New "Mormongate" Theories: A Reality Check

11/21/2007 09:47 pm ET | Updated May 25, 2011
  • Mark Blumenthal Mark Blumenthal is the Head of Election Polling at SurveyMonkey.

I was going to try to finish up my Disclosure Project post again today, but alas, there are more timely reports on the mysterious anti-Romney poll worth commenting on, a controversy some are now dubbing "Mormongate."

First, a summary of developments since I wrote about this story last week. A handful of conservative bloggers -- particularly Soren Dayton and Liz Mair -- have been theorizing that Romney supporters or the Romney campaign itself might be responsible for the calls. That speculation culminated in a widely cited article by the National Review's Mark Hemmingway, which argued that "evidence points" in the "general direction" of the Romney campaign as the culprit. Hemmingway's article drew a forceful response from both the Romney campaign and their pollster, Alex Gage of TargetPoint Consulting:

To set the record straight: TargetPoint Consulting has absolutely nothing to do with the calls in question. To be even clearer: TargetPoint Consulting has NEVER and will NEVER conduct a push-poll. TargetPoint is in the business of promoting Governor Romney, not manufacturing fantasy plots that involve smearing him.

Gage's denial is clear and unequivocal. However it is worth noting that nearly everyone involved, including Western Wats (the survey call center that allegedly made the calls in question) is eager to deny they conducted a "push poll." As I wrote last week, based on the reports of respondents, these calls sound more like a "message testing" poll than the classic "push poll" dirty trick (which is not a poll at all but calls that impersonate a real survey in an effort to spread a nasty message -- see my prior post). The questions worth asking, at this point, are who sponsored the calls and what did they say, exactly, about Romney's Mormonism?

Today, Dayton and Mair are asking why so many of the respondents that have come forward to report receiving the calls are either Romney supporters or paid Romney staff. The obvious explanation is that the Romney campaign directed supporters that were called to reporters. And in the latest development, TPM's Greg Sargent now reports that the Romney campaign confirms it did just that -- "referred reporters to two recipients of the calls without disclosing that the two were also on the Romney campaign payroll."

The two conservative bloggers are alleging that these ties to the Romney campaign imply something more sinister: Either a poll specifically targeted at Romney supporters in order to create a story, or perhaps an effort to get Romney staffers and supporters to lie to reporters about a non-existent survey.

For what it's worth, both allegations seem like a quite a stretch, but I want to answer one specific question raised by Liz Mair from my own perspective as a campaign pollster.

She wonders whether "including people on Romney's payroll or those publicly affiliated with the campaign in a call sample would be bad, and non-standard, practice." Not really. I cannot remember conducting any poll that attempted to exclude campaign staffers or those "publicly affiliated" with any campaign from the sample itself, as the registered voter lists typically lack any such information. Sometimes, when sampling from a list maintained by a political party, sample vendors offer the option of excluding those identified on the list as party activists or precinct captains, but that practice is far from standard.

Mair also goes on to cite a "top Republican pollster" who answers her question:

[N]ormally, people working for or associated with campaigns, and members of the media, are excluded from calling lists in the first place-- presumably because of bias that might be evident, which could raise questions about the accuracy of the result.

There may be some confusion here about what this particular pollster meant by "list." Some pollsters -- but by no means all -- begin their interview with what some describe as a "security screen." It asks if the respondent works for a political party, a campaign, a news organization, etc., with the aim of screening out such respondents from the final sample. I have always been skeptical that such screens accomplish what they aim to. My old firm typically used such screens only when a client specifically requested it. Pollsters may disagree about the merits of this procedure, but describing it as a universal or standard practice among campaign pollsters is just not accurate.

So should we be suspicious that most if not all of the respondents that have come forward are Romney supporters? I'm not sure we should. First, Romney supporters have the strongest motivation to come forward. They are likely more angry and upset and are more likely to want to report the calls to reporters or the Romney campaign. Second, as argued by Dayton and Mair, the Romney campaign has incentive to try to direct angry supporters to reporters in hope of generating a sympathetic story.

Third, we ought to think about the implications of the size of the sampled universe and the cooperation rate that pollsters are currently receiving from Iowa voters. Consider that the all time high Republican caucus turnout was little over one hundred thousand. Past caucus goers on the lists are the most active and committed Republicans in Iowa. Consider also that nearly every campaign and many different pollsters have been calling into Iowa in recent weeks, and that is on top of automated recorded calls placed by each campaign. Given that the best of surveys conducted under the best of conditions get response rates in the 20 to 30 percent range, and assuming that native campaign staffers and activists are probably the most likely to cooperate, the odds of getting a disproportionate number in the sample seems likely. The point here is simply that the odds of including a half dozen or so active Romney supporters (and even a paid staffer or two) in a sample of 600 or so Iowa Republicans do not seem terribly long to me.

So bloggers will speculate and dig further, as we always do, but I am not convinced from the facts before us allow the conclusion that pollster behind this survey intended to contact only Romney supporters.

Update: I had not seen it, but NRO's Mark Hemingway responded to criticism of his initial article today. His post includes a series of questions that TargetPoint refuses to answer beyond their blanket denial: "we had nothing to do with these calls." (via Sullivan).