Exceptionally Low Turnout Can Account For Polling Errors
At The American Prospect: Tweet In the home stretch, I wrote that midterm polling is far less accurate than in Presidential years. Today, in The...
Senate and House final snapshots, 2014
Tweet
Here are final polling snapshots for Senate races:
Put your own predictions in comments! Some more notes…
The calculations above will test the question of how well we can do with polls alone. As always, we did not do any house-effects corrections or fundamentals-based modeling. This is a polls-only snapshot.
Technical notes: The same methods were used as for the gubernatorial snapshots. The error bars are SEM of the polled demographic. To calculate win probability, I have incorporated an additional possible 2.5% error to account for polling error/bias. To see how many polls were used, see this spreadsheet.
House: Republicans win popular vote by 1.5 ± 2.0%, gain 8 ± 6 seats.
Please give your own predictions in comments. What surprises do you predict?
Topics:
I think, at the end of the day (which day may be in early January after a run off in Georgia), the dems hold the senate.
Probably 50/50, giving Joe Biden, finally, some real power (which his vice presidency hitherto has not provided).
From my lips to God’s ear…..
Republicans will take the Senate by a hair-thin margin and gain seats in the House. Baker will beat Coakley in Massachusetts.
There will be ridiculous overinterpretation of this as yet another Death of Liberalism, with Obama as tragic figure, and unwarranted predictions of a Republican sweep in 2016.
None of this is terribly hard to predict.
Matt, you write: “None of this is terribly hard to predict.”
Oh?
1) As ever, previously unknown factors were not included.
2) Dems poll low.
3) Cynicism has no specific correlation to fine odds prediction.
After much soul searching and cribbing from previous posters, I’ll make a pretty standard prediction.
R: AK, AR, CO, GA, IA, KY, LA (after run-off)
D: NH, NC
I: KS
There will be a lot of “dust,” but, when it clears, that’s how I see things.
As I see it, there may be three surprises, none of them happy for the Ds. KS, NH or NC may go R. As of now, I believe KS is the most likely.
As far as Orman, I believe it is more complicated. I suspect that Sen. Manchin (D-WV) will switch parties, particularly if NH, KS or NC go red and the Rs control the Senate outright. If that happens and Orman wins, expect Orman to caucus with the Rs. That means 54 Rs/ 46 Ds because of the Manchin flip. It gets even worse if either NC or NH flips, too.
I don’t think Ds recognize the severe downside risk that’s out there.
Thanks……….I think I am going to be sick now.
Not sure where you get the idea Ds don’t recognize “the severe downside risk that’s out there.”
Back in fall 2013 when people were talking about a massive Republican wave election, I calculated whether the Rs would have enough votes to convict Obama in a Senate impeachment trial if they won every single Senate election in 2014. I figured out that they didn’t; they’d be just barely short. Nor can they override presidential vetoes without Democratic votes. (And they’re not really going to win every single Senate election; some of the D seats up for reelection are quite safe.)
Having determined that, I decided there was a limit to how bad things could get.
Matt McIrvin, I like your thinking 🙂
My pick would be that that at least two out of AK, CO, GA, IA go Dem and Orman Caucuses with the Dems, my best guess is that it will be AK, and CO, I have been getting the feeling that the Rep are on to Sam and Nate and they have been pushing polls to the Rep to game the system.
“…they have been pushing polls to the Rep to game the system.” – J Gianni
Very possible though I think the polling community is lost enough w/o that sort of gaming due to the changes that have been mentioned such as heavy cell phone use, demographic shifts etc.
We shall see shortly though I don’t expect any of the polling organizations to admit they have problems.
I’ve seen nothing in the last six weeks to change the prediction I made to the NYT on September 27: The GOP will pick up AR, LA, MT, SD, and WV, plus one (but only one) of these three: AK, CO, IA. (These are all too close to call, even this morning, and I think a 2-1 split either way the most likely. As I’m a Democrat, I’ll make it a 2-1 split favoring the Democrats!) Meanwhile, the GOP lose KS and GA. That’s a net gain of 4 for the GOP, making the new Senate D+I 51, R 49.
I also predict Ativan prescriptions will be filled by the tens of thousands this afternoon.
thank you for reminding me to take my clonopin early.
I predict, hopefully, that the Dems hold the Senate with a 50/50 count, VP Biden holding the tiebreaker vote, due to Dems outperforming the polls, the polls under-polling minorities, and a stronger ground game in the Dems GOTV effort.
Are these Sam’s final predictions?
It would be nice to be able to read ALL the comments instead of just the last 5 or 10.
Never mind, Chrome seems to be having problems displaying all the text on a page. Maybe it is heavy traffic or ?
How did the pollsters miss Virginia so badly? No matter who wins, this election wasn’t even supposed to be close. The race tightened at the end but the absolute closest poll had a 7% Warner advantage. Something seems wrong, here.
While following this blog, I was pretty appalled at the biased data and especially the commentary here. I was hoping for a real discussion to gauge how the night was going to go, but found some of the predictions borderline insane and clearly driven by pure emotion. Next time, how about real analysis without the optimistic delusions?