فصل 83

83

The Bottom Falls Out

The evidence backs up the idea that there was a late shift away from me and toward Trump and third-party candidates. My support was strong in early voting across the country, and early-vote turnout roughly matched what our models predicted. But things collapsed in the final days and on Election Day itself.

In real time, it was hard to appreciate how fragile our position was. As I mentioned earlier, Joel Benenson’s polling showed a solid lead in the final week. Our data analytics team was also surveying thousands of people each night. “We have seen our margins tighten across the battleground states,” Elan Kriegel reported on November 3. But, he continued, “our national toplines have been +3 each of the last four nights.” We were up by the same 3-point margin in Michigan, Wisconsin, and Pennsylvania, he said. Democratic Senate campaigns and party committees were seeing similar numbers, and some were even more optimistic.

Exit polls would later find that voters who were still making up their minds in those final days broke strongly for Trump. In Pennsylvania, a state with no early voting allowed at all, the margin among late-deciders was 54 to 37. In Wisconsin, where 72 percent of people voted on Election Day, it was 59 to 30. In Michigan, where 73 percent of people voted on Election Day, it was 50 to 39. And the pattern extended beyond the Midwest. In Florida, late-deciders went for Trump 55 to 38. That late surge was enough to put all these states in Trump’s column.

Normally, campaigns have a decent sense of how undecided voters are likely to break, based on their past vote history and demographics. And history shows that most people who tell pollsters they’re considering a third-party candidate will “come home” in the end. In the final days of the 2016 campaign, the voters you would expect to return to the Republican Party did so. But that didn’t happen on our side. Many Democratic-leaning voters flirting with third-party candidates ended up actually pulling the lever for them. And some undecided voters we expected to ultimately choose us went to Trump instead or stayed home.

That included suburban moderates who might have voted for Republicans in the past but didn’t like Trump and had been looking for an acceptable alternative right up until the end. On Election Day, a lot of them held their noses and voted for him anyway. It’s revealing to compare the results in the suburbs of Denver and Las Vegas, where the vast majority vote early and I did well enough to carry both Colorado and Nevada, with the results in the Philadelphia suburbs, where nearly everyone voted on Election Day. The final Franklin & Marshall Poll in Pennsylvania, based on interviews nearly all conducted before October 28, found I had a 36-point margin over Trump in the four counties of the Philadelphia suburbs, leading 64 percent to 28 percent. By Election Day, I only beat Trump there by about 13 points. That loss of suburban support in the final week meant I couldn’t match Trump’s strength in rural areas and ended up narrowly losing the state.

Working-class white women also moved en masse in the final days. Trump led among this group nearly the whole campaign, but according to the NBC–Wall Street Journal poll, I had closed to within just 4 points during the October debates. Then, in the final week, Trump’s margin grew to 24 points.