The Anchoress
I’m the first to admit I am no statistician, and I ordinarily pay no attention to polling data until just before elections because polls – depending on how a question is phrased – can say almost anything you wan them to.
But Tom Elia looked at two polls that seemed like they used extremely wide margins, one for Ohio and one for Kentucky.
Ohio:
While both of the polls show Obama with large leads over McCain — one shows Obama with an 11-point lead over John McCain, and the other shows Obama with a nine-point lead over his Repub rival — the political breakdowns for the polls’ respondents are eye-popping.
Public Policy Polling, which has Obama with an 11-point lead in the state, breaks down the poll’s respondents thusly: Dem 55%, Repub 30% — a 25-point margin for the Dems. (A March PPP poll showing McCain with an eight-point lead over Obama broke down: Dem 48%, Repub 32% — a 16-point Dem margin.)
A SurveyUSA poll, which shows Obama with a nine-point lead in Ohio, breaks down the poll’s respondents like this: Dem 52%, Repub 28% — a 24-point margin favoring the Dems. (A March SurveyUSA poll showing McCain with a seven-point lead over Obama broke down: Dem 44%, Repub 34% — for a 10-point Dem margin.) — [bold emphasis mine – admin]
The poll is being lauded by some as proof that Barack Obama is experiencing a bounce in areas that have proven difficult for the Illinois senator.
Trouble is, the poll’s Kentucky respondents self-identify thusly: 57% Dem; 33% Repub — a 24% margin for the Dems. — [emphasis mine – admin]
So, what are we seeing, here? Are the big margins being used to tease the polling numbers in desired directions?
I sent both links to the newly revamped Election Projection because that site has always seemed pretty straight-up, readable and insightful, to me, and I asked Scott, “was Bill Clinton right? Is this really just some sort of ‘fairy tale’ being promoted by the press and manipulated in the polls?”
I don’t know the answer. But I think if the polls are going to run these sorts of margins, perhaps the question is a fair one?
UPDATE I: Well, Scott at Election Projection has looked at the numbers, and he says – “they ain’t happy”:
What I found is a staggering advantage for the Democrats.
During that time period, Americans cast 25.3 million ballots in 17 states, excluding Florida, and Democrats accumulated nearly 62% of them! Let me put that number into proper perspective. In 2004, the Democratic cut in those 17 states was just 51.2%. If we transpose those numbers to Ohio, where 2004 exit polls gave a 5-point partisan advantage to the GOP, we can start to draw some quantitative conclusions.
In 2004, Democrats edged Republicans by 2.5% in the group of primary states I’ve included in this test. This year, that advantage ballooned to 23.2%, a 20.7% change. It follows that, all things being equal, the partisan breakdown in Ohio would change from a 5 point edge for the GOP in 2004 to a Democratic advantage of almost 16 points this year. So how do the polls from PPP and SurveyUSA stand up to this test? Not too well. However – and this is a scary thought for Republicans – they are no more off the mark than a poll giving Democrats a six to eight point advantage.
Bottom line? Obama may not be winning Ohio by a dozen right now, but it’s hard to justify denying that he is winning. And that’s 20 critical electoral votes currently in Obama’s column that President Bush picked up in 2004. But there’s more. The election does not begin and end in Ohio. Many battleground states will fall into Obama’s column if the precedent set in January and early February is duplicated on Election Day. Unless something happens to quell the enormous enthusiasm permeating Democratic circles nationwide, the presidential election of 2008 may look a lot like the one in 1980 with a great communicator of a different stripe leading the way.
There’s your answer.
UPDATE II: Jim Geraghty is also writing about the polls.
And in an email Tom sends along these fun facts:
Back in February, a SurveyUSA poll in Texas showed the political break down favoring the Repubs by just 1%!
Since 1996, the Dems have not won a single statewide race in 74 tries. During that time, only nine Dem candidates received more than 45% of the vote (three in 1996, four in 1998); only five Dems have come within 5% points of their Repub opponents (one candidate, John Sharp, has accounted for two of those five races, Bill Clinton the other); only two have come within 2% points; only one has come within 1% point.
DJ Drummond says let not the heart be troubled as he looks at the polls.
Ad-Free & Subscriber Log-In