«

»

Nov 03 2012

Print this Post

Marist/NBC Polls & Other Fantasies

Share to Google Plus
Truth Has No Agenda…

There are those who look at things the way they are and ask why. I dream of things that never were and say why not.” ~Robert F. Kennedy

Who would ever believe that Kennedy’s stirring words would be subverted into the operating principle for a nationally recognized polling firm, Marist.

Indeed, after the two most recent state polls that Marist has conducted in conjunction with NBC News and the Wall Street Journal, the Romney campaign should consider a complaint to the Federal Election Commission (FEC) for an undeclared, in-kind contribution to the Obama campaign.

The fresh angst is generated by new polls by Marist in Ohio and Florida, showing President Obama leading, and in the case of Ohio, significantly. The conclusions are – simply stated – absurd, but give voice and hope to Obama backers, and perhaps wavering undecides, that POTUS still has the edge. However after a review of the polling internals the survey results simply collapse under their own weight.

Ohio

Marist states that President Obama leads Mitt Romney by an impressive six points, 51-45 percent.

But look at the Marist sample.

Marist’s model has +9 more Democrats in its sample than Republicans. But in 2008 wave election, as noted by @Numbersmuncher on Twitter,  the Democrats won nationally by +8, and the President won Ohio by only +5 more Democrats than Republicans.

Thus, for Marist’s poll to be correct, Democrats will have to more than double their turnout on Tuesday, from a baseline that was by far one of the best Democratic years in decades.

Does anyone believe that is in the cards?

Marist’s poll further states that 35 percent of Ohioans have already voted, and that the breakdown is 62-36 percent in favor of President Obama.

But if you look at the actual county tallies from Ohio as compiled by former operative and political analyst Adrian Gray a total of 979,170 early votes have been cast in Ohio, 54 percent for registered Democrats, 46 percent for registered Republicans – shaving a 26 point margin by Marist down to just 8 points based on actual data.

What’s more, the current early Democratic balloting in Ohio is down 181,275 from 2008. In contrast, GOP early balloting is up 76,000 from 2008. Add it together and the margin, 257,000 odd votes almost wipes out President Obama’s 262,000 vote victory margin in Ohio in ’08.

This leaves Marist’s last bit of data on Ohio. Marist projects that the Ohio voters who will turn out on Election Day (65 percent) plan on voting for Romney 52-42. Thus, if you correct for Marist’s flawed methodology and prodigious oversampling of Democrats in Ohio, and cut the Marist predicted Election Day advantage for the Republican nominee in half, Romney is headed to a solid win in Ohio.

Florida

The same problems dog Marist in Florida, but to a lesser degree.

Marist has Obama up by +2 in its poll, 48-47 percent. The Marist sample is D+2. As noted previously, in 2008, President Obama won nationally D+8. In Florida, Obama won by 3 points with a partisan make up of D+3. So again, Marist’s model forecasts a Democratic turnout in Florida close to the epic wave election of 2008.

As noted in my post yesterday, Gallup’s likely voter cohort for the electorate is actually R+1 to R+3. There is simply no way that the turnout on Tuesday is going to mirror 2008.

Marist claims that 63 percent of Floridians have voted, with Obama taking the early vote 53-46 percent.

Again, looking at the actual count based Adrian Gray’s analysis, Dems won the early vote in 2008 by 8 percent, close to Marist’s margin in the poll. But according to Gray, the actual Democrat lead now is only 2 percent. That 6% gain for the GOP represents 270,000 votes in a state Obama won by 236,000 in 2008. If the percentages hold over the weekend, Obama’s margin in Florida is already gone.

According to Marist, of the 37 percent of Florida voters who will cast ballots on Election Day proper, Romney is favored by a staggering 12 points, 52-40 percent.

So, as with Ohio, if you correct for Marist’s flawed methodology and Democratic over sampling, and cut the Marist projected Election Day Republican advantage in half – just to make it appear more reasonable, Romney is still on his way to a very solid victory in Florida.

In sum, sadly, coin flippers will have a better chance of getting Florida and Ohio right than Marist does with this flawed methodology and sampling.

We are going to be inundated with polls over the next 48 hours – most trying to capture a sense of momentum for one candidate or the other. Pay attention to Rasmussen, while the polls have swung in all directions since the conventions, his have been remarkably stable, and seem to have captured the true state of the race. Rasmussen was spot on with its prediction of the popular vote in ’08, and within one percent in ’04. He is credible.

One last general point as we roll toward the final 60 hours before Election Day.

There has been a lot of concern about a POTUS bump due to the President acting, well, presidential, in his tour of Hurricane Sandy ravaged NJ. Karl Rove even thought there would be a bump for Obama.

But Obama is gone and back on the campaign trail. The northeast doesn’t have any swing states.

Yet the area of devastation is daunting and the conditions of people in Staten Island, Brooklyn, lower Manhattan and up and down the Jersey coast are growing more desperate as weather gets cold, and fuel supplies, food and water literally run out.

POTUS got a bump from a photo op, but the reality of the damage to the tri-state area makes Sandy look very much like Obama’s Katrina.

That’s not going to help the President as voters go into the voting booths on Tuesday. Neither is using your vote as “revenge,” but that’s an entirely different blog post.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>