Obama is going to win, the question is by how much. There were concerns down south of Springfield for a while with Hillary polling well, but that appears to no longer be a problem. Obama won’t do as well down there as in Chicago, but he apparently has a healthy lead.
Survey USA listed the numbers today at 66 30 and that seems about right to me–the only question was they had African-Americans making up 22% of the voting group, yet in 2004, they were higher than that. The Survey USA poll that was closest to Obama’s final tally (and the only poll that close) had the percentage at 25%.
In fact, if you look at the Survey USA polls released across the board today, there’s a fairly significant drop in the percentage of African Americans included in the samples in all states with significant African-American populations. Most seem to be 3-5 points of African-Americans as a percentage of the population.
There are a few possible reasons for this. One is that African-Americans are less likely to vote come Tuesday. Given Obama is on the ballot that is unlikely and counter to everything we’ve seen this cycle.
Or it could mean that samples are having a hard time being polled over the last few days and they are undercounted. If that’s the case, it could be a good day for Obama.
Every poll everywhere is undercounting African-Americans, as best I can tell.
Do you have any grasp on why Clinton polls so much better in SUSA polls?
The numbers that stand out the most are the A-A numbers–at least 3-5 off what they had as the likely make-up of voters–and that might be lower than reality.
The other thing is that as an automated pollster, we still don’t fully understand how people respond to them. There could be some sort of bias who takes the call this close to the election–we haven’t seen that a ton in past elections, but the ever increasing number of robocalls could have something to do with it.
Just a guess on the second one. The gender numbers aren’t far off from what I can tell.