an influx of new voters in politically competitive electorate, when old voters’ preferences are
known.
To connect these theoretical implications to a more concrete example, we can consider a
senator representing two types of Southern states following the 1964 presidential election. In
states where Goldwater did very well, a senator can surmise that white voters have strong and
unified preferences on the right of the ideological dimension (certainly the second dimension
associated with civil rights, but also with the first dimension associated with other left-right
issues). Thus, the addition of black voters, even in large numbers, may not result in changes in
senators’ perceptions of the constituency median. Enfranchised white voters choosing
Goldwater over Johnson in large percentages are an indicator to the incumbent senator of the
large majority of white voters being on the right of the ideological dimension, and thus the
median voter being on the right (regardless of what percentage of newly enfranchised black
voters, a numerical minority, are added to the electorate). In contrast, in other states where
Goldwater fared poorly due to a larger variance in preferences among white voters (e.g., a state
like North Carolina where Goldwater lost, receiving only about 44%), the effect of even small
increases in black voter registration levels can greatly alter a senator’s ability to know the
location of the state’s median voter.
WHY THE U.S. SENATE?
Most empirical work on congressional representation has focused on the U.S. House,
with less work on the U.S. Senate, though this has been changing more recently. Work on U.S.
Senate representation has focused primarily on the effect of dual representation (Schiller 2000),
changes in membership and institutions (Sinclair 1989), and the effect of state size
(Oppenheimer and Lee 1999) on legislative behavior. Little work has focused on position-taking
8
Convention