As just discussed, temporal comparisons made by election commentators often implicitly assume that any change in electoral behavior indicates a change in the voters, overlooking the fact that the change may be in the alternatives between or among which voters choose. Did Republicans defect more than usual in 1964 (Goldwater) and Democrats more than usual in 1972 (McGovern) because their party identifications experienced a sudden weakening in those years? To some extent they did, according to ANES data, but the more important part of the story was that each party nominated a factional candidate who was strongly opposed by other factions of the party.

This is an easy example, not likely to be misunderstood, but the same consideration is present in every election. If an issue suddenly recedes or surges in importance, and/or there is a sharp change in electoral outcomes, it may not be the voters who are changing, but the candidates and the positions they take [Fiorina, Abrams, and Pope (2005), chapter 8]. Thus, while surveys consistently show abortion to be a second-tier issue for most Americans, some polls in 2012 showed a rise in importance during the campaign (e.g., Saad 2012). Did that reflect a change in voters’ priorities, or did it reflect the fact that out-of-the-mainstream comments by some Republican candidates made some Americans think, “Gee, these people are farther out on this issue than I had realized.”⁷

After presidential elections in which one party replaces another or congressional elections in which majority control shifts, one hears pronouncements that the electorate has shifted to the left or right. In part, that is true. Stimson (2004) shows that electoral reversals tend to occur at the end of a shift in voter “mood” in one ideological direction or another. But when a Democratic president takes office, mood shifts in a conservative direction, and when a Republican takes office, mood shifts in a liberal direction (Wlezien 1995). Here it appears that it is not voter preferences that are changing but the policies government produces. Loosely speaking, under Democratic administrations, centrist voters get too much of what they want, and under Republican administrations, they get too little (Stimson 2012).

As another example, consider the concept of “swing voters” (Mayer 2008, 2012). During the 2012 campaign there were frequent references to the small number of undecided voters (e.g., Epstein 2012). There are various ways of estimating the proportion of undecided voters – those who say they are undecided whom they will vote for, those who say they might change their minds, those who actually give different answers in a panel (Hillygus and Jackman 2003) – but I doubt that any method can give us a very precise estimate. The reason is that whether a voter will move between parties is to some extent endo­geneous: it depends on the candidates and issues.

In 1984 (Reagan v. Mondale), some surveys indicated that the number of swing voters was in the single digits, but 8 years later, nearly 20% of the electorate swung against both major parties and supported Ross Perot. Did the voters change that much in 8 years? Or more likely, did a large number of them arrive at a different decision when given a different set of choices? It is unlikely that strong partisans will swing, so their proportion of the electorate may give us a rough upper limit on the number of swing voters, but below that limit, different issues and candidates can produce wide variation.

The subject of independents is another major source of confusion in election commentary. A first source of confusion is their number: is it more than a third of the eligible electorate and thus the largest “party” in the US, or is it <10%, a number so small that it justifies the modern strategy of mobilizing the base as opposed to the traditional strategy of moving to the center? The second subject of confusion is the composition of the independent category. Are independents a mass of centrists (Killian 2012) or proto-libertarians (Boaz and Kirby 2010) waiting to be mobilized and move our gridlocked political system in a positive direction?

Personally, I wish they were, but they are not. Yes, some are ideological moderates not comfortable in either a too-liberal Democratic Party or a too-conservative Republican one. Yes, some are cross-pressured, like libertarians who are closer to the Republicans on economics and to the Democrats on social issues. But others are uninterested and uninformed about politics generally. And still others are alienated from politics entirely. One thing we can say with some confidence is that those who take the independent label when asked the standard question about party identification are a heterogeneous lot.

How many independents are there? Here political scientists bear a great deal of responsibility for the rampant confusion that exists. Political scientists traditionally divide independents into “pure” independents and “leaners.” The latter are independents who reply that they are “closer to” (note, not “think of themselves as”) one party or the other. There is ample evidence that pure independents are less informed and less involved politically than leaners (Keith et al. 1992), but not so the claim that leaners are actually “closet,” “covert,” or “hidden” partisans, a claim stated with great certitude by some political scientists:

The evidence for such assertions is remarkably thin. A careful reading of the “research” and the “numerous studies” shows that the empirical evidence is equally consistent with not one but two propositions: [1] leaners are covert partisans as often claimed, and [2] leaners use their vote intention or issue positions to answer the question about which party they are closer to.⁸ An analysis of ANES panel shows that in 56 of 60 comparisons, leaning Democrats and Republicans are less stable than the corresponding weak identifiers (as well as strong identifiers, of course) in their responses to the party ID item.

On average, leaners hold the same position from wave to wave <40% of the time, compared to 50% for weak identifiers. Moreover, the variance of leaners’ moves on the scale is greater (Abrams and Fiorina 2012). These findings are consistent with Brody’s (1978) and Miller’s (1991) contention that the direction in which independents lean is endogeneous.⁹ In addition, leaners are more likely to vote for third parties, to assert that no party represents their interests very well, and to rate the party toward which they lean lower than weak identifiers do. Moreover, in a number of respects, the policy views of leaners differ from those of weak identifiers – not greatly, to be sure, but consistently (Abrams and Fiorina 2012).

None of this shows that leaners are pure independents; they are not. But to the extent that the causal arrow is the reverse of that generally posited by many political scientists, equating leaners with weak partisans exaggerates the importance of partisanship, understates the importance of candidates and issues, and underestimates the potential change in the electorate that new candidates and issues could generate. This is a subject that cries out for more research that exploits the wealth of new data now available.

Some commentators write as if there is a direct correspondence between candidate choice in polls or elections and the underlying distribution of public opinion. In 2012, there were claims that the closeness of the polls indicated that Americans were deeply divided.¹⁰ Once again, closely-divided and deeply-divided are distinct concepts. If every voter were totally indifferent between the candidates, walked into the voting booth, flipped a coin and said “heads Romney, tails Obama,” the expected result would be a 50:50 election, but it would not indicate a polarized electorate.

Only additional public opinion data can shed light on that question. As Pope (2012) puts it, “When scrutinized closely, it becomes clear that Americans often vote in a highly partisan way, yet they are far less likely to think in a clearly partisan fashion.” If votes were divisible, many voters might want to vote something like 66% Democratic, 34% Republican or vice-versa, but the indivisible nature of the voting act prevents such behavior.

According to some political scientists “We all know that 2012 was a partisan, polarized election …” Kondik, Geoffrey, and Sabato (2012) Do we? Partisan probably – most recent elections have been partisan – but at the time of this writing, the data that will allow us to measure how deeply divided the American electorate was in 2012 are not yet available. I suspect that when the relevant analyses are performed, they will show an American citizenry that was less intense about its candidate preferences in 2012 than in 2008, and certainly one less intense than in 2004.¹¹

According to Markos Moulitsas (2012), founder of Daily Kos:

And from the other side of the political spectrum, conservative commentator and author Stanley Kurtz (2012) concurs:

Political journalists need to remember that most of the people they talk to professionally are abnormal, that is, that they are statistically far from the average. The political class comes from the tails of the distribution of American public opinion. Political scientists have documented this since the 1950s, when McCloskey and his associates (McCloskey, Hoffman, and O’Hara 1960) pointed out the differences between Republican national convention dele­gates and the Republican rank-and-file.¹² While we do not have comparable data that would enable us to track the degree of divergence over time, it is likely that the process of party sorting has increased this divergence, because as noted earlier, the sorting process is much farther along at the elite level than at the voter level.

Hard though it may be for political journalists to believe, of the approximately 61 million people who voted for Mitt Romney, more than 58 million did not watch Fox News last night. (I am making the generous assumption that all of the 2.6 million nightly viewers turned out and voted Republican.) Similarly, of the 65.5 million Americans who voted for Barack Obama, nearly 64 million did not watch any MSNBC news show last night.¹³

In the US, there is indeed a great divide within the political class – those Americans most involved with politics, including public officials, candidates, donors, campaign activists, interest group activists, and, lest we forget, the intellectuals and commentators allied with the parties. To journalists who spend their days interacting with people like this, the evidence of polarization is overwhelming.

Yet away from the politicized haunts where the national media is headquartered, the world is different. Most Americans are not ideologues and do not hold extreme views. They feel no need to impress their friends and acquaintances with the purity of their political views and the intensity with which they hold them. And they have no monetary or career incentives to behave like jerks.

I will write “probably” because it is always possible that research on the 2012 campaign will show the opposite. But this is most unlikely. As Diana Mutz (2012) notes, there is an “enormous chasm” between the beliefs held by journalists and the typical voter about the effects of campaign media vs. the findings of political communications scholars. The former believe that the effects of the media are major, even determinative, whereas academic research finds much smaller effects. I shall say no more about this, because given the long history of the disjunction, it is doubtful that academics could change journalists’ minds about this subject if they had a whole semester, not just a seminar. Who are they going to believe: academic researchers, or their own eyes and ears?

Beyond being told over and over by self-interested big-spending campaign consultants that TV ads are powerful influences on public opinion and voting behavior, part of the reason that political commentators ascribe such great importance to the media is the implicit assumption by many of them that Americans are dumb. Even though pre-school age children quickly learn to discount commercials on their TV programs, mature adults are supposedly unable to resist the influence of political media.

To be sure, voters are generally uninformed, and many try to avoid politics until the imminent approach of election day makes it impossible. Yet the collective electorate manifests a degree of knowledge and wisdom that gives those of us who have studied that electorate for decades some cause for optimism.¹⁴ If only I could say as much about the knowledge and wisdom of the political class.