit political decision-making to highly capable individuals (Brennan Reference Brennan2016). The exercise of political power is then justified only if those exercising it are the most competent feasible, as that will lead to the best outcomes. A weaker version of this argument contends that democratic institutions do need to reliably find ‘true’ answers to political questions, but just perform better than random (Estlund Reference Estlund2007). Some agree that democratic institutions are justified at least in part by their ability to reliably track good outcomes, but argue that the empirical evidence points to the epistemic benefits of group decision-making (Landemore Reference Chambers2012).
In our terminology, the strong instrumentalist position is concerned with reliability. For our purposes, we understand reliability as the ability of a political system to meet a set of minimal external standards of good governance that any exercise of political power must meet for there to be some obligations on the part of citizens to obey the law. We can see these principles invoked for various forms of technocratic delegation, such as to central banks, which would empower experts to run policy in accordance with objective, scientific knowledge. The value of reliability is one that all exercise of political power, whether democratic or not, has to reach – the value is entirely external to democracy, and so we can call this a ‘strong’ instrumentalist position insofar as it only values democracy as one of many possible instruments for realizing reliability. Thus, we could imagine a world in which democracy realized all sorts of values internal to democracy – ensuring everyone an equal say, holding representatives accountable, and so on – and yet democracy would still be unreliable, failing to meet strong instrumentalist standards, insofar as it led to outcomes that were below a threshold or inferior to an imagined alternative such as rule by experts.
Another version of instrumentalism – weak instrumentalism – focuses not on whether democratic procedures track external standards like truth or justice, but rather whether democratic procedures instrumentally realize certain political goals, like holding elites accountable, ensuring rotation of office without civil war, or dispersing political power. This view is still instrumentalist because it focuses on the harmful outcomes democratic institutions help to mitigate, such as elite capture, rather than the intrinsic goods of democratic procedures (Bagg Reference Bagg2018). But that harm is avoided because of the way that democratic institutions organize political power, and so the normative value in question is realized by our empirical knowledge of how democratic institutions function over time, rather than just the policy results that democratic institutions produce. Whether a politician is accountable is separate from whether democratic institutions realize good policy outcomes. While strong instrumentalism spells out some external ideal against which to judge democracy, weak instrumentalism focuses on the grave harms that can come from the abuse of political power. And while strong instrumentalism compares democracy to non-democratic alternatives, weak instrumentalism focuses on political goods that can only be produced through democratic regimes, proposing a normative theory of democracy that then matches up with a conceptualization of democracy that focuses on things like the rotation of power through election.
The core normative value for weak instrumentalism, then, is accountability. As with reliability, the concept of accountability can be spelt out in different ways. Accountability can mean a situation where politicians are forced through sanctions or institutional incentives to follow or track the views of the public, or it can mean that politicians face a genuine threat of losing power so that there is rotation in office, or it can mean some more complex ideal of answering to the views or needs of constituencies (for example, Ashworth and Bueno de Mesquita Reference Ashworth and Bueno de Mesquita2014; Achen and Bartels Reference Achen and Bartels2016). Nonetheless, we believe that accountability points to a cluster of views distinct from reliability – one can have reliability without accountability and accountability without reliability. This is so even as we may think that systems with accountability are also more likely to produce good outcomes. And as with reliability, while some views argue that the value of democracy is just accountability, we believe we can talk about weak instrumentalist political values – such as accountability, avoiding domination, ensuring the rotation of office, and so on – without necessarily thinking that these goods exhaust the value of democracy. Most mainstream political science views of the function of democracy, we would contend, fall into this weak instrumentalist camp.
Arrayed against these instrumentalist positions are those who think that democracy is justified by qualities inherent to the internal structure of democratic institutions, and not just either an external standard of reliability or some outcome those institutions tend to produce, such as accountable leaders. Take, for example, a simple majoritarian decision-making procedure, where everyone gets to vote directly on the outcome. In giving everyone an equal probability of being the decisive vote, the decision-making procedure treats everyone fairly. But this fairness is independent of the outcome that the procedure produces. Even if it leads to an outcome that you disagree with or fails to meet standards of reliability, we can still say the procedure treated you fairly. Thus, the procedure directly realizes the good of fairness, and so it is an intrinsic quality of the procedure. Some want to argue that the structure of certain democratic procedures, such as equal voting rights in a simple majoritarian procedure, unconditionally and directly constitutes certain normative values like equality (Kolodny Reference Kolodny2014; Viehoff Reference Viehoff2014). Others argue that the point is that certain features of democratic institutions, such as equal voting rights, symbolically represent some value, such as equal citizenship, which can then coexist with those institutions realizing a variety of other values, such as instrumental goods, such as good governance (Anderson Reference Anderson, Christiano and Christman2009; Christiano Reference Christiano2008).
From this perspective, democratic institutions help constitute the value of what we call inclusive equality through procedural fairness and symbolic recognition of equality. We call this inclusive equality because democratic institutions can produce inclusion in a polity on unequal terms – there may be unequal forms of inclusion, as well as forms of equality (such as basic moral equality) that are not relative to co-membership in a political system. The idea of inclusive equality is meant to capture a range of normative intuitions regarding the intrinsic value of democratic procedures. By intrinsic value, scholars mean a value of some institution or practice that is distinct from the causal effects of that institution on the world. So, equal voting rights, for example, are intrinsically valuable because such voting rights represent, or directly constitute, political equality between citizens. Put differently, we can evaluate whether democratic institutions realize inclusive equality independently of whether they realize reliability or accountability.
What level of citizen competence is required for democratic institutions to produce each of these values? To fully evaluate this question, we need evidence not only of individual-level competence but also of how different institutional mechanisms respond to different levels of competence. To some extent, we will examine this in our discussion of the empirical literature. Our goal is to mainly specify the kinds of relevant competence and generate some rough thresholds to guide the following discussion and refute some of the more extreme claims about citizen incompetence. We hope this can set an agenda for better connecting democratic theory, individual-level evidence about political knowledge and competence, and evidence of the interaction between individual-level competence and institutional and systemic dynamics.
While some strong instrumentalists like Brennan marshal the evidence about the lack of citizen competence, all that a defence of democratic institutions on strong instrumentalist grounds has to show is that democratic institutions are, in general, more reliable than alternatives in tracking objective truth or objectively good outcomes. This could be so even if people have quite low levels of political competence (Brinkmann Reference Brinkmann2018). Conversely, both the weak instrumental and intrinsic views are quite demanding, insofar as both point to the sort of competency required to make democracy work internal to democratic standards. In other words, while for strong instrumentalism, all that matters is that collective decisions are more reliable than alternatives over time, for weak instrumentalism and intrinsic views, we need a certain threshold of citizen competence even for democratic institutions to realize the values under consideration. Thus, for weak instrumentalism, if citizens lack the ability to connect political leaders’ actions to consequences, then democratic institutions will fail to realize the instrumental goods of elite accountability or dispersed power.
Things are more complex for views that insist on the intrinsic value of democracy. These views seem to discount the quality of democratic outcomes as compared to some external standards. This would seem to downplay the role of citizen competence, which is typically related to the ability to realize good political outcomes. Yet these views can be quite demanding about the sorts of attitudes participants in democratic systems have to have for those systems to realize the intrinsic values. For most accounts of the intrinsic value of democracy, the value democracy realizes is something like relating to each other as equals and so a sort of mutual respect reflected in fair procedures (Christiano Reference Christiano2008; Kolodny Reference Kolodny2014; Viehoff Reference Viehoff2014). Yet for those procedures to produce the goods of inclusive equality, participants in them also have to take an appropriate attitude towards each other. Thus, citizens require a high degree of competency to appropriately appraise, not external facts or the connections between political decisions and outcomes, but the attitudes and beliefs of those with whom they disagree (cf. Cramer and Toff Reference Cramer and Toff2017).Footnote 2 More minimally, they must believe that all citizens ascribe to the value of democratic institutions, as well as the norms of democratic procedures, and are willing to abide by the outcomes of such institutions.
Table 1 summarises these differing perspectives. Strong instrumentalism repeats Plato’s classic challenge to democracy: that it empowers individuals who do not have the objective knowledge necessary for good political decision making. Weak instrumentalism focuses on the knowledge required for Max Weber (Reference Weber and Livingstone2004) ethos of responsibility: Can we connect our political actions to their consequences? The worry is that certain features of political systems – the distance between ordinary voting and actual policymaking, the inability to observe multiple outcomes, and the specialized, often technical nature of policy making – all undermine the ability of ordinary citizens to accurately appraise what politicians are doing. The forms of competence required for weak instrumentalism will often overlap with those required for strong instrumentalism. For example, citizens may need to be able to gauge the causal effects of government policies on the natural or social world to know whether to hold politicians accountable. So, even as good outcomes and accountability are conceptually distinct, in practice, the forms of knowledge required to realize them will often overlap.
Theories of Democracy and Types of Political Competence
For those who value democracy for intrinsic reasons, the relevant form of competence echoes the Federalist Papers: Can people overcome their natural tendency towards factionalism so as to realize the goods of mutual respect that democratic institutions promise? Here, the relevant knowledge is of the motivations of the other side. An extreme version of this claim is made by Viehoff in his defence of the intrinsic value of democracy. He writes, ‘The intrinsic value of egalitarian relationships depends (among other things) on the equal concern that the parties have for one another. Where I treat the equal power that I have merely as an instrument for advancing my own good, without concern for the well-being of my friend or colleague, or the common good or internal justice of our group or community, submitting to our collective decision does not uphold a valuable relationship among equals’ (Viehoff Reference Viehoff2014, 373). His account suggests the high cognitive burden associated with the intrinsic view. Democratic procedures, Viehoff argues, may ‘lack authority because the citizens vote or act on the basis of the wrong kind of reasons’ (Viehoff Reference Viehoff2014, 373). This implies that citizens have knowledge, not only of the issues of the day and how they interact with the political system, but to some degree of the motivations of their fellow citizens and whether their votes are motivated by concern for the broader good and not just their own narrow self-interest. Otherwise, citizens would have no way of knowing whether democratic procedures produce authoritative outcomes.
Other versions of the intrinsic view are less demanding but still have fairly high cognitive requirements. Symbolic views, according to which fair procedures publicly recognize our equality, require knowledge of shared beliefs about the meaning of those symbols (Christiano Reference Christiano2008). Even more minimal intrinsic views, such as a view that equal voting procedures just treat everyone fairly, regardless of the attitudes of the participants (Waldron Reference Waldron1999), will, as a more empirical matter, require everyone to think that those they disagree with are committed to the fairness of the procedure – otherwise, individuals and groups will start to defect from political cooperation. We do not aim to settle on an exact account of the relationship between inclusive equality and co-citizen attitudes. But we do contend that part of citizen competency is citizens’ accurate knowledge of the intentions and goals of other citizens, as well as broadly shared commitments to the value of democratic institutions.
In all, then, the intrinsic view of democracy can presuppose quite demanding cognitive and motivational requirements. And from this perspective, the great threat to inclusive equality is forms of polarization that undermine citizens’ ability to view each other as equals. Citizens must have accurate knowledge of their co-citizens’ motivations and commitment to shared democratic norms. Yet this also opens up a range of complex questions, especially in societies where some citizens are not committed to the equal standing of some other citizens.
In sum, both democracy and competence are internally complex concepts. How we think about democracy has implications for how we think about citizen competence. Different conceptions of democracy lead to different requirements for citizen competence. While much of the debate about citizen competence takes responsiveness as the key democratic value, we have identified at least three different clusters of values that we may care about: strong instrumental values, weak instrumental values, and intrinsic values. And the ability of democratic institutions to realize each set of values could require distinct types of competencies as well as different thresholds.
Recent Evidence Against Citizen Incompetence
With this conceptual framework in mind, we now turn to recent empirical evidence relating to the nature of citizen competence. In short, this evidence shows that, across multiple dimensions, the incompetence of everyday people for democratic citizenship has been exaggerated. Below, we outline the extent to which popular depictions of democratic incompetence are challenged by empirical evidence.
Much of the evidence we present was gathered recently, subsequent to empirical political scientists becoming more interested in causal identification, pre-registration and analyzing larger, more comprehensive data sets than was previously possible or expected (so-called ‘big data’). This is not to say that present standards are flawless; the use of convenience samples, for example, may be more widespread now than before. But contemporary empirical social science is more insistent about measuring cause and effect than what preceded it, and has more tools at its disposal to do so. Similarly, most (but not all) of the evidence we describe comes from US-based samples. On the one hand, this represents a clear limit to our conclusions; on the other hand, we must note that much of the previous evidence often used to justify pessimistic accounts (for example, Converse Reference Converse1964; Taber and Lodge Reference Lodge and Taber2006; Achen and Bartels Reference Achen and Bartels2016) also relies heavily on US-based samples. Still, readers should keep in mind that much of the evidence discussed below does come from the USA.
With these limitations in mind, our conclusions apply across the different dimensions of competence that we have identified. That is, there is evidence that citizens are competent enough to realize the values of reliability, accountability, and inclusive equality, even as each requires different cognitive abilities and could be amenable to different sorts of interventions. We discuss each of these dimensions in turn, rehearsing common concerns and describing the state of the current evidence.
Reliability
People lack factually accurate knowledge
One of the central worries about democracy is that, given people’s ignorance of empirical facts, democratic procedures will produce outcomes that contradict or fail to act on objective features of the world, thus leading democracies to fall below a minimal threshold of reliability in addressing common issues. A lack of factual knowledge about politics can also be taken as evidence that people do not have the requisite knowledge to hold politicians accountable. While scholars have long been concerned that people lack sufficient levels of factually accurate knowledge, fresh evidence has cast doubt on that conclusion.
Earlier claims of widespread mass ignorance were at least partially attributable to measurement error. For example, it was long claimed that too few Americans are unable to accurately identify the Chief Justice of the Supreme Court. Yet, upon closer inspection, the evidence for this concern was based on poor survey practice. In their review of the American National Election Study (ANES) archives, which had conducted the surveys in question, Gibson and Caldeira (Reference Gibson and Caldeira2009) detected that many survey respondents who had provided verbal answers to this question were counted as providing incorrect answers, even though a fair reading of their answers offers reason to think that such respondents did indeed know the right answer.
Even if the extent of the overall public’s ignorance has been exaggerated, it is concerning that different groups (for example, males and females) appear differentially knowledgeable (a concern voiced by Delli Carpini and Keeter Reference Delli Carpini and Keeter1997 and Barabas et al. Reference Barabas, Jerit, Pollock and Rainey2014). Recent evidence tempers this concern. Perez (Reference Perez2015) demonstrates that the supposed disparity in political knowledge between whites and Latinos is largely explained by this dynamic; when Latino respondents answer Latino-themed questions, the knowledge gap narrows significantly. The same is true about the putative gap in knowledge between the sexes; when knowledge batteries are balanced with items for both males and females, the knowledge gap is no longer significant (Kraft and Dolan Reference Kraft and Dolan2023). Others (for example, Mondak and Anderson Reference Mondak and Anderson2004; Abrajano Reference Abrajano2015) have reached startlingly similar conclusions. As Perez puts it: ‘Large group deficits in political knowledge are artificially inflated by uneven item performance in diverse populations’ (Perez Reference Perez2015, 951).
Other work has clarified the means by which the prevalence of misperceptions has been exaggerated. Using panel data, Graham (Reference Graham2022) shows that those who endorse falsehoods typically do so with much more uncertainty than those who do not and that their responses are more unstable across survey waves. People who appear to be misinformed at one point in time are just as likely to be making poorly educated guesses that they revise, for the better, in the future. Popular accounts of a widely misinformed public often draw overly broad inferences from guesses given by uncertain survey respondents, whose answers are likely to change later.
A related response to concerns about the public’s level of political knowledge has taken a different approach and questioned the premise of measuring political knowledge in this way altogether. If, say, Gibson and Caldeira had never done their work, and we still (erroneously) believed that people were woefully unaware of the identity of the Supreme Court’s Chief Justice, why would that deficiency matter? Such questions may be no more than political trivia. Those who answer well may be well-suited to win matches of pub trivia, yet their preparation for democracy may be another matter entirely. Particularly in the age of the smartphone, committing political facts to memory seems less urgent – if it ever was (Delli Carpini and Keeter Reference Delli Carpini and Keeter1997; Lupia Reference Lupia2015). In all, owing to measurement error, question wording and instability, concerns about incorrect political knowledge are not in line with the evidence. That is not to say that citizens are perfectly informed or even as informed as some might desire. The evidence we have reviewed should not resolve all concerns about political knowledge. But ignorance does not undermine competence in a way to affect democracy’s ability to realize reliability, accountability, or inclusive equality.
People react adversely to conflicting factual information
The experience is typical: We attempt to convince an interlocutor of a position by providing them with new information, but our attempt fails. Perhaps our attempt even appears to strengthen their prior position. In the study of factual accuracy and politics, this phenomenon has a name: backfire. Related to motivated reasoning, backfire posits that the provision of factually accurate information will compel people whose political views conflict with that information to double down. Most famously, in ‘When Corrections Fail’, Nyhan and Reifler (Reference Nyhan and Reifler2010) show that providing US conservatives with factually accurate information about the War in Iraq caused them to become more convinced of falsehoods about the war. This could undermine the reliability of democracy, if it means democracy is persistently less likely to track truthful outcomes than other systems, as well as the accountability of political leaders, insofar as they may be able to mobilise the backfire effect to their advantage, and the mutual recognition of equal citizenship if we view our co-citizens as animated entirely by partisanship emotions and averse to plain evidence, rather than a sincere desire for truthful accuracy.
The backfire effect does not replicate. Plainly put, people typically respond to factually corrective information by becoming more accurate, irrespective of their prior political beliefs. Evidence for this conclusion is summarized by the cross-disciplinary team in Lewandowsky et al. (Reference Lewandowsky, Cook, Ecker, Albarracín, Amazeen, Kendeou, Lombardi, Newman, Pennycook, Porter, Rand, Rapp, Reifler, Roozenbeek, Schmid, Seifert, Sinatra, Swire-Thompson, van der Linden, Vraga, Wood and Zaragoza2020). Even Republicans who were shown corrections of falsehoods attributed to President Trump responded by becoming more factually accurate (Nyhan, Porter, Reifler and Wood Reference Nyhan, Porter, Reifler and Wood2019; Porter and Wood Reference Lewandowsky, Cook, Ecker, Albarracín, Amazeen, Kendeou, Lombardi, Newman, Pennycook, Porter, Rand, Rapp, Reifler, Roozenbeek, Schmid, Seifert, Sinatra, Swire-Thompson, van der Linden, Vraga, Wood and Zaragoza2020). The positive effects of factually corrective information on belief accuracy are not indefinite (Carey et al. Reference Carey, Guess, Loewen, Merkley, Nyhan, Phillips and Reifler2022), but do persist beyond immediate exposure (Porter and Wood Reference Porter and Wood2021). Meta-analysis further corroborates the efficacy of correcting misinformation (Chan et al. Reference Chan, Christopher, Kathleen Hall and Dolores2017). To be clear, it may still be possible that, in very select circumstances, people respond to factually accurate information by becoming less accurate. Yet the search for the backfire effect has mostly turned up empty (for example, Swire-Thompson, DeGutis and Lazer, Reference Swire-Thompson, DeGutis and Lazer2020).
Similar conclusions are reached by Guess and Coppock (Reference Guess and Coppock2020). As summarized by Coppock (Reference Robert2023, 3): ‘When people encounter new information, they don’t distort it to further entrench their preexisting views – instead, they are persuaded in the direction of that information.’ The distinction between the public being misinformed and uninformed (as described by Kuklinski & Quirk, Reference Kuklinski and Quirk2001 and others) is instructive here. Members of the public are not as misinformed as some fear, and quite clearly, they are capable of becoming better-informed.
Yet it is possible that group dynamics will still prove challenging for reliability. Especially since the rise of the Internet, scholars have expressed concern that groups of individuals bring out behaviours that are particularly problematic for democracy. For example, Sunstein (Reference Sunstein2002, Reference Sunstein2017) has repeatedly raised alarms about individuals’ penchant for selecting into groups of the like-minded, which he argues will exacerbate polarization and extremism. In its most worrisome form, homophily is said to create ‘echo chambers’, in which people exclusively surround themselves with like-minded individuals, and expose themselves to information only if it is congenial to their political preferences.
In two key respects, however, concerns about echo chambers appear to be overstated. First, some compelling evidence, based on individual-level web browsing data, shows that fewer people than popularly assumed are actually entrenched within echo chambers. While people ‘choose what they like’, and Democrats and Republicans have distinct media diets, ‘the preponderance of the content encountered is ideologically moderate’ (Guess Reference Taber and Lodge2021). At least on Facebook during presidential elections, this dynamic appears to shift, with social media users exposed to and interacting with content consistent with their ideological preferences (Gonzalez-Bailon et al. Reference Gonzalez-Bailon2023). Yet any observations about the effects of echo chambers are conditional upon the consumption of political content in the first place – and it turns out that this is surprisingly rare. Outside of elections, not as many people consume political content, and those who do are mostly not living in echo chambers of their own design.
Second, the evidence on echo chambers indicates that they may not be as pernicious as some fear, and that mitigating them does not help reduce the problems they are said to cause. In one set of studies, Becker, Porter and Centola (Reference Becker, Porter and Centola2019) randomly assigned participants to echo chambers, composed only of people from the same political party. They then assessed whether echo chambers would diminish or enhance factual accuracy over a range of politically contentious topics. Contrary to the most pessimistic expectations, factual accuracy increased more in echo chambers than in non-echo chambers. When surrounded only by like-minded partisans, people did not respond by rejecting empirical evidence. Instead, they became more accurate as a result, even when the topic did not flatter their partisan predispositions (for example, when Republicans were asked to provide accurate information about the economic record of the Obama Administration).
Other research on echo chambers has found that mitigating them does not repair the problems they are purported to cause. During the 2020 US election, researchers worked with Facebook to randomly assign people to see less like-minded content. The result: No detectable reductions in belief in misinformation and attitude polarization (Nyhan et al. Reference Nyhan, Settle, Thorson, Wojcieszak, Barberá, Chen, Allcott, Brown, Crespo-Tenorio, Dimmery and Freelon2023). This finding corresponds with Bail et al. (Reference Bail, Argyle, Brown, Bumpus, Chen, Hunzaker, Lee, Mann, Merhout and Volfovsky2018), who find that exposing people to counter-attitudinal information on social media – effectively breaking participants out of echo chambers – increases polarization. In sum, echo chambers are neither as common nor as deleterious as feared, and that reducing them does not have the positive effects one might expect. We have less reason than thought to fear that when citizens form groups, they will be led astray by group dynamics.
Overall, the evidence does not accord with the worry that citizens lack the factual information – or are so clouded by motivated reasoning – to enable democratic outcomes that track objective features of the world. Citizens have the cognitive capacities to, in general, ensure that democratic institutions realize reliability.
Accountability
People’s views and behaviours are captive to elite manipulation
So, people are neither factually ignorant about politics to the extent often assumed, nor do they double down on false beliefs while spending most of their time in pernicious echo chambers. But are they vulnerable to manipulation from political elites, such that they will be unable to hold them accountable? There is little doubt that people move in accordance with cues they receive from political elites. Political leaders have sway over their followers’ policy preferences, as well as their opponents’ (Lenz Reference Lenz2012). Attitudinal movement by elites in a particular direction can cause people to move in the same direction (Zaller Reference Zaller1992). These findings are taken to be one of the strongest arguments against the ability of democracy to hold elites accountable. Yet the evidence that elites can influence mass opinion does not support the stronger claim that people are captive to elite manipulation. Many – if not most – attempts by elites to manipulate political beliefs and behaviours are likely to either fail or have vanishingly small effects.
Consider the accumulated evidence on the effects of campaign messaging. Whether with television advertisements, direct mail, or door knocks, campaigns from all sides attempt to win supporters by delivering carefully crafted messages, often to highly targeted groups of respondents. Yet the evidence is this clear: These efforts work far less often than commonly assumed. In an exhaustive meta-analysis of experimental evidence, Kalla and Broockman (Reference Kalla and Broockman2018) demonstrate that the effect of most campaign messaging is a precisely estimated zero.
When such attempts do succeed, their effects can be short-lived. In a randomized experiment in partnership with a large gubernatorial campaign, Gerber and Green (Reference Gerber and Green2011) find that even well-made, professional television advertisements only shortly increase positive affect towards the candidate and that the effects are no longer detectable a week after exposure. Over fifty-nine experiments conducted with political ads during the 2016 election, Coppock, Vavreck and Hill (Reference Coppock, Hill and Vavreck2020) corroborate this conclusion, finding tiny, fleeting effects. In other words, the typical attempt by a political leader to win voters will fail; and even when such attempts succeed, they do so for a very short amount of time. People are simply not as gullible (Mercier Reference Mercier2020) as democratic pessimists sometimes fear.
Evidence from direct democracy further ameliorates the worry that voters’ preferences are simply shaped by elite cues. Going back to Bower and Donovan (Reference Bowler and Donovan1998), scholars have found that voters are reasonably good, although not perfect, at understanding ballot initiatives and voting in a way that reflects their preferences. In her comprehensive analysis, Colombo (Reference Colombo2018) finds that 70 per cent Swiss referendum voters can justify their choses and express an understanding of the issues. Similarly, Nai (Reference Nai2015) finds that low information voters are largely able to use information shortcuts to vote correctly (cf. also Milic Reference Milic2012). This echoes Lupia’s (Reference Lupia1994) earlier findings that information shortcuts enabled voters in complex ballot initiatives to make correct choices. Recent evidence, then, supports the view that when they face issues directly, voters are able to largely follow their own preferences and are not subject to elite manipulation. As always, the evidence does not reveal perfect voters. Many low-information voters still vote incorrectly in ballots, and Nai finds that more intensive campaigns reduce voter accuracy. The challenges are particularly acute in highly polarized, one-off referenda, such as those relating to the European Union. For example, studying Ireland’s vote on the Lisbon Treaty, Elkink and Sinnott (Reference Elkink and Sinnott2015) found significant misunderstandings among no voters and that their self-perceived ignorance influenced their rejection of the Treaty. But overall, the evidence from contexts where voters are making more repeated decisions on clearer, policy-related ballots portrays voters as largely capable of understanding the issues, coming to a conclusion based on their own values and information shortcuts, and then, indeed, becoming more knowledgeable of politics as a result of the availability of referendums (Bowler and Donovan Reference Brinkmann2002; Smith and Tolbert Reference Smith and Tolbert2004). All that being said, there are some circumstances under which political elites can change minds and behaviours. Using observational evidence, Sides, Vavreck and Warshaw (Reference Sides, Vavreck and Warshaw2022) find that large advantages enjoyed by one party in the quantity of political advertisements shown can redound to that party’s benefit. These benefits, however, are quite small at the national level and only become substantial at the local level. As the authors describe, the differences in effect size are likely attributable to differences in political knowledge about local politics, compared to national politics. People know more about national politics and are thus less affected by attempts to manipulate their beliefs about national politics.
Similarly, Kalla and Broockman observe that campaign messages focused on ballot initiatives tend to be influential and impact voters’ decisions. For both local candidates and ballot initiatives, it seems plausible that elite messaging fills gaps in political knowledge and affects behaviours as a result. This is not necessarily the stuff of elite manipulation, but of elites providing information that voters find useful. In the absence of knowledge, people listen to elites, but on topics about which they already have substantial reservoirs of knowledge, the effects of elite persuasion efforts run the gamut from tiny (Sides, Vavreck and Warshaw Reference Sides, Vavreck and Warshaw2022) to fleeting (Gerber and Green Reference Gerber and Green2011) to non-existent (Kalla and Broockman Reference Kalla and Broockman2018). In all, then, the evidence of elite manipulation is weak, and many instances of elite signalling may actually make up for the lack of factual information discussed above.
People cannot hold politicians accountable
One of the strongest allegations against democratic competence focuses on citizens’ ability, or lack thereof, to hold politicians accountable for past decisions. According to Achen and Bartels, this ability is not only limited; it is ‘blind’, insofar as it amounts to citizens blaming politicians for decisions squarely beyond their control. As they put it, ‘Incumbents will pay at the polls for bad times, whether or not objective observers can find a rational basis for blame’ (Achen and Bartels Reference Achen and Bartels2016, 118; cf. DeCanio Reference DeCanio2014, Reference DeCanio2015). They marshal an impressive set of examples that seemingly help make this point. Most infamously, they contend that then-incumbent president Woodrow Wilson was penalized by coastal voters in the 1916 election for shark attacks that had occurred the previous summer. This finding echoes related work on the influence that football scores have on vote share (for example, Healy, Malhotra and Mo Reference Healy, Malhotra and Mo2010). It is hard to imagine a plausible theory of democratic accountability under which citizens routinely credit and blame politicians for events that are clearly beyond their control.
Yet as subsequent scholarship has shown, citizens do not routinely behave in such a manner. For the vast majority of occasions, ‘irrelevant events’ such as football games and shark attacks wield very little power over citizens. Fowler and Hall (Reference Fowler and Hall2018) connect records on all fatal shark attacks between 1872 and 2012 to presidential election returns and find no evidence of a systemic relationship. This null finding obtains when they account for multiple shark attacks, when a president is running for re-election (as Wilson was), and when the attacks occur in close proximity to the election. If shark attacks harmed Wilson’s vote share in 1916, this pattern does not generalize.
The same is true of football games. While some evidence indicates that citizens punish incumbents when the home team loses a game (for example, Healy, Malhotra and Mo Reference Healy, Malhotra and Mo2010), more comprehensive evidence shows that this pattern does not generally hold. As Fowler and Montagnes (Reference Fowler and Pablo Montagnes2015) document, prior evidence purporting to show this effect is more likely a case of ‘researcher bad luck’ than democratic incompetence. Specifically, when Fowler and Montagnes account for the possibility that ‘college football teams in Democratic places happened, by chance, to win in years that were good for Democrats or vice versa’, they are unable to detect evidence of an effect on football scores and incumbent vote share. They repeat this exercise for professional football and reach the same conclusion (though see Graham et al. Reference Graham2023 for a defence of the initial conclusion).
If citizens were entirely incapable of holding politicians accountable, then policy would likely bear little resemblance to citizens’ preferences. Yet that does not seem to describe the relationship between preferences and policy accurately. In a comprehensive rejoinder to the pessimism around accountability exemplified by Achen and Bartels (Reference Achen and Bartels2016), Caughey and Warshaw (Reference Bowler and Donovan2022) marshals eighty-five years of data on public opinion in policy in US states to conclude that, slowly but surely, state policy comes to resemble the wishes of its citizens. As those who live in, say, Vermont become more liberal over time, so too do the policies of the state. This pattern holds across states. Critically, Caughey and Warshaw find that citizens’ influence over policy has increased over time; prior to 1965, the authors were unable to detect citizens having any effect on state-level economic policy. To be sure, the pace at which policy comes to match preferences is often too slow, with decades sometimes passing before policy changes in the direction desired by citizens. But Caughey and Warshaw show that, acting collectively, citizens hold their elected leaders accountable, rewarding them for bringing policy into line with their preferences and punishing them for not doing so.
Caughey and Warshaw have little to say on the role that political knowledge plays in holding politicians accountable. But perhaps the relationship between knowledge and accountability is less straightforward than popularly assumed. Ashworth and Bueno de Mesquita (Reference Ashworth and Bueno de Mesquita2014) describe a model which shows there are cases where lower information voters will incentivize politicians towards better behaviour. While we can wonder how frequent such cases are in practice, their argument nonetheless reminds us that accountability is about strategic interactions between voters and politicians, where the incentives politicians face matter, such that we cannot always draw straightforward conclusions based on psychological or behavioural findings. While there may be some occasions on which shark attacks or sports scores impede accountability, those would seem to be the exception, not the rule.
None of this is to say that all concerns about accountability are unwarranted. On the contrary, new evidence has raised new concerns. Consider, for example, the recurrent finding that people exposed to corrective evidence of false claims made by their co-partisans become more accurate without shifting their broader views, including toward politicians (Nyhan et al. Reference Nyhan, Porter, Reifler and Wood2019). Bisgaard (Reference Carey, Guess, Loewen, Merkley, Nyhan, Phillips and Reifler2019) presents evidence that, even when they accept unwelcome accurate information, people use such information in a way befitting their partisanship. In one experiment, participants randomly read positive or negative articles about the US economy. Across party lines, people responded to the economic evidence presented to them by regarding the economy as either positive or negative, consistent with what they’d read. Yet when Republicans read about a negative economy, they blamed then-President Obama, far more so than Democrats; and when Democrats read about a positive economy, they credited Obama, far more than Republicans. Information, it seems, does not fundamentally change who people wish to vote for and against.
None of this is to say that all concerns about accountability are unwarranted. On the contrary, new evidence has raised new concerns. Consider, for example, the recurrent finding that people exposed to corrective evidence of false claims made by their co-partisans become more accurate without shifting their broader views, including toward politicians (Nyhan et al. Reference Nyhan, Porter, Reifler and Wood2019). Bisgaard (Reference Carey, Guess, Loewen, Merkley, Nyhan, Phillips and Reifler2019) presents evidence that, even when they accept unwelcome accurate information, people use such information in a way befitting their partisanship. In one experiment, participants randomly read positive or negative articles about the US economy. Across party lines, people responded to the economic evidence presented to them by regarding the economy as either positive or negative, consistent with what they’d read. Yet when Republicans read about a negative economy, they blamed then-President Obama, far more so than Democrats; and when Democrats read about a positive economy, they credited Obama, far more than Republicans. Information, it seems, does not fundamentally change who people wish to vote for and against.
Whether this is welcome from a democratic perspective is unclear. On the one hand, we might wonder how coherent policymaking would be if people reacted to every piece of new information they came across by fundamentally shifting their views toward politicians. On the other hand, increased responsiveness to accurate information would likely be a boon for accountability. Either way, when it comes to accountability, the most pessimistic accounts are almost surely incorrect.
Inclusive equality
Polarisation controls how citizens view each other
In many democratic societies, polarization has increased dramatically (for example, Iyengar, Sood and Lelkes Reference Iyengar, Sood and Lelkes2012). By some accounts, people’s attachments to political parties have deepened over time (for example, Abramowitz Reference Abramowitz2010), and have come to shape many aspects of their identities (Mason Reference Mason2018); by other accounts, the lure of partisanship has weakened, perhaps especially in Europe (Mair Reference Mair2013). Either way, polarization inspires many concerns, including the fear that those who live in polarized societies are not capable of viewing those on the other side as equal citizens. There is little doubt that people are polarized, and that as polarization has increased, people increasingly view their political opponents with outright disdain, if not hatred. Affective polarization refers to the negative way that people regard their political opponents. To understand the depth of affective polarization, as well as the underlying trends, consider that the number of Americans who say they would be upset if a child of theirs married a member of the opposite party increased by sixfold between 1960 and 2010 (Iyengar et al. Reference Iyengar, Lelkes, Levundusky, Malhotra and Westwood2019). Polarization may influence job hiring, with holders of minority political views discriminated against (Gift and Gift Reference Gift and Gift2015). Even dating patterns appear to be affected by polarization (Huber and Malhotra Reference Huber and Malhotra2017).
But affective polarization is not an immutable condition. On the contrary, recent studies have offered a wide range of interventions that successfully weaken it. Exposure to nationalistic or patriotic cues can do the trick (Levendusky Reference Levendusky2018). So too can seeing corrective empirical evidence that undermines beliefs about one’s political opponents (Ahler and Sood Reference Ahler and Sood2018; Eroglu et al. Reference Eroglu2025). Correcting so-called ‘metaperceptions’ – perceptions about the perceptions of others – seems to be an especially fruitful strategy (Lees and Cikara Reference Lees and Cikara2021). Perhaps especially promisingly, a recent ‘megastudy’ meant to identify the most effective interventions among many possible candidates from across the social sciences found that nearly every tested candidate (23 of 25) succeeded at reducing partisan animosity (Voelkel et al. Reference Voelkel, Stagnaro, Chu, Pink, Mernyk, Redekopp and Willer2022).
Clearly, people are not hopelessly polarized against one another; rather, many low-cost interventions can ameliorate the problem. This shows that they are capable of meeting several aspects of the intrinsic school: They can be made to have accurate knowledge of their fellow citizens, which in turn can cause reductions in normatively problematic forms of polarization.
Citizens are willing to resort to violence in the face of disagreement
Among the most dire concerns made about democracy is that, when faced with disagreement, citizens will resort to violence against one another. Of course, refraining from the use of violence to solve disagreements is the foundation of inclusive equality, so the possibility of violence poses threats to the normative value of democracy as well as, of course, being intrinsically wrong. Recent events and polling data would seem to support this concern. The events of 6 January 2021, when Trump loyalists rioted in order to prevent the peaceful transition to power to his duly elected successor, were a stark reminder of the lure of violence even in long-standing democracies. Trump’s repeated assertions of a ‘stolen election’ only appeared to encourage even further violence, or at least rhetorical support for it in public opinion polls. According to one nationally representative 2021 survey, 10 per cent of Americans agreed that force should be used to reinstall Trump in the White House (Pape Reference Pape2022). Other findings were even more alarming, casting willingness to endorse violence as bipartisan. Roughly 33 per cent of Republicans and Democrats believed that violence could just be politically justified, with even larger numbers from both parties supporting violence if the opposite party prevailed at the ballot box (Diamond et al. Reference Diamond, Lee Drutman, Kalmoe and Mason2020). As some scholars have done (for example, Pape Reference Pape2022), extrapolating these figures suggests that tens of millions of Americans are ready to use violence against their political opponents.
Yet there are several reasons to doubt this is the case. Westwood et al. (Reference Westwood, Grimmer, Tyler and Nall2022) reevaluate surveys which seem to show large support for violence and find that the evidence is far weaker than described. Part of the weakness relates to imprecision; surveys described in Mason and Kalmoe (Reference Mason and Kalmoe2023) do not define ‘violence’, thereby permitting respondents to define the term for themselves. When Westwood et al. (Reference Westwood, Grimmer, Tyler and Nall2022) experimentally measure approval for specific types of violence, including murder, they find very little support. Another source of weakness relates to study participant attentiveness. When participants do not pay attention to surveys, they may randomly choose responses, creating illusions of trends where none exist. When Westwood et al. (Reference Westwood, Grimmer, Tyler and Nall2022) control for participant engagement, they again find little support for violence.
Other experimental evidence offers additional reasons to be sceptical about the mass appeal of violence. In a multi-wave experiment, Clayton et al. (Reference Clayton, Davis, Brendan Nyhan, Ryan and Wood2021) exposed a bipartisan panel of respondents during the 2020 campaign to a slew of election-related lies by Trump. Exposure to Trump’s claims about voter fraud – claims that he would echo through January 6 – powerfully diminished his supporters’ trust in elections while convincing them the election was rigged. Yet the same claims had no detectable effect on his supporters’ views toward violence. They believed his claims about voting fraud; those claims negatively coloured their views of democracy, but they were no more willing to embrace violence. Small numbers of people may be willing to engage in political violence. Finally, other work has shown that the rhetorical embrace of violence can be reduced in a straightforward fashion: by communicating that the other side is not supportive of violence (similar to the effectiveness of metaperceptions at reducing polarization). Multiple experiments show that the provision of corrective information to Democrats about the willingness of Republicans to endorse political violence, and vice versa, has sharp, long-lasting effects on both sides (Mernyk et al. Reference Mernyk, Pink, Druckman and Willer2022). This all implies that citizens have the capacity to sustain the knowledge and attitudes required for inclusive equality.
Connecting Evidence to Theory
In sum, many of the headline findings regarding citizen incompetence are either false or in need of significant qualification. Recent research, based on more sophisticated experimental methods or larger data sets, undermines the classical empirical foundations of the citizen incompetence thesis. At minimum, when these recent findings are accounted for, the evidence in support of the incompetence hypothesis could be characterized as mixed or inconclusive. However, by our reading, the accumulated empirical evidence suggests citizens are capable of being competent as judged by all three perspectives described in this article.
Consider first strong instrumentalism. Citizens have knowledge of objective and natural facts, and their store of both can be increased rather easily, suggesting no meaningful cognitive limits. Now, consider weak instrumentalism. Citizens have knowledge of actions and consequences – they are not incapable of accountability – and over time, leaders come to share their citizens’ preferences. It is important to note that the instrumentalist camps are generally regarded as the most pessimistic; yet, the evidence shows that this pessimism is not empirically substantiated.
The intrinsic value perspective is also supported. By and large, citizens do not reject democratic procedures and embrace violence. They do not live on the fringes and create echo chambers. And while their knowledge of their fellow citizens is not accurate at baseline, as with other forms of knowledge, it can be readily improved. People are polarized – though capable of being made less polarized – but despite this polarization, there is no widespread inclination toward political violence. Even amidst strenuous disagreement, many people seem to accept the rules of the game.
Yet for all that we want to push back against the generalized condemnation of citizen competence, we do not want to deny the very real threat to democracy that can emerge when even a relatively small group rejects crucial empirical facts, violently attack their political opponents, or otherwise systematically fall well short of an ideal of citizen competence. Even if, by-and-large, citizens around the world accepted the efficacy of COVID-19 vaccines, resistance from a small minority can pose a public health problem. The same is true of political violence.
The politics of climate change offers a vivid example of the way in which broader systemic forces, such as the information environment, and the behaviour of political elites, contribute to our present challenges. Even if, broadly speaking, citizens across democracies are responsive to facts in this domain, the capture of the US Republican Party and their media allies by climate change denialists could nonetheless seriously hamper the ability to respond to climate change appropriately. Experimental evidence makes clear that even those who deny climate change and/or align with the Republican Party can be made to believe in the science behind climate change if they are exposed to accurate scientific information – but that op-ed articles casting doubt on climate change can reduce support for policy change (Nyhan, Porter and Wood Reference Nyhan, Porter and Wood2022). The failures of the present may thus be institutional, relating to partisan cues and the media environment, rather than stemming from individual-level cognitive inabilities. Moreover, the incompetence of political elites, who consistently misperceive the preferences of ordinary citizens (Hertel-Fernandez, Mildenberger and Stokes Reference Hertel-Fernandez, Mildenberger and Stokes2019), is woefully underexamined, especially compared to the supposed incompetence of those ordinary citizens.
Although the bulk of the evidence presented here was gathered in the US, we have no reason to anticipate that different dynamics would be observable elsewhere. What cross-national studies have found on these matters is not meaningfully different from what we present. For example, corrections of misinformation improve belief accuracy not just in the United States, but in the Global South and elsewhere (Carey et al. Reference Carey, Guess, Loewen, Merkley, Nyhan, Phillips and Reifler2022; Blair et al. Reference Blair, Gottlieb, Nyhan, Paler, Argote and Stanfield2024; Porter and Wood Reference Porter and Wood2021). Fundamentally, our argument concerns the cognitive capacities of everyday people, without regard for nationality, ethnicity or other ascriptive criteria.
Discussion and Conclusion
Democracy is no doubt demanding in the cognitive burdens it places on ordinary citizens. It asks individuals to make informed decisions, not just about their immediate affairs, but about the collective well-being of their community. For many of our most pressing tasks, this means engaging meaningfully with disagreement about values as well as with complex scientific knowledge that today mediates much political action. But even as normative theories of democracy have become more internally complex and sophisticated, debates about citizen competence remain in a binary framework, evaluating the competence of individuals relative to an imagined unified baseline. We have challenged this on both conceptual and empirical grounds. For the most part, citizens are ‘competent enough’ – judged from several perspectives, evidence shows that they can be made to meet the standards of democratic competence.
What this means going forward, then, is that democratic theorists and empirical political scientists have an opportunity to develop a more differentiated account of the relationship between the values democracy realizes, the institutions and practices that make up a democratic system, and the epistemic infrastructure that supports citizens’ participation in democratic life (Herzog Reference Herzog2023). By implication, if we want to diagnose the ills of democracy, we should not look to individual-level psychological or cognitive shortcomings. The tendency, both in popular and academic commentary, to emphasize political or scientific ignorance, echo chambers, and misinformation may be a case of cherry-picked evidence that distracts from more significant explanations of democratic dysfunction. Given our overview of recent findings, we should be sceptical of accounts that emphasise individual cognitive incapacity. Rather, more credence should be given to explanations that focus on institutional factors like the outsized role of wealth, concentrated interests, and collective action problems. This is not to deny that rejection of scientific facts, election denialism, or political ignorance are not problems that should be addressed through vigorous efforts at correction and improvement. But it is to say that, by neglecting systemic forces, they may offer impoverished explanations for the distance between democratic ideals and lived reality.
We should still try to improve citizen competence when we can. In doing so, we should take the advice of Watts (Reference Watts2017) and pursue a solutions-oriented approach. But this approach should not focus on individuals; we think the evidence is clear that, at the least, individuals are largely capable of being made competent. Chater and Loewenstein (Reference Chater and Loewenstein2023) argue that behavioural scholars have thought too much about individuals and not enough about the systems in which they are a part. Considerations of citizen competence have suffered from the same problem. What is needed is an attempt to systematically conceptualize, measure, and evaluate interventions to improve competence.
Normative democratic theory can contribute to this by articulating the relationship between different aspects of democracy, understood as a system, to the different forms of competence. Recently, scholars have developed the concept of the democratic system to understand how different practices – voting, deliberation, protest, and so on – can each play different functions in a democratic society (Mansbridge et al. Reference Mansbridge, James Bohman, Thomas Christiano, Parkinson, Thompson, Warren, Parkinson and Mansbridge2012; Warren Reference Warren2017). Yet those different practices all require different forms of citizen competence – and stand in different relations to the values of reliability, accountability, and inclusive equality – and so it remains a task for political theorists to clarify the different types of competence that underpin a healthy democratic system.
How could empirical scholars measure and evaluate citizen competence at the systemic level? Doing so would require deploying multiple interventions at once, across large groups or municipalities, with the aim of drawing conclusions about what can be done at the macro-level to improve citizens’ competence. To understand how this might work in practice, consider a systematic study focused on climate change and competence. Rather than merely randomizing citizens’ exposure to scientifically accurate information, researchers could simultaneously randomize across multiple dimensions. At the same time, they could vary the internal incentives confronting newsrooms when deciding to cover climate change (following the model of King, Schneer and White Reference Saunders-Hastings2017); the information politicians have about their constituents’ views; and the information that citizens have about their fellow citizens’ views. This kind of work would often involve working with politicians and governments themselves (as proposed by Neblo et al. Reference Neblo, Minozzi, Esterling, Green, Kingzette and Lazer2017). Results from such a multi-level study could offer a systems-level understanding of competence than that which is available in the literature today, one that would speak to the values of accountability, reliability, and inclusive equality all at once.
In this article, we have advanced two arguments. First, citizen competence needs to be rethought in response to a more differentiated concept of democracy, one that distinguishes between reliability, accountability, and inclusive equality. Second, recent empirical evidence largely overturns the image of citizen incompetence that was developed in earlier research. Furthermore, this more recent research shows that citizen competence can also be improved through strategic interventions. We contend that worries about citizen competence are overblown. Insofar as democratic societies are suffering from political maladies, the causes are likely to be located elsewhere. Yet we do not think this conclusion warrants complacency. There is still room to improve citizen competence, but going forward scholars should examine interventions that operate at a systemic rather than individual level. As individuals, citizens are largely competent enough, but as societies, we have a way to go to ensure that the abilities of ordinary citizens help realize what we value.
Acknowledgements
For comments on an earlier draft of this manuscript, we thank Maxime Lepoutre. We also thank Daniel Nichanian and Thomas J. Wood for many relevant conversations over the years.
Financial Support
This research was supported by John Templeton Foundation, ‘The Political Economy of Knowledge and Ignorance’, Grant #61823, and was supported by the John S. and James L. Knight Foundation through a grant to the Institute for Data, Democracy & Politics at The George Washington University.
Competing Interests
None to disclose.
Open access
