There are tons of aspects of the two new polls released by the Maine People’s Resource Center (for which I work) and Public Policy Polling this past week that bear discussion, and there’s likely to be at least one more poll out this week. I’m currently planning a series of posts on what they mean for the races in Maine.
First, however, I’d like to delve a bit into one specific point: the discussion and criticism over party identification as represented in political polls. This is top of mind for me because of the reactions from two campaigns to MPRC’s latest results.
Lance Dutson, campaign manager for Republican U.S. Senate candidate Charlie Summers, tweeted on Wednesday that the “MPRC poll oversampled Dems, likely an effort to divert publicity from PPP numbers.”
The Kevin Raye for Congress campaign took things a step further, sending out a press release and blast email attacking what they said was a “deceptive poll vastly oversampling self-described Democrats, designed in a way that inflates the perceived strength of Democratic candidates in the upcoming election.” Raye campaign consultant Kathie Summers-Grice also told the MaineToday papers that not weighting by party identification to registration or turnout numbers was “preposterous” and “defies all logic” and made similar claims to the Bangor Daily News.
The MPRC poll did have the percentage of self-identified Democrats 10.5 points higher than the percentage of self-identified Republicans. This is not unusual, and the difference between these numbers and statewide party registrations is due to the fact that these are fundamentally different measurements.
The party that people identify with when asked about their preference on the phone is a fluid thing and often bears little relation to what box they checked on their voter registration card. The makeup of the electorate that actually votes is also different from registration percentages and changes from election to election.
As the Pew Research Center recently put it in report on party identification in political polls:
“While it would be easy to standardize the distribution of Democrats, Republicans and independents across all of these surveys, this would unquestionably be the wrong thing to do[…] In effect, standardizing, smoothing, or otherwise tinkering with the balance of party identification in a survey is tantamount to saying we know how well each candidate is doing before the survey is conducted.”
Harry Enten’s piece in The Guardian from last week on national polls and party ID is also instructive, as is Pew’s chart of changes in partisan identification since 1939 and this much more granular chart of polls over the past few years from Pollster.com. One look should tell you why weighting by party is a sucker bet.
But I don’t actually have to make an academic argument to prove that the complaints from the Raye and Summers camps aren’t sincere.
Both campaigns cited the PPP poll as a better example of voter opinion in Maine. Less than half an hour after his tweet, Dutson even sent out a fundraising email touting PPP’s results. The PPP poll had a 13-point gap in favor of Democrats, which I think shows rather clearly that their objections are political rather than factual. I would also note that the Raye campaign cited MPRC’s last poll, which had similar party ID numbers, in a July press release as evidence that they were gaining momentum.
A good way to judge whether a party ID gap is legitimate is to examine how well a polling organization’s results line up against actual electoral outcomes. As former LePage administration communications director Dan Demerrit noted in his column over the weekend, MPRC has an excellent track record:
“I tend to believe the Maine People’s Resource Center, or MPRC, has the more accurate findings. Over the last two years they have polled eight contests and, without fail, their findings among the published public polls have been closest to the actual results.”