Last week’s Franklin and Marshall College Poll raised some eyebrows (and some hackles) and is worth a closer look.
It was positive for Democrats, a lot less so for Republicans.
It said, for starters, that Gov. Wolf leads former state Sen. Scott Wagner by a whopping 19 points: 48-29.
Wagner’s campaign went a bit crazy: The poll is inaccurate, misleading and “a joke,” said Wagner spokesperson Andrew Romeo, who then stretched it a bit, labeling its findings the work of “left-wing pollsters.”
I guess, if your candidate is aggressive and Trumpy (and he is) your campaign should be, too — #FakePoll.
No surprise. Bad numbers hurt fund-raising. Best to raise holy hell and, at least, muddy the waters.
And, normally, I’d ignore the common practice of campaigns on the short end of polls complaining about polls. But this complaint is interesting — and instructive regarding the nature of polls.
Wagner’s camp issued a memo blasting the F&M poll for, among other things, the party breakdown of those questioned.
The poll is of registered voters, regardless of affiliation. It asked this question: Do you consider yourself liberal, moderate or conservative?
It found 27 percent liberal, 40 percent moderate, 33 percent conservative – pretty representative, in my view, of statewide ideology, and findings that don’t much vary in recent years.
The poll also asked a “self-ID” question: regardless of your registration, do you think of yourself as Republican, Democrat or independent?
Findings showed 51 percent think of themselves as Democrats, 41 percent Republicans, 6 percent independents. So, a 10-point edge for Democrats.
The actual major-party split, using current voter registration data, shows 48 percent of all registered voters are Democrats, 39 percent Republicans, a nine-point edge for Democrats.
Wagner’s camp isn’t arguing that a 9-point edge for Democrats is much different from a 10-point edge for Democrats. It is arguing that when people vote, especially in midterm elections, there’s nowhere near a 9- or 10-point difference in turnout between party voters.
It says the largest turnout disparity in the last three midterms was only a 2.3-point Democratic edge.
Therefore, using turnout in past elections produces a more accurate poll.
Wagner’s national pollster, McLaughlin & Associates, has Wagner trailing by seven points. Wagner’s campaign declined my request to view that poll.
Now I’d note candidate pollsters are paid (often dearly) by candidates. It’s not uncommon to see results more favorable than in independent polls.
And the difference here, according to F&M poll co-director Berwood Yost (he and Terry Madonna have done Pennsylvania polls for 27 years), is that the F&M poll was of registered voters, not likely voters, was not based on past turnout, and reflects only this moment in time.
“We’re not saying who’s voting,” Yost tells me — meaning that predicted or expected turnout, often a dicey element in polls (just think 2016), wasn’t a factor.
You can decide whether, nearly five months from an election, it ought to be.
Meanwhile, the same poll found Sen. Bob Casey with a 17-point lead (44-27) over U.S. Rep. Lou Barletta.
Barletta’s campaign issued a statement: “There’s a lot to be optimistic about in this poll for someone challenging a two-term incumbent senator who has been in statewide office for two decades. … As voters get to know Lou and his record of fighting for working families, this race will continue to tighten and Lou will win.”
No complaints about the poll.
Polls are a huge part of politics. There are good ones, bad ones, middling ones.
(ABC’s “FiveThirtyEight” grades hundreds of pollsters based on historical accuracy and methodology. It gives F&M a B- and McLaughlin a C-.)
And one thing to consider in any poll is “undecided” voters. In this instance, F&M says that’s 23 percent in the governor’s race, 28 percent in the Senate race.
That, and big Democratic leads, suggests Wolf and Casey, at least in name, are known to voters. Wagner and Barletta not so much. Hence Barletta’s “optimistic” outlook, a contrast to Wagner’s pugnacious reaction.
Finally, I’ve often questioned the value of horse race polls months away from elections. Do they predetermine outcomes? Do they (further) discourage voter interest? Do they help or hinder democracy?
Maybe we should take a poll.