Now that this crazy election in is our rear view mirror and almost all races have been called, it’s time to look at the projections here at EP and see how well I did. Of course, with Donald Trump’s major upset, the vast majority of election prognosticators like me must admit to missing the overall winner in the presidential election. However, it’s still useful to look at each individual state-level and district-level race to make a fuller evaluation of EP’s performance.
Election Projection’s final view of the presidential elections had Hillary Clinton winning the Electoral College by a 279-259 margin. That projection turned out wrong, yes, but it was closer than renowned guru Nate Silver’s fivethirtyeight.com prediction. He had both Florida and North Carolina going to Hillary Clinton. I projected them both to go to Donald Trump. RealClearPolitics nailed Florida and North Carolina but missed Nevada, predicting Trump would win the Silver State. I projected a Clinton victory there. What I got wrong – as did almost all the polls – were the fissures Trump was able to create in the blue wall. His wins in Wisconsin, Pennsylvania and, probably, Michigan were unexpected. And they represent my three misses in this year’s presidential election.
As in the presidential election, polls overstated Democratic strength in this year’s Senate elections. Thirty-four races were on tap with control of the Senate on the line. Democrats needed – as it turns out with Trump winning the White House – to pick up 5 Republican seats while holding onto all of their own. Most prognosticators gave the blue team better than even odds of accomplishing that feat. Election Projection did not. EP projected a net gain of 3 seats for Democrats, leaving them two seats short of the majority. Yet, even my projection turned out to be too pessimistic for Republicans. In a big surprise, Ron Johnson held onto his Wisconsin seat, beating former Senator Russ Feingold again even though he was losing by double-digits in several October polls. GOP incumbent Pat Toomey in Pennsylvania held off a very strong challenge from Democrat Katie McGinty. Finally, in the only race I incorrectly projected the Republican to win, Kelly Ayotte lost to Maggie Hassan in New Hampshire by a razor-thin margin.
House race projections here at Election Projection are unique across the internet in that I actually project a margin of victory in each competitive race. This year, there were 43 entries on my competitive race list. Thirty-five were held by Republicans; just eight by Democrats. With the red team wielding such a pronounced majority in the House, it stands to reason that EP – and almost everyone else – would project a shift in the Democrats favor. And so it was, but smaller than most expected. Election Projection’s final projected tally in the House was 230-205, a net gain of 17 seats for Democrats. As of this writing, a couple races have yet to be called. Assuming the rest of the votes don’t change the current outcome, Democrats will have taken 9 Republican seats while losing three of their own, a net gain of 6 seats. Election Projection missed the call in just 8 of those races. At 427 out of 435 correct calls, EP’s accuracy in the House was 98.2%, making 2016 my best House election yet.
Seats Incorrectly Called
There weren’t many statehouses up for grabs this year. Just twelve states hold gubernatorial elections during presidential years. Of those twelve, my final projections had 5 of them switching parties. Since three were projected GOP pickups, the overall tally was projected to shift in the Republicans’ direction by one net chair. One of the projected Democratic pickups, Indiana, was my only miss among the 2016 statehouse elections. Polls, once again, overstated the Democrat’s standing in this race.
State Incorrectly Called
This was the seventh election in which Election Projection has made projections. Looking at the overall picture, it was also EP’s most accurate. I correctly called 517 out of 532 races for a 97.2% accuracy rate. Despite having missed Trump’s shocking upset, that percentage is very gratifying. And even more so because of the makeup of the misses. Of 15 races I called incorrectly, my projection favored the Democrat in 14 of them . I think that makes it next to impossible to claim that my conservative ideology produces biased projections. It does raise an important question. Was there a way to decipher Trump’s victory and GOP strength given all the data available? I’m not so sure. As we saw in 2012, claims of skewed polls turned out to be famously misguided. Who could have known that this year those claims would be true? Such is the predicament I face heading into the 2018 elections (which are but 722 days from now, by the way).