It’s time once again to look back on what I screwed up in the past year — there’s really no better way to start the new year than going over your biggest mistakes from the old one. 2016 brought hard lessons in humility for a lot of political observers, and I’m no exception. The year wasn’t as bad for me as 2015, when Donald Trump could have handed me a “Make America Great Again” conductor’s hat and I still wouldn’t have seen the Trump Train coming. Still, there were plenty of articles I wish I had written differently. There were some articles I wish I had written that I didn’t, and some I did write that I wish I hadn’t.
One of my most basic problems: Putting too much stock in exactly what the polls were showing. We like to emphasize how imperfect polling is here at FiveThirtyEight. We were screaming from the rooftops that the polls were showing a close race between Trump and Hillary Clinton through the last couple weeks of the campaign. But if poll results for voters overall are inexact, results for subgroups — black voters or men, for instance — are even more so. Yet I still made a big deal about those subgroup results in “Trump’s Doing Worse Than Romney Did Among White Voters.” Trump was counting on white voters without a college degree to win the general election. Of course, that plan would be scuttled if he did significantly worse than Mitt Romney among white voters with a college degree, which seemed likely to happen, according to the polls. Trump, though, did better among white voters than Romney in key swing states. Let this be a lesson: If the polls are off — and they usually are — then some of the subgroup results are going to be off too. Tread carefully in crosstabs.
Another mistake I made: Putting too much stock in recent historical precedent. That’s what I did in “Trump Is More Unpopular Than Clinton Is — And That Matters.” Um, no, it didn’t matter. Trump had lower favorability ratings than Clinton, and historically the less-liked presidential nominee has lost. I did point out that “it could also be that the unusually large share of voters who don’t like either candidate makes these numbers less meaningful than usual.” So give me some credit for that, although that caveat probably came too late in the story. The bigger issue, in my opinion, is that I didn’t really explore the roughly 20 percent of the electorate who didn’t like either candidate. Had I done that, I might have seen the exact opening Trump had, which he eventually exploited.
I also gave history too much weight before the first general election debate. In “First Debates Usually Go To The Challenger,” I argued that Trump had a better chance at a post-debate bounce than Clinton. That didn’t end up happening. In fact, the exact opposite happened.
Perhaps my biggest problem was not the stories I did write, but the stories I didn’t write. For example, I should have updated my August story “Trump Is In Fourth Place Among Black Voters” in October or November. Had I done so, we would have published an article headlined something like “Trump Is Polling Better Among Black Voters Than Romney Or McCain,” which would really have been a service to the reader. Trump, although still polling poorly among black voters, was polling somewhat better than either John McCain or Romney in the final weeks of the campaign. Slightly outperforming with black voters probably helped Trump squeak by in Michigan.
Another error of omission: I never wrote an article about Iowa during the general election. I noticed that Iowa appeared to be leaning more Republican two years ago, in 2014, but I never wrote about the implications of that trend this year. I should have. Trump’s strength in Iowa was an overlooked sign of his strength in the Midwest, in Wisconsin and Michigan. And even on its own, Iowa’s shift toward Republicans increased Trump’s chances of winning the Electoral College while losing the popular vote. But while I was not writing that story, I did somehow find the time in August to write “Missouri Could Be A Swing State Again.” That’s what the polling suggested at that point, but the fact that I found time to write that piece and not one about Iowa is regrettably stupid.
One last piece I wish I had written: ANYTHING ABOUT THE DEMOCRATIC PRIMARY IN MICHIGAN! That’s right, I didn’t write a single line about that contest before Bernie Sanders’ stunning upset win. Of course, if I had written something, it would have said that Clinton was a heavy favorite in Michigan. Still, I should have acknowledged that there was a primary going on.
And let’s just stay with the primaries for a moment, because I screwed up plenty of times early in 2016.
Just a couple of months into 2016, it was already becoming clear that endorsements from elected officials were not carrying much weight in the Republican primary — even if they are traditionally a sign of success. Yet I kept trying to make the case that Marco Rubio had a good shot at the GOP nomination well into February (“The Party Is Deciding On Rubio“). Trump seemed to figure out early on that the normal rules didn’t apply as much in 2016. He recognized that making false “rigged” claims, like the ones he made after he lost the Colorado conventions and later in the general election, played well with some of the public. While folks like me were stuck in the minutia (“Trump Made A Mistake By Overlooking Colorado“), Trump was making thematic arguments that sold to voters.
In 2016, unlike in 2015, I didn’t dismiss Trump’s chances at every turn. In fact, I think I got the big picture right. The polls showed Trump as an underdog, but not a big underdog. The polls pointed to important Clinton weaknesses in the Electoral College and among non-college-educated white voters. All that came across loud and clear in my writing. Two months before the election, for example, I pointed out that Clinton had a problem in the Electoral College compared to the popular vote (“Why Clinton’s Electoral Map Isn’t As Good As Obama’s“). A few days before the election, I wrote “Trump Is Just A Normal Polling Error Behind Clinton.” That’s what really gives me solace about 2016: If you were reading my election analysis, you shouldn’t have been shocked by Trump winning or — more importantly — how he won.
Still, there’s a lot of lessons to learn from 2016. But take heart, there’s always 2017.1