November 5, 2012

On History and Other Factors, Final Dispatch Poll, a Claimed Toss-up, Really Calls Ohio Prez Race For Romney; Brown-Mandel Senate Race the Real Toss-up

Filed under: Taxes & Government — Tom @ 12:47 am

A month ago, when the Columbus Dispatch claimed that its ridiculous nine-point, 51%-42% lead for the incumbent in the presidential race between Barack Obama and Mitt Romney in Ohio showed that “Romney first must turn around his own campaign,” I wrote that what the paper had really demonstrated is that the race was tied. The 43%-35% Democratic Party skew explained virtually all if not more of the Obama margin. No one believes that this breakdown reflects what we’ll see among voters on Election Night.

Today’s Dispatch poll shows Obama with a two-point, 50%-48% lead, causing the paper to declare the race a toss-up. Though its sample is less skewed (40.0%-36.7% Dem-GOP in the presidential race, 40.6%-36.4% in the presidential race), its results really foretell a Romney victory in the race for the Buckeye State’s 18 electoral votes by at least the 2.8-point margin currently shown at uskewedpolls.com while rendering the race between incumbent Democrat Sherrod Brown and Republican challenger Josh Mandel a true toss-up.

Here are the key elements of the poll’s chart:

DispatchFinal2012PreelectionPoll

Besides the obvious partisan skew, there are a few other distortions.

Percentage of early voters. Slightly less than half of those surveyed by the Dispatch said that they had voted already or said they would before Election Day. That isn’t what’s happening on the ground. As of this evening, according to Gannett, 1.6 million Ohioans have already cast ballots. Even if that figure is missing the last few days of early voting, it’s unlikely that more than 1.9 million Ohioans, or one-third of the 2008 turnout of 5.7 million, voted early. The relevance is that the early voters polled by the Dispatch favored Obama by about 13 points; this component of Obama’s support is clearly overstated. It makes sense that it would be overstated, as one would expect people who set out to vote early to be more likely to answer mail surveys like the Dispatch’s.

Regional components. Let’s look at the swings in the Dispatch poll from late September to late October:

  • Northeast (over one-third of the sample, reflecting its component of the population): From 59-34 Obama to 54-44 Obama, narrowing Obama’s lead by 15 points. If Obama really fails to come out of Northeast Ohio with more than a 10-point cushion, he’s a virtual lock to lose Ohio.
  • Central: From 51-42 Obama to 52-47 Obama, narrowing Obama’s lead by 4 points.
  • Western: From 55-41 Romney to 58-41 Romney, a three-point Romney pickup.
  • Southwest: From 55-35 Romney to 61-38 Romney, a three-point Romney pickup. Soutwestern Ohio, with less than 10% of all respondents, is probably underrepresented in each poll.
  • Northwest: From 59-35 Obama to 51-47 Romney, a 28-point swing (only about 100 respondents each time, which is not enough to conclude on, but stunning nonetheless).

The lone remaining region, the Southeast, only made up about 3.5% of each polls’ respondents. If there’s hidden GOP strength on Election Night, it will come from this region if voters from these traditionally Democratic areas are as upset as they should be about Barack Obama’s war on coal.

Dispatch Polls’ history. The Dispatch poll, as is the case with so many other polling operations, has consistently shown lower Democratic Party candidate victory margins, smaller GOP candidate margins, and has occasionally missed its call as to who would win. Examples:

  • In 2008, it called the presidential race for Obama over John McCain by 6 points. The final result was a 4.5-point Obama victory. That difference alone, if repeated this time around and even before considering other factors, moves this year’s race from a 2-point Obama victory to virtually zero.
  • In 2004, it said that the Bush 43-John Kerry contest was tied. Bush won by two points.
  • Other races down the ticket have historically shown much larger variances. Its 2010 performance in missing the statewide races by an average of four points (each time underestimating GOP strength and incorrectly predicting the Attorney General’s race) wasn’t nearly as bad as its 2006 disaster (average miss of 12.5 points while blowing the call in the Auditor’s race). This should provide some additional comfort for Mandel as he confronts the six-point deficit the Dispatch claims he faces, especially considering it overestimated Brown’s victory margin over Mike DeWine in 2006 by 12 points.

All of the above, plus the plethora of data cited by Weapons of Mass Discussion blogger Matt Hurley over at FrontPage Magazine about the vast improvement in the GOP’s get-out-the-vote effort this time around compared to 2008, point to a sufficiently convincing Romney victory in Ohio and a likely squeaker breaking in Mandel’s favor.

Share

2 Comments

  1. I had a similar view of this poll. In fact it think the D-poll is kind of useless for predicting the Ohio statewide races because the sampling does not reflect the Ohio population distribution.
    For example, only 18% of the Ohio population lives in central Ohio (Franklin and 12 surrounding counties) yet 35% of the poll responses came from there. Likewise, SW Ohio (Cincinnati and the 16 surrounding counties) has 25% of the population and only 15% of the poll responses.

    The political preference of Ohio varies greatly with region so if your sample is not representative then you results are likely to be skewed and misleading.
    I also noticed if Obama really only carries NW Ohio by 10% and as well, he loses SW Ohio by 21% like the D-poll suggests — he is toast.

    I also agree the preference percentages for SE Ohio can’t be right — as pointed out the sample is too small. In 2004, G W Bush carried SE Ohio by 53 – 47% and even McCain the weakest republican in memory carried the region by 52 – 48% in 2008. I think Romney will take Ohio — I’ll be stunned if it goes any other way.

    Comment by Fargo44 — November 5, 2012 @ 7:22 pm

  2. Regarding comment #1 above: all of that is of course correct. They’re smart enough, and have been doing this long enough at the Discharge, to know those facts better than any of us. The real point is that these issues allow them to cook the outcomes reported before the elections as part of battlespace preparation. It would be interesting to search for how many times since that original O+9 poll that their resident leftist operative, Joe Hallett, has singingly reported that the choomster was campaigning in Ohio “with the wind at his back” – he really loves that phrase, redolent, as it is, of state media praise for Chairman Mao. As Limbaugh says, “democrats with bylines.” As the former McGovern pollster, and my personal favorite, Pat Caddell says, “our current media is a disgrace and a threat to the future of the nation.”

    Comment by sad democrat — November 6, 2012 @ 10:49 am

RSS feed for comments on this post.

Sorry, the comment form is closed at this time.