(Page 1 of 14) - The Determinants of Congressional Behavior: A "Weighted Averages" Model of Party and Constituency Influence on Members' Votes authored by Jessee, Stephen.

Weighing Their Options

PARTY AND CONSTITUENCY INFLUENCE IN

CONGRESSIONAL VOTING

STEPHEN JESSEE

Abstract. In this paper, I propose a model that estimates the relative weight
Senators place on party and on constituency when making voting decisions.
Unlike previous research, my technique simultaneously estimates these two
eﬀects and allows the estimated weights to vary across members. These esti-
mates have wide applications in testing theories about legislators’ incentives
and goals.

Congress scholars have long argued that legislators who are more loyal to their

party are more likely to be rewarded with various favors. In one of the most well-
known examples of this, Cox and McCubbins (1993) argue that parties are “cartels”
that help members overcome collective action problems. They claim that leaders re-
ward member loyalty with things such as preferred committee assignments. On the
other side of the coin, researchers have demonstrated that legislators who diverge
too sharply from their constituency’s wishes are more likely to be voted out of oﬃce
(Brady, Cogan Canes-Wrone 2002). If these sorts of arguments are assumed to be
true (at least on average), then members of Congress are constantly engaged in a
balancing act of sorts. It is probably safe to assume that nearly all members place
a fairly high premium on retaining their jobs from congress to congress. Therefore,
they should be wary of straying from their district’s wishes. But voting with their
party may provide them with goods such as committee assignments, party leader-
ship positions and other factors that fulﬁll their desires for institutional prestige
(Mayhew 1974).

Seen in this light, members are faced with a fairly straightforward optimization

problem: how much should they vote with their party and how much with their
constituency so that they can derive the most total utility. But previous work has
only investigated the rewards and punishments of the two behaviors in isolation.
If members are deriving utility from being near their party and if party position
is correlated with district position, then we could easily conclude that party is the
main determinant of member behavior when it might actually be district pressures.
By looking at each of these forces in isolation, we are likely to bias our results.
Therefore, I incorporate both party and constituency eﬀects into a hierarchical
model that takes the two forces into account simultaneously when estimating all
parameters, including the party and district eﬀects themselves. Levitt (1996) asks
a similar question, but assumes that these weights do not vary across legislators
(i.e. that all senators use the same weighting scheme). This greatly simpliﬁes
estimation, but the estimates he obtains are not useful for answering questions
about how members who weigh party more heavily in their decisions are rewarded
by party leaders or punished by voters, which is my aim here.

I construct a model that begins with member utility functions and builds up

from there, allowing each member to receive diﬀerent utilities from being close
to his party and from being close to his constituency. Alternatively, we can take