Yesterday, my inbox blazed with the news that Obama's standing among African Americans is plummeting in North Carolina. A new PPP poll was responsible for the crisis: It shows Obama with just 75 percent of the African American vote, down from 95 percent in 2008 and 87 percent in their May poll.
But let's keep this in perspective—African Americans aren't abandoning Obama. Nearly every other pollster continues to show Obama in the upper-80s or 90s among African Americans. Democrats always receive an overwhelming share of the African American vote and it's not realistic to expect that Romney will break new ground against a historic African American figure. So what explains PPP's results?
Nate Silver chalks up the poll to statistical noise. It's true that demographic subgroups experience higher margins of error due to small sample sizes, so that explanation is compelling.
The explanation would be completely satisfying if it was just one PPP poll that showed Obama underperforming among African Americans. (This was suggested by Jim Williams, a polling analyst at PPP, who said that the North Carolina result is not something PPP has "ever seen before.") But the odd thing is that I’ve seen a similar trend quite recently, in PPP's national tracking polls conducted for DailyKos/SEIU:
While one should always be cautious to rely on small demographic sub-samples, weeks of polling and the recent North Carolina poll might begin to point toward a broader issue with PPP's ability to poll African Americans.
Despite tending to underestimate Obama's support among a critical constituency, PPP continues to post top-level numbers more reminiscent of 2008 than the closely contested 2012 election. According to PPP's most recent polls, Obama leads by 8 percent nationally with 342 electoral votes.
Which brings me to my next question: How can Obama hold such a commanding lead in PPP's polling without overwhelming support from African Americans? Unlike every other pollster, PPP shows Obama performing exceptionally well among white Americans, which compensates for Obama’s exceptionally poor standing among African Americans (and often Hispanics). Even in yesterday's North Carolina poll, Obama received 36 percent of the white vote, up one percentage point from 2008.
It is difficult to reconcile Obama's supposed strength among white voters with diminished support among non-white voters, especially since most pollsters report precisely the opposite phenomenon.
PPP has a strong track record, so their numbers shouldn't be discounted. But their peculiar demographic subtotals reflect a pattern, and these underlying irregularities make it more difficult to accept their top-line numbers—which already deviate from most other national and state polls. We'll have to wait until November to see if they're right, but in the meantime, I'll be somewhat wary of PPP polling.