Edge in Polls May Not Tip House Seats

Democrats appear poised to win the House popular vote Tuesday by a wide margin, with national polls showing sustained disapproval of President Donald Trump — and yet the fate of the chamber is not a foregone conclusion.

On the day before the midterm elections, two vastly different outcomes remain easy to imagine. There could be a Democratic blowout that decisively ends Republicans’ control of the House and even endangers their Senate majority. Or there could be a district-by-district battle for House control that lasts late on election night and perhaps for weeks after.

The first would be interpreted as a repudiation of Trump, the second as another example of his political resilience. But the difference turns on just a few percentage points across dozens of House districts that remain exceptionally close, according to New York Times Upshot/Siena College surveys conducted the past few weeks.

After more than 10,000 interviews, the result, in the aggregate, is that Democrats and Republicans are essentially tied in the 30 districts rated as tossups by the Cook Political Report, with Democrats leading by about half a percentage point.

Democrats need to win only a handful of these tossup districts — perhaps as few as six — to gain the net 23 seats needed to take control, which is why they are considered favorites. But Democrats haven’t put them away. Instead, those races remain startlingly close. Each of the final 28 poll results in the tossup districts was within the margin of error, and 20 of the 28 were within 2 percentage points, a margin that pales in comparison with the typical measurement error in a poll.

With so many close contests, even modest late shifts among undecided voters or a slightly unexpected turnout could yield significantly different results, with very different consequences for the government and the future of the Trump presidency.

Overall, the polls comport with the growing consensus among operatives from both parties that Democrats are poised to gain about 35 seats in the House. If the Times/Siena polls were exactly right (they will not be), Democrats would gain 32 seats, assuming the two parties held the seats that were not polled.

All of the conditions remain in place for a wave election, like those that last flipped control of the House in 1994, 2006 and 2010. Democrats hold a commanding lead on the generic congressional ballot (which asks voters whether they intend to vote for Democrats or Republicans for Congress), including an 8-point lead in an ABC/Washington Post poll Sunday.

But the Republicans have considerable structural advantages in the House that the president’s party did not have to the same extent in previous wave elections. They are generally defending districts that voted for the president, a result of partisan gerrymandering and the tendency for Democrats to post lopsided and inefficient victories in urban areas.

The Republican geographic advantage is even more significant in the Senate, where Republicans are all but assured to retain control if they win just three of seven competitive seats where Trump won by at least 9 points in 2016.

The Democrats are poised to defy those structural disadvantages in the House because they have put so many Republican-leaning districts into play with a deep and exceptionally well-funded class of candidates. But Republicans can hope that partisan polarization is just enough to keep even strong Democrats from going over the top. The president has emphasized immigration and other hot-button issues down the final stretch, perhaps in an intentional effort to divide the electorate along the lines of the 2016 election.

If Republicans succeed in polarizing the electorate, they could take advantage of their underlying geographic advantages and hold down their losses in the House and gain seats in the Senate.

Alone, a more polarized electorate wouldn’t be enough for Republicans to hold the House. There are too many districts that voted for Hillary Clinton. To retain House control, Republicans would probably need some good luck on top of a more polarized map, or perhaps a broader 2016-like polling error. But a narrow Democratic majority might take weeks to become clear as California and Washington count late mail ballots.

The emphasis on a wide range of possible outcomes isn’t just a matter of hedging by pollsters and analysts after the shock of 2016, when Trump beat long odds. The uncertainty reflects the sparse data available and the unusually large number of competitive contests. In dozens of key districts, no more than a few polls have been done, making it hard to be confident whether (and where) Democratic or Republican candidates might hold a lead.

Final House polls have historically been less accurate than polls of statewide contests and presidential races. On average, House polls differ from final election results by about 8 percentage points. These polls tend to have relatively small sample sizes, increasing the margin of error, and voters tend to be less familiar with House candidates, meaning more “undecideds” until late in the race.

Democrats hope that these common sources of polling error might break their way this time.

Turnout is always uncertain, but it is more variable in lower-turnout elections, like a midterm, when even modest shifts in enthusiasm can transform the electorate. It is a particularly challenging question this year, in part because the turnout in recent midterm elections has been so low and so Republican.

This year’s early vote tallies already make it clear that the turnout will greatly exceed that of four years ago, but it is far less clear how that will translate to actual votes. It is equally unclear whether pollsters have been assuming the high-turnout, more Democratic electorate implied by early voting and the turnout in the special and general elections since Trump became president. During the past few weeks, the share of voters indicating that they are “almost certain” to vote has increased 10 points in Times/Siena polls, from 66 percent to 76 percent of the likely electorate.

In a few cases, the turnout in early voting has shown some polls were off in their projection of who will vote. Early voting in several Texas counties has already surpassed the projected electorate of a Times/Siena poll from early October, and Democratic Senate candidate Beto O’Rourke seems likely to benefit from the higher turnout.

A Monmouth poll of California’s 48th District in coastal Orange County showed the Republican incumbent, Rep. Dana Rohrabacher, leading by 2 points with an electorate in which registered Republicans outnumbered Democrats by 18 points. The actual Republican advantage in early voting is down to 12 points and continues to drop as turnout increases.

The final Times/Siena poll of California’s 48th shows Democrat Harley Rouda leading Rohrabacher narrowly with an electorate where registered Republicans outnumber Democrats by 10 points, an indicator of the extent that Republicans are struggling in the wealthy suburban bastions of 20th century conservatism.

Democrats can also hope that undecided voters will break their way. It is often supposed that voters will break against incumbents, and Republicans will be in trouble if there’s merit to that assumption. No Republican incumbent has eclipsed 50 percent in any of our polls of the tossup districts.

Undecided voters are also often thought to break toward the party out of power, and perhaps that’s especially likely with stupendously well-funded Democrats spending millions to increase their name recognition in the final stretch. It’s paying off for Democrats in the Times/Siena polls: On average, Democratic challengers are now known by 82 percent of voters in polls over the past 10 days, up from 60 percent in September.

Astonishingly, it’s enough for the Democrats to effectively deny the Republicans the advantage of incumbency that usually helps a party hold the House. In the battleground polls finished during the past 10 days, a larger share of the electorate now has a favorable view of the Democratic challengers than the Republican incumbents, with 48 percent of likely voters holding a favorable view of the Democratic candidate compared with just 46 percent who have a favorable view of the Republican incumbent.

All of this may be adding up to a late shift toward Democrats. The Times reported that both Democratic and Republican operatives see House polls as trending Democratic in the final days, and the last wave of Times/Siena polls are at least consistent with that possibility.

But late movement doesn’t necessarily predict the final result. Just ask Clinton, who gained slightly in the final polls taken in the weekend before the 2016 election.