John Sides reiterates an important point: the perception that Barack Obama “should” be winning by a huge margin (echoed today by David Brooks) is not supported by empirical evidence. The leading statistical models of presidential election outcomes forecast a narrow Obama win.

The consequences of this are actually more serious than most people realize. In the seminal work on mandates, Jim Stimson, David Peterson and two other political scientists argue that “mandates” are a collective interpretation of election results that carries an informational signal to nervous incumbents worried about re-election. As a result, members of Congress briefly shift their voting behavior in the direction of the perceived mandate (the three times in which this happened, Stimson et al argue, are 1964, 1980, and 1994). With expectations about Obama so high, there’s almost no way that anything short of an LBJ-esque landslide will be perceived as a “mandate,” which will make it harder for him to enact his legislative agenda.

Update 8/6 11:41 AM: A related story from the Financial Times: “Democrats anxious for Obama to widen lead”

I think one source of confusion on this issue is that the overall political environment is extremely negative for Congressional Republicans. Given the large number of retirements and competitive seats, particularly in the Senate, it looks like Democrats will pick up a number of seats in November. As a result, people assume that Obama has an equally strong advantage. However, as I argue above, the conditions that predict presidential election outcomes are much less definitive.