Over at OpenLeft, Chris Bowers has the following composite exit poll for the primary season so far, weighted according to turnout in each state.

The exit polls are from CBS, and the turnout numbers used in weighting were drawn from wikipedia. It does not include Florida or Michigan, and it also does not include Alaska, American Samoa, Colorado, Democrats Abroad, D.C., Hawaii, Idaho, Kansas, Maine, Nebraska, North Dakota, the U.S. Virgin Islands or Washington State, since there were no exit polls for those jurisdictions. Obama won the states without exit polls by a cumulative 70%-30%, but the higher turnouts in Michigan and Florida make the net effect of the missing states roughly a wash. Here are the numbers sorted by gender, ideology, partisan affiliation, and income:

Composite Primary Season Exit Poll To Date

The essential point that I think these tables show is that the Democratic Party simply is not as divided into opposing economic, gender, and ideological gaps as many have argued. Obama comes in with 51% in his best income group, and 44% in his worst. Clinton’s best income group is also 51%, and her worst is also 44%. Overall, that really isn’t much of a gap, and the ideological gaps are even smaller. Obama’s range is 48%-52%, while Clinton’s range is 43%-47%. Even in terms of gender the gap isn’t actually very wide. Obama does 8 points better among men than he does among women, while Clinton does 8 points better among women than men. I stand in awe at such massive division within the party.
Some might object, and argue that much larger gaps show up among white Democrats, since African-Americans of all stripes are heavily for Obama. Exit polls for all of the categories that further divide along race / ethnicity lines are too spotty to compile into a composite format. Further, I did not compile composite exit polls by race / ethnicity (I will do so at a later time). However, to this entire question I simply say “so fucking what?” Does the vote of a white low income Democrat count different than the vote of a black low-income Democrat? Are white low income Democrats voting along racial and ethnic lines anymore than African-American Democrats? Further, when one divides all of the above categories according to race / ethnicity, women and self-identified Democrats would be the only groups that remain larger than 30% of the electorate. At that point, margins of error start to go through the roof, and any results would be questionable, at best.
The largest gap is between Democrats, where Clinton narrowly leads, and non-Democrats, where Obama holds a substantial, 21% advantage. Some Clinton supporters are crowing about how the numbers among Democrats cancel out the “popular will” argument, and give superdelegates the right to over-ride the will the interlopers in the process. However, there is a major flaw in that argument: most of the voters who did not self-identify as Democrats are actually registered to vote as Democrats. For example, the roughly 225,000 New York voters who did not self-identify as Democrats are all registered as Democrats. Every single “closed” primary state, where only registered Democrats are allowed to vote, produced a substantial number of non-self-identified Democrats. These percentages of the electorate in closed primary states varied between 13% and 35% of the overall electorate, and Obama always did better among this group. In other words, Obama is almost certainly winning voters who are registered as Democrats, just not those who self-identify as Democrats. At worst, Obama is in a virtual tie among the former group.

Read the rest at OpenLeft.