Troublesome Survey USA Poll Of Last Night's Debate
by RT, Fri Apr 27, 2007 at 09:03:37 AM EDT
Survey USA has a pretty good rep in the lefty blogosphere. Its poll of last night's debate, however, is of very limited use. Let me explain why it's mostly useless, then I'll mention the one place where its results are actually usable.
On the left of your screen, if you click the link, is a dropdown where you can see the poll results either as percentages, or see the actual counts. Choose the latter.
Survey USA surveyed 1,250 South Carolinians about the debate. 403 actually listened to the debate, and they asked those 403 who they thought won or lost the debate. So far, so good. The problem: only 195 of those 403 were Democrats. 104 Republicans, 82 Independents, and 22 who (I'm guessing) didn't give a party affiliation were also included in the poll results.
The 403 are broken down by sex, race, age, and a number of other variables. But we don't care about most of these people. We want to know about likely Dem primary voters.
I'm assuming that SC has an open primary, but both parties' nominations will be contested, so there won't be much reason for Republicans to cross party lines and vote in the Democratic primary. Those independents that vote in a primary will be split between the two parties' primaries, so only a share of their votes should be counted. But in all those breakdowns by race, sex, age, and whatnot, the Dems aren't sorted out from everyone else. So those breakdowns are pretty much useless.
With one exception: in South Carolina, as everywhere else, black voters are overwhelmingly Democratic. Possibly even more so than everywhere else, because the SC GOP is very, very conservative. (Remember their furor over the Confederate flag on the state capitol.) Not a very black-friendly party at all.
So it's safe to assume that the black respondents of the poll reasonably accurately represent black SC Dems. And by subtracting their numbers from the total Dem numbers, we should be able to get a (rather weak, considering the numbers) look at white SC Dem preferences. 144 blacks listened to the debate, and 51 other (presumed white) Dems listened to it.
Of the 144 blacks, 81 felt Obama won, 30 felt Clinton won, 16 weren't sure, 6 each for Edwards and Biden, and no more than 2 for anyone else.
That's the big news: when blacks see or hear Obama, he can pull them away from Clinton. If Obama can win a solid plurality of black voters, he should be able to win the nomination, so this is good news for him, and bad news for Clinton.
It's also bad news for Edwards: as the Pew poll showed, Edwards isn't picking up much of any black support, and you can't win the Dem nomination without a decent share of the black vote. Since he wasn't able to change many minds last night amongst African-Americans, he's got a BIG problem.
At least 43 nonblack Dems thought Hillary won, so Hillary took the lion's share of that category.
But due to a bad poll design, Obama's ability to impress black South Carolinians is the main takeaway from this poll.