Jackson Lears on polling:
By the 1930’s, opinion pollsters believed they had discovered a cohesive mass audience—“the American public”—and a modal personality type—“the average American.” Since many Americans shared the pollsters’ naive faith in numbers, they accepted social scientists’ statistical constructions as accurte descriptions of themselves. The desire to fit in reinforced the normative power of statistical aggregates. As one of George Gallup’s interviewers observed at the end of the decade, “eight out of ten [respondents], after answering a question, will either ask directly what most people said about it or will remark indirectly, ‘I suppose nobody else said that.’ They are delighted if told that everybody said it. It makes them feel that they are right.”
Something For Nothing, 234-235.
I’m not sure what to think of the current polls, but I do wonder just how much polling influences elections. If Bush is ahead going into the election, does this mobilize more Kerry voters? Would a slightly trailing Bush bring a higher Republican turnout? It’s hard to say, but my hunch is that voter attendance is affected.