How to read polls
According to the Washington Post today, 49% of evangelical Christians in Iowa voted for Huckabee, 19% for Romney, 11% for Thompson, and 10% for McCain; nothing is said about how the other 11% voted. This means, of course, that 51% of them did not vote for Huckabee. Nevertheless, our friend E.J. Dionne speaks of his victory in Iowa as “the revenge of evangelical Christians who had been taken for granted by the GOP establishment and decided to vote for one of their own.” Even though six out of ten Iowan Republicans declared themselves born-again, Huckabee got only 34% of the vote. So I don’t see how Dionne’s conclusion follows; there seems to be a good deal of diversity among evangelical Christians in Iowa. That 19% voted for Romney I find striking. Or am I missing something?
In an editorial, the Post asks people “not to get carried away by the results [advice the editors might send to their reporters and columnists!]. In both parties, caucusgoers were a small and unrepresentative sample of a small and unrepresentative state.” (Perhaps someone could parse the use of the word “unrepresentative” in these two phrases.) If this is the case, why is it taken so seriously? Why should it have led at least two of the candidates to discontinue their campaigns, among them Sen. Biden who complains about the difficulty he had in getting the media to cover his efforts? (In the Post’s charts today, results are given only for three of the Democrats and four of the Republicans.)
Is anyone thinking about a better way of going about this?