Wednesday, October 31, 2007

UI Pollster Stands By Methodology

UI Pollster Stands By Methodology

Reaction to Monday's release of the latest University of Iowa presidential poll has focused almost as much on polling methodology as on the results.  But David Redlawsk, the UI political scientist who conducted the poll, stands by the screen of likely caucus goers that critics called too broad.


"We think the critics are missing what we are doing," Redlawsk told Iowa Independent, "but we think that's mostly our fault. at some point I said -- and I think it is in the press release this way too -- that Edwards supporters are more likely to caucus than Obama's."


The poll (complete results) showed a close race on the Democratic side, with Hillary Clinton at 29 percent, Barack Obama at 27 percent and John Edwards at 20 percent.  The Republican results showed a strong lead for Mitt Romney at 36 percent, with  Rudy Giuliani and Mike Huckabee at 13 percent and Fred Thompson at 11 percent.


Redlawsk explained the poll's three-step screen of likely caucus goers.  "First, we ask people if they caucused in 2000 or 2004. Technically this isn't done as a screen, but it allows us to understand the real likelihood of folks caucusing this time around, since probably the best predictor of caucusing is having done it before.


"Second, we ask people to indicate their likelihood of caucusing this time on a 4 point scale, very, somewhat, not very, and not at all," Redlawsk continued. "We drop the not at alls completely - they're done. We then ask the remaining three levels -- 1's, 2's, and 3's -- which party they plan to caucus for. Anyone saying 'don't know' or some party other than Republican or Democratic is also dropped -- if they can't name the party, they're probably not caucusing."


Redlawsk said at that point the 1's and 2's are considered "likely" caucus goers, and the 3's are "potential" caucus goers.


"Edwards' supporters are more likely to be 1's -- in the VERY category, while Obama has more 2's," he said


The poll did not look at the possible impact of different caucus dates.  "At the time we were in the field, the Republicans had set Jan. 3 and the Democrats had not yet moved," said Redlawsk.  "We simply did not have time to query the Democrats on different date options, since the survey was already too long.  And frankly, I assumed the Democrats would end up on the 3rd in any case."


Plenty of people say they are still "somewhat likely" to change their mind but fewer than 10% said they are very likely to do so, said Redlawsk. "Republicans though are noticeably less firm than Democrats."

No comments:

Post a Comment