On Tuesday, FiveThirtyEight hosted its second chat on Facebook, with our editor in chief, Nate Silver, and senior political writer Harry Enten. Nate and Harry discussed the 2014 midterm elections, including our Senate forecast and some of the more interesting gubernatorial races. We’ve pulled in some of our readers’ thoughtful questions below; the content has been lightly edited for clarity. (Politics isn’t your thing? We held an anything-goes chat in July.)
Clark Hedrick: How would you distinguish “wrong” prediction estimates from the impossibility for numbers to pick up very late-breaking momentum?
Nate Silver: The fact that there’s some time (usually 2-3 days) separating the “final” polls of an election from the election date itself can sometimes matter — perhaps the New Hampshire Democratic primary in 2008 was an example. (Most polls were conducted before the final Obama-Clinton debate.) In theory, this is accounted for in the uncertainty in our estimates. We have it lucky, though — there are a number of countries that ban any polling from being released in the final 24 to 72 hours before an election.
Frank Massaro: Is there any explanation for why the model forecasts higher probabilities for the Republicans to obtain 50 and 52 seats than 51 seats?
Harry Enten: Good question. Blame Greg Orman in Kansas. Orman has said he’ll caucus with whichever party has the majority in the next Senate. So, if Orman wins and one party already has a majority, they will get a “bonus” seat. It’s possible, however, Orman’s decision will determine who has the majority. For a full explanation of how our model is treating Kansas, see here.
Jonathan Arndt: Do you think that the trend of younger people dumping their landlines in favor of cellphone only has had a skewing effect on polling numbers toward Republicans? I used to get called quite often when I had a landline but never on my cellphone (not complaining).
NS: This was a pretty gigantic problem in 2010 and 2012 and may have accounted for some of the over-performance of Democrats vs. the polls in those cycles. However, the vast majority of survey firms are including cellphones now or conducting polls online.
Chris Bytof: Wisconsin governor?
HE: I have thought for a while that Gov. Scott Walker was the slight favorite for re-election. It should be noted though that the polling has tightened since then. Still, the information that I’m presented suggests that Walker has a better chance of winning than losing. It will almost certainly be closer than in 2012 or 2010. What that exact percentage is is another question. He is far from a shoo-in. [Editor’s note: Marquette University Law School released a survey on Wednesday showing Walker up 7 percentage points on Democrat Mary Burke.)
Jess Milcetich: Who’s going to win the Maryland gubernatorial race? The media keeps saying the race is extremely close. Is it really?
HE: I don’t really think it’s that close. It’s certainly not 2010, but it’s not a nail biter if you believe the polls. Larry Hogan has led in exactly ZERO polls this year. This includes his own polling. The average of nonpartisan-sponsored surveys this month has Anthony G. Brown +8. He’s not a lock, but he’s clearly the favorite.
Justin Buck: What would the election results be if everyone could register and vote online? Basically wanting to know, if everyone in the U.S. actually voted, how different would our government look?
NS: One way to estimate this is to compare polls of all adults and likely voters. They suggest the vote might be a net of 3 to 6 points more favorable to Democrats if everybody in the country voted. With that said, political coalitions and strategies might change a great deal if we had close to 100 percent turnout — there’d be less pandering to older voters, for instance.
Colleen Brooks: North Carolina currently has four times the early voters in 2014 compared to 2010. Will that affect your figures at all in the Senate race?
HE: Most of the polling indicates that Kay Hagan has a small but steady lead. The early voting numbers seem to be confirming that. She is getting people to the polls who didn’t vote in 2010, which will be key for her to win. Our belief is that the polling (which can actually ask early voters whom they voted for) will pick up on any effect.
Jordan Price: My question is about the stability of Republicans’ overall chance of winning a Senate majority. Despite the fact that the FiveThirtyEight model has been running simulations for nearly two months, and that Election Day is now only one week away, the GOP chance of a Senate takeover is still, and has almost always been, in the low 60 percent range. Does the fact that Democrats have held on to a mid-upper 30 percent chance of keeping the Senate — even now with, just one week left for an upset — suggest that the political environment has improved for them enough to offset the GOP’s continuing advantage of remaining slightly ahead later and later in the game?
NS: This deserves a longer reply, but I think it tends to suggest that the overall political mood is a lot more ambivalent than in 2010. If we were dealing with a 2010-type environment, the Senate would be a slam dunk for the GOP.
Hunter Chenevert: Any real chance for Wendy Davis to keep it close in Texas?
HE: What’s your definition of “close”?
Hunter Chenevert: Say, 7 points?
HE: We’re in the business of being very careful about saying something has “no chance” of happening. So, yes, there is a chance Davis loses by single digits. But 7 points? That’s optimistic if you’re a Davis supporter.
Visit the FiveThirtyEight Facebook page to see the whole chat.
