On a monstrous Monday (a new track record as we finally hit the century mark with
107 polls), my attention focuses on only about a dozen of them.
We have, at this point, made clear time and again the odd (and growing) disconnect with what the national polls have shown and what the state polls show. Currently, that disconnect sits at about three percentage points, with an average taken from the national polls stubbornly sitting at a Romney edge of about a percentage point, and an estimate of the national vote based on the variety of state polls consistent with an Obama lead of roughly two points.
Today, we look at the national polls themselves. Even within that relatively small data set, we have now seen a couple of weeks of incredibly wide variance, with the average day showing a 5-7 point range between the most optimistic numbers for the president, and the most optimistic numbers for the challenger. Today is no exception, with a six-point spread between the most optimistic Obama poll and the most optimistic Romney poll.
With Rasmussen edging back towards the mean a bit, at this stage the national polling disparity is mostly a case of Gallup vs. the world. Which it has been for a while, actually. It is worth looking into it a bit to explore why that is the case.
More on that after the jump. For now, though, on to the (crapload) of numbers:
PRESIDENTIAL POLLING:
NATIONAL (ABC/WaPo Tracking): Obama 49, Romney 49
NATIONAL (American Research Group): Obama 48, Romney 48
NATIONAL (Gallup Tracking): Romney 51, Obama 46 (LV); Obama 48, Romney 48 (RV)
NATIONAL (IBD/TIPP Tracking): Obama 45, Romney 44
NATIONAL (Ipsos/Reuters Tracking): Obama 48, Romney 47 (LV); Obama 51, Romney 41 (RV)
NATIONAL (Pew Research): Obama 47, Romney 47 (LV); Obama 47, Romney 45 (RV)
NATIONAL (Politico/GWU Battleground): Obama 49, Romney 48
NATIONAL (PPP Tracking): Romney 49, Obama 48
NATIONAL (Rasmussen Tracking): Romney 49, Obama 47
NATIONAL (UPI/CVoter): Obama 48, Romney 47
CALIFORNIA (USC/LA Times): Obama 54, Romney 40
COLORADO (American Research Group): Romney 48, Obama 47
FLORIDA (CNN/ORC): Romney 50, Obama 49 (LV); Obama 52, Romney 46 (RV)
FLORIDA (PPP): Obama 49, Romney 48
INDIANA (Pharos Research Group): Romney 55, Obama 42
KANSAS (Jayhawk Poll): Romney 56, Obama 36
MARYLAND (Baltimore Sun): Obama 55, Romney 36
MASSACHUSETTS (Univ. of New Hampshire/Boston Globe): Obama 52, Romney 38
MINNESOTA (Mason Dixon for the Minneapolis Star-Tribune): Obama 47, Romney 44
MINNESOTA (St. Cloud University): Obama 53, Romney 45
MISSOURI (Mason Dixon): Romney 54, Obama 41
MONTANA (Pharos Research): Romney 50, Obama 43
NEBRASKA (Omaha World-Herald): Romney 52, Obama 38
NEBRASKA (Pharos Research Group): Romney 58, Obama 39
NEBRASKA 2nd DISTRICT (Omaha World-Herald): Romney 49, Obama 44
NEVADA (CallFire/Faith Horizon): Obama 50, Romney 46 (LV); Obama 51, Romney 46 (RV)
NEW HAMPSHIRE (Grove Insight for Project New America--D): Obama 47, Romney 44
NEW HAMPSHIRE (Lake Research for USAction--D): Obama 47, Romney 42
NEW HAMPSHIRE (PPP): Obama 49, Romney 47
NEW JERSEY (Philadelphia Inquirer): Obama 51, Romney 41
NEW MEXICO (Albuquerque Journal): Obama 50, Romney 41, Johnson 5
NEW YORK (SurveyUSA): Obama 62, Romney 33
NORTH CAROLINA (Elon University): Obama 45, Romney 45
NORTH CAROLINA (Rasmussen): Romney 52, Obama 46
NORTH DAKOTA (Pharos Research): Romney 55, Obama 38
OHIO (Gravis--R): Obama 50, Romney 49
OHIO (Mellman Group for Americans United for Change): Obama 49, Romney 44
OHIO (Pharos Research Group): Obama 49, Romney 46
OHIO (PPP): Obama 51, Romney 47
OHIO (Rasmussen): Romney 50, Obama 48
OHIO (Univ. of Cincinnati for the Ohio Newspapers Group): Obama 49, Romney 49
OREGON (Hoffman Research--R): Obama 47, Romney 42
PENNSYLVANIA (Philadelphia Inquirer): Obama 49, Romney 43
TENNESSEE (Middle Tennessee State Univ.): Romney 59, Obama 34
TEXAS (Univ. of Texas/Texas Tribune): Romney 55, Obama 39
VIRGINIA (Garin-Hart-Yang for Priorities USA PAC): Obama 49, Romney 46
VIRGINIA (Gravis Marketing--R): Obama 48, Romney 48
VIRGINIA (Washington Post): Obama 51, Romney 47
DOWNBALLOT POLLING:
CT-SEN (Rasmussen): Chris Murphy (D) 51, Linda McMahon (R) 45
FL-SEN (Mason Dixon): Sen. Bill Nelson (D) 47, Connie Mack IV (R) 44
FL-SEN (PPP): Sen. Bill Nelson (D) 50, Connie Mack IV (R) 42
FL-SEN (Rasmussen): Sen. Bill Nelson (D) 49, Connie Mack IV (R) 46
FL-SEN (Voter Survey Service--R): Sen. Bill Nelson (D) 49, Connie Mack IV (R) 44
HI-SEN (Ward Research for the Honolulu Star-Advertiser): Mazie Hirono (D) 57, Linda Lingle (R) 35
IN-SEN (Pharos Research Group): Joe Donnelly (D) 47, Richard Mourdock (R) 46
MD-SEN (Baltimore Sun): Sen. Ben Cardin (D) 50, Daniel Bongino (R) 24, Rob Sobhani (I) 14
MA-SEN (Univ. of New Hampshire/Boston Globe): Sen. Scott Brown (R) 47, Elizabeth Warren (D) 47
MO-SEN (Kiley and Company for the McCaskill campaign): Sen. Claire McCaskill (D) 53, Todd Akin (R) 39
MO-SEN (Mason Dixon): Sen. Claire McCaskill (D) 45, Todd Akin (R) 43
MT-SEN (Pharos Research): Sen. Jon Tester (D) 48, Denny Rehberg (R) 47
NE-SEN (Omaha World-Herald): Deb Fischer (R) 48, Bob Kerrey (D) 45
NE-SEN (Pharos Research Group): Deb Fischer (R) 50, Bob Kerrey (D) 47
NE-SEN (Public Opinion Strategies for the Fischer campaign): Deb Fischer (R) 55, Bob Kerrey (D) 39
NM-SEN (Albuquerque Journal): Martin Heinrich (D) 50, Heather Wilson (R) 42, Jon Barrie (IAP) 3
NY-SEN (SurveyUSA): Sen. Kirsten Gillibrand (D) 64, Wendy Long (R) 22
ND-SEN (Pharos Research Group): Heidi Heitkamp (D) 50, Rick Berg (R) 48
OH-SEN (Gravis--R): Sen. Sherrod Brown (D) 48, Josh Mandel (R) 47
OH-SEN (Pharos Research Group): Sen. Sherrod Brown (D) 50, Josh Mandel (R) 43
OH-SEN (PPP): Sen. Sherrod Brown (D) 53, Josh Mandel (R) 42
OH-SEN (Univ. of Cincinnati for the Ohio Newspaper Group): Sen. Sherrod Brown (D) 51, Josh Mandel (R) 47
PA-SEN (Philadelphia Inquirer): Sen. Bob Casey (D) 49, Tom Smith (R) 42
TN-SEN (Middle Tennessee State Univ.): Sen. Bob Corker (R) 59, Mark Clayton (D) 21
TX-SEN (Univ. of Texas/Texas Tribune): Ted Cruz (R) 54, Paul Sadler (D) 39, Others 5
VA-SEN (Gravis Marketing--R): George Allen (R) 48, Tim Kaine (D) 46
VA-SEN (Washington Post): Tim Kaine (D) 51, George Allen (R) 44
WI-SEN (Rasmussen): Tommy Thompson (R) 48, Tammy Baldwin (D) 47
MO-GOV (Kiley and Company for the McCaskill campaign): Gov. Jay Nixon (D) 55, Dave Spence (R) 33
MO-GOV (Mason Dixon): Gov. Jay Nixon (D) 48, Dave Spence (R) 42
MT-GOV (Pharos Research): Steve Bullock (D) 47, Rick Hill (R) 44
NH-GOV (PPP): Maggie Hassan (D) 48, Ovide Lamontagne (R) 44
NH-GOV (Rasmussen): Ovide Lamontagne (R) 48, Maggie Hassan (D) 46
NC-GOV (Elon University): Pat McCrory (R) 52, Walter Dalton (D) 38
NC-GOV (Rasmussen): Pat McCrory (R) 54, Walter Dalton (D) 35
ND-GOV (Pharos Research): Gov. Jack Dalrymple (R) 63, Ryan Taylor (D) 34
CA-07 (PPP for Credo): Ami Bera (D) 46, Rep. Dan Lungren (R) 46
FL-22 (PPP for Credo): Patrick Murphy (D) 48, Rep. Allen West (R) 47
HI-01 (Ward Research for the Honolulu Star-Advertiser): Rep. Colleen Hanabusa (D) 52, Charles Djou (R) 41
HI-02 (Ward Research for the Honolulu Star-Advertiser): Tulsi Gabbard (D) 73, Kawika Crowley (R) 8
IL-08 (PPP for Credo): Tammy Duckworth (D) 54, Rep. Joe Walsh (R) 40
IL-08 (We Ask America--R): Tammy Duckworth (D) 55, Rep. Joe Walsh (R) 45
IL-10 (We Ask America--R): Rep. Bob Dold (R) 54, Brad Schneider (D) 46
IL-11 (We Ask America--R): Bill Foster (D) 50.4, Rep. Judy Biggert (R) 49.6
IL-12 (We Ask America--R): Bill Enyart (D) 51, Jason Plummer (R) 46, Paula Bradshaw (G) 4
IL-13 (We Ask America--R): Rodney Davis (R) 50, David Gill (D) 45, John Hartman (I) 4
IL-17 (We Ask America--R): Rep. Bobby Schilling (R) 52, Cheri Bustos (D) 48
MD-06 (Baltimore Sun): John Delaney (D) 42, Rep. Roscoe Bartlett (R) 41
MN-08 (OnMessage for the Cravaack campaign): Rep. Chip Cravaack (R) 50, Rick Nolan (D) 40
MN-08 (PPP for Credo): Rick Nolan (D) 48, Rep. Chip Cravaack (R) 44
MT-AL (Pharos Research): Steve Daines (R) 52, Kim Gillan (D) 45
NE-01 (Omaha World-Herald): Rep. Jeff Fortenberry (R) 65, Korey Reiman (D) 24
NE-02 (Omaha World-Herald): Rep. Lee Terry (R) 47, John Ewing (D) 43 (LV); Rep. Lee Terry (R) 48, John Ewing (D) 42 (RV)
NE-03 (Omaha World-Herald): Rep. Adrian Smith (R) 64, Mark Sullivan (D) 24
NH-01 (PPP for Credo): Rep. Frank Guinta (R) 48, Carol Shea-Porter (D) 47
NY-19 (Public Opinion Strategies for the Gibson campaign): Rep. Chris Gibson (R) 49, Julian Schreibman (D) 39
ND-AL (Pharos Research): Kevin Cramer (R) 55, Pam Gulleson (D) 41
RI-01 (OnMessage for the Doherty campaign): Brendan Doherty (R) 45, Rep. David Cicilline (D) 39, Others 6
TN-04 (Public Opinion Strategies for the DesJarlais campaign): Rep. Scott DesJarlais (R) 49, Eric Stewart (D) 36
A few thoughts, as always, await you just past the jump ...
A sign of how detached Gallup has been from the rest of the polling world can be summarized by this tweet from earlier today:
Gallup finds Mitt Romney leading 52% to 46% among early voters. They manage to be totally different even with early voters.
— @Taniel via web
Hard to imagine, but it is a fair summation of a larger issue with Gallup. While their polling among the larger universe of registered voters has been close to the national average, their likely voter numbers have been in some kind of alternate universe.
Today's national average, without Gallup, would be Obama +0.1. With it, Romney +0.4. And that impact is muted slightly, because there were a total of four non-tracking polls in the mix today. That is because Gallup is three points away from any other pollster today, clinging to their finding that Mitt Romney is en route a sizable popular vote victory of 5 percentage points.
How has Gallup's likely voter screen managed to (apparently) lurch so far off of the fairway? Well, that is a tougher question to answer than you think, for a reason explained about week ago by Pollster founder Mark Blumenthal:
Ultimately, solving the mystery of Gallup's currently divergent result is difficult, particularly since it does not routinely disclose the demographic or attitudinal composition of its likely or registered voter samples.
However, longtime Daily Kos community member
D Wreck compares the Gallup numbers with the Census Bureau's voter participation data, and sees a clear pattern of undersampling of Latino and African-American voters, presumably on the premise that they will not turn out. Political Science Professor
Alan Abramowitz put a specific figure on it (through interpolation), saying that the LV model for Gallup assumed only 20 percent of the electorate would be nonwhite. That would be the lowest figure since 2000, a relatively low-turnout affair. Virtually no pollsters, as it happens, have predicted a white turnout of 80 percent in this election. Indeed, the over-under seems to be at about 75 percent.
Furthermore, though they do not weight for it, Andrew Sullivan noted that Gallup's LV screen assumes an electorate which will have more self-identified Republicans than Democrats. While that is possible, one supposes, full disclosure demands that it be pointed out that a presidential electorate with more Republicans than Democrats has not occurred (according to exit polls) at any point dating back to 1976. Indeed, Democrats have had a lead of at least a couple of points on the party ID score in every election dating back to that election, with one exception (it was evenly split in 2004).
So, in summary, Gallup's LV numbers are different for a very simple reason. They are seeing a very different electorate than their colleagues. It doesn't mean, necessarily, that they are wrong. It is possible, of course, that they have the electorate pegged, and everyone else is falsely seeing a more Democratic and less white electorate.
One problem for Gallup, of course, is the simple First Rule of poll analysis that I have said time and again: when you have one answer, and everyone else has a different answer, it is unlikely that everyone else is wrong. Another problem? The last time Gallup employed an LV screen, in 2010, they missed, and missed horribly. Their final poll had Republicans beating Democrats by a 15-point margin in their national congressional vote estimates. The final margin was...six points. Their RV screen (R+4) wound up being much, much closer to the pin.
In other polling news...
- Today's Polling Wrap, as noted in the open, is a new track record, with 107 new surveys both at the top of the ballot and downballot. However, given the weather-related trauma that is being inflicted on the east coast, expect volume to reduce substantially. Already, three separate national pollsters (PPP, Gallup, and IBD/TIPP) have announced that they are suspending their national tracking polls. One would expect that more will follow. What's more, a number of pollsters are based in New Jersey. They include (though this is not an exhaustive list) Gallup, SurveyUSA, Rasmussen, and Monmouth. Even though this is my gig, there are clearly more important things in polls. Thoughts and prayers to our readers and friends who are in harm's way.
- If people aren't trying to read tea leaves from samples of 500-1000 voters in polls, they are busy trying to do so from the statistics on absentee and early balloting in the states that offer those options. As always, Professor Michael P. McDonald is one of the big go-to guys on the subject. Here was his assessment of the state of play on that subject from this weekend. It's worth a read, and he is worth following, as he constantly updates his analysis.