On the day before Election Day, I created the first polling-based Campus Citizen forecast for the U.S. Senate elections. How’d I do and how’d I do it? Should preeminent electoral forecaster Nate Silver of FiveThirtyEight fear for his job? No, but if current returns hold we correctly predicted all but 2 of the 34 races that held their general election Tuesday. FiveThirtyEight called all but 3 of the 34 races correctly. This is an overview of all of the elections that our forecast identified as the most competitive, with the ratings of either Tilt Democratic or Tilt Republican.
Arizona – Correct (so far)
I forecasted this race as Tilt Democratic with a 0.1 percent lead for Rep. Kyrsten Sinema (D-Ariz.) over Rep. Martha McSally (R-Ariz.). This race has yet to be called, but Sinema is currently ahead in the returns by the kind of narrow margin our forecast predicted. If Sinema wins, this race would be a pickup for the Democrats as it was formerly held by retiring Sen. Jeff Flake (R-Ariz.).
Florida – Incorrect (so far)
I wrote on Monday, “Nelson has recently pulled ahead in the polls, but his re-election is far from his certain in a state infamous for its close elections.” Florida is still too close to call, with a legal battle brewing over recounts in the state. I rated this as tilt Democratic, but Gov. Rick Scott (R-Fla.) is ahead in the results so far.
Indiana – Incorrect
Ironically, one of my missed calls was my home state of Indiana, which I ranked as Tilt Democratic with Donnelly boasting a narrow 1.4 percent lead in the polls. Republican candidate Mike Braun won in a slight upset victory by 7 percent, a margin no poll accurately called. One poll from Fox News, conducted from Oct. 28 to Oct. 30, showed a Donnelly lead of 7 points and was thus off by 14 points in the margin if current returns hold. Porter County’s results are disputed with the Porter County Board of Commissions requesting the FBI to investigate alleged violations of electoral law. The amount of votes in Porter County would not flip the race to Donnelly, but the final margin is still unknown.
Missouri – Correct
I forecasted Josh Hawley (R-Mo.), the sitting attorney general for the state, would prevail over incumbent Sen. Claire McCaskill (D-Mo.) to win this race. Hawley was ahead 1 percent in the polling average which was weighted only by sample size, recency, and methodological quality. FiveThirtyEight incorrectly predicted that McCaskill would win, in part because of their aggressive polling adjustments in the margin which I voiced concerns about on Oct. 30. However, Hawley won by 6 percent over McCaskill, a much larger margin than predicted in this forecast.
Nevada – Correct
Rep. Jacky Rosen (D-Nev.) defeated incumbent Sen. Dean Heller (R-Nev.), as predicted by the Campus Citizen forecast. Rosen won by 5 percent, a greater margin than her less than 1 percent polling lead in the weighted average. A similar polling error happened during the 2016 presidential election, where Hillary Clinton won by 3 percent more than what a simple average of the polls in Nevada suggested was a narrow Trump lead. Rosen first entered politics when she won a U.S. House seat in 2016, previously working as a computer programmer.
Texas – Correct
Sen. Ted Cruz (R-Texas) narrowly won the election over Rep. Beto O’Rourke (D-Texas) as the Campus Citizen predicted. O’Rourke performed the best of any Democrat in Texas since 1990, when former Gov. Ann Richards won the Texas gubernatorial election. O’Rourke also received more votes than Democratic presidential nominee Hillary Clinton during the 2016 presidential election, notable for the fact that far fewer voters tend to turnout for a midterm election. Exit polls also suggested that O’Rourke is positively seen by Texans even though he lost the election. 52 percent of respondents say they had a favorable opinion of him, while 42 percent said they had an unfavorable opinion. This was a higher overall and net favorable rating than the victor of the election, Sen. Ted Cruz.
West Virginia – Correct
Sen. Joe Manchin (D-W. Va.) won his re-election bid over Republican nominee Patrick Morrisey (R-W. Va.), the attorney general for the state as predicted. Manchin won by 3.2 percent, slightly less than his weighted polling average lead of 3.8 percent.
My Senate forecast for the Campus Citizen used the same principles as my Congress Compass forecast, which accurately predicted 96 percent of the House races if current results hold. 366 polls were weighted by recency, sample size, and methodological quality fore the forecast, but the margins of the polls were never adjusted or skewed. And it’s for this reason that I was able to correctly nearly every Senate race, with the exception of Florida (so far) and, ironically, Indiana.