Pennsylvania has not voted for a Republican presidential nominee in 24 years, when George H.W. Bush carried the state narrowly in 1988. In subsequent cycles, the Keystone State frequently felt like 'the one that got away' for Republicans; polling would look close and tightening in October, only to go blue in November, thanks to Democrats' formidable firewall in Greater Philadelphia. This year, it appeared that Pennsylvania wouldn't even be in the conversation for the GOP ticket. Barack Obama won the state by double-digits in 2008, and numerous public opinion polls showed the president maintaining -- or even expanding -- his advantage. Indeed, a CBS/NYT/Quinnipiac poll released at the tail end of September indicated that Obama led Mitt Romney by a whopping 12 points, 54-42 percent. Game over, right? Perhaps not. A string of October polls suggests the president's lead is tenuous, sitting within the margin of error.
The first pollster to detect this dynamic was Susquehanna Polling and Research, which measured a two-point, margin of error race in the state last month. When the firm released its September data, outside observers dismissed it as an outlier and moved on. Chief pollster Jim Lee penned a compelling memo defending his methodology and assumptions, but it garnered little notice. When a separate Susquehanna poll, conducted for a media client, produced a nearly identical result this month, a few more eyebrows perked up. Then came a perception game-changer: Sienna College's published a presidential poll showing three-point race, within the margin of error, with many undecided voters still making up their minds. And, crucially, the majority of the survey was conducted before last week's presidential debate. Could Pennsylvania be much more competitive than the laugher many anticipated? I spoke with Susquehanna's president and founder, Jim Lee, about his organization's findings. He doesn't see a massive or unexpected Romney surge in Pennsylvania; he sees terribly flawed polling assumptions from some of his competitors.
Susquehanna is a Harrisburg-based, Republican-affiliated polling firm founded in 2000. They serve numerous campaigns as well as media clients. "We only really poll for Republican candidates, but the media uses us and respects us because we're accurate," Lee says. "In our public polling, we call them like we see them, based on the data," even if it means bad news for the GOP. The outfit's track record over recent cycles has been fairly impressive. In 2008, their final survey showed Barack Obama leading John McCain "by around eight points," Lee says. The Democrat won by ten. In 2010, Susquehanna slightly underestimated the margin of Republican Tom Corbett's victory, but nailed the high-profile Senate race. "We were the only pollster who got that one right. Everyone else had [Republican Senator Pat] Toomey leading pretty comfortably, but our data showed that the race was getting closer. I even presented our information to Toomey, who gave me a lecture about polling methodology. We had better information than his internals," Lee recalls. An aggregation of polls from RealClearPolitics at the time showed most surveys projecting a 5-7 point win for Toomey. Only Susquehanna anticipated the late tightening and correctly predicted Toomey's slender two-point win. How did Lee and his team pull it off? By studying historical data and poring over a huge volume of state-level data. Based on this information, they determined a likely Democrat vs. Republican turnout model, then weighted their poll samples accordingly. They've done the same this year, concluding that Pennsylvania's electorate will likely be in the D+6 range in 2012, two points less Democratic than 2008.
"Our polling has been validated and vindicated because our D-to-R ratio is much more appropriate than other pollsters'," Lee explains. Some firms keep under-sampling Republicans, by as much as 7 or 8 points. That's a big deal. Nationally it's tough, but on a state-by-state basis, you can use data to weight samples based on likely turnouts, and we do. We know this state. These other firms showing the president with gigantic leads here have been massively under-sampling Republican voters. They have Republican turnout polling even lower than registration levels in this state -- which almost always understate Republican turnout. It just isn't believable. We've been polling at 48-42 D-to-R, for a D+6. The political landscape has clearly shifted." Another data point Lee finds relevant is that in his statewide polling, the incumbent routinely fails to hit or exceed 50 percent in any of what he regards as the three "key" measures: Job approval, favorability and head-to-head. "Voters are pretty polarized at this point. He was at 52-34 favorability in 2008, today it's 47-47, and it's been stuck there for months. That's just one example of why we've been looking at a much closer race than a lot of people were suggesting."
The Romney campaign has indicated that they're monitoring the situation in Pennsylvania very closely, but they continue to downplay expectations there. Romney has devoted relatively few resources to the state, focusing much more heavily on other battlegrounds. (The nominee did deliver a speech in Wayne, Pennsylvania in late September). Keystone presidential aspirations may again prove to be a mirage for the GOP, but the notion that the state will be an easy layup for Barack Obama looks more dubious than ever. If the Obama campaign feels compelled to step up its efforts and expenditures to lock the state down, that could be an organizational concession of significant erosion elsewhere. Stay tuned.
UPDATE - New Pennsylvania poll: Obama 47, Romney 45. That makes three separate polls within the last week showing a two-to-three point contest. Maybe Susquehanna wasn't such an outlier after all.
UPDATE II - Is Romney within six...in Connecticut?