We close the week in Massachusetts, where a new PPP automated shows a close race and a student group at Emerson College is getting into automated polling. The new entrants are likely to fuel yet another debate about automated polling, party weighting and the seemingly futile search for best practices in the polling world. This is the HuffPost Pollster update for Friday, May 3, 2013.
PPP SHOWS CLOSE RACE IN MASSACHUSETTS With Luke Johnson: "Rep. Ed Markey (D-Mass.) enters the general election to fill John Kerry's Senate seat with a relatively slender 4-point lead over Republican rival Gabriel Gomez, according to a poll released Friday. Markey is supported by 44 percent of likely Massachusetts voters, while Gomez is backed by 40 percent, the Democratic firm Public Policy Polling (PPP) found...While Massachusetts' electorate is heavily Democratic -- President Barack Obama has a 53 percent approval rating in the state -- the PPP poll found Gomez attracting some cross-party support from Democrats and independents, making the race more competitive." [HuffPost,Pollster chart]
STUDENT GROUP POLLS MASSACHUSETTS SENATE RACE - National Journal polling editor Steven Shepard: "The first post-primary poll in next month's Massachusetts Senate special election was released Thursday, but while the survey carried the name of a prominent Boston university, it wasn't conducted by the school or its faculty. The automated poll was conducted by a newly-reinstated student group on campus...Most university polling outfits -- like Quinnipiac and Marist -- are run by the colleges and led by faculty...The Emerson College Polling Institute, on the other hand, is headed by president Felix Chen, a junior at the school. 'Last year, I took a class, and the professor talked about how to do research, how to do public opinion polling,' Chen told Hotline On Call on Friday." [National Journal]
Also shows Markey leading - Their survey gives Markey a six-point lead over Gomez (42 to 36 percent). Interviews of 793 voters were conducted on May 1, 2013 using an automated, recorded voice methodology that called voters in landline telephone households selected from a list of registered voters. [ECPS release, results, methodology]
Republican advisor - More from Shepard: "The Emerson students do have a faculty adviser in Republican consultant Spencer Kimball, who said he teaches full-time at the college. 'The students have basically learned the skill sets to be able to write, produce and analyze the polling,' Kimball said. He described his role in the latest poll as relatively hands-off. 'The students wrote it and analyzed it,' said Kimball. 'I read over their press release for spelling and grammar.'"'
Reached by HuffPollster, Chen described Kimball as "an advisor" who the group consults "if we run into some polling problems." He stressed, however, that students were responsible for "all the polling, all the results, all the press releases...so I think we're really non-partisan. We're trying to give the best results to the public."
Who pays? - While automated polling is inexpensive, there are non-trivial costs. Chen says they purchase samples from Aristotle International (the firm used by PPP and others). A vendor, who Chen did not disclose, performs the automated calling. Who covers those costs? "Our [Communications studies] department supports the student organization," Chen says, "and we also have alumni supporting. I feel like they are really passionate about the hands-on approach of the students."
Party weighting?- Shepard: "Chen said the survey was weighted by gender and party-identification (a controversial practice). And he acknowledges that the automated methodology the poll uses undersamples younger voters, something he says is affecting all pollsters." [National Journal]
Strictly speaking, the student survey asked voters about their party registration (whether they are registered to vote as Democrats, Republicans, independents or unaffiliated), not their party identification (how they think of themselves). Another Massachusetts pollster that has asked both questions on the same polls finds "between 20% and 24% of each [registered] party group members do not identify as members of that party." [MassIncPolling]
As of 2012, the official breakdown of all Massachusetts voters shows 36 percent are registered as Democrats and 11 percent as Republicans, with 53 percent unenrolled. The Emerson student survey reports a party breakdown that is 43 percent Democratic, 15 percent Republican and 42 percent "independent." [MA Sec. of Commonwealth]
Their weighting target? Chen explained that the students compared their results to the party identification estimates for individual states published by Gallup, and concluded that their own unweighted number for Democrats was too low. Chen: "We just realized that all of our poll is based on landline and there is a Republican bias for landline to do the survey, so we added 6 percent of Democrats into the party identification." [Gallup]
Unfortunately, as Shepard implies, pollsters are divided on whether to weight by party and, if so, how. Some survey researchers will howl at the methodological choices made by the Emerson polling group, but then, most would have similar complaints about PPP and other well-known automated polls. We're a long way from consensus about best practices in the world of modern pre-election polling.
FRIDAY'S 'OUTLIERS' - Links to more news at the intersection of polling, politics and political data:
-Ken Cuccinelli's campaign "prebuts" a still-in-the-field Washington Post Virginia poll. [Examiner]
-Harry Enten notes that Republicans are now more civil libertarian than Democrats. [Guardian]
-Frank Newport finds few signs that sequestration has affected most Americans. [Gallup]
-Ken Walsh profiles Joel Benenson's Obama campaign "Ethnography Project." [US News]
-Arizona pollsters BRC, using a non-standard approval scale, find a net negative approval rating for Jeff Flake. [BRC]
-Harvard University political scientist Ryan Enos asks journalists and politicos to complete a five minute survey. [@RyanDEnos]