Here’s the raw numbers from SurveyUSA, 10/28-11/3/2012, for KSDK-TV, KSHB-TV, KSPR-TV and KYTV-TV, 589 likely voters:
If the election for President were today, would you vote for … (choices rotated) Republican Mitt Romney? Democrat Barack Obama? Or one of the other candidates?
Mitt Romney (R) 50%
Barack Obama (D) 43%
Missouri will also elect a Governor. If the election for Governor were today, would you vote for … (choices rotated) Republican Dave Spence? Democrat Jay Nixon? Or Libertarian Jim Higgins?
Dave Spence (R) 39%
Jay Nixon (D) 48%
Jim Higgins (L) 5%
Missouri will also elect a United States Senator. If the election for US Senate were today, would you vote for … (choices rotated) Republican Todd Akin? Democrat Claire McCaskill? Or Libertarian Jonathan Dine?
Todd Akin (R) 36%
Claire McCaskill (D) 51%
Jonathan Dine (L) 8%
So let’s go into the details
The first detail worth noting is that on a survey with the sample of 38% Republicans and 31% Democrats, Claire McCaskill leads by 15 points. Yes, that is possible.
SurveyUSA finds McCaskill leading by 24% (55-31) with Independents. She wins 67% of Moderates and 20% of Conservatives (81% of the electorate). She wins 47% of landlines and 61% of non-landlines. SurveyUSA uses a system where “Respondents not reachable on a home telephone (28% of likely voters) were shown a questionnaire on their smartphone, laptop or other electronic device”.
That landline number is 47-40 Claire. That’s important, compared to the landline-only Public Policy Polling results, where Claire leads 48-44. While the whole cellphone/landline debate hasn’t exactly produced any real consensus about the impact of being landline-only or being landline/cellphone/others. The SurveyUSA numbers move closer to the PPP numbers when you compare their numbers on the exact method uses to survey voters.
(for reference: SurveyUSA landline numbers for President were 55/40 Romney and Nixon 47/Spence 45)
Another difference between SurveyUSA and Public Policy is the partisan composition of their universes.
SurveyUSAs sample: 38R/31D/29I (as noted)
PPP’s (.pdf) sample: 36R/33D/32I
So Claire’s numbers are better on a +7R SurveyUSA than a +3R PPP? Yes, that’s possible.
Public Policy says that Claire and Akin are tied with Independents and that Akin wins 79% of Republicans, instead of the 67% that SurveyUSA put in the Akin column.
If you get wild and creative, you could combine the Public Policy sample, and SurveyUSA party numbers, then you get the following numbers:
President: Romney 48, Obama 46
Senator: Claire 54, Akin 35
Governor: Nixon 50, Spence 39
If you combine the SurveyUSA sample and Public Policy numbers, you get:
President: Romney 54, Obama 43
Senator: Claire 45, Akin 45
Governor: Nixon 50, Spence 46
So that sets up a fun little universe of possibilities if you mix your drinks. (Although the Romney v. Obama numbers from Mason-Dixon were the same as Blunt/Carnahan numbers (54/41). But that’s Mason-Dixon.)
If Claire McCaskill finishes ahead of Jay Nixon, the SurveyUSA method is going to be vindicated and/or we will have wound up in a world we didn’t expect to be in back on August 7th.
If she doesn’t do that well, then the universe continues undisturbed. Because it’s not like the varying success of non-landline methods has stopped Reuters/Ipsos, YouGov, and other online-friendly pollsters. Sometimes pollsters like SurveyUSA get a direct hit (McCain 48, Obama 48). Sometimes they miss slightly (Claire 51, Talent 42). Polling is sort of like a science.
The Presidential race in Missouri is pretty much two campaigns who only run ads in the parts of Missouri whose TV stations cover Iowa. But in all likelyhood, the better Obama does, the better the rest of the Democrats do in the election and the Missouri Democratic ticket probably runs slightly ahead of Obama.
In 2012, We live in a very surreal state. And in a few days, we’ll begin the process of forgetting Todd Akin, Dave Spence and Mitt Romney.