The Irrational Electorate
One of the best-selling political books of the 2008 election season has been Just How Stupid Are We? a report on “the truth about the American voter” by popular historian Rick Shenkman. Shenkman’s little book presents a familiar collection of bleak results from opinion surveys documenting some of the many things most Americans don’t know about politics, government, and American history. “Public ignorance,” he concludes, is “the most obvious cause” of “the foolishness that marks so much of American politics.” Lest this pronouncement seem dispiriting, an obligatory hopeful coda offers anodyne proposals for civic improvement.
Never mind whether the additional civics courses and “democracy parties” Shenkman proposes are really going to stem the tide of public ignorance. The reader’s first response to Shenkman’s indictment should be: So what?
Does it really matter whether voters can name the secretary of defense or know how long a senate term is? The political consequences of “public ignorance” must be demonstrated, not assumed. And that requires focusing not just on what voters don’t know, but on how what they don’t know actually affects how they vote. Do they manage to make sensible choices despite being hazy about the details of politics and government? (Okay, really hazy.) If they do, that’s not stupid—it’s efficient.
Obviously, what counts as a “sensible choice” is itself a matter of legitimate disagreement. Shenkman seems to think that since “foolishness . . . marks so much of American politics,” voters must be making stupid choices. However, most analysts have aspired to judge voters by less subjective standards—criteria grounded in specific notions of procedural rationality, or in voters’ own values and interests, or in comparisons with the behavior of better-informed voters who are similar in relevant ways. Moreover, such analysts have recognized that what really matters is not whether individual voters go astray, but whether entire electorates do. A lot of idiosyncratic behavior can be submerged in the collective verdict of 120 million voters.
According to Shenkman, “The consensus in the political science profession is that voters are rational.” Well, no. A half-century of scholarship provides plenty of grounds for pessimism about voters’ rationality.
When social scientists first started using detailed opinion surveys to study the attitudes and behavior of ordinary voters, they found some pretty sobering things. In the early 1950s, Paul Lazarsfeld and his colleagues at Columbia University concluded that electoral choices “are relatively invulnerable to direct argumentation” and “characterized more by faith than by conviction and by wishful expectation rather than careful prediction of consequences.” For example, voters consistently misperceived where candidates stood on the important issues of the day, seeing their favorite candidates’ stands as closer to their own and opposing candidates’ stands as more dissimilar than they actually were. They likewise exaggerated the extent of support for their favorite candidates among members of social groups they felt close to.
In 1960, a team of researchers from the University of Michigan published an even more influential study, The American Voter. They described “the general impoverishment of political thought in a large proportion of the electorate,” noting that “many people know the existence of few if any of the major issues of policy.” Shifts in election outcomes, they concluded, were largely attributable to defections from long-standing partisan loyalties by relatively unsophisticated voters with little grasp of issues or ideology. A recent replication of their work using surveys from 2000 and 2004 found that things haven’t changed much in the past half-century.
The intervening decades have seen a variety of concerted attempts to overturn or evade the findings of the classic Columbia and Michigan studies. In the 1970s, for instance, some scholars claimed to have discovered what the title of one prominent book called The Changing American Voter, a much more issue-oriented and ideologically consistent specimen than the earlier studies had portrayed. Unfortunately, further scrutiny revealed that most of the apparent improvement could be attributed to changes in the questions voters were being asked rather than a remarkable elevation of their political thinking. When voters were asked the old questions in the 1970s, their responses displayed no more consistency or sophistication than the responses from the 1950s described by the authors of The American Voter.
In the 1990s political scientists took a different tack, acknowledging that voters were generally inattentive and uninformed but denying that the quality of their political decisions suffered much as a result. A spate of books and articles with optimisticsounding titles such as The Reasoning Voter and The Rational Public argued that voters could use “information shortcuts” to make rational electoral choices even though they lacked detailed knowledge about candidates’ policies and platforms. These “shortcuts” could take many forms, including inferences from personal narratives, partisan stereotypes, and endorsements or other “cues” from trusted people or groups.
Unlike the analogous literature in psychology, this first wave of scholarship on political cues and “information shortcuts” stressed their potential value while paying little attention to the ways in which they could lead voters astray. In one of the most colorful examples of an “information shortcut,” political scientist Samuel Popkin suggested that MexicanAmerican voters had good reason to be suspicious of President Gerald Ford in 1976 because he didn’t know how to eat a tamale—a shortcoming revealed during his Texas GOP primary campaign against Ronald Reagan, when he made the mistake of trying to down one without first removing its cornhusk wrapper. According to Popkin, “Showing familiarity with a voter’s culture is an obvious and easy test of ability to relate to the problems and sensibilities of the ethnic group and to understand and care about them.” Obvious and easy, yes—but was this a reliable test? Would Mexican-American voters have been correct to infer that Ford was less sensitive to their concerns than Reagan? I have no idea, and neither does Popkin.
Lacking any objective standard for distinguishing reliable cues from unreliable ones, some scholars have simply asked whether uninformed voters—using whatever “information shortcuts” are available to them—manage to make similar choices to those of voters who are better informed, as the literature on “information shortcuts” suggests. That is what I did in a 1996 study, “Uninformed Votes,” which examined presidential elections from 1972 to ’92. Based on statistical analyses of votes cast in each election by well-informed and less-informed voters with similar characteristics, I assessed how closely voters’ actual choices matched the votes they would have cast had they been “fully informed.” I found that the actual choices fell about halfway between what they would have been if voters had been fully informed and what they would have been if everyone had cast their ballots on the basis of a coin flip.
In How Voters Decide, political scientists Richard Lau and David Redlawsk analyzed the same elections using a less demanding criterion for assessing “correct” voting. (They took each voter’s partisanship, policy positions, and evaluations of candidate performance as givens, ignoring the fact that these, too, may be subject to errors and biases.) They found that about 70 percent of voters, on average, chose the candidate who best matched their own preferences—a result, the researchers said, that left them “pleasantly surprised.”
Lau and Redlawsk raised, but did not really attempt to answer, the more consequential question: “Is 70 percent correct enough?” Answering that question requires a careful assessment of the extent to which “incorrect” votes skew election outcomes.
Optimism about the competence of democratic electorates has often been bolstered (at least among political scientists) by appeals to what has been dubbed the “miracle of aggregation”—an idea formalized in a mathematical demonstration by the social theorist Condorcet more than 200 years ago. He showed that if several jurors make independent judgments of a suspect’s guilt or innocence, a majority are quite likely to judge correctly even if every individual juror is only slightly more likely to reach the correct conclusion than he would simply by making a choice based on a coin flip. Applied to electoral politics, Condorcet’s logic suggests that the electorate as a whole may be much wiser than any individual voter.
The only problem with this elegant and powerful argument for the efficacy of majority rule is that it may not work very well in practice. Real voters’ errors are quite unlikely to be random and statistically independent, as Condorcet’s logic requires. When thousands or millions of voters misconstrue the same relevant fact or are swayed by the same vivid campaign ad, no amount of aggregation will produce the requisite miracle—individual voters’ “errors” will not cancel out in the overall election outcome.
In addition to assessing how well each individual voter’s choice matched his or her hypothetical “fully informed” choice, in “Uninformed Votes” I provided estimates of how well each overall election outcome matched what it would have been if every voter had been fully informed. The average discrepancy between the actual popular vote in each election and the hypothetical outcome if every voter had been fully informed amounted to three percentage points—more than enough to swing a close election. In four cases—1980, 1984, 1988, and 1992—the differences between actual and hypothetical election outcomes were large enough to provide strong evidence that “errors” by millions of individual voters did not entirely cancel out. These departures from “fully informed” election outcomes revealed a systematic bias in favor of incumbents, who generally did substantially better than they would have if voters had been fully informed, and a smaller bias in favor of Democratic candidates. Clearly, the “miracle of aggregation” is not sufficiently miraculous to render voters’ ignorance politically irrelevant.
Studies of this sort make it pretty clear that political ignorance matters—not only for individual votes, but also for election outcomes. Thus, this research undermines the notion that “information shortcuts” or sheer aggregation can compensate for voters’ shortcomings. Subsequent work has shed light on how some of the powerful political “heuristics” used by ordinary voters contribute to the problem. For example, a team of psychologists led by Alex Todorov established that candidates for governor, senator, or representative who are rated as “competent” by people judging them solely on the basis of photographs are considerably more likely to win real-world elections than those who look less competent. Brief exposure to the photographs—as little as one-tenth of a second—is sufficient to produce a significant correlation with actual election outcomes. A follow-up study showed that the electoral advantage of competent-looking candidates is strongest among less informed voters and those most heavily exposed to political advertising.
The ideal of rational voting behavior is further undermined by accumulating evidence that voters can be powerfully swayed by television advertising in the days just before an election. A major study of the 2000 presidential election by Richard Johnston, Michael Hagen, and Kathleen Hall Jamieson tracked prospective voters’ responses to changes in the volume and content of campaign ads as well as to news coverage and other aspects of the national campaign. Their analysis suggested that George W. Bush’s razor-thin victory hinged crucially on the fact that he had more money to spend on television ads in battleground states in the final weeks of the campaign.
A team of scholars from UCLA elaborated on this analysis in an attempt to clarify how long the effects of advertising last. They found that most of the effect of any given ad on voters’ preferences evaporated within one week, and that “only the most politically aware voters exhibited . . . long-term effects.” (Of course, the fact that the most engaged voters were susceptible to long-term effects of advertising may itself be troubling, but at least they responded to a considerable accumulation of arguments over the course of the campaign rather than solely to the last arguments they happened to hear before stepping into the voting booth.) In another study, the same authors found even shorter half-lives for advertising effects in a variety of state-level and congressional races. A third study, by a different team, also found only ephemeral advertising effects in the early stages of a Texas gubernatorial race. A major ad buy produced a seven-point increase in voter support for the featured candidate a day after the ads aired, but no discernible effect two days later. The authors noted that this “pattern of abrupt change and equilibration” in voter intentions in response to campaign advertising “appears to be inconsistent with a model of rational learning.”
These and other recent studies offer abundant evidence that election outcomes can be powerfully affected by factors unrelated to the competence and convictions of the candidates. But if voters are so whimsical, choosing the candidate with the most competent-looking face or the most recent television ad, how do they often manage to sound so sensible? Most people seem able to provide cogent-sounding reasons for voting the way they do. However, careful observation suggests that these “reasons” often are merely rationalizations constructed from readily available campaign rhetoric to justify preferences formed on other grounds.
Consider the role of Social Security privatization in the 2000 presidential election. It was a huge issue, the focus of more than one-tenth of all campaign-related television news coverage and about 200 ads on a typical television station in a battleground media market in the last week of the campaign. By Election Day, there was a strong statistical relationship between voters’ views about privatization and their presidential choices—just as one would expect if voters were pondering this important issue and casting their ballots accordingly. However, a detailed analysis by political scientist Gabriel Lenz found very little evidence that people actually changed their vote because of the Social Security debate. What happened, mostly, was that people who learned the candidates’ views on privatization from the blizzard of ads and news coverage simply adopted the position of the candidate they already supported for other reasons. The resulting appearance of “issue voting” was almost wholly illusory.
Findings such as these have led some political scientists to discount the role of “issue voting” in elections. Where else can one look to find support for the idea that voters are making rational choices? Perhaps they rely on a straightforward judgment about whether the country seems to be on the “right track” or “wrong track,” as pollsters often put it. Incumbents do, after all, tend to prosper in elections when times are good and suffer when times are bad. In an influential 1981 book, Retrospective Voting in American National Elections, political scientist Morris Fiorina attributed the electoral significance of economic booms and busts, successful or unsuccessful wars, and favorable or unfavorable social conditions to the fact that even uninformed citizens “typically have one comparatively hard bit of data: They know what life has been like during the incumbent’s administration.” The less they know about the details of policies and platforms, Fiorina reasoned, the more likely they are to rely upon “retrospective” voting as “a cost-cutting element” in deciding how to vote.
Fiorina’s theory struck political scientists as plausible, if not entirely edifying, because it seemed to demand much less of voters than the old-fashioned, unrealistic view that they should follow the news, formulate policy preferences, study the candidates’ platforms and records, weigh the relative importance of cross-cutting issues, and render a considered verdict regarding the best future course of government. Instead, they need only judge whether things are going well or badly. How hard can that be? Alas, my Princeton colleague Christopher Achen and I have produced a series of studies suggesting that even unheroic-sounding retrospective voting may be much harder than it sounds.
For one thing, voters’ perceptions may be seriously skewed by partisan biases. For example, in a 1988 survey a majority of respondents who described themselves as strong Democrats said that inflation had “gotten worse” over the eight years of the Reagan administration; in fact, it had fallen from 13.5 percent in 1980 to 4.1 percent in 1988. Conversely, a majority of Republicans in a 1996 survey said that the federal budget deficit had increased under Bill Clinton; in fact, the deficit had shrunk from $255 billion to $22 billion. Surprisingly, misperceptions of this sort are often most prevalent among people who should know better—those who are generally well informed about politics, at least as evidenced by their answers to factual questions about political figures, issues, and textbook civics. If close attention to elite political discourse mostly teaches people to believe what the partisan elites on “their” side would like to be true, the fundamental premise of books such as Rick Shenkman’s—that a more attentive, politically engaged electorate would make for a healthier democracy—may be groundless.
Even when voters do have an accurate sense of how things are going, they tend to be inordinately focused on the here and now. For example, studies of economy-driven voting almost invariably find that voters are strongly influenced by economic conditions during the election year, or even some fraction of it, but mostly ignore how the economy performed over the rest of the incumbent’s term.
That shortsightedness is not just a psychological quirk; it has significant political consequences. Over the past 60 years, there has been a marked partisan disparity in the timing of income growth, with Democratic presidents presiding over more overall growth (especially for middle-class and working poor people), but Republicans presiding over more growth (especially for affluent people) in presidential election years. Thus, voters’ economic myopia has produced a substantial Republican bias in presidential election results—a bias large enough to have been decisive in three of the nine Republican victories since World War II: in 1952, 1968, and 2000.
No Republican boom seems to be forthcoming in this election year, and John McCain will be punished at the polls as a result. Whether the current economic distress is really President Bush’s fault, much less Senator McCain’s, is largely beside the point.
Voters have great difficulty judging which aspects of their own and the country’s well-being are the responsibility of elected leaders and which are not. In the summer of 1916, for example, a dramatic weeklong series of shark attacks along New Jersey beaches left four people dead. Tourists fled, leaving some resorts with 75 percent vacancy rates in the midst of their high season. Letters poured into congressional offices demanding federal action; but what action would be effective in such circumstances? Voters probably didn’t know, but neither did they care. When President Woodrow Wilson—a former governor of New Jersey with strong local ties—ran for reelection a few months later, he was punished at the polls, losing as much as 10 percent of his expected vote in towns where shark attacks had occurred.
New Jersey voters’ reaction to shark attacks was dramatic, but hardly anomalous. Throughout the 20th century, presidential candidates from incumbent parties suffered substantial vote losses in states afflicted by droughts or wet spells. Shenkman argues that “‘throw the bums out’ may not be a sophisticated response to adversity but it is a rational one.” However, punishing the president’s party because it hasn’t rained is no more “rational” than kicking the dog after a hard day at work.
While voters are busy meting out myopic, simple-minded rewards and punishments, political observers are often busy exaggerating the policy content of the voters’ verdicts. The prime example in American political history may be the watershed New Deal election of 1936. Having swept into office on a strong tide of economic discontent in 1932, Franklin Roosevelt initiated a series of wide-ranging new policies to cope with the Great Depression. According to the most authoritative political scholar of the era, V. O. Key, “The voters responded with a resounding ratification of the new thrust of governmental policy”—a stunning 46-state landslide that ushered in an era of Democratic electoral dominance.
The 1936 election has become the most celebrated textbook case of ideological realignment in American history. However, a careful look at state-by-state voting patterns suggests that this resounding ratification of Roosevelt’s policies was strongly concentrated in the states that happened to enjoy robust income growth in the months leading up to the vote. Indeed, the apparent impact of short-term economic conditions was so powerful that, if the recession of 1938 had occurred in 1936, Roosevelt probably would have been a one-term president.
It’s not only in the United States that the Depression-era tendency to “throw the bums out” looks like something less than a rational policy judgment. In the United States, voters replaced Republicans with Democrats in 1932 and the economy improved. In Britain and Australia, voters replaced Labor governments with conservatives and the economy improved. In Sweden, voters replaced Conservatives with Liberals, then with Social Democrats, and the economy improved. In the Canadian agricultural province of Saskatchewan, voters replaced Conservatives with Socialists and the economy improved. In the adjacent agricultural province of Alberta, voters replaced a socialist party with a right-leaning party created from scratch by a charismatic radio preacher peddling a flighty share-the-wealth scheme, and the economy improved. In Weimar Germany, where economic distress was deeper and longer lasting, voters rejected all of the mainstream parties, the Nazis seized power, and the economy improved. In every case, the party that happened to be in power when the Depression eased went on to dominate politics for a decade or more thereafter. It seems far-fetched to imagine that all these contradictory shifts represented well-considered ideological conversions. A more parsimonious interpretation is that voters simply—and simple-mindedly—rewarded whoever happened to be in power when things got better.
Stupid? No, just human. And thus—to borrow the title of another current bestseller, by behavioral economist Dan Ariely—“predictably irrational.” That may be bad enough.
This article originally appeared in print