Party control and political agendas
At the 2010 annual meeting of the Midwest Political Science Association, I presented a paper in which I used a time-series clustering algorithm to identify eras in Congress based on the substantive nature of the Congressional agenda. I found that it was possible to correctly identify changes in party control in the Senate and House, based only on the change points in the time series of agenda focus in multidimensional space, with a reasonable degree of accuracy. Further, I found that different majority parties had statistically distinguishable agenda patterns, and that full knowledge of the time and attention devoted to each major topic predicts, with a high degree of accuracy, the majority-holding party in both chambers.
The paper on which the talk was based can be found here.
The slides for the talk may be seen below. I found that in explaining the idea of clustering agendas over time, it was useful to make an analogy to identifying seasons based on weather patterns, as depicted on slides 12-15.