Polling the Pollsters
Wednesday, September 29, 2010
The post-primary polls concerning the New York gubernatorial race left everyone confused.
The first one to come out, by Quinnipiac University, shocked most New Yorkers by presenting Cuomo with only a six point lead over controversial Carl Paladino.
A few days later, the Siena Research Institute released their own poll, showing Cuomo with a comfortable 33 point lead.
Then the Marist Institute for Public Opinion (MIPO) came out with their own...showing Cuomo with a moderate 19 point lead.
WNYC's Azi Paybarah delved into the soup last week--and today the three pollsters were guests on The Brian Lehrer Show to discuss their different numbers. Maurice "Mickey" Carroll, the Polling Institute Director of Quinnipiac University; Dr. Don Levy, Director of Siena; and Lee Miringoff, Director of Marist.
How could basically the same question asked of basically the same number of New Yorkers over basically the same few days produce such different results?
The pollsters basically said the devil is in the details. The wide gap between Siena and Quinnipiac's numbers came from differences in terminology: Siena polled "registered voters" while Quinnipiac and Marist polled "likely voters." The "likelies" are a subgroup based on a winnowing process of filtering registered voters through the sieve of questions like "How much do you care?" "How much do you support your candidate?" "How likely are you to vote?" and then putting all the data in an algorithm that spurts out a number.
Miringoff said "likely voters" polls are a more accurate measurement of the probable electorate, even if it's not entirely crystalized yet who those likely voters actually are. But, he warned that researchers should not switch from "registered" to "likely" polls too close to the general election date, because the public may mistakenly interpret the new numbers as a swing in opinion, when they really just signify a change in the polling recipe.
The other big difference in these three polls was "The Lazio Effect." Quinnipiac didn't include Lazio (who at that point was still in the race), while Siena and Marist did. If you assume Lazio supporters will all now support fellow Republican Paladino, that gives Paladino a much bigger edge.
Carroll, Miringoff and Levy also discussed the "enthusiasm gap": 51 percent of registered Republicans say they are very enthusiastic about voting in the New York state election, but only 34 percent of registered Democrats say they are.
In the end, Carroll presented some clarity: "What all these polls say is that this is a real election, which we hadn't thought before!"
What do you think? How much weight do you give polls? What do you think about the media's coverage of the numbers? Have you ever been part of a poll?