Notes on the Q Poll
The Quinnipiac University pollsters showed that they aren’t just surveyors of the political landscape today as they cleverly stretched their Connecticut polling results into at least two days worth of news articles instead of the the traditional one. On Tuesday, they dropped the news that the Governor has come off the peak of her approval numbers (though remaining high) and that Chris Dodd numbers are as stagnant as the CT economy. Today, they released the Obama/McCain head-to-heads.
There aren’t many clever ways to spin the topline numbers. The newly-minted presumptive Democratic nominee has a 21% lead over Republican presumptive nominee John McCain, with 56% saying they’d go with Obama versus 35% for McCain.
When you dig into the numbers, you don’t have to be qualified to park a remote control car on Mars to figure out what is going on. President Bush is weighing in at 19%-78% approve/disapprove - basically, John McCain is running a marathon with a dead elephant tied on his back. Even Republicans have soured on the 43rd President, with 45% expressing their disapproval.
There is something a little curious about the Q Poll though. It is widely regarded in Connecticut political circles as the “gold standard” polling outfit. What they say matters, mostly because they are the only ones talking in the Nutmeg State. But if you take a look in Pennsylvania, Florida, and Ohio, you can see a pattern that seems to reflect a roughly 3-5% ‘Quinnipiac bump’ for Barack Obama.
In Pennsylvania, Real Clear Politics has the RCP Polling Average at Obama +7.7 over Sen. McCain. Q’s last poll had Obama up by 12 points, though Rasmussen has it at just +4. Since February 19, RCP has the results of 28 polls posted - Quinnipiac has done eight of them. The numbers have been all over the yard - some with McCain up, some with Obama up, others having the two Senators at basically even. But all eight Q polls have had Obama up.
Ohio is a different state and a different side of the same coin. The Buckeye State seems poised to be Waterloo once again for one of the two candidates and there have been 24 polls since August 2007, according to RCP. QU has done seven polls, and four have had McCain up - but in six of the cases, the numbers have been just enough different from the other pollsters to be interesting. While other polls showed a bit of a rise for McCain in late March, Q’s did not. They are consistently lower than the others.
Florida is a lousy comparison because most of the other pollsters are using much smaller sample sizes. But if you do compare the two most recent that have similar samples - the other poll being a Strategic Vision poll - you see that QU is Obama +4 while Strategic Vision is McCain +8.
At the end of the day, this is going to probably be a difference in QU’s polling model - perhaps trying to model the anticipated wave of newly-registered Democrats. Regardless, in the context of the current case, 3-5 points has only a minimal impact on 21% worth of spread. But it is worth keeping an eye on.
It is also important to take the numbers seriously, but not too seriously. It is barely July, the extent of the political battles between the two campaigns haven’t - for most normal people - even registered on the radar screen, and the ground-pounders aren’t out in force yet. It will come.














July 3rd, 2008 at 1:27 am
The big debate is whether to reweight voters’ party identification to match the actual registration numbers or to simply take the party ID offered by the sample as the correct weighting.
Quinnipiac appears — in Connecticut at least — to be using the most recently published voter registration statistics (Oct 07) to weight their samples. That is, Republicans were 21.2% of the voters in Oct 2007, and they are 21.2% of the sample in the handful of questions I grabbed from their limited crosstabs.