We took legislators' left-right policy preferences as measured by their now-standard Nominate scores, the square of the Nominate score, and their degree of risk as measured by the Cook political report. We then estimated the probability of voting for the bailout as a function of these three variables and found all of them to be highly significant (details available on request).
The picture of the findings looks like this:
From this we can make three observations:
1) There were partisan effects: Democrats were more likely to vote for the bailout than Republicans, all else equal. Even the most liberal Democrats from safe seats voted for the bill at about a 30% rate, while the most conservative Republicans voted at 10% or less.
2) The vote failed due to a defection from the wings; both liberal Democrats and conservative Republicans voted against the bill. This is an unusual, but not unique, voting pattern -- we'll examine the reasons for it in our next post.
3) Electoral risk played a role; as members faced tougher challengers, their probability of voting for the bill fell. Going from a safe seat to one that leans your way, and then from leaning to toss-up, costs about 20% likelihood at each step.
(HT Eugene Volokh)