Pollsters: 1/5 of Ohio Vote Already In! Me: Bull

Knowing PPP would have something wacky, I decided to cut right down to the most absurd point highlighted by the firm: 19% of respondents have already voted and they are breaking 3-1 Obama. That seems to spell certain doom for Romney.

Is it actually true? Have just under a fifth of Ohio voters actually voted?
Facts tend to get in the way of such fantasies.
CUYAHOGA COUNTY (Largest County in the State, overwhelmingly Democrat).
TOTAL REGISTERED VOTERS: 928,798 (which, FTR, is down over 180,000 from 2008, and with the registration deadline past in the state, it won't climb from here).

Of this total, 229,794 have requested absentee ballots.
Of this total, 31,233 have returned them, 13.59% of the total requested. Red flag #1.

That total requested accounts for a third of the total number of votes counted in 2008. So fractions of fractions now. Red flag #2.
Assuming lower turnout of 670,000 (closer to 2008 which ironically had fewer votes than 2004), that 31,233 drops to less than 5% of the total vote from the largest county in Ohio. A base county for President Obama and the Democrats. A core part of his Early Vote strategy.
Sure, its just one county, but it is also one of the most Democratic (and as the press meme is going, they have the advantage here), one of the more enthusiastic on voting, and will account for 1 in every 9 ballots cast in Ohio. If we were seeing the levels claimed, we would see it here, and we aren't. Flag #3.

But what about In-Person voting?
Statewide, about 60,000 Ohioans (out of around 8million registered) have voted in-person.

If we are to believe the hype that 18 to 19% of Ohians have already voted as per the surveys from Marist and PPP are claiming, where are the voters? If the Democrats are turning their ballots back in droves, where the hell are they?

Per the SOS' own press release, 1.1 million voters have requested (not submitted, as we see above, that is not the case at all) absentee ballots, and 59,000 have voted in-person. That totals to just around 20% of the 2004/2008 total vote, so on it's face the 18/19% statistic being bounced around and "found" by polling firms like Public Policy Polling could be true, except for two things: counties are still mailing out these requested ballots so it is impossible for those voters to have sent them back; and on the county level the return rate yields a number in the mid-single digits, if that. A third of what PPP and Marist's respondents are claiming.

The smaller the # of respondents for a given question/sub-question, the higher the margin of error. It is how you can wind up with polls giving a Republican 45% of the black vote: if that sample size out of the larger polled group is small, wackiness can ensue. It is why ultimately these subsamples need to be taken with a grain of salt, and not paraded about as significant when the actual numbers on the ground don't match up.

Anyone saying the race is over based on "19% said they already did and PPP said they are breaking 76-24 Obama ELEVENTY OMGZ!" is trolling, and on a very pedestrian level. Want to know how many people have voted? Check the counties themselves, or better yet, for a more concentrated spot of information, check out ningrim's spreadsheet here (requests only).

Beyond the usual smattering of poll junkies and politicos who will gobble and ooze with delight over garbage, any pollster hyping the statistic is either ignorant of the actual statistical numbers reported or they are deliberately pushing a very, very steamy pile.

But nobody would push something that obviously isn't true. That would be the work of a hack, not a reputable pollster.

posted by CAC at 04:07 AM

| Access Comments