« About Mystery Pollster | Main | More on Party ID »

September 27, 2004

The Odd Specialization

I’m slowly digging out from under the many comments and emails last week. Thanks to everyone who stopped by. I’ve got a long list of questions to answer and topics to cover (though please, send more if you have them). Bear with me if it takes a bit to get to your questions.

When I wrote up who I am and what I do last week, a lot of you were surprised at the notion of "Democratic” and "Republican” pollsters. "An odd specialization,” noted commenter Shannon L. Astute emailer KW asked:

Polls conducted by unbiased sources are most trusted by the public, why would you want to proclaim yourself to be a democratic or republican pollster? Don't campaigns want authentic data too? Does this limit your employment opportunities (if I were a republican candidate, why would I hire a democrat pollster?)

Campaigns certainly want authentic data, but they almost always hire pollsters who take clients of only one party or the other, just like the consultants who produce television and direct mail advertising.

"Trust," as commenter Rob put it succinctly, is one big reason for the specialization. I’ll let him finish the thought:

To design appropriate questions for surveys, campaigns can't help but reveal something of their strategies and tactics to pollsters. Given the possibility for damaging leaks, party consistency is at least one way to be a gain a bit more comfort -- though leaks still occur.
I’ll put it more strongly. Modern campaigns, by necessity, typically reveal everything of their strategies to pollsters. Pollsters are typically part of the management team that designs the campaign’s strategy. We use the surveys to test and refine the campaign's messages as well as the attacks we expect from our opponents.

Another reason is choice. Most of us started as political activists working on campaigns. We got into polling work because we want to help our side win, so we don’t mind limiting ourselves to just one side. Does it limit other opportunities, such as corporate research, interest groups or news outlets? Probably, though that’s why Democratic and Republican pollsters often team up, like Democrat Peter Hart and Republican Bill McInturff for the NBC/Wall Street Journal poll, to counter the perception of bias.

Ultimately, we know our first job is getting the numbers right rather than spinning the results, and that is the approach we take privately with our clients.

Disclaimer: Academic studies of publicly released internal campaign polls have shown evidence of bias. The polls done by Democratic firms typically seem a few points more favorable to the Democratic candidates, the Republican polls similarly favorable to Republicans. But most internal campaign polls never see the light of day, and we don’t see those patterns in the whole of our work. As such, we believe the finding of bias occurs because campaigns are absolutely biased about which poll they choose to release. Candidates want to tout surprisingly good news and bury the unexpectedly bad. Hence, the publicly released polls seem skewed. Caveat Emptor (see this paper by University of Oregon Professor Joel Bloom for a good review of the issue).

Update: I neglected to mention that Republican pollster Todd Rehm also posted a thoughtful comment on this subject (scroll down to the end of the comments). Please note that contributions from any of my colleagues lurking out there are always welcome.

Related Entries - Pollsters

Posted by Mark Blumenthal on September 27, 2004 at 01:51 PM in Pollsters | Permalink


The comments to this entry are closed.