During the afternoon of the presidential election, I exchanged emails with a friend in the White House to try to assess how things looked on their end. I heard back that they were cautiously optimistic about the presidential race but increasingly concerned about the prospects for winning the Senate. I told my friend that things looked good in Pennsylvania from everything I could observe. Turnout was very high in my suburban Philadelphia precinct, and reporting from friends in the city indicated excellent voter participation. There was no sign in the southeastern part of the state that Clinton was about to suffer a 10 percent or greater drop-off of Obama’s 2012 support in 23 counties, and a 10 percent or greater drop-off on Obama’s 2008 support in 45 counties.
The Clintons got the news first that something was desperately wrong from an operative in Florida.
Around 7:45 on election night, when Hillary Clinton and her aides still thought they were headed to the White House, troubling news emerged from Florida. Steve Schale, the best vote-counter the Democrats had in the state, told campaign officials they were going to lose the biggest battleground in the country. Yes, Clinton was doing well in some places, but Donald Trump’s numbers in Republican areas were inconceivably big.
“You’re going to come up short,” Schale said, stunning aides in Brooklyn who were, until that moment, comfortably cradled in the security of their own faulty analytics.
People talk a lot about “the faulty analytics,” but it’s important to realize that Clinton met reasonable targets in her areas of strength. Compared to Obama’s 2012 performance, she netted about 400 more votes out of Pittsburgh and Philadelphia combined, and in the Philly suburbs, she netted 5,796 more votes out of Delaware County, 26,097 more votes out of Chester County, and 34,376 more votes out of Montgomery County than Obama had in 2012. This compensated for her modest 1,243 net underperformance in the more working class Bucks County suburb.
Compared with Obama’s landslide 2008 election, she impressed by netting 6,331 more votes than he did in Pittsburgh’s Allegheny County, which more than offset the 3,482 fewer votes she netted in Philadelphia County.
When we talk about “faulty analytics,” therefore, it’s important that we understand, regardless of turnout, Clinton was netting votes at a clip that should have assured victory. I don’t know that they thought they were going to do much better than they did in blue areas and I don’t think that’s where there was a failure of modeling or analysis.
The failure seems to have been in not realizing that Trump was going to win red areas by “inconceivably big” margins.
I keep coming back to this because it doesn’t seem to be truly appreciated.