Ideology, Party, and Voting in the U.S. Congress, 1959–1980

Current methods of roll-call analysis have practical as well as theoretical shortcomings. We propose here a method based on a spatial theory of voting that overcomes these problems. We apply metric multidimensional unfolding to interest-group ratings of members of Congress in order to obtain a Euclidean spatial configuration of congressmen. Each roll-call vote is then mapped into the configuration of members in a way consistent with spatial theory. Based on 190,000 ratings issued from 1959 to 1980, our empirical analysis demonstrates that a single liberal-conservative dimension accounts for more than 80% of the variance in the ratings. A second dimension, associated with party unity, accounts for 7% of the variance. Approximately 86% of all roll-call voting for the 22 years of our study is consistent with a simple one-dimensional spatial model. The votes that best fit the liberal-conservative dimension are drawn from the government management, social welfare, and foreign policy areas. The votes that best fit the two-dimensional configurations are drawn from the agricultural area.