Poll Analysis Skewed

The AP has released a nationwide story that Obama's support among "working class whites" is "faltering." I submit that the poll upon which this argument is based is faulty. For one thing, nobody asked me my opinion. But in all seriousness, let's take a look at the numbers behind this poll:

The article asserts: "53 percent of whites who have not completed college viewed Obama unfavorably [in April], up a dozen percentage points from November."

The raw data to support this claim is the following:

The AP and Yahoo received 401 total answers on this portion of the survey.

Very favorable: 62
Somewhat favorable: 126
Somewhat unfavorable: 75
Very unfavorable: 73
Don’t know enough to say: 63
Refused / Not Answered: 2

The poll reported based on the 148 people who had somewhat unfavorable or very unfavorable ratings for Obama.

Overall, there were 277 white, non-Hispanic voters included in the survey. Overall, there were 183 non-college educated people in the survey.

Interestingly, if you divide 148 by 277, you get 53%. However, I suspect that the AP-Yahoo pollers actually cross-referenced the white, non-Hispanic voters with the 183 non-college educated people in the survey and arrived at some subset of the 183 non-college educated people who also happen to be white, non-Hispanic. 53% of 183 is 97 voters. So of the 148 people who had unfavorable ratings, a maximum of 97 of them were white, non-Hispanic.

If there was a 12% shift from November to April, that means that a maximum of 12 voters changed their minds (based on around 100 voter sample).

I would hate to see Obama's Presidential campaign suffer a beating in the press because of a maximum of 12 people out of the whole country. Of course, it only took 12 jurors to spark riots in the street in LA back in the 1990s.

I hope people are able to see past the pack mentality of the press.