In its 2012 American Values Survey, the Pew Research Center identified a new and profound dividing line in American politics. “Americans’ values and basic beliefs,” Pew concluded, “are more polarized along partisan lines than at any point in the past 25 years.” Partisan identification, Pew added, “has now become the single largest fissure in American society, with the values gap between Republicans and Democrats greater than gender, age, race or class divides.”
Pollsters should bend over backwards to make sure their sampling methodology reflects this insight. Oversample the adherents of one party or the other, Pew’s research tells us, and you might as well toss the entire poll onto the trash heap.
Perhaps this is why President Obama’s standing in the polls is higher than one might expect, given the steady stream of negative economic news. Pollsters may be getting this all-important criterion of valid polling — the correct ratio of Republicans to Democrats — wrong.
How would the media have reacted if the headline on the most recent Washington Post/ABC News poll had been “Romney Up 5 over Obama; Enjoys Clear Edge on Handling of Most Policy Issues”? Imagine, moreover, the hyper-spin if the poll had also found that close to 60 percent of Americans disapprove of the president’s overall job performance and that similar supermajorities disapprove of his handling of such core issues as the economy, health care, and immigration.
The psychology of the entire presidential race would shift in a nanosecond.
Well, of course, that wasn’t the case. The Post poll found the two candidates in a dead heat — 47 percent to 47 percent. Obama’s job-approval rating remained stuck at 47 percent, and his disapproval checked in at a disappointing, but politically manageable, 49 percent. Slightly more than half of Americans disapproved of his handling of the major policy issues.
But lost in all the analysis is that the poll may have included far too many Democrats and too few Republicans in its sample; 33 percent of those surveyed were Democrats, and only 24 percent were Republicans. That cuts against the voluminous data on partisan affiliation collected by the Gallup Organization.
According to Gallup, Democrats outnumbered Republicans nationally by up to nine percentage points for a considerable spell. But over the past few years, the GOP has erased that advantage. In the most recent Gallup data, from June, the two parties are at absolute parity, with 30 percent of Americans identifying themselves as Democrats, 30 percent as Republicans, and 39 percent as independents. This partisan parity remains largely in place even when Gallup adds “leaners” (independents who lean toward one party or the other) to the GOP and Democratic numbers.
Reconfiguring the Post poll to reflect Gallup’s findings would dramatically alter the top-line numbers. Romney’s strength would rise significantly as Obama’s fell. Romney, it turns out, may actually be ahead by as much as four or five points.
Other widely cited polls have the same unfortunate tendency to oversample Democrats. Case in point: the Kaiser Health Tracking Poll. It often finds majority support for Obamacare and, more generally, widespread public support for an aggressive governmental role in health care. Its most recent survey found not only that voters were happy with the Supreme Court’s decision to uphold Obamacare but also that a majority (53 percent) wants to either keep the law as it is or expand it.
If this sounds odd and out of sync with numerous other polls on Obamacare, consider that Democrats outnumbered Republicans in the Kaiser sample by more than ten percentage points — 32.1 percent to 21.9 percent. Oops.
So, Mr. or Ms. Conservative, before you scratch your head at the next poll that has your liberal cousin doing handsprings, check first to see whether the ratio of Republicans to Democrats is close to what Gallup has found. If it isn’t, pay no mind. It’s just another exercise in “Garbage in, garbage out.”
— Michael G. Franc is vice president of government studies at the Heritage Foundation.