The Monkey Cage

Nate Silver analyzes the effect of health insurance PAC contributions on Senators’ support for a public health insurance option. He finds a significant effect, controlling for ideology and state health care spending per capita. He estimates that 9 more Senators would support a public option if not for PAC spending. He concludes:

To be clear, not all of the opposition to the public option is the result of special interest money…But the money is why, even with 59-60 votes in the Senate and a President with high approval ratings, Democrats are facing an uphill battle on the issue.

This is by far a more sophisticated analysis than you typically see in news reporting. But it still raises a few questions.

First, the dependent variable is not really “support for the public option” but “as of this moment, stated support for the public option.” So a Senator who hasn’t taken a position publicly (Warner) is coded the same as someone who is opposed (Landrieu). It would be interesting to disentangle these groups. To be dorkily precise, the model should be a multinomial logit (support, oppose, no stated position) rather than a binary logit.

Second is another modeling issue. Silver estimates the effect of PAC spending for different levels of ideology. He shows that the effects of spending appear strongest among liberal Democrats, weaker among centrist Democrats, and virtually non-existent among Republicans. So it appears that there is an interaction between ideology and PAC spending: the impact of the latter depends on the former. However, it doesn’t appear that the model includes an interaction term, and that would be a more rigorous way of investigating any interaction. I will confess that it seems improbable to me that liberal mainline Democrats — the group probably most committed to a public option a group probably reasonably committed to a public option — would be so swayed by PAC contributions. [I thank commenter Dan Miller for catching this mistake. I will add that Silver finds that $120,000 in contributions is enough to reduce the probability of a mainline Democrat’s supporting the public option from .80 to .10. That is a massive effect, which strikes me as too large to be entirely credible.]

The remaining issues are the standard criticisms of any simple analysis of campaign contributions and their effects on policy positions. First, given the millions of dollars needed to run a Senate campaign, it’s not clear why a few tens of thousands from health insurance PACs would be all that influential. Second, PACs tend to target resources at members of Congress who already tend to support their views. In other words, there is a potential endogeneity problem in the model that Silver constructs. Does health insurance PAC money make senators less likely to support the public option, or are Senators who tend to oppose or at least be skeptical about government health care programs more likely to get contributions from health insurers and HMOs? (I note that 538 commenter Satya hits this on the head. Comments threads can be good things.)

This doesn’t mean that PAC spending is irrelevant here — or anywhere — but I am much more hesitant than Silver to say things like “the money is why.”

[Update: See also Brendan Nyhan’s reaction here.]