The vast majority of congressional Democrats have, whether deliberately or not, adopted the rationally self-interested strategy of siding with the popular Obama against the unpopular congressional GOP. But a substantial minority of congressional Democrats have consistently done the reverse—siding with the unpopular congressional GOP on the size of stimulus, on the public option, on the Consumer Financial Protection Bureau’s authority over car dealers, on taxes, and on the merits of the energy status quo.The reason this has happened, at least in most cases, is because while the polls Yglesias cites are national soundings, Blanch Lincoln is elected by the people of Arkansas, and Ben Nelson is elected by the people of Nebraska, and Obama and the Democrats are less popular in those places. The problem is that for Lincoln, there's really no way out; she likely will alienate swing voters in her constituency by supporting the national Democratic agenda, but if she dissents from it and it fails, Democrats and Obama generally suffer, which will hurt her as well. That's why, during the health care debate, I and others suggested that the best-case scenario for marginal Democrats involved the bill passing without their vote. That's probably not true for every bill, but it is probably true for most bills that pass against the unified or near-unified objection of the Republicans. The obvious problem for Senate Democrats in 2009-2010 is the combination of unified rejectionist Republicans insisting on filibustering everything and a 59/41 or 60/40 partisan split, meaning that on issue after issue marginal Democrats are forced to take tough votes.
In other words, marginal Democrats in the Senate haven't been irrational, and I'm not even sure they've been overly fearful (after all, most of them did stick with the Democrats on quiet a few votes). It's just that their incentives aren't always the same as John Kerry's or Barbara Boxer's incentives.