WITP 5(01).book(WITP_A_315126.fm)

In this article, we discuss the design of party classifiers for Congressional speech data. We then examine these party classifiers’ person-dependency and time-dependency. We found that party classifiers trained on 2005 House speeches can be generalized to the Senate speeches of the same year, but not vice versa. The classifiers trained on 2005 House speeches performed better on Senate speeches from recent years than on older ones, which indicates the classifiers’ time-dependency. This dependency may be caused by changes in the issue agenda or the ideological composition of Congress.