I will remind, the Columbus Dispatch poll showed the race in Ohio at 45/45 (tie) the last time around.
I have a problem with the Columbus Dispatch poll, because it is a mail-poll. That being said, The Columbus Dispatch, which has been polling issues and events in Ohio since 1916, picked the winner in 6 of the last 7 presidential cycles (2008, 2004, 2000, 1996, 1992, 1988, 1984) and it showed a statistical tie in 2004. So, in this time frame, it has a 6 for 7 accuracy record.
In 1984, the Columbus Dispatch poll from September had Reagan up over Mondale by +16. I am still looking for the final poll results (they are behind an Elyria Chronicle paywall). Reagan won by +18.76 in Ohio (almost identical to his national margin, as is often the case with Ohio). If the +16 holds, then the final poll was off by 2.76 points, similar to 1996, 2004 and 2008
In 1988, the Columbus Dispatch poll from October has Bush 41 up over Dukakis by +4 (47 / 43). In the final Columbus Dispatch poll in November, Bush 41 was up over Dukakis by +6 (53 / 47). Bush 41 won Ohio by +10.85%. The poll was off by 5 points.
In 1992, the final Columbus Dispatch poll had Clinton up over Bush 41 by +1.5. Clinton won by +1.83%. The poll nailed the results in that year.
In 1996, the Columbus Dispatch poll from September had Clinton up +9 over Dole (just as the current one has of Obama over Romney), 52 / 43. I have yet to find the data for the end poll, but the NYT from Nov. 1 1996 reported that the margins for the last poll stayed at +9. Clinton won Ohio by +6.36% in 1996. The final poll was therefore off by 2.6 points, very similar to 2004 and 2008.
In 2000, the Columbus Dispatch poll from the beginning of September had Bush up +6 over Gore. At the beginning of October 2000, the Columbus Dispatch had Bush at +5 over Gore. The final Columbus Dispatch poll from 2000, at the beginning of November, had Bush at +10 over Gore. Bush won Ohio by +3.51%, so, though the poll called the winner, it overstated Bush's win by 6.5 points.
In 2004, the Dispatch's last poll was on 09/22, just a week before this one came out in 2012. It showed Bush up +7 over Kerry. Interestingly enough, the poll BEFORE that one in 2004 from the Columbus Dispatch, showed a 46/46 (tie). Sound familiar? The final poll from the Columbus Dispatch in 2004, shortly before election day, was a 50/50 tie. Bush won OHIO by +2.11%, so the poll was off by 2 points.
In 2008, around this time (10/05), the Columbus Dispatch had Obama up by +7 over McCain, but the poll before, from 08/22, had McCain up +1 over Obama. The Dispatch put out a later poll that it did not put out in '04, on 11/01/2008, showing Obama up +6 over McCain. Obama won Ohio by +4.58%, so the poll called the winner correctly, but was off by 2.4 points.
This means we have historical precedent of the Columbus Dispatch end-poll being off by around 2 points for 3 of the last 5 presidential cycles. But it correctly called the winner outright in four of them and had an absolute tie in 2004.
From the article, some important information to take away:
"A new Dispatch Poll shows him trailing President Barack Obama in bellwether Ohio by 9 points, 51 percent to 42 percent.
A surge of Democratic support for Obama has transformed the race since the first Dispatch Poll had the two dead-even at 45 percent just before the Republican National Convention in late August.
The survey is the fifth major poll — from The Washington Post to Fox News — of Ohio voters in a week to show the president ahead by 5 to 10 points. He also leads in surveys of most of the remaining swing states.
Obama’s rise comes at an especially fortuitous time for the Democrat: Ohioans begin casting early ballots in two days...
...Experts say as many as 40 percent of Ohio voters will cast an absentee ballot before Nov. 6.
For the first time in state history, most of the state’s 7.8 million registered voters have been mailed an absentee-ballot application, and they’ll get another in early October. As of last week, more than 723,000 already asked for an absentee ballot."
That is it in a nutshell: as in 2008 and 2004, when both Obama and Bush (respectively) reached the apex of their numbers in the Columbus Dispatch polling, Obama may have reached his apex her, with +9. But the kicker is that early voting has now started, which means those votes will be "locked-in" for the respective candidate. And it will be massive early voting this time - MORE than in 2008. This means that there could be no better time for the President to enjoy early voting in Ohio than right now. Believe me, the Romney is surely wishing that he was up at least +4 or +5 right now, for then they would enjoy the advantages of early voting. The other kicker is that Obama is over the 50% mark.
Here was the breakdown of the poll (more text afterward):
Now, before anyone screams about the partisan breakdown of this poll, listen to what the Dispatch says:
"In August, almost exactly the same number of Democrats and Republicans responded to the Dispatch Poll. But after the mail-poll ballots went out this time to registered Ohio voters chosen exactly the same way — at random by a computer — more Democrats returned the poll forms than did Republicans. The breakdown: 43 percent Democrat, 35percent Republican, 20 percent no party affiliation, about 2percent Libertarian and less than 1 percent combined for the other parties officially recognized in Ohio: Constitution, Green and Socialist.
Republicans have complained that, in recent Ohio history, Democrats have seldom if ever enjoyed such an advantage at the ballot box indicated by recent polls. They are correct.
If the 2010 Ohio governor’s race had seen the same partisan breakdown, Democrat Ted Strickland would have been narrowly re-elected instead of losing by 2 points to John Kasich, the new poll indicates.
In the final 2008 Dispatch Poll, a Democratic edge of just 4.5 percentage points among poll-takers helped Obama to a survey lead of 5.4 points; the actual result was Obama by 4.6 points.
But whether there’s a serious, across-the-board breakdown among several independent state and national polling organizations, or that simply more Democrats are planning to vote than Republicans, there’s no way to interpret the results as good for Romney."
Further muddying the waters is that Ohio does the most unique voter registration classification in the world. Voters who register for the first time and even give party identification are listed as "unafilliated" until they have actually voted in an election. From GE to GE, they must re-register. So, Ohio's VR stats look like this:
|State VR||DEM||GOP||IND||NO AFF||Total||DEM %||GOP %||IND %||NO AFF %||Margin||Mar %|
|Info from OHIO SOS 06/15|
I also discussed this in the voter registration thread from June, 2011, which you can read HERE or HERE.
So, whether you want to question the demographic / political make-up of this poll, it brings me to one of my beefs with mail-in polls. You mail them out and what comes back is what you get. Then again, when you do telephone interviews with people and 85% of them hang-up, you also get what you get...
It is however, telling that Republicans who were willing to respond in August were suddenly not so interested this time around. That is not exactly a ringing endorsement of Mitt Romney. And very untypical of the respondents to the Columbus Dispatch poll.
The main point is that this poll for Ohio has been far more on that it has been off. And to my knowledge, a +9 is the highest margin a Democrat has ever gotten in this poll since Bill Clinton in 1996. I have not yet researched farther back than 1984, but as time allows, I will add previous presidential cycles to this study.