On Sunday, CBS News released new polling (conducted by YouGov) in Florida, Arizona, and Texas. THE WEEKLY STANDARD’s SwingSeat Senate Forecast Mode has eaten that data and produced new forecasts for each state’s senate race. Here’s what the model thinks:
Florida is a true toss-up
SwingSeat puts Rick Scott’s chances of toppling incumbent Democratic Sen. Bill Nelson at 48 percent. That’s about as close as a race can get. The YouGov survey put Scott ahead by five points, and recent polls by the Florida Chamber of Commerce and Florida Atlantic University also have Scott ahead. But when we’re this far out from the Election Day, the model takes a long view of polling. Many voters haven’t tuned into the 2018 election and polling is often more sparse this early in the cycle. So the model is also thinking about earlier polls that had Nelson ahead. Taken together, these polls suggest an extremely tight race.
The other interesting detail in Florida is how the fundamentals contrast with the polling. The fundamentals—which is election nerd shorthand for everything that isn’t a head-to-head poll—strongly favor Nelson. He’s a sitting senator (they tend to do better than non-incumbents) in a purple state (other purple state Democrats like Bob Casey Jr. are polling well) in a midterm where Democrats could win in a wave.
The eventual result may converge on the fundamentals (I’m still experimenting with different ways to blend fundamentals and polling). Or maybe Scott is a good enough candidate that this race will stay competitive through Election Day. Either way it’s best to think of this race as a toss-up—which is what the Cook Political Report, Sabato’s Crystal Ball, Inside Elections and RealClearPolitics say.
Ted Cruz looks solid in Texas
The YouGov poll also put Cruz ahead by 10 points in Texas. That poll, along with the others we’ve seen, translate to an 86 percent win probability for Cruz.
An 86 percent probability isn’t a certainty (at various points in the 2016 election, Hillary Clinton had a similar win probability), but it’s a real advantage. It’s better than the chance of flipping a coin twice and not getting heads twice in a row. And a result like this isn’t so far off what we would expect from an incumbent Republican senator in a state as red as Texas this year.
Bad polling for Republicans in Arizona
In Arizona, YouGov tested Democratic Rep. Krysten Sinema against her possible Republican opponents—Republican Rep. Martha McSally, Kelli Ward, and former Maricopa County Sheriff Joe Arpaio. Sinema beat all three of them, with McSally and Ward both outperforming Arpaio.
The model believes that Sinema has a roughly 80 percent chance of winning this race, even when it assumes that McSally (who has been, on average, outpolling her primary opponents) wins the nomination. That isn’t quite where my gut is on this race. I think McSally has a better than 20 percent chance of taking the seat. But that’s OK—statistically powered forecasting is just one of multiple methods of thinking about these races, and analysts are allowed to question their models.
The overall estimates are basically unchanged
This new round of polling didn’t change the GOP’s overall chances of holding the Senate much. They’re currently at 70 percent, which is right about where they were when the model debuted last week.
Put simply, the model already knew that Cruz was ahead in Texas, that Florida was close, and that Sinema had been polling well in Arizona. If we got a new poll in Montana or Indiana, the overall estimates might change more. But for the most part, Sunday’s polls reinforce what we already knew about the state of the race.
Other data points from last week
The model also ate some polls from North Dakota, Wisconsin, West Virginia, Utah and elsewhere last week.
North Dakota is the only poll of that bunch that really moved the needle. We don’t have a ton of data from the Peace Garden state, so the Mason-Dixon poll showing Republican Kevin Cramer up by four points gets a lot of weight. The model currently gives Cramer a roughly 2-to-1 advantage there, but I expect some combination of changes to the model or subsequent polling to move that estimate closer to 50-50.
In Wisconsin, a new Marquette poll shows Democratic Sen. Tammy Baldwin ahead of Republican Kevin Nicholson by double digits. The model estimates that Baldwin has a roughly 90 percent win probability. The model assumes that Nicholson will win the GOP primary (he’s currently ahead in the polls), but if Leah Vukmir is subbed in for Nicholson, the GOP win probability inches up by a little bit.
In West Virginia, the model is still high on Joe Manchin. Monmouth University, a very accurate pollster that the model trusts, showed him ahead of Republican challenger Patrick Morrisey by seven points. That didn’t move the Democratic win probability too much (it has him with an 84 percent win probability) because Manchin was already polling well.
West Virginia is another state where I think the model is a little high on the Democratic candidate (and working in fundamentals would almost assuredly decrease his win probability). But it’s worth considering the possibility that the polls are right. There aren’t many candidates who are able to win unfriendly territory as frequently and convincingly as Manchin has, and the model tends to like candidates who poll as strongly as he has at this point in the cycle.
Finally, Mitt Romney is about as close to a certainty as we can get in Utah. A Salt Lake Tribune poll put him ahead by almost 40 points in the general election. Maybe more importantly, it shows him leading by an even larger margin in his primary race. In April, Romney narrowly lost to little-known Republican candidate Mike Kennedy at the state’s convention, forcing both men to run in the primary. Some might have seen that a bad sign for Romney (e.g., it’s Trump’s party now, not Romney’s etc.). But the this poll reinforces the notion that the state’s larger primary electorate likes Romney and that he shouldn’t have too much trouble winning both the primary and the general.
A few other polls were released on Monday (e.g. the University of Texas poll and Emerson), but those didn’t make into the Sunday night update. They’ll be reflected in tomorrow morning’s forecast. Check out SwingSeat here.