Assessing the Mechanisms of Senatorial Responsiveness to Constituency Preferences
| Author | Matthew K. Buttice,Benjamin Highton |
| DOI | 10.1177/1532673X15607300 |
| Published date | 01 September 2016 |
| Date | 01 September 2016 |
| Subject Matter | Articles |
research-article2015
Article
American Politics Research
2016, Vol. 44(5) 767 –793
Assessing the
© The Author(s) 2015
Reprints and permissions:
Mechanisms of
sagepub.com/journalsPermissions.nav
DOI: 10.1177/1532673X15607300
apr.sagepub.com
Senatorial
Responsiveness to
Constituency Preferences
Matthew K. Buttice1 and Benjamin Highton2
Abstract
This article analyzes the relationship between U.S. senators and their
constituencies over the entire period of time that senators have been selected
by direct election. Focusing on preference change within states, we identify
three mechanisms that might produce responsiveness in senators’ ideological
locations. We find that it is not merely the case that responsiveness is
produced by party representation. Replacement of one senator with another
of the same party facilitates responsiveness, too. And, even without electoral
replacement, individual senators appear to adjust their ideological locations in
response to changes in their electorates’ preferences. We also investigate how
the mechanisms of responsiveness changed with the erosion of Democratic
dominance in the South and as the parties grew stronger over time.
Keywords
representation, constituency responsiveness, senators, Congress
Over the past 100 years, the partisan and ideological complexions of many
state electorates have changed considerably. This is true of all the former
Confederate states along with many states outside the South, including Utah,
1Office of the Independent Monitor, City and County of Denver, CO, USA
2University of California, Davis, USA
Corresponding Author:
Benjamin Highton, Department of Political Science, University of California, Davis,
CA 95616-8682, USA.
Email: bhighton@ucdavis.edu
768
American Politics Research 44(5)
Vermont, Massachusetts, and Arizona. Most other states have experienced at
least some change. Whether due to generational replacement, patterns of
migration, or the evolving opinions of those who remain, preference change
is a common phenomenon. This article addresses the consequences for politi-
cal representation, specifically the responsiveness of elected officials to con-
stituency preferences. Through a consideration of the entire period of time
U.S. senators have been constitutionally required to be selected by direct
election, we investigate the mechanisms of responsiveness by focusing on
longitudinal change within states rather than the more commonly used
approach of cross-sectional analysis.
Theoretically, we identify three distinct mechanisms through which an
electorate’s partisan and ideological preferences may influence the ideologi-
cal locations of senators. First, as preferences change within a state, voters
may use elections to replace one senator with another from a different party.
Second, electoral replacement may involve replacing a senator with another
from the same party. And, third, in anticipation of future elections, senators
may respond directly to changing preferences, thereby preempting the need
for either of the replacement mechanisms to produce responsiveness.
Empirically, we take advantage of the change in state electorates over time
to estimate how the mechanisms contribute to responsiveness. We trace, for
example, the evolution of the Vermont electorate’s preferences and how it
relates to which party won Senate elections in the state, differences between
senators of the same party in the state, and how individual senators adjusted
their ideological positions in response to changes in the state.
This article’s contributions derive from analyzing all three mechanisms
over a longer time period than in previous studies. We find evidence for sub-
stantial overall responsiveness, which is produced by the three hypothesized
mechanisms. Elections matter as a means of replacing one senator with
another, but senators also appear to respond directly to the preferences of
their electorates. All three pathways facilitate democratic accountability and
responsiveness.
Background
Article I, Section 3 of the U.S. Constitution specifies that a state’s senators be
“chosen by the Legislature thereof.” The 17th Amendment to the Constitution—
ratified in 1913—altered the process by requiring that senators be “elected by
the people thereof,” thereby instituting the direct popular election of senators.
In contrast to elections for the House where states are divided into geographi-
cally distinct districts, both senators from a state are elected from the same
constituency—the statewide electorate—making Senate elections “at large”
Buttice and Highton
769
contests. And, although there are two senators elected from each state, there
are separate elections for each one that typically occur in different election
years.1 As a result, U.S. senators may be characterized as being elected through
single-member at-large elections. Another important feature of the electoral
system for the Senate is that the geographic boundaries of the constituencies
are fixed through time. They are never subject to redistricting as they are for
the House.
The single-member at-large electoral system with fixed constituency
boundaries means that one can study the responsiveness of senators’ ideo-
logical positions to constituency preferences over an extended period of time.
As states change and evolve, we can observe whether there are differences in
senatorial behavior and the mechanisms that account for those differences.
Our analysis of “responsiveness” follows from Achen (1978), which pro-
poses a straightforward model of representation where an elected representa-
tive’s ideological location is determined by constituency preferences:
s = α + β× c + ε.
(1)
In this setup, β indicates how much change in the ideological location of a
senator (s) occurs when constituency opinion (c) changes.2 This aspect of the
relationship between an elected official and the constituency has received
extensive attention among scholars, more commonly in a cross-sectional
framework where differences across constituencies are related to differences
across elected representatives.3 Instead, our focus here is similar to research
on “macropolitics” (e.g., Erikson, MacKuen, & Stimson, 2002) that analyzes
the same electorate or “polity” over time (often an entire country) to deter-
mine how mass preferences relate to the ideological positioning of political
leaders and institutions, collective or “dynamic representation” (Stimson,
Mackuen, & Erikson, 1995).4 Here, too, there is a difference with the present
study because rather than observing a single polity through time, we observe
representation for 100 Senate seats across the 50 state electorates.5
Theory and Existing Evidence
To elaborate our theoretical approach, consider Vermont, which is an espe-
cially useful state to examine because it has undergone major change in its
electorate’s preferences over time. As we illustrate below, in the early to mid-
20th century, presidential voting in Vermont was consistently more
Republican than in the country overall. Since then, there has been a steady
shift toward the Democrats, to the point that presidential voting in Vermont
in recent elections has consistently been among the most Democratic in the
770
American Politics Research 44(5)
country. Theoretically, we expect change like that which has occurred in
Vermont to produce change in the ideological representation a state receives
in the Senate for three reasons.
One mechanism of responsiveness is through party representation. Over
the entire period that senators have been directly elected, on the main dimen-
sion of party conflict over social welfare policy the Democratic party has
been more liberal than the Republican party (McCarty, Poole, & Rosenthal,
1997, 2006).6 Given the evidence that partisanship and policy preferences/
ideology matter for voters (e.g., Ansolabehere, Rodden, & Snyder, 2008),
then as Vermonters became more Democratic and liberal, one would expect
an effect on the party of winning Senate candidates.7 Earlier in the time
period covered by our study we expect to find more Republican (and there-
fore conservative) senators in Vermont. Later in the period, we expect fewer
Republican and more Democratic (and therefore liberal) senators from
Vermont. Parties are thus one mediating mechanism of responsiveness of
senators’ ideological locations to constituency preferences.8
The degree to which party representation contributes to responsiveness
depends on at least two factors. First, the magnitude of the association
between constituency preferences and the party of the candidate who wins an
election influences party representation, with a stronger relationship produc-
ing greater responsiveness. In the absence of a connection, then given the
different pools from which candidates of different parties are drawn and/or
party effects in general, one would still expect substantial differences in the
ideological locations of Democratic and Republican senators. But, those dif-
ferences would be unrelated to constituency preferences and therefore not
contribute to the type of responsiveness considered in this article. Related, if
a state only elects senators of the same party, then the prospects for party
representation as a mechanism of responsiveness are severely limited.9 In
such a political context, party representation could occur if the ideological
locations of the...
To continue reading
Request your trial