If Bias Doesn’t Matter …
Posted on | October 27, 2012 | 46 Comments
. . . why is Dan Hodges cheering Nate Silver for “singlehandedly dismantling the myth of Mitt-mentum”? That is to say, if the reporting of poll data doesn’t influence voters one way or the other — if there is no bandwagon psychology effect created by such media narratives – then what does it matter whether journalists believe one candidate or another is winning?
Good luck answering that, and good luck to all the liberals who believe that counterfactual cheerleading can prevent Obama’s defeat if, in fact, we’re already on the other side of a “preference cascade.”
When Gallup showed Romney widening his lead to 5 points — and at 51 percent, just 10 days before Election Day — Donald Douglas anticipated that Nate Silver’s head would explode. But the Grand Swami will find a way to rationalize this, and the Graveyard Whistling Choir will cheerfully parrot whatever rationalization Silver provides them.
- -
Only one poll in October has shown Obama as high as 50 percent, and if Silver shows a sudden enthusiasm for John Zogby, you’ll know why.
As for me, I’m more traditional in thinking that an incumbent president who is persistently polling below 50% in October is on his way to becoming an ex-president. Romney’s margin in the Real Clear Politics average is a slender 0.9%, but . . . we’ll see. Gallup’s topline number involves pushing of “leaners,” and we don’t know whether Romney will maintain or extend this advantage.
Look at the Investors Business Daily poll for something interesting: Obama leads by 2.3 points in the topline, but the sample is D+7 and Romney leads by 10 points among independents.
We’ve seen this pattern so often in so many polls this year that I think it’s time to ask, “Is this what brand damage’ looks like?”
Did the post-2004 unpopularity of President Bush and his party so damage the GOP’s reputation that voters who might have once identified themselves to pollsters as Republicans are, instead, now thinking of themselves “indepedendents”? Other than persistent sample bias (or “weighting” of the samples by pollsters) it’s hard to think of any other explanation of this trend. This has shown up in poll after poll — a remarkably high advantage for Democrats in terms of partisan identification (e.g., 38% D vs. 31% R in the IDB poll) and a double-digit advantage for Romney among independents.
If my “brand damage” theory explains this phenomenon, then the topline numbers in the polls are more or less valid. If, on the other hand, the D/R split is tilted to the Democrats’ advantage due to some kind of artificial sample bias, then Romney’s lead is being underestimated. Put the D/R split closer to parity, and Romney’s double-digit lead among independents would guarantee a big win — 53% or more in the popular vote.
We could have a big discussion about “brand damage” and the failure of the GOP to build a stronger sense of partisan loyalty among Republican-leaning voters, but I’m content to let that wait until after the election. There is a definite problem, I think, but the discussion of the problem will be fundamentally shaped by the election results, and we don’t yet know how that will turn out.
Wisconsin: Romney 49, Obama 49
New FL poll shows Romney up 5, 51/46
Or do we? My gut hunch is that Obama has really already lost, and that the only question is whether the margin of his defeat will be decisive enough that everyone can deal with it, or whether we’re going to get the Nightmare Scenario:
92% (!) of Obama voters think Obama is going to win. Obviously these folks get their news from the mainstream media, aren’t paying attention to recent poll results, and/or are just bad at math. But more importantly, if the current “preference cascade” continues and Romney wins, as appears more likely every day, a significant minority of that clueless 92% is likely to engage in Occupy-style riots and endless false accusations of voter fraud and “election stealing.” And the irresponsible media (e.g., Dan Rather, MSNBC) will have only itself to blame.
Please, Lord, not that. If the election is close enough that the fringe left can claim it was “stolen,” it’s almost worse than losing.
Almost, I said. The fact that David Axelrod is talking about how Obama could win without Ohio is, I think, highly significant.
Maybe Nate Silver can give odds of (a) Obama winning Ohio and (b) Romney winning nationally by 5-pt margin gallup.com/poll/157817/el…
— Robert Stacy McCain (@rsmccain) October 27, 2012
Because I just can’t see both (a) and (b) being true. mobile.twitter.com/rsmccain/statu…
— Robert Stacy McCain (@rsmccain) October 27, 2012
@jimmiebjr This mystic MSM faith in Obama’s kryptonite Ohio mojo? I ain’t buying it. Not if Romney’s winning indies and +5 nationally.
— Robert Stacy McCain (@rsmccain) October 27, 2012
@vodkapundit @jimmiebjr Not making any predictions myself, you understand, but Obama down 5 in Gallup 10 days before the Election Day?
— Robert Stacy McCain (@rsmccain) October 27, 2012
@vodkapundit @jimmiebjr Well, I think I just heard Candy Crowley humming a few arpeggios, IYKWIMAITYD.
— Robert Stacy McCain (@rsmccain) October 27, 2012
“KEEP CALM AND FINISH HIM.”
– Bryan Preston
UPDATE: This morning’s results tick the RCP average up slightly — Romney is now a full point ahead — and look at the latest four tracking polls:
Rasmussen Reports 10/24-10/26 ……… Romney 50, Obama 46
ABC News/Wash Post (10/22-10/25) … Romney 49, Obama 48
Gallup (10/19-10/25) ………………………. Romney 51, Obama 46
IBD/TIPP (10/20 – 10/25) ……………….. Rommey 45, Obama 47
These four have results through Thursday, and the average of them is Romney 48.8%, Obama 46.8%. Thus the average of the four most recent polls has Romney +2, which is twice the RCP average that (as of today) includes results that date back as far as Oct. 15. And something else you’ll notice from the RCP average: In general, the bigger the sample, the greater the Romney lead. Ranking them by sample size:
Gallup (2,700 likely voters) ………… Romney +5
Rasmussen (1,500 LV) ……………….. Romney +4
Monmouth (1,402 LV)) ………………. Romney +3
ABC News/Wash Post (1,382 LV) … Romney +1
Politico/GWU (1,000 LV) ……………. Romney +2
IBD/TIPP (957 LV) …………………….. Obama +2
Associated Press (839 LV) …………… Romney +2
NBC/WSJ (816 LV) …………………….. TIE
Wash. Times/Zogby (800 LV) …….. Obama +3
CBS News (790 LV) ……………………. Obama +2
Does this mean anything? I’m not sure, but just ask yourself this: If the opposite were true — if the polls with the largest samples all favored Obama, whose lead was larger in the more recent results — don’t you think Nate Silver would be shouting it from the rooftops?

Pingback: The Lamentations of Their Women | Daily Pundit
Pingback: Philly Inquirer Poll Has Obama Dropping To Under 50% | The Lonely Conservative
Pingback: “The Children of the Future” Sing Creepy Songs of Dear Leader Obama (Video) and Saturday Links!
Pingback: (Not exactly) This Week in Automotivators, October 22-28 « The TrogloPundit
Pingback: Mark October 30 on Your Calendar: Democrat Panic Hits Pandemic Stage : The Other McCain