Last night, Matt Yglesias looked at some polling numbers Chris Bowers had posted and concluded that

59 percent of Democrats believe that John Edwards is proposing to withdraw all US forces from Iraq within nine months. 71 percent believe that Barack Obama is proposing to do this. And 76 (!) percent believe Hillary Clinton is proposing to do so. Needless to say, none of them are, in fact, proposing anything of the sort -- though I wish they would.

His point--that the electorate has no clue at this point what any of the candidates' policy positions are--is true enough. But what struck me was not merely that the poll respondents were wrong overall, but that they were wronger still in their comparative assessments of the Democratic contenders--that is, that they thought Hillary was the least hawkish and Edwards the most, with Obama somewhere in the middle. This is, of course, the exact opposite of the candidate's respective positioning on Iraq. One explanation would be that Edwards is a white man, so voters assume he'll be more conservative than a black man or white woman, etc., etc. But given that respondents rated the candidates' dovishness more or less in proportion to their overall popularity--Hillary first, Obama second, Edwards third--I suspect there's something else going on here. The Democratic electorate, which favors withdrawal, probably isn't choosing which candidate it likes on the basis of policy positions; it's ascribing its favored policy positions to the candidates it already likes on the basis of name recognition and other unrelated attributes. That is to say, rather than the candidates' popularity being a function of their positions, their perceived positions are, at this point at least, a function of their popularity. --Christopher Orr