My research journey this semester was one full of twists, turns, and surprises. I began the semester finishing up on article collection for the Obamacare media project, and before long had transitioned into collecting group data for my team’s project. My original intention was to pursue a project investigating framing of the Affordable Care Act (ACA) by elites and average Americans but terminated that project when I ran into the brick wall of unavailable data.
Then Professor Settle steered me towards a new project, one that I embraced and made my focus for the semester: measuring the ideology of states.
This subject was one I had experience with. Throughout the Obamacare team’s quest to identify a coherent research focus we kept stumbling across the need to figure out how to measure the ideology of states. I had read several articles on the matter and found the subject interesting. To me the challenge of measuring the political climate of a state represented a fascinating opportunity to test my ability to take a complex phenomenon and condense it down into working measure. I embraced the challenge and ran with it.
The first step was to look at how scholars had operationalized state level ideology in the past. My review of the literature turned up, among others, two landmark studies that coincide well with the competing theories on how to develop a good measure. The first of these studies was published by Robert Erikson and his colleagues in September 1987 issue of the American Political Science Review. This article laid out an approach that relied on disaggregating national polling results to get state-level indicators of partisanship in states. The second, published in 1998 in the American Journal of Political Science by William Berry and his colleagues, presented a methodology focused on aggregating various indicators of elite ideology within states. These indicators include interest group evaluations of elites, partisanship of the state legislature, and more. I chose to follow a methodology modeled after Berry and his colleagues, largely because of the data available and the kind of analysis I wanted to conduct.
Having determined my procedure, I gathered my data. I chose to focus on four indicators of state ideology: party of the governor, partisan makeup of each state’s upper and lower houses, and the average DW-Nominate score of each state’s U.S. Senators (all data was from August 2009, the time frame for the articles collected by the Obamacare team). I had collected data for these variables earlier in the semester, and they seemed to be relatively strong predictors of ideology. The way these measures were structured/operationalized, each score was between -1 and 1 with -1 (exception: the party of governor was coded such that a state with a Democratic governor received a -.25 and a state with a Republican received a .25). I then aggregated the four measures for each state and divided by four to arrive at my ideology score for each state.
To verify my results I compared them with my comparison variable, Presidential vote share for the 2008 election (as this election was closer to the 2009 time frame for my other data). My comparison variable was coded such that states that went blue in 2008 would receive a negative score between -1 and 0, while those that went red would receive a positive score between 0 and 1. The scatterplot, which I unfortunately was unable to upload, showed a strong positive correlatino between the two variables.
I arrived at the conclusion that my methodology, while not perfect, was a step in the right direction in terms of measuring ideology within a state. Granted there is significant room for improvement in this research design. For instance, my analysis relies on the assumption that Presidential vote share in 2008 serves as a valid comparison measure for the ideology scores I came up with. I believe, however, that my work this semester can serve future members of the SNaPP Lab and the Obamacare team specifically with future research.