The consensus view regarding the likely outcome for the midterms is that Democrats regain control of the House, and Republicans add seats in the Senate. That strikes me as largely correct, though I leave allowance for late-breaking developments—like the debate over birthright citizenship—to shape the precise contours of the results. But an under-explored trend bears looking into: the consistent underperformance of Republican candidates in middle- and working-class America over the last month.
This tendency is distinct from the one drawing the most attention—the possibility that the furor over the Brett Kavanaugh hearings might have energized previously dormant Republican voters. By definition, the Kavanaugh factor would affect only turnout, not attitudes about the Republican Party or President Trump. If it’s real, we would see differences in the composition of the likely voter pool in polls across the country, as more Republicans line up to cast ballots. There is some evidence that it’s happening, but it is far from clear how much it might contribute to GOP hopes.
Instead, both national- and district-level data show that Trump and the GOP are slightly better thought of than they were a month ago. This means that districts and seats that looked to be slightly in trouble for the GOP in late September look more favorable today, though the overall national picture is little-changed. But this movement has not yet translated to higher support for the Republican candidates in targeted races.
The New York Times-Siena polls show this quite clearly. As of November 1, the Times had completed or almost finished two polls, one in September and one in October, in 23 competitive House seats. Trump’s job approval had stayed steady or increased in 20 of them, and the Republican share on the generic House ballot had improved in 18. Regardless of how the individual Republicans were faring, the ground underneath them had shifted slightly in their favor over the ensuing month.
The polls also show that Republican chances rise or fall depending on these national trends. Fifteen of these GOP candidates were running behind the generic ballot score, and another six were running even with or only one point ahead of it. Only two, Pennsylvania’s Brian Fitzpatrick and Texas’s Will Hurd, were running two or more points ahead of their party. This means that millions of dollars are being spent to promote individual candidates, while the vast majority of voters are looking at their party.
This symmetry extends through every category of seat. All five Republicans running in seats where a majority of voters want a Republican House hold leads, and six of eight lead in districts where between 47 and 49 percent want a GOP House, and the GOP leads the Democrats. On the flip side, nine of the ten Republicans running in seats where Republicans run behind the Democrats and are getting 46 percent or less of the vote are behind themselves. Most Republicans with narrow ballot leads have room to grow among undecided voters who like Trump and the GOP, giving them a better chance of prevailing than looking simply at their own numbers would suggest.
The same pattern holds for 15 other House seats that the Times polled for the first time in October. Republicans trail in four races on the generic ballot, and all four GOP candidates are behind or tied. In another five races, the Republican is tied or only narrowly ahead, but the GOP is ahead and at or close to a majority of the vote on the generic ballot. Democrats may look to be in good position based only on the congressional head-to-head, but to win, they need to convince people who want a Republican-controlled Congress to vote for a Democrat. Only Washington’s Republican congresswoman Jaime Herrera Beutler is running significantly ahead of her party; everyone else has room to grow.
The same pattern holds true in public Senate polls. I looked at every poll completed between October 20 and 31 in six key races—Arizona, Florida, Indiana, Missouri, Montana, and Nevada. The Republican nominee runs behind Trump’s job-approval rating in nearly every poll, regardless of the state, and runs even with him in all other polls. The polls vary in showing the Republican or the Democrat ahead, but all reveal that quite a few Trump supporters are not yet sold on the local Republican candidate.
One can’t be sure how these voters will decide. But in 2016, undecided voters who liked Republicans broke to Trump, even if they had grave reservations about whether he could do the job. These “reluctant-Trump” voters decided the election, and it was the analysts’ failure to anticipate this late break—rather than poll failure itself—that confounded the public.
I’m not yet sure where I’ll come down on the precise outcome of the race, but I do know this: some Trump backers, perhaps those who are not loyal, rock-red Republicans, still have time to make up their minds. How they vote will likely determine whether November 6 results in a blue wave or another surprising election night.
Henry Olsen is a senior fellow at the Ethics and Public Policy Center and an editor at UnHerd.com. He is the author of The Working Class Republican: Ronald Reagan and the Return of Blue-Collar Conservatism.