Theory, Media, and Democracy for Realists

ABSTRACT Democracy for Realists delivers a long-overdue attack upon apologetics for American political realities. Achen and Bartels argue that the “folk theory of democracy” is not an accurate description of democracy in the United States and that without a greater degree of economic and social equality, democracy will remain an unattainable ideal. But their account of the gap between ideal and actual relies too heavily on the innate cognitive limitations and biases (particularly intergroup bias) of our psychology. These are important, but they provide only an incomplete understanding if we do not incorporate a focus on information and ideas. Media-effects research and comparative media studies provide evidence that the ignorance of the U.S. voter is only partly caused by cognitive limitations, and they point the way to a more effective way to bring social psychology into political science.

and lobbying. With "sentiments and views" (Madison et al. [] , ) arising naturally from their socio-economic status or group membership, underrepresented groups could then donate and lobby appropriately. But if the sentiments and views of the relatively poor, for example, arise naturally from their lack of property, it is hard to see why mass democracy has failed where campaign contributions and lobbying muscle would succeed. A penniless candidate committed to implementing the solutions desired by the poor could be elected by them; a politician successfully lobbied to reject the solutions desired by the poor could be voted out of office. Or so it is (borrowing from economics) assuming perfect information. If we do not assume perfect information, nor that political-economic interests (and how to advance them) are self-evident, then to whom contributions should be made or for what one's hired lobbyists should lobby are questions just as difficult as for whom to vote.
Herein lies the major weakness in Democracy for Realists: it lacks a place for information. Its theory of group politics is an undeniable improvement upon the political-science status quo. But without incorporating information into our understanding of political behavior, we get a misleading picture: namely, that the lamentable yet irrefutable ignorance of most U.S. citizens is the result of inherent cognitive limitations. If this were the only cause of their ignorance, amelioration would be impossible until the advent of genetic engineering capable of transforming an onaverage lazy and apathetic species into information-devouring news junkies. Luckily, while most failures are orphans, the ignorance of the average U.S. citizen has several fathers-and human nature is only one of them. To see why, and how this affects Achen and Bartels's diagnosis of really existing democracy, requires that we give serious consideration not only to the groupish psychology on which they focus but to information, to ideas, and the question: Why do people know what they do know about politics?

Ignorance Theory
A survey of the research on widespread political ignorance nets so many pithy barbs that it's hard to choose only one; but my choice is from Scott Althaus (, ): "If ignorance is bliss, then the pursuit of happiness seems alive and well in American society." Of course, ignorance is not so much a pursuit as it is our default state. We may be the knowing hominid, but we enter the world in a state of nearly infinite ignorance. What we Homo sapiens do to merit our name is collect large amounts of information throughout our lives, which nonetheless merely move us from complete ignorance at birth to nearly-complete ignorance when we die. That we necessarily make only meager progress is because the realm of information is infinite.
Information theory defines its object as the reduction of uncertainty; the "information" transmitted by a landline or wi-fi signal is a reduction in the listener or recipient's uncertainty about some aspect of the world (Pierce , ). I say "I love you" on the phone, the soundwaves I produce are transformed into compact signals, and the listener receives information: a reduction in her uncertainty about my feelings. Or, possibly for good reason, my "I love you" conveyed no information whatsoever, despite being faithfully reproduced by the listener's phone. Herein lies the slipperiness of information: it is separable from meaning (Hidalgo , xvi). "I love you" could have had no meaning, if it were the dozenth time it was said in a row; or it could have had any number of unintended meanings, produced by collision with limitless other information the listener might have (for instance, if I tend to profess love when I have something to hide, or if I regularly use those words as a goodbye). As John Mayfield (, ) explains, "meaning emerges from interactions between system states. If there are no interactions, there is no meaning. For meaning to be present, particular states of one system must have particular effects on another system"-as when information we receive over the phone changes our understanding of the speaker.
If information is physical, instantiated in electrical pulses, neuronal connections, and patterns of ink on paper, then its communication requires its transportation (Peters , ). Information will not transport itself through the ether into the brains of all those who would benefit from it. Information of relevance to one's personal life is readily at hand, if not always collected; but information of relevance to politics is almost always at a great distance. Few people live in capital cities, are close friends with top officials, or receive daily briefings from congressional aides; even the few who do must supplement this first-hand information with second-or third-hand information from the news media, or from others who (in most cases) got their information from the news media. Therefore, in addition to innate cognitive limitations, we have to think about the logistics of information: how pieces of information with political relevance are generated, recorded, transported, and finally absorbed-and whether they were damaged or transformed in the process.
We begin with a blank slate of political information-barely smudged, perhaps, by psychological predispositions or "elective affinities" that make some ideologies more attractive than others. We can fill the slate only with the information to which we are exposed over time. Both our cognitive limitations and restrictions on the supply of available political information guarantee that we will forever be ignorant of the vast majority of politically relevant information out there. Ignorance is our default state; being informed requires the transportation of information. Only if we confuse ourselves with omniscient beings does ignorance become a puzzle in need of a solution. If ignorance is our default state, and information is physical and needs to be transported across vast distances, then the question is not why citizens are so ignorant, but how is it that they know anything at all about politics. Why are U.S. citizens more knowledgeable about politics than they are about any number of other possible topics, such as theoretical physics or Yukio Mishima's oeuvre? That the average citizen knows that politics even exists, and can at least formulate a few sentences describing what it is, is what needs explaining.
Here is one explanation. Unlike most of what comprises the enormous realm of information, politics is widely perceived to have at least some relevance to people's own lives. Moreover, even a social network exclusively comprising the politically apathetic is a rich source of politically relevant information ("How many people do I know who are unemployed or uninsured, who are being hurt by high taxes, who benefit from social programs," etc.); and a wealth of political information is just a mailbox, newsstand, remote control, or mouse click away. In other words, there is a vast logistical infrastructure devoted to transporting information about politics, a supply network that reflects demand for such information, even if one judges the demand insufficient.
The logistical infrastructure is what we call the media or the "means of communication": institutions from conversation to blogs to the News Corporation to the New York Times, all organized to collect, process, and distribute political information across the world. Thus, if we want to understand what is left when we subtract omniscience from our conception of human nature, the media must be our point of departure.

The Media
The folk theory of democracy imagines a citizenry of philosophers, or ladies and gentlemen of leisure, who enjoy spending their days reading books and newspapers, debating their contents in salons. "The reality," Achen and Bartels point out (, ), is quite different. Human beings are busy with their lives. Most have school or a job consuming many hours of the day. They also have meals to prepare, homes to clean, and bills to pay. They may have children to raise or elderly parents to care for. They may also be coping with unemployment, business reverses, illness, addictions, divorce, or other personal and family troubles. For most, leisure time is at a premium. Sorting out which presidential candidate has the right foreign policy toward Asia is not a high priority for them.
To borrow again from economics, sorting out which presidential candidate has the right foreign policy is a cost, which the benefit must exceed in order for such sorting to occur. Such costs are relatively easy to calculate: for instance, x hours spent researching on the Internet or at the library, plus the price of a computer and Internet connection or parking and the risk of late fees. But without information on potential benefits-from the infinite benefit of avoiding death from nuclear war, to the minor benefit of a slight increase in GDP-the cost-benefit analysis simply cannot be undertaken. In that case, the benefit of knowing about presidential candidates' foreign policies defaults to zero, to join the limitless number of other pieces of information that are assumed by default to have benefits lower than their costs of attainment. With benefits vague, distant, and impossible to calculate without relevant information, the only way to increase the likelihood that people will choose to absorb political information is to reduce its cost; to provide it "cheaply," easily, accessibly. As Thomas Ferguson (, ) points out, "it is not necessary to assume or argue that the voting population is stupid or malevolent to explain why it often will not stir at even gross affronts to its own interests and values. Mere political awareness is costly … it will not be supplied or often even demanded unless some sort of subsidy" is provided. Subsidies for political information come from two sources: in commercial news media, from advertisers willing to provide the subsidy in exchange for the attention of desirable audiences, and in public-service news media, from governments willing to provide the subsidy to attain the goal of an informed citizenry.
Achen and Bartels do touch upon the media, but on the whole the topic is left under-examined. They mention Markus Prior's "mo' media mo' problems" research on the introduction of cable television having no impact on average levels of political information, which suggests, when combined with the contemporary increase in educational attainment, that current levels of political ignorance may be an intractable problem (Achen and Bartels , -). On the other hand, they note in passing that media coverage rather than the actual incidence of crime drives public perception of the latter (ibid., ); that the media serves as a conduit to fill essentially hereditary partisan identities with ideological beliefs (ibid., ); that people's political beliefs are cobbled together with information gleaned at first or second hand from the media (ibid., ); and that a free press communicating to a literate public is what undergirds political accountability (ibid., ). Pedro Gomes (, ) goes so far as to argue that "it is increasingly the case that for something to be recognized as real, it must first be mediatized." But if the news media are powerful, the power is not being well used: "the problem is not that voters are necessarily irrational, but that most voters have very little real information, even about crucially important aspects of national political life" (Achen and Bartels , ).
Whether political ignorance is a tractable problem or not, then, depends on whether the news media can be more effective at transmitting political information. Achen and Bartels (, ), despite acknowledging the power of the media, seem to lean toward skepticism. They observe that many scholars continue to express idealistic hope that … improved mass media [among other reforms] … might bring public opinion into closer correspondence with the standards of the folk theory. But in sober moments most acknowledge the repeated failures of all those prescriptions.
Not quite. To paraphrase G. K. Chesterton on Christianity, media reform has not been tried and found wanting; it has been found difficult and left untried. Or, it has not been found at all, in that "improv[ing] the mass media" is a reform off of most people's political radar.
Certainly, media reform would be a difficult project. As Walter Lippmann ([] , -, ) observed nearly a century ago: Those who are now in control have too much at stake, and they control the source of reform itself. Change will come only by the drastic competition of those whose interests are not represented in the existing news-organization. … In a few generations it will seem ludicrous to historians that a people professing government by the will of the people should have made no serious effort to guarantee the news without which governing opinion cannot exist.
In Latin America, media reform has certainly proven difficult, but it is at least on the political agenda. Pascual Serrano (, ) describes the goal of media reform efforts: We are faced with a new challenge: to find a way for citizens to reclaim our right to information through the State, from which we need to demand the enforcement of its duty to guarantee it. We, citizens, must give power to the State, and the State, for its part, must give us control. This is the true freedom of the press in a democracy.
This recalls Ferguson's (, ) "Golden Rule" as it applies to the provision of information in democracies: "In politics, you get what you pay for. Or someone else does." The alternative to government as media sugar daddy is not free sugar; someone must provide the information subsidy, and the alternative is private advertisers. The evidence, discussed below, suggests that the latter do a poorer job than the former, except in cases where the government is able to exert influence over the media it funds.
But to suggest that media reform is capable of making a dent in widespread public ignorance entails first that the media produces appreciable effects, and that one of these effects could be to better inform the public -in other words, that human cognitive limitations do not make such ignorance irremediable, no matter what media system is put in place. Achen and Bartels do not address this question directly, but they would seem to agree that the media produces appreciable effects while doubting that one of these effects could be a more politically knowledgeable public. Yet the accumulated evidence provides strong support for both conclusions.
First, the effects. Like many political scientists, Tim Groseclose (, ) was a confirmed media-effects skeptic; in line with rational-choice theory, he supposed that the assumed-omniscient individual utility-maximizer from orthodox economics actually described human beings, such that they would be able to calculate media bias and correct for it in their processing of information. Hence the effects of media bias would net to zero for each individual. His conversion experience was driven in part by a highly formalized experiment, a numerical game of "strategic information transmission," which produced results inconsistent with rational-choice theories and suggested human fallibility instead (ibid., -). This, and not decades of research and hundreds of studies, was what proved to him that the media can have effects. For those steeped in media-effects research, Groseclose's conversion experience makes one imagine a tourist visiting the Louvre and exiting hours later with the firm conviction that art is simply beyond human capabilities-a conviction later reversed when the tourist sees some graffiti, ponders it, and decides that art is possible after all.
For generations, political scientists have been misled by the "minimaleffects" research tradition, which focused primarily on whether the broadcast media had short-term, persuasive effects during election campaigns (Katz ). Today, the idea that the media produce only minimal effects is no longer tenable (e.g., Bryant and Oliver ; Preiss et al. ). The implications are profound. I happened to be born to Republican parents and so developed a liking for the Republican Party; I happened to learn one view of history in school, happened to prefer one ideological variety of media, and happened to develop a conservative political ideology. At no time was I ever exposed to fairly comprehensive summaries of all political ideologies in existence, choosing the one that seemed most reasonable. The realization that even our most cherished beliefs are accidental and arbitrary can be unsettling, and may goad us into adopting a healthy amount of epistemological humilityor, to fight against the realization itself. Who would want to face the possibility that all of their cherished beliefs and ideas about politics are (or very well may be) wrong? This is, however, the unavoidable conclusion of media-effects research.

Ignorance as a Media Effect
Achen and Bartels (, ) do not disregard the central role in public opinion and political behavior played by ignorance: "most voters have very little real information, even about crucially important aspects of national political life." They cite surveys showing that half of Republicans but less than a tenth of Democrats still believed in Iraqi WMDs  months after the invasion, and over two thirds of Republicans but just over one third of Democrats believed that Saddam Hussein had supported al-Qaeda (ibid., -). On the question (asked in ) of whether the Reagan administration presided over a reduction or increase in inflation, the pattern of delusion was reversed: almost half of surveyed Republicans gave the correct answer, but less than a tenth of surveyed Democrats. In both instances, large percentages of the population held false beliefs. This could be the effect of cognitive dissonance reduction: Democrats avoiding the dissonance from believing that "Reagan is bad" and that "Reagan reduced inflation" at the same time, and Republicans avoiding the dissonance from believing that "Bush is good" and that "Bush was wrong about Iraq" at the same time. Or it could be more psychologically subterranean, with motivated reasoning working subconsciously to mold evaluations to favor one's in-group. But it could also be a simple matter of cognitive consistency, with incoming information consistent with one's priors receiving low scrutiny and inconsistent information receiving high scrutiny (Ditto ).
Thus, in the case of inflation, pertinent information may have come from daily life as much as from media reports; people might have heard on TV that the consumer price index had risen more slowly in recent years than before, but salient memories of paying high health-care bills or college tuition may have loomed larger in their mind's eye. Economic issues are "obtrusive"information about them comes from daily life, the media is not the sole source-and media effects are weakest for obtrusive issues (McCombs , ). Yet even on obtrusive economic issues, information from the media can buttress that from one's own personal experience, and tie it to perceptions of the national economy and evaluations of the president (Mutz ). In the case of Iraq, pertinent information for all but a small fraction of the populace must have come from the media; it is an "unobtrusive" issue, information about which cannot be gleaned from one's daily life. Few people could chat with the prime minister of Niger about yellowcake over lunch, browse through Iraqi (or U.S.) intelligence records at work, or discuss Saddam Hussein's opinions on Islamic fundamentalism with members of al-Qaeda at a bake sale. To get information from Iraq, Niger, or even Langley and D.C. to American citizens requires Walmart-scale logistics, which only the news media provide. And what the media provided was a conduit through which the Bush administration sent misleading or false information without significant critical scrutiny (Lewis , -), which was picked up and believed by majorities of TV news viewers, with Fox viewers most receptive and NPR/PBS listeners/viewers least receptive (Kull et al. ). Most revealingly, in European countries with different media systems, Bush administration claims were treated less charitably and were submitted to greater scrutiny; armed with this information, Europeans were more opposed to the war, irrespective of the fact that several ruling parties in Europe supported it, and partisan motivated reasoning would be expected to affect European as well as U.S. conservatives (Verhulst and Walgrave ). Motivated reasoning is not omnipotent, and can be constrained by exposure to information (Kunda , ). Demand-side biases are important, but supply matters as well.
Nonetheless, the fact that even deeply held political convictions are the result of historical accident, and that our political knowledge comes in large part from the news media to which we are exposed, does not entail that a different media system could produce higher average levels of political information. (Nor unbiased information.) It remains a possibility that our innate cognitive limitations set such a low ceiling for average levels of political information that even an optimal media system would produce much the same results. This, however, is not supported by research in comparative media studies.
Of course, using cross-country comparisons to answer whether differences in media systems can produce a more knowledgeable citizenry is difficult. Several cultural and (non-media) institutional factors are implicated, which arguably could better explain differences in average levels of political information across countries than differences in media systems. But this is a familiar challenge in social science, and scholars in this area have tackled it.
First, possibly the most salient difference between media systems is degree of commercialization: how much of the news media is statefunded public-service media, and how much is for-profit commercial media? In contemporary Western Europe, all media systems are hybrids, with varying levels of commercialization; but all have a relatively strong public-service media sector, and none are as commercialized as the U.S. system. Viewers of commercial media in Europe and the United States can then be compared with viewers of public service media in Europe (in the United States, public-service media viewers are harder to find, given PBS's miniscule market share) to find differences in levels of political information, controlling for demographic variables. Hundreds of studies and statistical analyses have focused on this question, and the clear result is that public-service media outperform commercial media in nearly every aspect of providing political information (Cushion , , ). Public service-heavy media systems outperform predominantly commercial media systems across the board (Aalborg and Curran ). Public service media systems are better at providing hard news, better at covering elections, and better at producing a knowledgeable citizenry that votes (Cushion , -, , ).
Unsurprisingly, the gap is widest in the area of unobtrusive issues. Tests of international-affairs knowledge have demonstrated that on average, people in the United States are more ignorant than the publics of all European G nations, and strikingly so: for instance,  percent of American respondents answered only one or none of a five-question knowledge proxy correctly, while  percent of Germans answered all or four of five questions correctly (Dimock and Popkin , ). A comparison of the United States and Switzerland found those in the United States to be significantly more ignorant about world affairs than the Swiss, and that this difference is partially attributable to the greater supply of international news in Switzerland's public service media system (Iyengar et al. ). (On some hard-news questions, Swiss high school dropouts performed better than U.S. college graduates.) Television news in the United States not only provides less information about the world than European TV news, but two-thirds of it is focused on countries with heavy U.S. military or diplomatic involvement, rendering much of the rest of the world invisible (Curran et al. ). This same pattern of more knowledge about the world following a greater supply of international news was found in a study of eleven countries across five continents, with more-commercialized media systems performing worse than public-service media systems (Aalberg et al. ). What international news does appear in the U.S. media tends to be tightly constrained to the perspectives of political elites; as Mikhail Alexseev and Lance Bennett (, ) argue, this "may remove much chance for the public to become actively involved in policy issues in ways that might define new policy options, produce more informed opinions, or stimulate higher levels of citizen participation." Overall, the evidence shows that public-service media do a better job than commercial media in providing foreign coverage (Curran et al. ). TV news in public service-dominated European countries offers an average of between  and  minutes of foreign coverage per day, while the two leading commercial broadcasters in the United States provide a combined daily total of four minutes (Brekken et al. , ). Although some have argued that the U.S. media provide such limited coverage because Americans are less educated and familiar with the rest of the world, Christian Kolmer and Holli Semetko (, ) rightly ask: "But what came first-the lack of education or the lack of information in U.S. television news?" Most likely, it is the lack of information. Levels of interest in international news are high in the U.S., more so than in many other countries with a greater supply of international news (Aalberg et al. , ). However, there is a glaring mismatch between what news editors choose to cover and what the public wants (Tai and Chang ).
In fact, people in the United States report higher average levels of interest in politics and closer attention to both domestic and international news than Europeans (Aalberg and Curran ). Why then are U.S. citizens far more ignorant about domestic and international politics than Europeans? Some factors include the greater degree of distrust Americans have for their government and their media system, as well as the greater degree of economic inequality in the United States. But a large part of the explanation must be the differences between the U.S. commercialized media system and Europe's public-service media systems, with their heavily funded, market-leading public broadcasters. A commercial channel, to stay in business, must maximize revenue; it may discover that providing ample coverage of public affairs maximizes revenue, but if it does not, it will have to do otherwise. For commercial channels, providing ample political information and pluralism is an externality, and goes beyond and sometimes against rational, profit-maximizing considerations (Fu ). A public-service channel, on the other hand, is under no pressure to maximize revenue, and can aim at providing ample public affairs coverage without regard to the bottom line. This basic fact of the political economy of media has left its mark. "Thus, in [the European] system, the citizen watching a popular channel needs to actively choose to avoid information about public affairs. In the [U.S.] system, the citizen watching a popular channel needs to actively seek out this information" (Aalberg and Curran , ).
Even if-in a counterfactual world-U.S. audiences actually preferred what they are currently being offered, this would not provide justification for it. As Cass Sunstein (, ) explains, without access to an alternative media system, "the broadcasting status quo cannot, without circularity, be justified on the basis of [current] preferences. Preferences that have adapted to an objectionable system cannot justify that system." In any case, it makes little sense to think of culture as stable and unchanging, with fixed preferences the media can only adapt to but not influence (Williams , ).
Although levels of education are a primary factor in how much political information people pick up from their media environment, TV news helps to reduce knowledge gaps between those with high and low levels of education (Jerit et al. ). Hence, the media have an independent role in promoting a (relatively) knowledgeable citizenry, in addition to the educational system. The policy-specific information the media provides is particularly important in making political decisions, as a series of experiments demonstrated: even (and especially) among those with high levels of general political knowledge, exposure to policy-specific information produces a significant influence on political judgments (Gilens ). Another experimental study found that the effects of education and political sophistication are greatly reduced, if not eliminated, by exposure to specific, highly diagnostic policy information (Kuklinski et al. ). Identity factors such as class and personal history can also be overwhelmed by a lack of information, as one study found: as a group, the highly informed held a variety of different opinions consonant with their backgrounds, while the uninformed showed little difference in opinion despite having a variety of differences in background (Geddes and Zaller , ). In other words, without information about a policy, we are unable to turn our predispositions into dispositions-regardless of how our predispositions have been formed. Both supply and demand are important.
Greater income equality is important as well, since political knowledge gaps between those with high and low levels of education are significantly smaller in more equal countries (Gronlund and Milner ). However, a study of fourteen EU member states found that even after controlling for a battery of other factors (gender, education, age, income, ideology, and political interest), in ten of the countries a preference for public-service media was still strongly correlated with knowledge about politics (Holtz-Bacha and Norris ). Using statistical techniques to mimic a real-world experiment, a study of six countries representing North America, Europe, and Asia concluded that exposure to public-service media increases political knowledge to a greater extent than exposure to commercial media, but only where funding and other mechanisms guaranteed public broadcasters independence (Soroka et al. ).
Levels of political knowledge also correlate strongly with political interest; however, in public-service media systems, this correlation is far weaker than in commercial media systems. As Shanto Iyengar and colleagues (, ) explain, in commercial systems "political knowledge depends heavily upon political interest; in public-service systems, however, it is possible for the less interested to overcome their motivational handicap because of the greater availability of news programming." The greater availability and supply of hard news in public-service media systems means that those without high levels of education or political interest are inadvertently exposed to hard news, and thereby become informed about the political realm (Hahn et al. ). This applies with particular force to ethnic minorities: in the commercialized U.S. system, minorities are both less exposed and less knowledgeable about hard news, while in the more public service-oriented British system, there are no such gaps in knowledge between ethnic minorities and the ethnic majority (Curran et al. ).
Since greater knowledge is associated with greater political participation, it makes sense that more-informative forms of media correlate with higher rates of political participation. A comprehensive study of  democracies found that countries with public-service media systems had higher levels of voting than countries with commercialized media systems. For every  percent increase in audience share for public broadcasting, there is a . percent increase in voter turnout-and among advanced democracies, a . percent increase (Baek , -). Greater density of media options is also correlated with "correct" voting (choosing candidates that match voters' policy preferences) across dozens of countries (Lau et al. ). Political and ideological pluralism in the media system (particularly for newspapers but also for television) also produce more political participation (van Hempen ).

The Group Theory of Politics
Information requires transportation if that about which it reduces uncertainty is out of the range of our eyes and ears. Since most of the stuff of politics is out of that range, understanding public ignorance about politics requires an understanding of the logistics of delivering political information (the supply side) and an understanding of our cognitive limitations (the demand side). Evidence from comparative media studies demonstrates that our cognitive limitations and the demands of daily life are not the only causes of public ignorance; in addition, how media systems are structured makes a contribution. Hence Achen and Bartels's pessimism about the possibility of instantiating the folk theory should be tempered a bit.
Another reason for temperance is that Achen and Bartels (of necessity, of course) focus on only one aspect of political psychology, intergroup bias, or our innate groupishness, which is but one psychological tendency or force among many others. In-group bias may be transformed in the presence of other factors, such as a tendency toward system justification, as when members of disadvantaged groups display out-group favoritism toward privileged groups as a means of avoiding negative affect or maintaining consistency with their just-world beliefs, or both (Jost et al. ). Nor are its effects uniform. Among those who develop either a highly complex or a rudimentary style of thought, national identity is less categorical relative to those who develop a style of thought of intermediate complexity; and (national) in-group favoritism decreases as thinking style becomes more complex (Rosenberg and Beattie, in press). Psychology has not so much "moved away" (Achen and Bartels , ) from studying intergroup bias as it has begun to explore other psychological forces in addition. Research on intergroup bias continues, for instance on how it affects our use of language to differentially describe in-groups and out-groups (Annoli et al. ), and how the use of these subtle linguistic cues in the media contributes to the development of prejudiced opinions (Geschke et al. ).
A more fundamental problem with too great an emphasis on "groupishness" is that it may lead to a neglect of the continued causal impact of ideas and information. Groups, and the psychological glue that helps them cohere, are important for the way that they influence the ideas people develop and the information people are exposed to, but group membership does not determine political ideas; rather, it triggers a psychological force affecting (among others) the ideas one will develop, and most importantly, it influences the selection of information to which one is exposed. Other things being equal, in-group bias will make us more likely to vote for candidates sharing our partisan group identity or our ethnic, religious, gender, sexual, and other identities. But other things are never equal in a complex system. The information to which we have been exposed and the ideas we have developed can overwhelm (or reinforce) in-group bias operating on these identities. We have to keep the informational and psychological content conceptually separate, even (or especially) if they are most commonly intertwined. This is a subtle yet important shift of focus. To illustrate, consider the following passage: Citizens tend to adopt the view of the parties and groups they favor. If they are unusually highly engaged in politics, they may even develop ideological frameworks rationalizing their group loyalties and denigrating those of their political opponents. Sometimes they even construct convenient "facts" to help support their group loyalties. The reasoned explanations they provide for their own beliefs and behavior are often just post hoc justifications of their social or partisan loyalties. Well-informed citizens are likely to have more elaborate and internally consistent worldviews than inattentive people do, but that just reflects the fact that their rationalizations are better rehearsed. (Achen and Bartels , ) Yes, we tend to adopt the view of the parties and groups we favor-but how? Intergroup-bias research shows that our perceptions and opinions are influenced to favor groups of which we are members, and research on conformity demonstrates the power of peer pressure. But to adopt the views of the parties and groups we favor requires the transmission of information about these views, which are not self-evident. So it is not enough to point to psychological dynamics affecting information transmission or processing; we must look at how the media, teachers, and social networks transmit information about the views of the parties and groups we favor. For those "unusually highly engaged in politics," we are dealing with a much greater amount of information, which may include justifications for one's group loyalties, but also for one's political values and opinions. For these unusual people, the "reasoned explanations they provide for their own beliefs" are not merely justifications of group loyalties but expressions of a partial, biased accumulation of political information substantial enough to be deemed an ideology. It is because their worldviews are internally consistent and elaborate-they are filled to the brim with information accumulated over years-that their arguments-cum-rationalizations seem better rehearsed.
Achen and Bartels get closer to integrating the epistemological and the psychological with the following observation: "For most people most of the time, social identities and partisan loyalties color political perceptions as well as political opinions" (emphasis mine). Psychologically, that is all they do: color or influence. Epistemologically, they do far more, since they structure the way that information spreads: I get my news from Fox because my fellow Republicans endorse it, or I like it better because its messages fit with what I already believe and value; or I get my news from the World Socialist Web Site because my left-wing friends introduced me to it, and its writers think as I do, etc.
If we treat ideas as mere epiphenomena of group memberships, even implicitly, we will miss what is most interesting and important about politics. For instance, Hillary Clinton's primary nomination was won due to her overwhelming advantage among black primary voters, securing over  percent of the black vote while splitting the white vote with Bernie Sanders. This inspired much contemporaneous commentary about Sanders's problems with nonwhite voters, and why black and Hispanic communities were turned off by the economic focus of his messaging. A year later, however, nationally representative polls revealed that Sanders had become significantly more popular among nonwhites than whites (e.g., Easley ). This would be puzzling from a simple group-theory-ofpolitics perspective, since this seemingly seismic shift in these groups' opinions was unaccompanied by any shift in Sanders' political positions or rhetoric akin to the Roosevelt administration's outreach to black voters in the s (Achen and Bartels , ). But if we consider intergroup bias and the groupishness it creates to be merely one force or tendency operating among several others, our focus might shift to the ideas generated by group members, and the information they receive that influences their ideas. We would have to treat the group of nonwhite voters as less homogeneous, look at the minority of that group who were Democratic primary voters, and see what information they were exposed to and what ideas drove their votes; we would then look at the larger group, and search for differences between them and the group of primary voters over time, particularly in their information exposure and the ideas they generated. And there would be yet more to research, since we still would not have examined black Republicans, let alone differences within that group.
To deny that ideas arise naturally from group memberships is not to say that some ideas can come close to doing so. If my ethnic identity is black, and I observe that most of those who share my ethnic identity vote for Democrats, with that one piece of information I can generate the idea that Democrats look after the interests of "people like me" and vote accordingly. Or, if I happen to develop an interest in politics-from family, friends, teachers, or just reading a particularly inspiring book-I may absorb a great deal of information about politics and generate a wealth of ideas explaining why Democrats (and what kind of Democrats) are the best option for my community, and the country as a whole. Alternatively, the information to which I am exposed through education, media, or social networks might lead me to generate ideas supporting Republicans (e.g., Sowell ) or opposing both parties in favor of a radical alternative. Ideas and information are always the proximate cause of voting decisions, and even if the ultimate causes are psychological biases, selective exposure to information, ideologically segregated social networks, etc., a political science deserving of the name must attend to information and ideas, not just psychological biases.
Take shark attacks as instances of blind retrospective voting (Achen and Bartels , -). Perhaps, as Achen and Bartels claim, voters in New Jersey beach townships were indeed irrationally punishing President Wilson in the way that a drunk might kick his dog out of anger at his boss. But it is also possible that they had rational ideas (even if we want to judge them as incorrect) about what Wilson should have done but did not do to help. While there was no permanent federal disaster assistance agency at the time, over the previous century there had been dozens of federal laws providing ad hoc assistance in cases of natural disaster (Moss , ). Did the New Jersey voters know this? Did they think that Wilson should have called for similar legislation-or mobilized more Coast Guard or even Navy vessels to kill sharks, or offer bounties for each shark that fishermen caught? What arguments could be found in local newspapers that voters might have adopted, and what arguments might residents have created on their own or during conversations? Dismissing their voting behavior as plainly irrational may stop research into such possibilities. But there is no "scientific" reason to do so. Indeed, the groupish psychology that Achen and Bartels equate with "science" offers no help in explaining the anti-Wilson vote-unless we are to think that the voters identified with human beings as the "us" and sharks as the "them," and accused Wilson of siding with the sharks. Achen and Bartels have in fact turned up a case in which, at least arguably, we must turn to voters' ideas and information if we are to explain the phenomenon scientifically.
More broadly, what this and any number of other examples demonstrate is that to understand politics, at the same time that we keep in sight the demand side-the psychological limitations and biases that affect how we absorb information and generate ideas about politics out of it-we also need to attend to the supply side. This means attending to the ecology of information (or ecologies, as they differ between groups), the specific information that is most widely available and accessed, and the specific ideas that are most widely shared. And since we are most interested in the ecology of political information, we must focus primarily on the system of distribution for political information, the news media. We must ask what are the most common bits of information or facts transmitted by the news media, and what are the most common framings or packagings of those facts. Like media reform, this approach has not been tried and found wanting; it has never been tried, and it would be a shame if Achen and Bartels's call for a reorientation of public-opinion research around political psychology had the side effect of preventing it from ever being tried.