A fascinating series of votes just concluded in Arkansas’ state legislature. A year ago, Medicaid expansion was enacted in the state. What is remarkable is that not only is it one of the  more conservative states in the US, both chambers of the legislatures are controlled by Republicans, and the vote that required a 3/4 supermajority. Those are serious obstacles to Medicaid expansion at a time when the Affordable Care Act is fairly unpopular, especially with conservative Republicans.

The key to overcoming these hurdles was the decision by Democratic Governor Mike Beebe to seek Medicaid expansion by placing new enrollees onto the private health care excange rather than simply expanding traditional Medicaid rolls in the state. This experiment was ratified by Obama administration, despite misgivings. With this deal in hand, Beebe was successful in obtaining approval from the Republican legislature.

The appropriations bill containing the private option is up for reauthorization every year. Given how unpopular the ACA and Medicaid expansion are with conservatives, you’d expect the vote to go down resoundingly. Especially if the supermajority threshold was set at 75% of both legislative chambers. Instead, after a few tough votes in the state House (where the measure passed but just below the 75 vote threshold) and a single one in the state Senate, the measure passed, and is on its way to being signed by Beebe.

So, with a majority of Republicans joining all Democrats in voting for the measure across both chambers, party is not a very good guide for helping us know why people voted as they did. Instead, I argue ideological considerations would be central, and that the Republican divide would largely be one of (relative) moderates and (relative) conservatives.

I estimated a simple bivariate model of the final passage votes, with our state legislative ideological scores as the sole predictor. Below is a visualization of the predicted probabilities coming out of the model (the blue line), with actual votes as dots at 1 (Yes) and 0 (No), colored by party (Democrats blue, Republicans red). The blue crosses 50% at around a score of 1.75 or so; so those more liberal than that score are roughly predicted to vote No, and those above Yes.


The model does quite well, correctly predicting approximately 83.4% of the votes, much better than a model that looked solely at party. So most of the vote could be explained by the ideology of the state legislators voting. That’s perhaps not surprising, considering how ideologically-charged health care policy is in the United States.

It does get a number of the Republican votes wrong (those No votes who are between about 1 and 1.75 on the ideological scale, and a few Yes votes who are above 1.75). So ideological considerations are not the sole determinant of this vote, nor of all votes. But they’re likely the biggest.

I am currently writing an academic paper on the votes in the states on Medicaid expansion and establishing state-based health insurance exchanges (here’s a very old and rough version). What I find is that Republicans are deeply divided over both policies in the states wherever these votes have occurred, despite the unanimity of Republican opposition in Congress to the ACA. And that division is largely explained by the ideological stances of legislators: current Republican moderates and moderate conservatives vote in favor of these policies, while hard core conservatives vote against. Lots of fascinating things are happening in health care policy at the state level, so paying attention only Washington, DC is deeply misleading.

The fascinating thing about this vote is that it underlines how typically progressive policies like ACA expansion can take root even in such inhospitable soil. The Arkansas example and continued flexibility by the Obama Administration’s HHS Department can presage future expansion in the 25 remaining states that have not yet done so. And it looks like other Republican-dominated states like Pennsylvania, New HampshireMissouri, South DakotaUtah, and others are directly emulating Arkansas’ example.