As candidates jostle to make the cut for the first GOP presidential debate this week, the McClatchy-Marist Poll has temporarily suspended polling on primary voter choices out of concern that public polls are being misused to decide who will be in and who will be excluded.
The Marist Institute for Public Opinion, which conducts the national survey, said the debate criteria assume too much precision in polls in drawing a line between candidates just a fraction apart, presume that the national polls being averaged are comparable, and turn the media sponsoring most of the polls from analysts to participants.
“It’s a problem when it’s shaping who gets to sit at the table,” said Lee Miringoff, director of the Marist Institute.
Fox News, which is co-sponsoring the debate in Cleveland, announced weeks ago that it would limit participation in the nationally televised prime time event to candidates who ranked in the top 10 of an average of five national polls up to 5 pm on Aug. 4.
“Such polling must be conducted by major, nationally recognized organizations that use standard methodological techniques,” Fox said.
Fox News said Sunday that it was confident people will find the debate criteria fair.
“When the results are released, everyone will see that common sense and fairness prevailed,” Michael Clemente, executive vice president of news at the network, said in a statement to Politico.
Several public national polls are being unveiled in advance of the Tuesday deadline.
The new McClatchy-Marist Poll did test general election voter choices between Democrat Hillary Clinton and each of 17 Republicans. But this time it dropped the questions about who voters would choose today in a GOP primary or a Democratic primary, preferring instead to test how much voters like or dislike each candidate.
Miringoff, who earlier issued a top 10 list of reasons not to use polls for the top 10 debate slots, said a main objection is that sponsors assume that polls are more precise than they really are. Some averages break down each number to tenths of a percentage point, suggesting an accuracy belied by the margin of error of plus or minus several percentage points.
On Friday, for example, Politico media writer Dylan Byers tweeted that one new poll by Rasmussen had pushed Gov. John Kasich of Ohio 0.5 percentage points ahead of Gov. Chris Christie of New Jersey in an average of public polls.
Several candidates at or near the cut line have been reaching for publicity, arguably in search of moving up their poll numbers to make the prime time debate stage.
“It’s a bad use of public polls,” Miringoff said. “It asks public polls to have a precision that ignores the margin of error. There’s a big distinction made where there’s no statistical difference.”
He’s not alone.
“Don’t use decimals,” said polling experts Sheldon R. Gawiser and G. Evans Witt in their book, A Journalists Guide to Public Opinion Polls. “Since these results are always subject to error, using decimals implies a precision that does not exist.”
Gawiser is the retired director of elections at NBC News. Witt is the CEO of Princeton Survey Research, a much respected polling firm.
Why sit out this time and not others?
“The first debate is perhaps the most telling. It’s the first time on the platform for everybody,” Miringoff said. “And the only woman candidate, the man who won 11 primaries last time, may be sitting in the audience,” he added, referring to Carly Fiorina and Rick Santorum.
Then there’s the averaging of five different polls.
“Do you round each poll or do you round the sum of the polls?” Miringoff asked. “Some polls are different. Some use cell phones, some don’t. It’s putting apples and oranges in a big basket.”
Finally, there’s the idea that the polls themselves are forcing candidates near the make-or-break line to strive for attention, particularly on Fox News.
“It’s making candidates change their behavior. Kasich is trying to get a big bounce. Rand Paul has a video with a chain saw. Lindsay Graham is hitting cell phones with golf clubs,” Miringoff said.
“Now the public polls are affecting the process they’re supposed to be measuring.”