What makes a poll or pollster "partisan?" It's an important question for poll watchers that's getting harder to answer.
Many news organizations avoid reporting on polls sponsored by political partisans, for reasons that are a matter of common sense. "Be wary of polls paid for by candidates or interest groups," the AP Stylebook warns; "their release of poll results may be done selectively and is often a campaign tactic or publicity ploy."
The National Council on Public Polls expands on that advice, explaining that when an interest group sponsors a poll, its sponsorship "may have swayed the question wording, the timing of the poll, the group interviewed and the order of the questions." The council recommends that journalists "carefully examine the poll to be certain that it accurately reflects public opinion and does not simply push a single viewpoint."
For all those reasons, our colleagues at The Hotline have long labeled partisan polls -- those sponsored by campaigns, party committees or pollsters who do most of their work for candidates -- with a "D" or "R" to designate their affiliation.
Those labels are important. My colleague Charles Franklin, a political scientist at the University of Wisconsin, studied publicly released horse race surveys from the 2000 and 2002 elections and found that polls identified as partisan by The Hotline tended to skew in favor of their candidate by about 2.9 percent and against their opponent by roughly the same amount.
But this straightforward partisan-or-not labeling that was easy to apply a decade ago, when most "partisan" polls in the public domain were paid for by either a candidate or party committee, is getting much more difficult.
The first wave of change comes from partisan pollsters who release their own surveys without formal campaign or party sponsorship. The best examples are the polls conducted by Democratic pollster Stanley Greenberg under the auspices of an organization known as Democracy Corps and, more recently, those conducted by Republican pollster Whit Ayres for the group Resurgent Republic.
Public Policy Polling, a firm that conducts automated, recorded voice polls and works primarily for local Democratic candidates, adds a slightly different twist. It routinely conducts and releases polls into the public domain for marketing purposes, surveys it pays for out of its corporate revenue without sponsorship by any outside group.
All of these pollsters clearly have partisan affiliations and Democracy Corps and Resurgent Republic receive partisan funding, so it is easy to label their surveys with a D or an R. Still, they typically release complete surveys, rather than selected questions, and both Democracy Corps and PPP release surveys on a more routine basis. "For this new breed of partisan pollster," Franklin notes, "there may be a 'house effect' in favor of their side, but it is much less due to selective release than is the case with campaign polls."
Those are the easy cases. Consider some of these tougher examples:
• Rasmussen Reports: "Because we value our independence and credibility," its Web site says, "Rasmussen Reports cannot be hired to conduct a poll for anyone." Yet the Democratic-affiliated site Think Progress found that Scott Rasmussen, the firm's founder and president, was paid $141,000 in 2003 and 2004 by the Bush campaign and the Republican National Committee for items listed as surveys or voter data.
In an e-mail, Rasmussen insists any candidate work has been handled since 2003 by a separate "field service" known as Pulse, but he concedes that early on "some of the revenue went to Scott Rasmussen, Inc." and "one or two commissioned polls may have been released under the Rasmussen Reports brand name." He adds that since 2003, Pulse has done roughly the same amount of work for Republicans as Democrats.
Still, Democrats such as Illinois Senate candidate Alexi Giannoulias now attack Rasmussen as "pro-Republican," citing political observers who see big house effects favoring Republicans in Rasmussen's polls. So should they be labeled as partisan?
• Research 2000: Not long ago, Research 2000 conducted the vast majority of its public polls for newspapers and local television stations. But about three years ago, the company started polling for the liberal Web site Daily Kos. Now, according to Research 2000 President Del Ali, roughly 80 percent of the publicly released polls conducted by the company are for Daily Kos or other progressive sites or organizations like the Progressive Change Campaign Committee.
Daily Kos is a Web site with a distinctly liberal point of view that actively promotes specific candidates, but it is not a political action committee. Should we consider it partisan? And should we label Research 2000 as a partisan pollster like Greenberg or PPP given the volume of work it does for progressive Web sites and organizations?
• Fox News/Opinion Dynamics: If we consider Daily Kos or Research 2000 partisan, then what about the polls done for Fox News? It maintains a similar relationship with a theoretically independent pollster, yet Fox News programming has a point of view that is frequently as conservative as Daily Kos is liberal. And as critics like bloggers Nate Silver and Steve Benen frequently argue, their questions often show a similarly conservative slant. So should we label them as partisan?
I could go on, but the point should be clear: The line between partisan and nonpartisan polling gets blurrier every day.