Voting booths set up inside a polling station in Christmas, Florida on November 8, 2016.
Poll: Dems have advantage with likely voters
02:13 - Source: CNN
CNN  — 

First things first: The theme song of the week is Roundball Rock by John Tesh from the NBA on NBC.

Poll of the week: The New York Times and Siena College have released multiple House polls over the last week. One of the more prominent results was in Wisconsin’s 1st Congressional District, where House Speaker Paul Ryan is retiring.

It found Republican Bryan Steil leading Democrat Randy Bryce by 50% to 44%. That lines up well with where most organizations rate the race (lean Republican).

What’s the point: The polls produced by Siena and the Times generally point to a House where Democrats have a slight advantage, though Republicans could still retain control. Some of the polls, like the one in Wisconsin’s 1st District, have the Republican ahead, while others, like in Minnesota’s 3rd District, have the Democrat up.

As Real Clear Politics elections analyst Sean Trende noted on Twitter in regard to the polls, they point to a Democratic net gain of 30 seats. This estimate is largely in line with the FiveThirtyEight polls-only (or “lite”) estimate of a gain of 34 seats for the Democrats. Both are somewhat more than the net gain of 23 seats they need to take control of the House.

These district polls, however, tend to paint a better picture for Republicans than other indicators such as the national generic ballot polls and the fundamentals such as fundraising. Those suggest a gain of over 35 seats for Democrats.

In other words, there is disagreement.

What’s important to keep in mind is that district polls at this point in the cycle can underestimate the party benefiting from a wave election.

I went back since the 2006 election and looked at how much the polls from roughly within a month of this point in the cycle performed. (That is, polls completed from about 52 to 82 days before the election.)

The immediate thing that jumps out is the side that has won the national House popular vote has always done better on Election Day than the polls indicate right now. The average overperformance was a little over 3 points.

Democrats did better than their polls in 2006, 2008 and 2012. Republicans did better than their polls indicated in 2010, 2014 and 2016.

The district poll overperformance seems especially large in wave years. Back in 2006, the average district poll had the Democrats trailing by 1.5 points. The result in those districts polled ended up being Democrats winning by 4 points. That’s a bias of 5.5 points against the Democrats.

Jump ahead to 2010. When Republicans were ones riding the wave, they were down by 3 points in the average district poll right now. They ended up winning in those districts by 1 point. That’s a 4-point overperformance for the Republicans.

The 2014 midterm election tells the same story. Republicans did 5.5 points better than the district polls suggested they would at this point.

Now it would easy to believe that the reason the district polls aren’t predictive at this point is because polls themselves are inaccurate. That doesn’t quite seem to be the case.

Look at these same years, but only at the polls within the final 10 days of the elections. The polls have less than a point bias in 2006 and 2010. In 2014, the district polls – like all the national and state polling – underestimated the Republicans (though the underestimation was 2 points fewer than the polls two months out).

Put another way, the district polls were pretty good in the end, but can be a lagging indicator of what other measurements are projecting two months out from the election.

It’s of course no guarantee that the district polls at this point will be a lagging indicator this year. But if there were a party that probably has more upside from the district polls, it’s the Democrats, given that other factors are more favorable for them.