On April 2, 2016, eight months before he was elected, four months before he secured the Republican nomination, and one month before his GOP rivals dropped out of the race, I made this prediction on Business Insider:
- Donald Trump would win the election and become president.
- He would need only a minority of the popular vote to do so (gaining a majority in the electoral college).
- He would do it by increasing his support from white voters.
- And — based on a theory by an obscure Republican election strategist from Maine named Matthew Gagnon — Trump would take Pennsylvania, Ohio, Michigan, and Wisconsin. Those are traditionally Democratic states that haven’t been won by a Republican since the 1980s.
That is exactly what just happened.
To be clear, I also reiterated the view of the”experts” at the time, which was that this scenario was extremely unlikely to happen. I should have put some money on a Trump victory but I did not. Although I found my prediction interesting as an intellectual exercise, I didn’t believe my own eyes, even as it came true right in front of me.
Here is how it happened.
Back in April, it was simply not conceivable that Trump would win. Or at least, not conceivable to anyone outside the Trump camp. The experts all agreed: You cannot become president if you have angered women and non-white minorities. You need to appeal to all the demographics to get a majority of votes. A couple of colleagues pressed me not to publish my theory because they thought it was foolish.
But the Trump campaign was the only thing everyone was talking about, so it seemed useful to discuss whether there was any logistical way he could get the White House despite the odds.
That is when I found this Quora post from Matthew Gagnon. You should go read it. It is the Nostradamus text of modern Republican politics. And it came from someone who didn’t want Trump to win the primaries. “Since I didn’t like Trump, I hoped my opinion would be taken more seriously,” he told me.
Although it was written in March 2016, Gagnon’s Quor answer reads now like a post-election analysis. Aside from a few stats about exactly how many electoral college votes Trump would get, Gagnon nailed every aspect of this election eight months before it happened — all the way down to the way the pre-vote polls would be wrong.
The gist of it is about how angry white, working class voters are in many states:
“Think about it. You made good money. You provided for your family. Suddenly a bunch of low wage workers from what you consider a third world hellhole (whether that is fair or not is not the point) steal your jobs when the company you work for offshores your factory. Republicans cheer. Democrats cheer. They both love free trade. You hate it, and you blame it for your poor economic circumstance.
“The number of these voters who have been depressed, ignored, dismissed, and outright ridiculed is tremendous. And there are a lot of them in states like Pennsylvania, Ohio, Michigan, and Wisconsin. A lot.
“That means that these states — particularly Pennsylvania and Michigan — which have been Republican targets for nearly thirty years to no avail, have a very real possibility of flipping.
Flip they did, all four of them.
“Every race is winnable for any candidate. Every single one. It bothers me when people say somebody ‘has no chance’ — it is literally never true,” Gagnon told me. “So I thought the readers might enjoy getting a more sober analysis.”
“I myself come from Maine’s second district, so I lived with and grew up around the voters who threw the election to Trump this year, and know them well. I think most of the elites, media, opinion makers and political operatives don’t understand either these people, or the phenomenon that has empowered them politically, nor do they understand Trump’s appeal and ability to realign the politics of this country,” Gagnon added.
Back in April, I thought “nice theory — but no data.” So I went looking for stats that might show whether white voters in those states could turn the whole election toward Trump.
Nate Silver’s FiveThirtyEight site has an invaluable interactive map that uses data from the 2012 election and current demographic data to let you run various scenarios of your own. Basically, you use a set of sliders to adjust the levels of support from whites, blacks, Asians, and so on, in order to get different results.
When I began playing around with the map, my assumption was to calculate what was the least amount of votes from 2012 Trump had to shift in his favour to win. It turns out that the smallest adjustment you can make to create a speculative Trump victory was to move the “non-college educated white” vote favouring Republicans from 62% in 2012 to 69% in 2016. Furthermore, the other demographics are so much smaller than the white bloc, that it requires much bigger shifts on those variables to change the result.
Whites may be in decline in America, but they are still the largest voting bloc. Shift them and you’ve shifted the entire election. This was how that speculative pro-Trump map looked right before the election:
Note specifically that the map gives Gagnon’s angry white states to Trump, and adds Florida to the mix as well.
Although President Obama’s 51%-47% victory over Mitt Romney in 2012 felt like a landslide at the time, the winning margin was only 4% of all votes cast (5 million votes out of 127 million cast). Trump didn’t have to swing 4% — he simply had to swing enough of that margin in the angry white states to get control of the electoral college.
It’s worth re-noting that Hillary Clinton actually got more votes nationally than Trump, she simply got them in the wrong states because the distorting effect of the electoral college favoured Trump’s angry white states.
So now I had a nice theory and some data showing that Trump could actually win with a relatively slight movement in votes.
Then, when I wrote my November 4 recap, I noticed that FiveThirtyEight had changed its data on white voters. Previously, in April, the site said non-college whites formed 33% of the vote in 2016. But by November FiveThirtyEight updated its data. It now said: “Non-college-educated whites skew older and rural and will be 46 per cent of eligible voters in 2016.”
Forty-six per cent. That’s nearly half of everyone casting a ballot.
I do not know why that change was made. But the change was chilling for anyone who assumed Clinton would win. When I first started looking at this in April, my notion was, “what is the whacky/scary/crazy/outlier scenario that would have to happen for Trump to win?” The new FiveThirtyEight data put us well within that scenario.
The data also adjusted the effect of the sliders on FiveThirtyEight’s map. Now, Trump only had to shift his level of under-educated white support from 60% of their votes to 65%. On the old numbers the shift he needed was 62% to 69%, or seven points. The newer data showed he only actually needed a five-point shift in that demo.
Gagnon was right. The data was there for anyone to see: Trump had it in the bag as far back as March.
This is an editorial. The opinions and conclusions expressed above are those of the author.
NOW WATCH: Briefing videos
Business Insider Emails & Alerts
Site highlights each day to your inbox.