Unlike every other Congressional or state legislative iPhone application out there, TexLege includes roll call vote analysis to evaluate and compare the partisanship in voting for individual Texas House Representatives and Senators.
I calculate the partisanship scores using all the roll call votes taken in House since 1991, and since 1999 for the Senate. However, in order allow legislator score comparison from one session to another, I apply a statistical transformation on the data. In a sense, a legislator’s current voting record score is influenced by their scores in previous sessions. This is, in my professional opinion, much more desirable than the traditional full-on aggregation of roll call votes into one massive overly smoothed index. The way I analyze roll call votes ensures that scores calculated for prior sessions do not change with the addition of subsequent legislative sessions. I’ve always had a difficult time understanding why a change in a legislator’s partisanship in 2000 should have any impact on his partisanship back in 1973. (It shouldn’t!)
Don’t interpret these scores as a measure of ideology. That’s bad science. Many interest groups and politicos commonly make this mistake. Rather, these scores are a measure of partisanship within the voting record only. Partisanship in voting and in ideology are two very different things.
You shouldn’t compare scores between House members and Senators, the scales are not the same. I’m working on a technique that will put both chambers in the same scale, but I haven’t finished it yet. However, you can compare scores between members of the same chamber, or from one legislative session to another.
For more information:
The roll call partisanship analysis originates from my dissertation titled “The Power of the Texas Speaker: Maintaining Influence and Governing over a Divided House”, available from UMI/ProQuest or through your university’s library. I use Dr. Keith Pool’s W-NOMINATE software to generate scores for a single session. I then apply Dr. Gerald Wright’s modified Groseclose-Levitt-Snyder transformation to iteratively regress individual legislator scores from one session to the previous session.