Another claim that Pennsylvania is full of racists

First, Ed Rendell claimed that racist, conservative whites wouldn't vote for Obama (just like they didn't support Lynn Swann). Then Daily News columnist Fatimah Ali declared their would be a "full-fledged race and class war" if Obama loses. Most recently, it was Congressman Jack Murtha claiming western PA was full of racists, then backing off that position, claiming that he doesn't really believe they're racists, just rednecks.

Now, columnist John Baer is repeating the charge that McCain is depending on the "cracker" vote in Pennsylvania, using curmudgeons like Rendell and Murtha as his neutral experts. His man-crush on Obama aside, Baer is generally a respected journalist.

But Ryan Shafik already debunked this "racist white conservative" nonsense, when he noted that Lynn Swann (a black Republican challenger) outperformed Rick Santorum (a white Republican incumbent) in the supposedly redneck, cracker parts of the state.

Furthermore, Baer needs a remedial lesson in understanding polls (as do many). When a poll has a "margin of error" of 3%, this is a margin of error of +/- 3% for each candidate. That is, if a poll shows Obama 50% and McCain 45%, with a margin of error of 3%, that means Obama is likely to be between 47% to 53% and McCain 42% to 48% among all voters. Note that the range applies only at the time the poll was taken, as voters do tend to change their views over the course of the election

This margin of error implies a 95% confidence interval - meaning that the pollster can say with 95% certainty that a candidate is within the margin of error from what the poll reported. Of course, this implies a truly random sample. Often times polls have sampling error, which skew the result. Michael Barone has a good article in today's Wall Street Journal explaining the many ways polls can make errors.

Barone will be the guest speaker at our Speaker Franklin Awards Dinner, giving his post-election analysis.