My column for this week asks a tough question: What defines a "partisan" poll? A dozen years ago, the answer was easy: A poll paid for by a campaign or political party or conducted by a pollster that works for candidates of a particular party or the party itself. But with the rise of new media, the line between partisan and independent polling gets blurrier every day. Please click through and read it all.
The phrase "I could go on" near the end of the column was more than rhetorical device. While liberals often criticize questions they consider leading or biased in Fox News polls, I have often heard the similar complaints from conservatives about how polls from the other networks are conducted or reported. And if we label the polls from Fox as partisan given the point-of-view of much of their programming, should we apply similar labels to virtually all newspapers based on their editorial page endorsements?
We run into similar difficulties if we focus on the partisanship of the pollster rather than the sponsor of any given survey. As noted in the column, that line is relatively easy to draw for campaign pollsters, like Stan Greenberg of Democracy Corps or Whit Ayres of Resurgent Republic. Even PPP, which does most of its client work for Democratic campaigns, is easily classified in this way. The issue isn't one of "bias" but of partisan and economic interests disclosed: Like the campaign pollsters, PPP does most of its client work for the candidates of one political party. But what if the pollsters work is a mix of partisan and non-political clients? But how many polls for partisan clients does it take to merit the "D" or "R" label in our charts?
While these are difficult questions, they present important practical considerations for Pollster.com. We want to be a source for all available public polls, yet we also want to provide readers with sufficient information to make judgments about which polls to trust and the tools to filter out surveys they consider untrustworthy or unfairly slanted. I would like to hear from readers -- in comments or by email -- with thoughts about whether and how we can expand beyond the simple "D" or "R" labeling in terms of describing partisanship and filtering in the charts.
Follow Mark Blumenthal on Twitter: www.twitter.com/MysteryPollster