February 16, 2008


Democrats and the Politics of Identity (Blake D. Dvorak, 2/16/08, Real Clear Politics)

My colleague Jay Cost, as someone who not only knows what an ordinary least squares regression analysis is, but also knows how to perform one, has calculated that each of the states Obama has won since Super Tuesday played heavily to at least one of Obama's demographic strengths: states with either large African-American populations or, somewhat counter-intuitively, "homogeneously" white populations; states with high median incomes for white voters; states with low Hispanic populations; and states with low union membership.

Using this formula, Cost noted that it is possible to forecast the upcoming contests. So, for instance, Obama should do well in Oregon (homogeneously white) while Clinton should do well in Kentucky (low median white income). The model does not account for momentum, however, which Obama might have picked up by winning the last eight contests. But the larger point is that, given the demographic makeup of the remaining states, one can predict, with a reasonable degree of accuracy, just which candidate should do well in which state. In fact, on average, the remaining states favor Clinton slightly. Good news for her tear ducts.

But bad news for a Democratic Party that can no longer deny that it has a serious problem of identity politics.

...what do you call a party that's trying to nominate a candidate who can only carry majority black/upper middle class states?

Posted by Orrin Judd at February 16, 2008 8:24 AM
Comments for this post are closed.