The Analysis of Variance in Party Loyalty: The Eighty-seventh Congress

This is a study of the impact of constituency pressures on
the United States House of Representatives in the Eighty-seventh
Congress. It is an attempt to measure by quantitative methods
the relative effectiveness of pressures on congress from the
various types of constituencies normaily associated with one party
or another. Political pressure is defined here as a force which
brings about distinctive patterns of voting behavior on the part
of the representatives. whether the force is applied through conventional
forms of coercion or through the appeal of loyalties and
ideologies. The two sources of pressure which are the major concern
of this research are the representative's party affiliation and
his constituency. The former is held constant while the latter
is permitted to vary in order to observe more closely the differing
impact of certain constituency pressures on members of the Democratic
and Republican parties.
The dependent variable in this study is party loyalty
which is the per cent of the times a congressman voted with his
party on a set of one hundred and fifteen party unity roll-calls.
These roll-calls are those on which a majority of one party opposes
a majority of the other party. The independent variables are gross
measures of broad demographic, socio-economic, and political characteristics
of congressional districts. The theory is that these
factors reflect significant differences between the Democrats and
Republicans with regard to electoral centers of support. Consequently, it is hypothesized that variations in a congressman's constituency
would be related to variations in legislative voting
behavior as measured by party unity roll-calls .
The step-wise multiple regression technique is the means
of testing the major proposal. It indicates that a proportion
of the variance in party loyalty can be explained by variation in
constituency. But the proportion explained for the Democrats
(34.7%) is much greater than that for the Republicans (11. 8%).
The threshold concept is proposed as an explanation for this. It
is also suggested that new research is needed to clarify the types
of constituencies most likely to associate with the two major
parties.
Thus, the theory that constituency pressures can explain,
to some extent, legislative voting behavior has been tentatively
validated. However, the findings indicate that a more complex,
multi-variate, theory of legislative behavior is needed before
accurate predictions can be made. Therefore, the constituency model
must be expanded to include other factors such as psychological and
institutional ones.