With just under two weeks to go until the election of 2016 finally and mercifully passes into the history books, polls consistently show Hillary Clinton leading by a comfortable margin, albeit one which may be narrowing slightly. However, political polling has been relying more and more on the demographic assumptions of the models used to compensate for ever decreasing response rates to telephone polls. That is to say, instead of purely measuring responses, the polls have become more and more a reflection of the assumptions used to take the measurements (emphasis added below). Without going down the conspiracy theory rabbit hole (please, don’t), it’s perfectly reasonable to wonder whether this election is the beginning, or possibly even the middle, of a major realignment in American politics.
Keeter noted that some nonresponse bias does exist: less-educated people and people of color are less likely to participate in polls, and people who aren’t politically active are less likely to answer political surveys. But, he concluded, “most of these biases can be corrected through demographic weighting of the sort that is nearly universally used by pollsters.”
Every election cycle the winning party proclaims their victory is evidence of a mandate that has forever altered the political landscape, but this usually is not the case. That said, this election is so different from recent contests in several remarkable ways: both major party candidates have huge negatives and emerged from especially acrimonious primary fights. Also, a fifth-party candidate who isn’t even on the ballot in all 50 states is within reach of possibly winning electoral votes in Utah, which has not happened since 1968–also, interestingly, a year of crisis and realignment for the two major parties. Donald Trump’s success or failure will turn on whether people who have not been engaged in politics–precisely the people who tend to be underrepresented in polls–are motivated by his populist demagoguery to participate for the first time in this election.
To understand why such a realignment could affect the accuracy of polling we can look to the study of the stock market. The blog “Political Calculations” has done considerable research into the behavior of markets, which generally remain within a few standard deviations of a long-term stable trend line–until they don’t. Such deviations outside of the stable trend are called a Lévy flight. These events happen when some unexpected news suddenly changes the assumptions and expectations of the market such as last year’s panic over the true depth of China’s economic slowdown which established a new trend in the market.
Other events, such as the Brexit shock, are transitory. Then, after a short freak-out, investors sat down, had some tea, and remembered that Great Britain is still one of the top five economies in the world. Now markets are keeping calm and carrying on in proper British fashion. Indeed, after last year’s China uncertainty and our crude oil price war with Russia had all settled down, we have remained on a fairly stable trajectory since March of this year, Brexit notwithstanding. The question is then, whether this bizarre election is a momentary blip or a major realignment? In the latter case, what if the models and demographics are just plain wrong?
Nate Silver’s popular FiveThirtyEight model uses state-level data from 1972 to the present. He admits that this is not much data–only 12 presidential elections’ worth. Moreover, the period from 1972-2014 has been a pretty stable political order, so what if that stable order is breaking down? Nate Silver also gave odds as of Wednesday morning that Trump only had something like a 15-17% chance of winning the election, but that’s actually a lot better odds than the Lévy flights that we see in the market that go outside of three standard deviations, the odds of which are less than 1%. A political Lévy flight in which Trump wins a shocking upset landslide like 1980 is really only between one or two standard deviations, which is to say, pretty unlikely, but also not really that preposterous. To put it differently, if you were offered one million dollars to play Russian roulette, even a one in twenty chance of losing (which is two standard deviations) would be too uncomfortable for any sane person to risk those odds.
In gambling, where each toss of the dice is a discrete Bernoulli trial, it is mere superstition to talk about a particular combination being “due.” At the craps table, the odds of rolling snake eyes–one in 36, to be precise–are always the same, each and every time. Politics is not a Bernoulli trial. Politics is like any other natural process from earthquakes and volcanoes to weather patterns and the behavior of markets, where each observation is affected by the preceding outcome. For a timely example, just look at this year’s World Series which will end one of the longest droughts in sports history regardless of who wins.
Thus, just as we can speak about being “overdue” for a major hurricane making landfall in Florida like this year’s Matthew or the end of either the Cubs’ or the Indians’ legendary losing streak, we can also speak of being “overdue” for a political realignment. Given the enormous upheavals in the postwar era which ushered in the Baby Boomer generation, it is perfectly reasonable to assume that as the younger generations increasingly assert their political influence, we will also have enormous upheavals and dislocations which even if they are foreseeable, remain unpredictable and chaotic.
At a fundamental level, we can also look at the historical shifts in the political landscape which occur pretty regularly. Turbulent and bitterly close elections are fairly common, such as 1800, 1828, 1836, 1844, 1860, 1876, 1888/1892, 1912/1916, 1948, 1960, and of course, the violent and chaotic assassinations and riots of 1968. Many of these years have come to demarcate political eras and also coincide with each successive generation. To wit, this is almost certainly the last election where we will have two Baby Boomers running at the top of the two major parties. It may also be the last election where the two major parties are the Democrats and the Republicans.
One thing is for certain: even if Hillary Clinton wins this election as most pollsters are predicting, the animal spirits of our nation’s political life are restless and violent. If not this this year, maybe it won’t be until 2020 or 2024, but the old consensus that lasted from about 1972-2008 is gone and something else is going to take its place. In this election, we have two especially terrible options, but if you think the idea of voting for Donald Trump is unpleasant now, consider that in the next era of politics which is rapidly drawing upon us, we could very well have even worse options to choose from in every election for decades to come. If a realignment is really happening, the greatest danger of Hillary Clinton being elected is that whoever runs against her in the next election will say the same things that Trump does, but he or she will actually be good at it.