The GOP Has Their Own Controversial Robopollster

Steven Shepard

At a meeting of prominent party pollsters hosted by the Republican National Committee this spring, one veteran suggested the committee formally advise candidates against contracting with companies that used automated technology to conduct surveys.

On its face, that suggestion made sense: Automated, or interactive-voice response, polls don't call cell phones, and thus they seem to run afoul of guidelines proposed this spring by the RNC and the National Republican Congressional Committee. But automated polls don't require human interviewers, so they are much cheaper to conduct than live-caller surveys. And since survey costs are increasing rapidly -- as Americans move away from traditional landline phones and are generally less receptive to participating in polls even if they are reached -- automated surveys are being used with increasing frequency by campaigns and the parties' campaign committees. Both the NRCC and the Democratic Congressional Campaign Committee began conducting their own interactive-voice response surveys in the last election cycle as a supplement to the millions of dollars in live-caller polls they commission.

It also didn't help that the pollster who made the suggestion at the RNC pollster meeting did so while sitting only a few feet from Brock McCleary, the 2012 polling director for the NRCC who has now established his own robopolling firm, Harper Polling. McCleary didn't say a word; he didn't need to, given that the suggestion seemingly came out of left field in a context in which automated polls weren't even being discussed.

The uncomfortable incident served to underscore the tension that exists between pollsters who want to innovate and experiment with emerging techniques and those who are comfortable using older, heretofore reliable methods of survey research, a tension that is the subject of a new story I wrote for National Journal magazine, published this week and available to National Journal members. The story outlines how the RNC and NRCC have tried to mandate certain methodological standards in an effort to address some of the party's inaccurate 2012 polling -- and the degree to which GOP campaign pollsters are cooperating with the effort.

Harper's McCleary was the only pollster invited to the meetings who primarily uses interactive-voice response, or automated, phone calls. In multiple interviews earlier this year, he balanced defending his methodology -- because they must use automatic dialers, IVR surveys are banned by federal law from calling cell phones -- with the NRCC's recommendations. (The RNC protocols included a specific carve-out for automated polls.)

"The most accurate pollsters in the country [were] PPP and Purple Strategies" during the 2012 election cycle, he said, naming two prominent automated pollsters. "I'm certainly seeing that there's an incredible amount of interest" in automated polls so far in the 2014 cycle.

IVR polls have been in the spotlight this week, as the Democratic firm Public Policy Polling has been engaged in a spirited back-and-forth about its controversial methodology with Nate Silver of ESPN and Nate Cohn of The New Republic. But Silver's and Cohn's criticism of PPP has been mostly limited to unique quirks involving their business practices and methodology, not an overall debate about automated polling (though one could certainly argue that some of PPP's methodological quirks stem from the inherent limitations involved with IVR surveys).

The debate among Republicans is more basic: Should the party be supporting automated polling, even as the growth of cell phones and abandonment of landlines means those polls are reaching a smaller segment of the electorate? And is there an inherent contradiction in inviting an automated pollster to participate in meetings asking live-caller firms to increase their costs by calling more cell phones?

McCleary, for his part, supports calling more cell phones in general, a recommendation made by both committees. But he downplayed the effects that would have in increasing accuracy. "Even as the lone IVR pollster of the group, I am in support of calling more cell phones," he said. "I just think you need to have a realistic idea of what that means in terms of accuracy."

Other pollsters dismissed automated polling as a viable path in the future, as the population continues to move away from landline telephones. "The one thing about the IVR polls, that's going to be a tough industry for them, because they can always do it cheaper, but only sometimes they'll be right, because they can't do the cell phones," said GOP pollster John McLaughlin.

McCleary's participation in the process, along with the business he's started taking from the NRCC thus far this cycle, has led to some consternation among other, more traditional GOP pollsters. "It's sending a very conflicting message," said one GOP pollster who spoke on the condition of anonymity. "The NRCC is telling their polling vendors to [call] a specific number of cell phones. Then they're going around and using IVR polling from their former political director."

"You toe the line carefully here," the pollster continued. "You don't want to piss off folks and say the NRCC is just helping out one of their guys and giving them business."

McCleary stressed that the NRCC recommendations came from meetings with the pollsters themselves, and that they applied only to live-caller polls, not automated ones. "Party committees [were] doing IVR polling prior to this cycle," he said. "We still need to do the two methodologies, not in competition, but complementary to each other."

The GOP criticism of McCleary and Harper hasn't been confined to not-for-direct-attribution quotes. This spring, after a Harper poll showed GOP nominee Gabriel Gomez trailing Democrat Edward Markey by 12 points in the special election for a Senate seat in Massachusetts, Brad Dayspring, the communications director for the National Republican Senatorial Committee, said the Harper survey "might as well have been written in crayon."

Markey won by 10 points.