Thursday, March 20, 2008

Poll Warning: Be on the Lookout for Bad Numbers, Analysis

By Stuart Rothenberg

This looks like another excellent cycle for Democratic Congressional prospects, as the party benefits from strong fundraising, an impressive crop of recruits and a wind at its back that stems from a damaged GOP brand and voters’ desire for change.

Maybe that’s why two polls stand out like sore thumbs.

The first is a Feb. 9-10 Penn, Schoen & Berland Associates poll for Beth Hafer (D), the daughter of former Pennsylvania state Treasurer Barbara Hafer (D). The younger Hafer is running in Pennsylvania’s 18th district for the right to take on Rep. Tim Murphy (R) in the general election.

The survey showed Hafer at 22 percent of the vote in the primary, with the rest of the field in the low and middle single digits. Her “favorability” stood at 35 percent, while the candidate with the next highest favorables was businessman Steve O’Donnell (11 percent favorable).

I’ve seen plenty of polls like this released on both sides of the aisle, and I almost always think they are meaningless. In the case of Hafer, that’s especially true.

Like the early national presidential polls that showed Sen. Hillary Rodham Clinton (D-N.Y.) and former New York Mayor Rudy Giuliani (R) way ahead in the contest for their parties’ presidential nominations, the Hafer poll measures little more than name recognition.

But in this case, the name recognition almost certainly isn’t Beth Hafer’s — it’s her mother’s. Barbara Hafer ran for statewide office (and was elected) a number of times, and Western Pennsylvania voters certainly know her or her name. No wonder someone with the name Hafer was leading a field full of other candidates who had little or no name recognition.

The Hafer campaign (and the polling firm) can argue, of course, that the snapshot nevertheless says something significant about the state of the race: that Beth Hafer starts off with the advantage, and the other candidates need to increase their name recognition just to have a chance of overtaking her. That is undeniably true.

But if any of Hafer’s primary opponents run well-funded, serious campaigns for the nomination, this Hafer poll is almost meaningless. The state of the race will change quickly, making this mid-February poll quickly obsolete.

A far greater problem is a late February/ early March poll in Michigan’s 7th district. Actually, it isn’t the EPIC-MRA poll for the Detroit News and WXYZ-Action News that was the problem as much as it was the story about it that appeared in the News and bears at least some responsibility for a rash of misleading e-mails about the survey.

The newspaper story began with the very reasonable observation that Republican Rep. Tim Walberg could have a tough re-election race this year. The trouble came in the next paragraph, which noted that Walberg is in a “statistical dead heat when voters are read biographical information about him and challenger Mark Schauer.”

Three paragraphs later, the article quotes an EPIC-MRA pollster saying that he’d have “grave concerns” if he were the Congressman that an “unknown Democrat ... could match up so competitively.” The statement is particularly odd because one of the reasons Democratic insiders are excited about Schauer’s bid is that they note he already represents one-third of the Congressional district in the state Senate and is popular among those voters.

Two paragraphs after that, the article notes that when “voters had no other information than the two candidates’ names and party,” Walberg held a 51 percent to 40 percent lead.

In other words, in the initial ballot test, Walberg was over 50 percent and held a double-digit lead. This is the ballot test that really matters, and it is always reported as how the race stands.

So-called second ballots, which involve a retest of the candidates after additional information is provided to respondents, are much poorer measures of candidate strength and much less reliable predictors of the future. That’s because they introduce information into the survey that voters don’t necessarily have — and may never have.

Yet the Detroit News story led with the second ballot, creating a misimpression about the state of the race, at least as the poll found it exists today.

Interestingly, the Schauer campaign sent out a news release crowing “Schauer Leads Walberg in New Independent Poll,” even though that is demonstrably untrue. Walberg held an 11-point lead in the survey. I’d hope the folks at the Schauer campaign know the difference between a first ballot and a second ballot.

One of Schauer’s consultants distributed the article by way of an e-mail titled “Detroit News poll shows Schauer, Walberg in dead heat.” Sorry, but the race isn’t a “dead heat,” and anyone in the business should know that. All that anyone can say about the second ballot is that if and when voters get more information on the two candidates, they may find the Democrat appealing.

Actually, Walberg’s numbers in the race are far from terrible, given the problems the GOP is facing these days and given the numbers that some other Republicans are getting. He isn’t safe, of course. Not by a long shot. This is definitely a competitive race.

The bottom line here is that nobody involved in the story looks good. The Detroit News article did a terrible job of presenting the numbers and drawing implications from the survey. The EPIC-MRA pollster, assuming he was quoted correctly, failed to put the poll into context. And the Schauer campaign intentionally misled observers.


This column first appeared in Roll Call on March 17, 2008. Copyright 2008 © Roll Call Inc. All rights reserved. Reprinted with permission.