Princeton Election Consortium

A first draft of electoral history. Since 2004

Meta-Margins for control: House D+1.0% Senate R+4.2% Find key elections near you!

Two Futures: 45 or 50 Democratic Senate seats…but not in between?

November 4th, 2018, 4:00pm by Sam Wang


Four years ago, I pointed out that close Senate races all tend to fall in the direction of one party or the other. Since then, the idea has stood up pretty well. It implies two very different possible futures. There are a few races I will be watching on Tuesday to figure out which is ours.

The overall pattern above is consistent with two ideas: (1) Polling misses a bit of enthusiasm for the winning side in Presidential election years. (2) Polls also don’t catch energy that goes against the President’s party in midterm years.

The chart shows the range of possibilities in races where the polls showed the race within 3 percentage points. 2014 fits the pattern, 2016 less so.

In 2014, the President’s party (Democrats) had a midterm election go hard against them beyond what polls indicated. That was the lowest-turnout election since 1942. This could help account for the systematic polling error, which I estimated as being more than five points.

In 2016, the outcome favored Republicans, but less so. Compared with final polls, the unexpected winners were Republicans in Pennsylvania, Wisconsin, and Indiana. Democrats still won in Nevada and New Hampshire.

An overall error as small as two percentage points would lead to fairly different outcomes, at either end of the range:

Future #1: Democrats get to 50 seats. This future corresponds to the idea that there is hidden support for the out-party, i.e. the Democrats. That would lead to 50 Democratic/Independent seats.

Arguing in favor of this idea is Trump’s high unpopularity. Arguing against this idea are the reports of high turnout, which is a situation that pollsters ought to be able to deal with. If there is a polling error, I would guess it will not be nearly as large as 2014. But a two-point error is well within the range of possibilities. Which gets Democrats to 50 seats. This possibility is more likely than people realize, I think.

What happens in a 50-50 divided Senate? Last time this happened, in 2001, there was power sharing. But those were very different times, and the Senate is far more polarized now. Think of the possibilities: Pence stuck in the Senate all the time to break ties, or Democrats attempt to recruit Murkowski to their side. That would be quite a show.

In addition to this 50-seat scenario, as an outside possibility I will be watching the Tennessee race between Phil Bredesen (D) and Marsha Blackburn (R). Tennessee is a low-turnout state even in Presidential years. This year, early voting shows increased turnout among Tennessee voters under the age of 50. So there’s some chance of a hidden bonus there. Nonetheless, Blackburn leads in polls by a median of 5 points, which is a fair amount. If Bredesen pulls off a win, it would boost the odds substantially that Democrats can get as high as 51 seats. It’s a long shot.

That’s a contrast to Texas, where early voting has passed the 2012 Presidential numbers and might get to 2016 levels. Does this mean there’s a hidden bonus for one side or the other, or does it mean polling will be more accurate? I think polls are accurate and Cruz will win, but we will find out.

Future #2: Democrats end up with 45-46 seats. Conversely, the Arizona Senate race has shown a slim lead of about 3 points for Kyrsten Sinema (D) over Martha McSally (R). If McSally pulls out a win, that would suggest the polling error favors Republicans, and a bad overall outcome for Democrats in the Senate.

I guess there is also Future #1.5, a median-ish outcome of 47-49 D’s. That’s not nearly as good as Republicans had hoped for a few years ago. But it’s also not a dream outcome for Democrats.

Tags: 2004 Election · 2006 Elections · 2008 Election · 2012 Election · 2014 Election · 2016 Election · 2018 Election · Senate

6 Comments so far ↓

  • Marco

    My expecations are probably wishful thinking but I see, based on polls, AZ going Blue, ND going red, with IN, FL squeaking to the Blue side.

    MO is on the razor’s edge and TN is the key. If, for some miracle, enthusiasm and Taylor Swift carry TN, we are going to have a Blue Senate.

    But I rather buy a piece of swamp land in FL than count on this outcome. I’d be ecstatic with a Blue House, and that is a real possibility. Crossing fingers and toes…

  • TC

    Given the polling, the 2016 result of Trump’s election was not a shock, though some were surprised. Similarly, given the 2018 polling , Democratic control of the Senate would not be a shock. Whether it will happen, who knows but remains a quite real possibility. That’s true of a fairly wide range of outcomes however, as with Trump’s election, given the polling.

  • Marco

    The $ 64,000 question is how pollsters are modeling the actual voters to asses their sample. They did not do too well in 2016 (hence the averaging of polls to get a better idea). Did they improve this year? From an article I read this morning:

    “the pollster, emphasized the race is truly up in the air on Tuesday night.

    Voter registration this year is double what it was in 2014, and the number of people who have already voted absentee is nearing double the number that year, according to the registrar’s office.

    “They’re not getting picked up in our samples because we’re drawing samples based on past voting behavior,” he said.

    We shall see Wednesday morning…

  • TC

    Yes, to assessing the samples. There are many levels to that: assessing people’s voting intentions and abilities correctly, and then making sure you don’t misrepresent or misunderstand your own assessments, and those of others.

    As regards in which places and directions are there most likely to be polling misses or systematic misses: on the one hand, who knows? on the other hand more signs seem to point in a certain direction than in another.

    No one should be surprised if Democrats win a house seat in Oklahoma and win the governorship there, though the polls point to Republicans. I haven’t seen much of anything on these possibilities, though I haven’t been actively looking. Residents of Oklahoma, the heart of the historical progressive populist movement, have greatly improved voter turnout, and this was one of a handful of states last year where teachers either struck or threatened to strike to good effect: Oklahoma, West Virginia, Arizona, Colorado, Illinois, Pennsylvania, New Jersey. Richard Ojeda has been much cited as having a chance to win a House seat in West Virginia, and if the teachers can’t bring him through it will not be for lack of trying. A similar situation, though less noted in Oklahoma: Kendra Horn (House race) and Drew Edmondson (Governor race). The polls could be spot on in giving the edge to the Republicans in these three races, but no one should be surprised if the votes go to the Democrats in not only these races but in many races that the polls put as much closer.

    So many things don’t get asked or answered. Does the increased quantity of the youth vote this year not only make its own direct impact (presumably Democractic) but does it also affect the quality of the more senior votes? And if so, which way? This new election cycle, a lot of such questions could be asked and answered. Most aren’t, which is not necessarily a bad thing. My hunch is that if they were asked and answered, then they would point in certain direction more than polling has shown, though to what magnitude of effect is harder to figure.

  • Nettoyeur

    I vote in Tennessee (though at the moment I am working outside the US). The GOP candidate for Senator, Marsha Blackburn with few if any legislative achievements to her name, is currently and has been for years full-on Trumpkin and Birther. Diane Black took a similar tack in the GOP gubernatorial primary this year, with the claim that pornography was the root cause of school shootings —and she lost to the pragmatic farmer/businessman/engineer Bill Lee. The Democratic Senatorial candidate, Phil Bredesen, is a pragmatic and popular former governor, not unlike Bill Lee or the present governor, Bill Haslam. Blackburn is ahead in the polls, but the outcome of this contest could be a surprise, especially if turnout is strong.

Leave a Comment