D-leaning PPP is out with a new poll of the Ohio governor's race today that shows incumbent Democratic governor Ted Strickland trailing Republican John Kasich 42-37%. http://www.publicpolicypolling.com/...
It's not the margin, it's the motion, as they say. Strickland was 20 points ahead one year ago, and his approval rating was then Olympian. 37% for an incumbent is abyssal, regardless of where the opponent stands. PPP has Strickland's approval rating now at just 33% among all voters, and just 53% among Democrats, the latter being worse than he was rated by all voters just one year ago. Strickland's Ohio approval rating is now lower than Obama's, the "black man" whom Strickland once said could not win the presidential contest in this state.
Imagine that. PPP -- and remember they are Democrats -- now labels Strickland
"an unusually unpopular incumbent."
But wait, didn't news headlines just a month ago boast of a Strickland "resurgence" or "renaissance" (like the mythical Nuclear Renaissance that Strickland pitches), based on a Quinnipiac poll that had the Ted-meister trouncing Kasich by a whole five points?
To round out the recent published polling, Rasmussen threw cold water on the Ted-head midwinter mini-jubilee with a poll in early March that had Kasich beating Strickland by a whopping 11 points. http://www.quinnipiac.edu/... Despite the margin difference, that poll was in surprising agreement with PPP on the level of Strickland's support in the state -- just 37-38%.
We await the new Ohio Poll, which has the best track record in Ohio. But throughout the latter half of 2009, the Ohio Poll closely mirrored Rasmussen's polling of the Strickland-Kasich race, and we can expect the same to hold true now.
So what is with Quinnipiac? Well, first of all, of the four major pollsters of this race, only Quinnipiac samples all registered voters. PPP, Rasmussen, and the Ohio Poll all screen for Likely Voters. Thus, all of the polls could theoretically be accurate if indeed Democrats are less likely to vote this year, a situation confirmed by other national polling and by Strickland's very low approval ratings among voters in his own party.
And if we look to polls as some prefiguration of the election, then polls of all registered voters are increasingly irrelevant, because, duh, likely voters are more likely to vote.
But that explanation alone would be giving Quinnipiac too much credit. This diary is, in fact, a case study in trash polling -- polling that gets the dynamic of a race wrong because of methodological problems. Bad polling can have all sorts of unfortunate consequences, including the inducement of false hopes or overconfidence, encouragement of bad campaign strategy, and the diversion of donations to unworthy candidates. All of these potential consequences are evident in the case at hand. Quinn, in Ohio, has gotten just about everything wrong.
The race for the Ohio governor's seat merits attention because of incumbent Ted Strickland's precipitous decline in approval and support over the past year, and Ohio's obvious importance as a swing state for 2012 and beyond (if there is a beyond).
Ohio is an especially challenging state to poll because Ohioans cherish their status as the nation's belwether, and frequently or intentionally defy expected patterns. As a perfect microcosm of the United States, Ohio includes classic large cities, and densely settled countryside that is part Appalachian, part midwestern, and part east-coast. Polling models derived from more homogeneous states are almost guaranteed to get Ohio wrong.
The typical polling error in Ohio, mirroring the typical popular misconception, is to treat Ohio like an urbanized state, because Ohio has so many cities -- the big three (Cleveland, Columbus, Cincinnati), the medium five (Dayton, Toledo, Youngstown, Akron, Canton) and many small cities.
In reality, though, Ohio is one of the least urbanized states in the nation demographically, meaning that a higher percentage of Ohioans lives outside the cities compared to other states. California, Utah, Nevada, and Arizona are all much more urbanized than Ohio.
That's important because urban and rural areas dramatically differ in partisanship patterns, election turnout, phone accessibility for polling, and willingness to participate in polls. These city-country differences tend not to matter in highly urbanized states like California, or very rural states like Iowa. Only in a state like Ohio, where city and country are balanced and mixed-up, do the polling differences have large distortional effects.
That is very clearly the case with recent Quinnipiac polling of the major Ohio races, and the problem is evident from both internal and external inconsistencies.
The external inconsistency is that Quinnipiac has deviated substantially from Rasmussen, Ohio Poll, and now PPP assessments of the same races, which closely track each other. Even if one wishes to dismiss Rasmussen (which would be arbitrary since Rasmussen has a very good track record in Ohio), the Ohio Poll (short for the University of Cincinnati/Ohio Newspaper Poll) enjoys the most consistent reputation for polling Ohio races with accuracy.
The Ohio Poll, Rasmussen and PPP all have Kasich leading Strickland in the governor's race by 5-11 points. Only Quinnipiac now gives Strickland a 5 point lead.
The innacuracy of that Quinn result is made clear by internal inconsistencies. For example, Ohio approval ratings for Obama and Strickland have closely tracked each other, according to all pollsters, for obvious reasons. However, in the most recent Quinn poll, Strickland's performance is rapidly rising while Obama's approval is markedly falling. That result is nonsensical.
Even more nonsensical, as part of its polling of the U.S. Senate race, Quinn asked respondents to what degree they blame "the government in Columbus" for current problems. (Why they did not ask this question as part of the gubernatorial poll remains elusive and suspicious.) The answers to this question varied largely by region. Southeast Ohioans (Appalachians) are, as expected, by far the most likely to blame the state government. However, according to Quinn's gubernatorial poll, the voters of this same region are the most enthusiastic backers of the incumbent governor. How can the residents of one region selectively blame state government while also increasing their support for the incumbent in Columbus?
The answer is: They can't. The polling is trash, based on faulty methodology.
To further elucidate this point, let's look more at the regional crosstabs for the governor's race. Regional breakdowns are crucial, because one of the few constants in politics is the relative leanings of regions. That holds true nationally and within Ohio. As an example, Obama's fortunes may rise or fall, but we expect that he will always do relatively better in Vermont than in Utah, with Ohio falling someplace in between. If a poll comes out showing Obama's approval as higher in Utah than in Vermont, we'd know it was a load of crap. Regions don't flip that way.
Most pollsters, including Quinnipiac, divide Ohio into five regions, with little meaningful variance in the definition of boundaries. (PPP does not publish regional cross-tabs.) There are three more urban regions -- NE, SW and Central -- and two very rural regions -- NW and SE.
A constant in Ohio politics is that the Northeast, including Cleveland, is the most strongly Democratic, by far, while the Northwest, with the sole exceptions of urban Toledo and Wood County, is the most strongly Republican. The Southwest is generally Republican, with the exception of downtown Cincinnati. Central and Southeast Ohio are swing regions that usually determine the outcome of statewide races. On the national analogy, think of the Northeast as Vermont, the Northwest as Utah, and the Southeast as Ohio.
If you look at the Ohio Poll and Rasmussen regional crosstabs for the current gubernatorial race you find that these relationships hold true. That is, Strickland leads by a large margin in the Northeast, Kasich leads comfortably in the Northwest, the Southwest, and also Central Ohio, where his former congressional district lies. The Southeast
is a toss-up, and even as the race shifts, these regional relationships hold true. That is, we can rank the regions in terms of Strickland's strength as follows:
- NE
- SE
- Central
- SW
- NW
Simply reverse the order for Kasich's strength. This ranking mimicks the pattern of Ohio results in all recent presidential elections. Two regions with sequential rank may occasionally switch position, but otherwise the pattern is invariant.
Now look at the regional crosstabs in Quinnipiac's 2010 gubernatorial polls. Astoundingly, we find that Strickland's best consistent performance is in the Northwest region, which is Kasich's best region in every other poll. We also find that Strickland is doing so incredibly well in the anti-Columbus Southeast that it has eclipsed his performance in the true-blue Northeast.
So the Quinnipiac regional ranking for Strickland, averaging its last few polls, looks like this:
- NW
- SE
- NE
- SW
- Central
This is madness. But it's a methodical madness, because the regions that Quinn is getting wildly wrong are the two heavily rural regions. Something in Quinn's methodology is causing its polls to have a very strong Strickland bias in rural areas -- compared to other polls and Quinn's own crosstab questions.
What could that problem be? Well, Quinn doesn't publish its methodology, but it's not difficult to make an educated guess based on some peculiarities of Ohio's political culture. In virtually all of rural Ohio, Democrats disproportionately register compared to Republicans. But Independents outnmber both parties combined in registrations, and those Independents tend to vote Republican. Thus, in a rural county like mine (Pike), registered Democrats may outnumber registered Republicans by 3 to 1, yet the county votes Republican in recent presidential elections.
Pollsters seldom sample Independents according to their true registration numbers, because they don't vote in primaries, and frequently have lower turnout overall. Therefore, there is a temptation to sample voters on the basis of the ratio of party registrations. That tends to work in urban areas, but not in adjacent rural areas. Quinn is doing something like that in Ohio, and that's why it's producing a large consistent bias in favor of Ted Strickland in Ohio's rural regions.
Simply put, Quinn is calling too many rural Democrats, and too few rural Republican-leaning Independents. This bias matters little in states that are more urbanized, and in Ohio is corrected by those pollsters who apply likely voter screens.
Which is all to say that those Ted-head celebrations in February were escapist fantasies. That was clear even at the time to anyone who goes out and talks to people in real life. (I'm thinking of opening a recovery clinic for "progressive" bloggers in Pike County, where the patients will be required to check their laptops at the door.)
Why has One-Term Ted of Ohio gone from being Mr. Popularity -- a shoe-in for the Vice-Presidency if Hillary had made the Oval Office -- to now being "unusually unpopular"?
If you don't know, you haven't been paying attention. Mr. Strickland has literally gambled his way out of more billion-dollar budget deficit fiascos than can be easily counted, he's led perhaps the second most corrupt state government in recent memory (and at least some from the most corrupt administration landed in jail), he's waged a campaign of destruction against Ohio's natural environment, he's backed more failed phantom development projects than you can throw a buckeye at, and he's shown far more commitment to playing Party Boss than to serving as governor of the state.
Here in Strickland's home province (the Tikrit of Ohio, as they say), it's hard to find a third cousin or former neighbor of Ted, who doesn't have something extremely unpleasant to say about the man. In the one-word association game, say Ted Strickland and the response you most often get back is: "liar." That's not a sentiment easily changed or turned around, even if the party's fortunes do rise.
Briefly put, Ohioans have smartened up about Ted Strickland. Now if the party leaders smarten up, they'll find a way to get him off the ballot. (Pssssst -- I hear that Lee Fisher will be looking for a new gig after May 4.)
And don't come at me with any of that "Kasich is worse" crap. Kasich isn't ruining my party. Kasich isn't removing any rational alternative from Ohio politics. John Kasich is simply the certain product of the Strickland reelection campaign.
In the film Annie Hall, Woody Allen says:
Life is divided into the horrible and the miserable. That's the two categories. The horrible are like, I don't know, terminal cases, you know, and blind people, crippled. I don't know how they get through life. It's amazing to me. And the miserable is everyone else. So you should be thankful that you're miserable, because that's very lucky, to be miserable.
This has become the meme, theme, and ream of the Strickland reelection campaign, hence of the entire Ohio Democratic Party. I expect the Party will rent busses, or VERY slow trains, plastered with the slogan:
YOU'RE LUCKY YOU'RE MISERABLE SO VOTE FOR STRICKLAND:
HE'S LESS HORRIBLE THAN THE OTHER GUY!
Not surprisingly, this has failed to win voter enthusiasm, and Ted Strickland will be trounced by John Kasich in November, if the governor stays in the race that long.
Let's urge him not to, because he will doom many down-ticket races in the process.
P.S. I don't hate Ted Strickland. I voted for him in 2006. I hate what he's done to the state of Ohio.
UPDATE: Denial-fest commenters have proven their skill at: a. employing foul language, b. saying it ain't so with no counter-evidence, c. name-calling, and d. wishful thinking which is really best directed at wishing that Strickland were the kind of Democrat he ain't.
What I don't see in the negative comments is ANY counter-argument as to why what Ive said about the Quinn polls is in error.
As to WHY Ohioans have, with many good reasons, turned against Ted Strickland (and no, it's not because of the economy, it's entirely because Ted really is a habitual liar) see my blogroll.
It's because Ohioans have very good reason to unelect Mr. Strickland that he will not enjoy the bounce that other honest and competent Democrats will enjoy from HCR.
And for that reason, I believe that Jennifer Brunner can and will beat Rob Portman, at the same time that Strickland loses to John Kasich.
I too wish that Mr. Strickland were a different human being.