This paper addresses the problem of using roll call voting data to ascertain the ideal points of legislators in a multi‐dimensional space. After identifying difficulties inherent in factor analytic studies of roll call data, I present a simple formal model of legislative voting that resolves these difficulties. Finally, I derive some implications of the model and use them to outline a technique for estimating legislators' ideal points.
Notes
I thank Peter H. Aranson, James D. Laing, Peter C. Ordeshook, and Howard Rosenthal for their helpful comments at various stages of this research.