Serving Two Masters
| Author | Hans Noel,Seth E. Masket |
| Date | 01 March 2012 |
| Published date | 01 March 2012 |
| DOI | 10.1177/1065912910388188 |
| Subject Matter | Articles |
Political Research Quarterly
65(1) 104 –123
Serving Two Masters: Using Referenda
© 2012 University of Utah
Reprints and permission:
sagepub.com/journalsPermissions.nav
to Assess Partisan versus Dyadic
DOI: 10.1177/1065912910388188
http://prq.sagepub.com
Legislative Representation
Seth E. Masket1 and Hans Noel2
Abstract
Studies comparing the ideological leanings of voters and elected officials are often hampered by the lack of a common
measure. The authors use legislative referenda—on which state legislators and voters both vote on the same issue—as
bridging observations to develop a common measure for both. They use this measure to help assess two theories
of legislative representation, the well-known dyadic model and a partisan model that assumes legislators are also
accountable to a collective party agenda. Examining referenda votes during several sessions of the California Assembly,
the authors report several findings that are consistent with the partisan model. They find that legislators are significantly
more ideologically extreme than the median voter in their districts. They also find that members of the majority party
are considerably more extreme relative to their districts than members of the minority party are and that the majority
party becomes even more extreme the longer it maintains control of the chamber.
Keywords
political methodology, political organizations, parties, representation, electoral systems, ideal-point estimation
How representative are our representatives? In their long
her or his district, leveraging votes on legislative refer-
history examining this central question in representative
enda. This measure could be used to address a number of
democracy, political scientists have grappled with two
representation questions, from class bias to variations in
thorny issues, one theoretical and one methodological.
responsiveness at election time. We focus on the question
First, whom should we expect legislators to represent if
of partisan representation.
they are effective representatives? Second, how can we
The article proceeds in three parts. In the first part, we
tell whether they are, in fact, representing them?
spell out the theoretical expectations of party representa-
This article engages both of these questions. The latter
tion. In the second part, we address the methodological
issue is empirical: It is very difficult to get good measures
issues associated with developing a common ideological
of the preferences of voters that can be compared to the
measurement for both voters and legislators. In the third
activity or preferences of legislators. We generally know
section, we describe the results of our analysis, and we
which legislators are more or less liberal, relative to each
conclude with reflections on what our findings tell us about
other, and likewise for constituencies. But we cannot map
the relationship between government and the governed.
them to each other.
The former issue is deeper and theoretical. Scholars of
representation have suggested a wide range of norms for
Political Parties and
the relationship between voters and elected officials. We
Representation
believe one has been undertheorized. Dyadic representa-
Perhaps the dominant framework for understanding the
tion models, in particular, are hard to reconcile theoreti-
relationship between elected officials and their constitu-
cally with political parties. The notion of responsible party
ents is known as dyadic representation (see, e.g., Dexter
government, for example, imagines the electorate hold-
ing the entire party, not a single representative, account-
able for its collective activity. And yet an attempt by a
1University of Denver, CO, USA
2
party to be collectively responsible may undermine indi-
Georgetown University, Washington, DC, USA
vidual representation.
Corresponding Author:
We attempt to address these two concerns with a new
Hans Noel, Georgetown University, Department of Government,
methodological approach. We develop a measure of the
Intercultural Center 681, Washington, DC 20057-1034
ideological divergence of an individual legislator from
Email: hcn4@georgetown.edu
Masket and Noel
105
1957; Erikson 1971; Miller and Stokes 1963; Ansolabehere,
favor. These policies might be “moderate,” but empiri-
Snyder, and Stewart 2001; Erikson and Wright 2005).
cally they do not appear to be. Both the Democrats and the
This concept supposes that each member is paired with a
Republicans, especially in recent decades, clearly favor
constituency that elects her or him and assesses the degree
noncentrist policies. To sum up, the pure form of dyadic
to which legislators and constituents agree on questions
representation occurs when the observed behavior of a
of policy. There is some disagreement as to the exact nature
legislator matches district preferences, and the pure form
of the constituency that members represent. As Fenno
of partisan representation occurs when the behavior of all
(1978) noted, there are multiple constituencies within any
legislators of the same party matches a collective prefer-
given legislative district, and while there is a general nor-
ence that is more extreme than that of the median voter.
mative view that members should be serving the interests
Of course, we do not observe either pure form. Nor do
of all district residents, some have found that they often
we expect them. Elections can hold legislators account-
represent only the more active citizens of their own politi-
able, and so we expect some tendency toward dyadic rep-
cal party (Fiorina 1974; Gerber and Lewis 2004) or per-
resentation. But legislators are also beholden to party
haps only their wealthier constituents (Bartels 2008, 2009).
organization, so we expect that tendency as well. Loyal
Dyadic representation might be expected because elec-
party members, that is, seem to serve two masters.
tions hold legislators accountable. If a legislator were to
On one hand, they want to achieve the policies of their
drift too far from the preferences of her or his district,
ideologically more extreme party. On the other hand, they
another politician would challenge her or him for reelec-
want to please the demands of their more centrist voting
tion and win. However, since voters are not uniformly
constituency. To the extent that the party is powerful, we
attentive and informed, it is possible for legislators to
expect the party as a whole to manage this tension, in
shirk their districts, either representing a particular sub-
large part through their control of nominations. They seek
constituency (as above) or else representing no one at all.
out candidates who are extreme enough to deliver on the
A second, less well-developed framework for repre-
party’s ideological goals once in office but are not so
sentation is that of partisan representation. Under this
extreme as to incur the wrath of voters. Once the party
model, strong legislative parties may undermine dyadic
agrees on such candidates, they can often ensure that
representation to the point where legislators are voting as
those candidates win the party’s nomination by channel-
solid partisan blocs (Bullock and Brady 1983; Wright
ing key resources—endorsements, funding, expertise—to
1989; Clinton 2006). On any given vote, a constituency is
the preferred candidates and depriving other candidates
represented poorly. The party, however, will be well rep-
of those resources.
resented, and a constituency’s preferences are honored to
This same logic will continue inside the legislature. To
the extent that they voiced them through their partisan
the extent that the party leadership pressures members
vote in the previous election. These models have varied
into voting the party line, they will relax that pressure
in their explanatory power across different times in his-
for members in more moderate districts, allowing those
tory and even across different states at the same time
members to generate more moderate voting records
(Wright et al. 2009). Partisan representation might be
(cf. Lawrence, Maltzman, and Smith 2006). But the aggre-
expected because party organizations exert pressure on
gate tendency will still be to get policy that is more extreme
legislators or because legislators are themselves partisan.
than the district would want.
We return to these mechanisms more below.
Under an unfiltered dyadic representation, the legisla-
Partisan and dyadic representation are distinguishable
tors should be beholden to their district. Under a reasonable
insofar as the preferences of the district and the prefer-
model of partisan representation, legislators will make ges-
ences of the party differ.1 It is possible for representation
tures of responsiveness to their districts, but they will also
to appear partisan simply because the legislative districts
be pulled toward their more extreme party. This brings
are themselves partisan. Legislators may, in fact, be hon-
us to our first prediction: to the extent that parties are
oring dyadic representation if the median voters in their
strong, legislators’ voting behavior should be more extreme
districts are uniformly extreme. The alternative is that
than voter preferences. Downsian pressure notwithstand-
legislators are voting in partisan blocs despite represent-
ing, the party...
To continue reading
Request your trial