George Washington University political scientist John Sides believes the mainstream media hand wringing about whether Obama will unify the Democrats is without merit.

Sides argues the exit polls are unreliable predictors of Democrats’ ultimate behavior in November. And, he offers some research to support his contention. First, analyzing the American National Election Study data from the University of Michigan, which has conducted polls since 1952, he shows increasing Democratic unity over the past decade in presidential voting behavior. "The party loyalty of Democrats has been increasing over time and has essentially hovered at 90% since 1992," Sides writes (And Republicans are similarly loyal to the Republican nominee.) So if history is any guide, 90% of Democrats should come around and support Obama by November--irrespective of their attitudes today.

Second, Sides also presents data from the Annenberg National Election Study that tracked the percent of Democrats saying they would vote for Al Gore between December 1999 and November 2000. It supports Sides’ contention that as the campaign progresses, voters’ preconceived notions get reinforced, and both parties "come home" to their respective nominee. Gore’s numbers among Democrats improved from 78% to 87% during the course of the campaign.

I agree with Sides and expect that when Senator Clinton drops out, Obama will get a bump in the polls. The wild card in the equation, however, is John McCain’s traditionally better-than-expected performance among Democrats. That’s the big mitigating factor. Democrats will no doubt rally around Obama in the months ahead, but I wonder if McCain’s cross-party appeal (compared to other more recent GOP presidential nominees) will attenuate the Illinois Senator’s party unity numbers somewhat below historical levels.