Help me keep The Crosstab ad free. Subscribe to premium content and join the community on Patreon for just $2.
Become a Patron now!

The Crosstab — Data and Politics Newsletter — May 6, 2018

Welcome, I’m G. Elliott Morris. Happy Sunday! Here’s my weekly newsletter about politics and elections that puts the news in context with public opinion polls, political science, other data (some “big,” some small) and looks briefly at the week ahead. Let’s jump right in.

Thanks to all the generous supporters who make this newsletter possible via Patreon. Patrons who pledge the five dollar tier or higher are listed here (in no particular order) as both a reward and to express my gratitude for their generosity that enables me to write this weekly newsletter:

Cameron David Ben Austin Katy Dan David Justin Joshua Ed Christian Essem Calvin Christina Guillermo Ed Sandeep Jon Justin Hunter Josh Erik Nik Krishan David Logan Bob Darcy Andrew Joseph Malcolm Tal Eric Daniel Gail Alan

Become a Patron!

Note: This week’s newsletter is a bit shorter than usual due to my rather hectic final exams schedule last/this week. I hope you’ll forgive me, and I’ve tried to make it up by releasing a good hint at what’s in store for 2018 U.S. Senate forecasting. Next week I’ll have the full, usual content!

LEAD STORY: It’s Early, but the Race for the Senate Looks Like a Tossup

Election analysts have been in a frenzy over the state of the 2018 U.S. House midterms recently. With special elections to congressional districts shifting up to 20 points toward Democrats, this is no surprise. But the same indicators of a large Democratic wave in the House also show a blue-ish picture in the Senate. Using national indicators and current polling averages/election ratings, the state of the Senate race is a ever-so-slightly-red tossup.

Using the same method I use to predict U.S. House elections, I calculated expected outcomes for Senate elections in 2018. Though I have not yet released a public model using this method (stay tuned) what I can say now is that things are… close. Rather close indeed. The process used to make the predictions in the table below is as follows:

  1. Combine a state’s partisan lean with whether or not an incumbent is running in the district
  2. Add in the current swing in the national environment implied by the generic congressional ballot and presidential approval.
  3. Average each baseline prediction with the polling average in the seat, giving more weight to the polling average. If there are no polls to average, I use the average of margins implied by ratings from Cook Political Report, Inside Elections, and Sabato’s Crystal Ball. I nudge these measurements to the left by 2% to match projected movement in the generic ballot from now until election day.

Obviously, things will changes by the fall, but this aids in gaining understanding of how these races all play together in the fight for control of the chamber.

State Incumbent Polling/Rating (%) Prediction
FL Dem. D+4 D+7
NV Rep. R+1 D+1
TX Rep. R+4 R+6
MO Dem. D+2 D+1
IN Dem. D+12 D+8
MT Dem. D+15 D+10
TN OPEN D+10 D+2
ND Dem. D+3 R+2
WV Dem. D+1 R+6

If these predictions panned out in November, the Democrats and Republicans would have an even 50-50 chamber, with Vice President Mike Pence serving as GOP tiebreaker.

There are a few things going on here. First, you can clearly see the effects of the incumbency advantage (read more here) on the Senate elections in North Dakota and West Virginia. Although Donald Trump won the latter by a massive 43 point margin in November, we should expect Joe Manchin (Dem.) to lose reelection by just 22 points there in the fall based off incumbency and the national environment alone. In the West Virginia prediction you can also see a lot of the heavy lifting being done by current polling in the state. Although the baseline prediction is R+22, the average of ratings (D+1) lifts that expectation up to R+6.

This creates for several moving parts in the forecasting process. First, if one Senator has a larger-than-average (8% on the margin) bonus from their incumbency advantage, they will do better/worse than the quick prediction above shows. This is likely the case in Montana, where current Senator Tester (D) has a “Likely Democratic” rating to win reelection.

Then, there’s the generic ballot and presidential approval. If either measure goes up or down in aggregate, then the predictions for each state change. And though the effects of movement in both could cancel out — the generic ballot has shown Democrats “gaining” ground recently while Trump has improved ever-so-slightly in his approval ratings — the generic ballot has more weight in the equation.

Finally, there is state-level polling. Polls of the Senate races at the state level are a big deal; good news for Democrats in Tennessee, Arizona, and Texas is solely responsible for the tossup nature of the national Senate contest. If Democrat Phil Bredesen in Tennessee, for example, was doing just half as good as he is in current polling, the probabilities might look a little bit different.

So there you have it: a quick look at what the data say in the race for control of the Senate in 2018. Though this is a useful guide, keep in mind that factors can (will) change in the (near-term) future that will change our assessment of the individual contests themselves. With that being true, it’s best to evaluate all of these predictions with a healthy dose of uncertainty. As a general rule, nothing within a single-digit range of outcomes should be treated as anything above a Lean contest for the favored candidate, at best, and a Tossup, at worst.

More on Senate prediction, soon — if you’re a fan of the current U.S House forecasting model, I think you’re going to like what’s in store.

Echelon Insights: The Secrets of Voter Turnout

The team at Echelon has a must-read report on voter turnout that is key to understanding what dynamics are at play in the 2018 elections.

Nicholas A. Valentino and Kirill Zhirkov : Blue is Black and Red is White? Affective Polarization and the Racialized Schemas of U.S. Party Coalitions

The new study from Valentino and Zhirkov finds that racial resentment partisanship is causaing affective polarization in the American public. Michigan State’s Matt Grossman summed it up on Twitter:

Racial resentment increasingly tied to more positive affect toward your party & negative affect toward other, in a way that other attitudes are not (they also show that its influence is greater for those who implicitly tie Dems to blacks & Reps to whites)

Here’s the key graphic from the paper:

David Byler (TWS): What’s the GOP’s Magic Number in the House?

That’s it from me this week! I’ll be back next Sunday to recap the week in data and politics. In the meantime, follow along with me on twitter and read my news posts on my blog.

G. Elliott Morris avatar
About G. Elliott Morris
Elliott is a ("big") data journalist and data scientist who specializes in American politics, public opinion, and predictive analytics. He is a government, history, and computer science student at The University of Texas at Austin and worked previously for the Pew Research Center and Decision Desk HQ.
comments powered by Disqus