In his best-selling book The Black Swan, probabilist Nassim Nicholas Taleb warns against the need for easy narratives to explain the unexpected. Given how unexpected the result of this Tuesday’s election was, it is worth taking some time to review what Taleb calls “the narrative fallacy.”
According to Taleb,
The narrative fallacy addresses our limited ability to look at sequences of facts without weaving an explanation into them, or, equivalently, forcing a logical link, an arrow of relationship, upon them. Explanations bind facts together. They make them all the more easily remembered; they help them make more sense. Where this propensity can go wrong is when it increases our impression of understanding.
Yesterday, I reviewed New York Times exit polling data to try to look at what we actually know about who voted for president-elect Donald Trump or Sec. Hillary Clinton. The results, as I noted, were often surprising.
The reason they are surprising is precisely because of what Taleb gets at here: we have a tendency to want everything to fit into neat-and-tidy narratives. But reality rarely works that way, especially in the case of unexpected events. Donald Trump’s win was a Black Swan event to many.
By nearly every poll, Hillary Clinton was the favorite to win (and, of course, it appears she did win the popular vote). The most cautious index I saw was FiveThirtyEight, where to their credit they stressed the probabilistic nature of their forecast. They basically put the odds at 2-to-1 in favor of Clinton, and they even said that her chances were not as good as President Obama’s reelection in 2012. Even so, they too have been reeling at the inaccuracy of, again, basically every poll.
So Trump won. It wasn’t outside the realm of possibility (obviously now), but it was certainly unexpected to most, even some of those who gave him a better chance than others. How did it happen?
Ah! There’s the catch! “How did it happen?” is the question everyone wants to have the answer to, no matter if they voted for the winner or one of his opponents. I have seen explanations that run the gamut of apocalyptic to miraculous and many in-between.
Economic Class?
Was it due to economic class division that favored Trump? Partly, but as I noted yesterday, Clinton still won low income voters, according to the NYT exit polls (we will get better data in the days and weeks ahead). People generally qualify this as “white working class” support for Trump. Again, there is definitely something to that, but other indicators clearly mattered too.
Race?
Was it due to issues of race, from policing to immigration to white identity politics? The vocal alt-right support for Trump was well (too well?) covered by mainstream media. His restrictionist immigration policy was the flagship of his campaign. He declared himself the “law and order” candidate. Certainly that was all part of his win. People who listed immigration as their top issue voted in large part for Trump, for example. But according to the data Trump also won a greater share of black, Latino, and Asian American voters than did Romney in 2012. This explanation would satisfy many, but it can’t be the whole story. (One problem is clearly that people assume black, Latino, and Asian Americans are all homogeneous groups. But Cuban and Vietnamese Americans, for example, have long voted majority Republican.)
Bernie Sanders?
Was it due to Clinton failing to connect with Senator Bernie Sanders’ supporters? There is something to this too. Independent voting was especially high among lower age groups, who favored Sanders in the Democratic primary. (Although, being a member of that group I would stress that not all of them voted in the Democratic primary or even then supported Sanders.) News of DNC support for Clinton during the primaries broke during the presidential debates as well, seeming to confirm the claim of many Sanders supporters that the DNC, beyond their super-delegate system, had its hand on the scale. But one would still expect a much higher turnout for Green party candidate Dr. Jill Stein, for example, if that brand of politics was really a sine qua non for those voters. Maybe they only cared about one or two issues in the end (such as opposition to international trade), and Clinton didn’t offer them as much as Trump. Maybe, but we really don’t know.
#NeverTrump?
Was it due to the Clinton camp failing to reach out to the concerns of disaffected, #NeverTrump conservatives? That was probably a missed opportunity, but either they weren’t as #NeverTrump as they claimed or they weren’t as large a voting block as some imagined. Even in Utah, where independent social conservative Evan McMullin was on the ballot and had seen some strong polling numbers, Trump still won the state handily with 47% of the vote. McMullin, who had been polling above Hillary Clinton, came in third with only 20% in his home state among a voting demographic that looked a lot like him (white, Mormon, socially conservative).
Pro-Life?
Was it due to white evangelical and other pro-life voters who were concerned with the Supreme Court voting for Trump? Again, in some cases for sure. I’ve talked with those voters before and after the election. And Trump improved over Romney notably among Roman Catholics, for example, but his numbers with white evangelicals were actually about the same as Romney. As I said yesterday, “White evangelicals voting Republican are not an anomaly.”
Terrorism or Islamophobia?
Was it due to fears over terrorism or, worse (and not at all the same), Islamophobia motivating people to vote for Trump? Trump certainly won those voters who listed terrorism as their top concern. But he also improved over Romney among voters who are non-Jewish, non-Christian, but also not “no religion” (unhelpfully listed as “something else” by the NYT). That category has to include American Muslims. I’d love to see a breakdown of the numbers to know for sure — maybe a huge Hindu or Buddhist turnout for some reason favored Trump? maybe Mormon falls into this category (some demographers would include it among Christian) — but at the moment this narrative is too simple as well. Perhaps of note in this regard: my wife was electronically exit-polled and had to answer “other” for her religion because the options she was given was either Protestant Christian or Roman Catholic Christian. We are Eastern Orthodox Christians.
Anger at the Establishment?
Was it anti-establishment anger that fueled Trump’s win? The outsider vs. “the Establishment”? Federalism vs. centralization? Again, Trump won voters who were angry at the federal government — so that is definitely a factor — but he did not at all do as well as Romney did in 2012. Clinton actually improved over Obama’s 2012 numbers among angry voters.
Third Parties?
Was it due to third party voting? This seems the least likely to me. Jill Stein didn’t do very well, and as Johnson’s polling numbers declined over the last month, Trump’s, not Clinton’s, rose. In the final tally, it appears Johnson underperformed the final polls as well, suggesting that many who said they’d vote for him instead of Trump changed their mind at the last minute. Indeed, as I pointed out yesterday, Trump voters were more likely to have come to the decision to vote for him in the last days and weeks of the campaign. Thus, if there is any story here, it is not that third party voters took votes from Clinton, but rather that there were too few third party voters to draw votes away from Trump in the end.
Concluding Thoughts
And of course, there are many other narratives out there and to come in the future, many of which will be valuable and true to some degree, but never total.
As moral psychologist Jonathan Haidt has pointed out, while we often think we use our reason to understand how things work, we more often use it to justify what we already felt or thought about any given topic. Humans are free rational animals, after all, with passions and appetites, not robots. What we cling to as reasons that explain things are often just self-justifications. Add to that the Christian conviction that “all have sinned” (Romans 3:23) and are corrupted by sin, and self-justification and self-deception should be things we expect and actively endeavor to guard ourselves against — an ancient Christian ascetic practice known as watchfulness.
Thus, while fascinated by the many factors that appear to have affected the result of the 2016 presidential race, I’m also trying to resist the need to have an easy explanation. Depending on the person, casting the result as wholly good or wholly disastrous may be comforting, but doing so simply does not reflect the complex composure of those people, equally created after the image of God as you and me, who actually voted in this country. As Nate Silver pointed out at FiveThirtyEight yesterday, if only 1 in 100 Trump voters chose Clinton instead, the narratives we’d be hearing this week and in the months to come would be very different, even though the actual margin of victory for her would be nearly as close.
Another lesson here is to remember economist F. A. Hayek’s conviction, as Jordan Ballor touched on yesterday, that there are real limits to what can be measured, and thus to what we can plan on based on what we measure. That is not to say that data is unimportant — obviously I think it tells us a lot — but only that its predictive power is limited when there are so many important but non-quantifiable factors that affect real life. Really doing so requires humility on our parts, a virtue that ought to characterize any Christian’s life, no matter if he or she is happy with Tuesday’s election results or not.
