2
is unidimensional and whether the explanations for the low-dimensional policy space of
congressional politics are applicable to state legislatures.
The basic point of interest here is the remarkable unidimensional liberal-
conservative character of floor politics in Congress (Poole and Rosenthal 1997).
Historical analyses of the U.S. Congress have revealed a remarkably stable
unidimensional structure with only a few periods of higher dimensionality interspersed.
The two primary periods in which Congressional voting patterns revealed higher
dimensionality occurred during the Civil War era and a period in the 1950-1960s. These
also correspond to periods in which the two-party system was undergoing great
transition, which suggests that political parties play a critical role in structuring and
stabilizing voting alignments. This argument has been developed in historical analyses of
congressional roll call voting by David Brady and his collaborators and offered in a
somewhat different fashion by Aldrich and Rohde as the theory of conditional party
government (Brady, Cooper and Hurley 1979; Brady 1985; Cooper and Brady 1981;
Aldrich 1995; Rohde 1991). For our purposes here, our interest turns to their focus on
the linkage between the strength of parties and the structure of roll call voting. These
theories differ in the stress given to changes in the parties’ constituencies versus the
independent effects of members’ issue preferences. Their conclusion is the same
however: highly differentiated and opposing parties yield a single dimensional
configuration of roll call voting. The patterns of roll call voting are not as clear when the
parties are less central to legislative process, or where the preferences within the parties
are quite heterogeneous.
Convention