Blogger Paul Brewer offers a useful clarification of my earlier post about the new Pew poll, in which I noted that its results were skewed because it featured a Republican sample of only 27 percent. He points out that in previous recent Pew polls, the number of respondents who claim to be Republicans has never been higher than 30 percent. How, therefore, could I say the sample was skewed? Well, in the first place, polls on political fallout are meaningless unless they measure Americans who say they vote, which this one did not — and the results show it, since the common finding in polling around the November election last year was that self-identified Republicans made up 37 percent of the electorate. In the second place, Pew says it deliberately oversampled blacks — who vote Democratic by a margin of 9 to 1 — in order to get a large enough number of them to measure their attitudes. Even though Pew says it then harmonized its data with census data, once again, the results are politically meaningless because it mixes the 60 percent of Americans who did vote last year with the 40 percent who didn’t (and presumably, never will, since 2004 had the highest turnout numbers in recent history). Pew’s data are consistent only with Pew’s previous data, not with the truth about the political damage done to the president.