Those of you who actually know me in real life know that I had spent a few years on a book manuscript tentatively titled The Price of Partisanship. By the time I didn’t get the tenure, I was running up against the fact that something about the underlying logic did not make sense.
The basic logic behind the original argument was fairly simple and very consistent with the conventional wisdom, up to a point. First, the party in power manipulates the agenda to bias the outcome in its favor. Once the bias is introduced, parties increasingly become homogenized ideological groupings who are drawn to them because they become synonymous with ideological poles. As parties polarize, its agenda setting powers are used even more systematically to further pursuit of even more polarized agenda. Put differently, partisan agenda setting begets, sustains, and redoubles polarization. As the parties’ agendas grow ever more polarized, however, more and more voters become indifferent between the status quo and either party, and this discourages turnout. However, this process has a natural stopping point since the large body of nonvoting potential voters in the middle that polarization creates provides an opportunity for an enterprising politician to stake out the middle ground and take the election away from the polarizing parties. As it were, polarization can only be sustained if everyone does it, and at some point, there has to be a breaking point. Polarization enabled the ideologues who’d take over the parties an opportunity to advance policy goals in the short to medium term, while the breaking point represented the long term cost of the gains from partisanship.
So far, so good, but there were a few logical problems that didn’t wash. First, the large long term patterns in voter turnout, critical to the argument, did not bear much resemblance to the implication of the argument. Parties polarized fairly rapidly from late 1980s onward, accelerating in mid to late 1990s. If anything, polarization brought about an increase in aggregate turnout, not a decrease: 1970s and 1980s, with relatively low degrees of polarization, was characterized by generally lower turnout than the polarized 1990s and 2000s. There was a good reason for this, of course: voters knew who was who in politics much better: voters in 1970s actually needed to know something specific about a given congressman, say, to evaluate him as a candidate. The Democrat and Republican labels offered so little information that they provided no useful clue for what to think about him. Gaining and disseminating information particular to a candidate was both difficult and costly for voters and candidates alike. It was not simply a matter of money: if anything, money did not matter all that much. The really critical components of a successful campaign for Congress, in particular, were time, personal characteristics and accomplishments, and distinct legislative records around which the information was going to be built on, which were sparse commodities. If the turnout patterns are an indication, they didn’t add up to THAT much. While there is some reason to believe that excess agenda setting might be behind this–but, truth be told, I don’t think it is possible to marshal any empirical evidence for it. Those who are left out, again, are not necessarily for or against the agenda that is being enacted: their interests are simply orthogonal to what is being done. Some of their wants may be deferred. Some may be enacted. But on the whole, they are not so much being defied as much as ignored. This is an interesting corollary to Mayhew’s observation in 1974: being on the losing end in Congressional politics is not so bad, as it gives the losers the opportunities to position-take and define themselves. If you are simply ignored, you can’t even say that you are an enemy of those in power with clarity and consistency.
The logic does have an implication that goes beyond aggregate turnout: it is not so much how much turnout there is, but who turns out. Clearly defined party labels where Democrat corresponds neatly to “liberal” and Republican corresponds neatly to “conservative” would bring out ideological voters. Voters who don’t care much for ideology would not be so impressed. Have we reason to believe that the correlation between turnout and ideology has been changing systematically since, say, 1980? This is actually a very difficult question to answer because patterns in turnout have not been studied nor pertinent data been collected very systematically in a manner that makes it easy to draw linkage between political orientation and participation. Current Population Survey provides the best source of data on turnout, but is largely devoid of relevant “political” information, for good reason, since it is administered by the Census Bureau. Data with rather less fidelity is available via Pew, but the information concerning the earlier period is rather spotty. There are only two really systematic studies on turnout between 1980 or so and today, Who Votes? by Wolfinger and Rosenstone and Who Votes Now? by Leighley and Nagler, who, fortuitously provide book ends to the period of interest to me. Their arguments suggest that my broad suspicions hold: whereas the voting and the nonvoting electorates around 1980s were very similar in their political orientations, they are quite different today, with the gap between the ideologues and non-ideologues growing sharply since. The linkage between the Congressional parties and their legislative activities and the turnout, however, cannot be drawn with any meaningful statistical significance (literally and figuratively) due to sparseness of the data.
Compounding the problem further is that the apolical voters who drop out of the electorate will not be neatly defined along a neatly defined liberal-conservative axis. On one hand, this makes the task easier: Socio-economic-demographic data is far easier to find on the voters and nonvoters and, indeed, Leighley and Nagler have done a wonderful job of painting who these non-voters are in terms of these variables: the poorer, less educated, more ethnic, male, and less interested in politics overall. I have done some preliminary work that suggests that these are the critical missing voters that allow the Republicans to punch above their weight in House elections: safe Republican districts elect their congressmen with large winning margins and relatively few votes due to many voters who stay out of the electorate, especially in presidential elections. Extrapolating from the districts’ overall demographics and those of the voters who do turn out suggests that the gap between voters and nonvoters (again, mainly in presidential elections) is larger than in Democratic ones.
This is, of course, quite similar to that of so-called “missing white voters,” but with an important political twist. The finding by Leighley and Nagler is that, unlike when Wolfinger and Rosenstone wrote their book, today’s turnout is increasingly predictable on socio-economic-demographic variables. My tentative findings were that this pattern is starker in the districts won by Republicans with larger margins, that is, the voters who do not vote in those districts, at least in presidential elections, are distinctly different compared to those who do.
In my original conceptualization, these missing voters should have been “moderates” in the spatial sense who are disillusioned by growing partisan polarization, giving opportunity to the “centrists” to freeze out parties that do polarize. In this sense, Clintonian triangulation is a great idea that ensures permanent electoral majority. I had to abandon this idea–even though, apparently, my erstwhile political science brethren have not–because it became painfully obvious that there is no such thing as either a unidimensional policy space or “the median voter,” but I had to invent a whole new framework of logic to fit this argument into in lieu of a simple spatial model, which, quite frankly, remains incomplete yet–although thinking about the logic behind missing white voters has brought me much closer to completing that puzzle. The bottom line is that the missing voters are largely apolitical, who enter into politics only rarely (duh!), and, if they do, it would be in response to signals that need not lie in the same dimension at all, even assuming it is “spatial” in the first place. The good news is that this fits rather well with applications of decision theoretic models coupled with some neat new ideas in political behavior. The bad news is that I have no empirical evidence to link what turns off these voters has anything to do with “politics,” let alone partisan “polarization,” which I am myself defining in an unconventional fashion. So this is, in the end, just an assertion based on a rather flimsy evidence and much unconventional thinking: today’s politics has created an unequal playing field, where politics is typically contested on an increasingly unidimensional realm comprehensible only to those who like politics, in the Conversian sense of a clear ideological dimension, which reinforces itself if left to its own devices. But, as the process continues, it leaves aside people who don’t fit this one political dimension neatly and don’t process politics that way–which can only be shown indirectly, as we know that people who drop out of the electorate happen to coincide with people who are not as “ideological.” Too many apolitical people left out of the electorate, however, provides fuel for an occasional forest fire where unconventional candidates armed with programs that don’t fit the politics as usual enter the fray. If the campaign is successful, we may have a critical election and a realignment. If not, the politics as usual returns. Such events, of course, will be sufficiently rare that no amount of data will “predict” them.
In this sense, 2016 has been a wonderful campaign season. I hope I will still have an audience for this book, if and when it gets finished.