Elections Update: Return of Special Elections

Welcome to our update on the elections on Sunday, October 28th! At 10:30, the classic version of the FiveThirtyEight House forecast gave Democrats one in six (85%) chance of taking control. The Lite version gave them one in 7 (77%) and Deluxe, one in five (82%). These correspond to the figures we have seen all week.

Going down the list of seats that may switch, you may notice familiar names. Just in time for Halloween, several districts that have held special elections in the past 18 months have come back from the dead. And the similarities are strange: based on the classic version of our House forecasts, the expected margins of voting in these districts announce very close to the margins of victory of the special elections. In two of the districts, even the same candidates are in the running.

This week, we received new polls in three of these contests, confirming that they were again very present.

• If you are like me, you probably never wanted to hear the words "Georgia 6th" after the 2017 special elections that broke the record for the most expensive congressional elections in history. But we have two new polling stations this week. Republican pollster JMC Analytics teamed up with Democratic firm Bold Blue Campaigns to probe the district from Oct. 13-18; Republican Karen Handel (winner in 2017) led Democrat Lucy McBath between 49% and 45%. A few days later, Thirty-Ninth Street Strategies, a poll conducted on behalf of the McBath campaign, said Handel was leading with only 1 point. (Thirty-ninth Street did not have any primitive voters in its sample.) Once adjusted for internal effects, including the fact that this last ballot was an internal democrat, both Republican Lead 5 points. It's pretty consistent with our overall race ranking as a "lean Republican".
• We had three polls this week in the Montana congressional race (which outgoing Republican Greg Gianforte won for the first time in May 2017, even though a journalist was slammed the day before the vote), but they did not both seemed. The first, a poll of Montana State University-Bozeman, revealed that Gianforte was leading the former Democratic State representative, Kathleen Williams, between 48% and 40%. But he only polled the registered voters and unlikely. The survey was conducted over three weeks in late September and early October (usually a survey is in the field for a few days). It was not the last data. A competing MSU survey – the Billings campus – gave Gianforte a 3-point lead among potential voters from October 8-13. Finally, the University of Montana had the best news for Williams, giving it 45.8 percent to 45.3 percent lead. But the results of this poll were favorable to the Democrats, and the new poll actually represented a sharp decrease from the 13-point lead of Williams in the latest UM survey. As a result, none of the three polling stations managed to convince our model that it was not another "Republican lean" race.
• Finally, the district that will organize the last special election is also the closest. The 12th district of Ohio confronts once again Republican Troy Balderson (outgoing president) to Democrat Danny O'Connor (Green Party candidate Joe Manchik, playing the role of potential spoiler). End Citizens United, a pro-democracy group, released a survey, led by Clarity Campaign Labs, which had Balderson at 48 percent and O'Connor at 46 percent. A few days later, the O'Connor campaign dropped a survey by GBA Strategies that said the race was tied at 47%. However, as both polls were partisan, they probably overestimate O'Connor's position and Balderson improved in our forecasts earlier this week. However, it's still only a precarious 5 out of 8 chances to win.

What can we learn from the fact that these districts (plus 8th Arizona, which was not polled this week but held special elections earlier this year) are once again competitive? Obviously, it's a good sign for the Democrats. As you may recall, the 2017-18 special elections were held in heavily red districts and states that are generally not closed, but the Democrats still made them competitive; The Democrats surpassed by 16 points on average the proportion of supporters of the Five Thirtieth out of eight. If Democrats can match that in November, they are very likely to win the house.

But it's not automatic: there are some districts we did not mention that held special elections last year, like the 4th in Kansas and the 5th in South Carolina. According to our model, although Democrats are barely gaining two points in 2017, they are both safe for the GOP. And of course, even if the Democrats match their performance in special elections in the upper districts, it does not matter – this time, they really have to win. Fortunately for them, there are enough less red districts to win the House anyway. But a real gangbus performance for the party probably involves the selection of some of these "lean Republican" seats.