-,,'-r‘é??;1.gjl,

' 'r.

  
        

é;

’e’x‘

 

“ $.44

‘38..

n 1:...
gig;
.5813: v

i

«c. . ..

u;

“15».
IL _..o

 

u ..
. c

lb.‘ ’1

.41.:-

mi

*
.

1'”

 

.1 :4
315

 

6:376 ‘ ‘
25:59.; «4.49"

’3‘.

 

THESIS

_ l
.14; (I)

This is to certify that the
dissertation entitled

Pressures, Priorities, and Legislative Behavior:
Roll-Call Voting in The House of Representatives

 

presented by

Suzanne M. Gold

LIBRARY
Michigan State
University

 

 

 

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in Political Science

 

 

 

Major Professor’s Signature

3 I‘M-r 2007

 

Date

MSU is an afiinnative-action, equal-opportunity employer

 

' PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/C|RCIDaleDue indd-p.1

PRESSURES, PRIORITIES, AND LEGISLATIVE BEHAVIOR:
ROLL-CALL VOTING IN THE HOUSE OF REPRESENTATIVES

By

Suzanne M. Gold

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Political Science

2007

ABSTRACT

PRESSURES, PRIORITIES, AND LEGISLATIVE BEHAVIOR:
ROLL-CALL VOTING IN THE HOUSE OF REPRESENTATIVES

By

Suzanne M. Gold
In this three-essay dissertation, I test the bonds of representation between members of the
House of Representatives and their constituents by employing the 2000 National
Annenberg Election Survey, which has yet to be fully exploited in political science.
First, I argue that district preference, party pressure, and national preferences have a
significant and independent impact on representative voting behavior. Because of
progressive ambition, members of the House balance national preferences with party
pressure and district preference when voting on the floor. Next, I posit that representative
are responsive to district preference on salient issues but not responsive to district
preference on non-salient issues. Looking at the 2000 election cycle, where education
was the most salient issue, and defense spending was one of the least salient issues, I find
district preference on education has a significant and independent effect on
representative’s education spending roll-call votes. Finally, I illustrate that incongruence,
or mismatch, between district preference and House member roll-call behavior entices
quality challengers and causes a decrease in incumbent vote share. More specifically,
representatives that are considered extreme, consistently vote the party line, and deter
quality challengers resulting in an increase in vote share because voters actually prefer

more extreme incumbents.

This dissertation is dedicated to those who supported me and believed in me - even when
I did not believe in myself.

iii

ACKNOWLEDGEMENTS

First, I would like to thank my friends and family for staying by my side through this
long process. Thank you for your support, understanding and faith in me. I also would
like to thank my better half, Mark, who provided the love and encouragement I needed
to keep going. Next, I thank my PIPC colleagues, especially Greg Robinson, Mike
Crespin, and Yael Shomer who were all important contributors to my growth as a
scholar. I give a special thanks to Alissa Galford for her editing skills and extremely
quick turn around. Finally, I would like to thank three professors who contributed to my
academic development. First, to Burt Monroe whose support was invaluable. He has
made me a better Political Scientist and a better colleague. Second, Nathan Monroe took
me under his wing when there was no one else there. Finally, I would like to thank Dave
Rohde for providing me with opportunities I otherwise would not have had. I never

would have made it through without his training, encouragement, and kindness.

iv

TABLE OF CONTENTS

LIST OF TABLES ............................................................................ VII
LIST OF FIGURES ........................................................................... VIII
CHAPTER 1 — STRIKING A BALANCE: CONFLICTING PRESSURES AND
LEGISLATIVE VOTING, 1994-2004 ..................................................... 1
Introduction ........................................................................... 1
Party .................................................................................... 3
District ................................................................................. 5
Progressive Ambition ................................................................ 6
Combining Pressures ................................................................ 7
Striking a Balance .................................................................... 8
Data .................................................................................... 8
Measures .............................................................................. 9
Results ................................................................................. 9
How Are Members Voting? ......................................................... 10
Estimating the Model ................................................................ 1 1
Value of the Mixed Model .......................................................... 12
Conclusion and Discussion ........................................................ 14

CHAPTER 2 — EDUCATION V. DEFENSE: ISSUE SALIENCE AND THE 106TH

CONGRESS .................................................................................... 24
Introduction ........................................................................... 24
Congressional Decision Making ................................................... 27
Voter Preferences, Salience, and Representative Behavior ..................... 28
Education and Defense Spending in the 106th Congress ........................ 31
Data .................................................................................... 31
Measures .............................................................................. 32
Results ................................................................................. 33
Issue Salience and Legislator Behavior ........................................... 33
Additional Issues ..................................................................... 35
Issue Salience and Overall Behavior .............................................. 36
Conclusion and Discussion ........................................................ 38

CHAPTER 3 — EXTREMITY AND IN CONGRUENCE: THE IMPACT OF PARTY

AND DISTRICT PREFERENCE IN HOUSE ELECTIONS, 1994-2000 ............ 44
Introduction ........................................................................... 44
Voter Information and House Behavior ........................................... 46
Previous Research on Electoral Punishment ...................................... 47
Extremity, Incongruence, and House Elections .................................. 51
Data .................................................................................... 52
Measures .............................................................................. 52

Results ................................................................................. 54

Phase 1: Enter the Quality Challenger ............................................. 55
Phase 2: Decrease in Vote Share ................................................... 57
Conclusion and Discussion ......................................................... 62
APPENDIX .................................................................................... 70
REFERENCES ................................................................................ 71

vi

LIST OF TABLES

Chapter 1
Table 1 - Expectations for Models of Roll-Call Influence ................................. 17

Table 2 - Party, District, and National Forces on Roll-Call Behavior 1994-2004. . 1 7

Table 3 - District Preference and Roll-Call Behavior (without Party) 1994-2004. . 1 7

Chapter 2

Table 1 - Education Spending and Legislative Behavior .................................. 41
Table 2 - Defense Spending and Legislative Behavior .................................... 41
Table 3 - Additional Issues, Salience, and Legislator Behavior .......................... 42
Table 4 - Issue Preference, Salience, and Overall Legislator Behavior .................. 43
Chapter 3

Table 1 - Extremity, Incongruence, and Quality Challengers (1994-2000) ............ 64
Table 2 - Probability of Drawing A Quality Challenger (1994-2000) .................. 64
Table 3 - Extremity, Incongruence, and Vote Share (1994-2000) ....................... 65
Table 4 - Predicted Values of Vote Share (1994-2000) ................................... 65

Table 5 - Extremity, Incongruence, and Vote Share (1994-2000) with Party of the
Representative ................................................................................. 66

vii

LIST OF FIGURES

Chapter 1

Figure 1 — Partisan Model ..................................................................... 18
Figure 2 - District Downsian Model ......................................................... 19
Figure 3 — National Downsian Model ........................................................ 20
Figure 4 — Mixed Model ....................................................................... 21
Figure 5 — Data Plotted Against Mixed Model ............................................. 22
Figure 6 — Estimated Mixed Model .......................................................... 23
Chapter 3

Figure 1 - National Downsian/Moderation Model ......................................... 67
Figure 2 — Extremism/Party Model ........................................................... 68
Figure 3 — District Partisanship and Presidential Vote, 2000 ............................. 69

viii

Striking a Balance: Conflicting Pressures and Legislative Voting
1994-2004

“What is the primary purpose of a political leader? To build a majority. If voters care
about parking lots, then talk about parking lots. ”

-Newt Gingrich
Introduction

As political leaders, members of the House of Representatives wear many hats.
Members of the House are representatives of their parties who are constantly trying to
work the majoritarian system, while concurrently acting as transplanted locals from their
districts. Toeing both the party and constituency lines creates a delicate juggling act for
members of the House. Stay in step with both, and members can help ensure their seat
safety. However, when balancing these two often times conflicting pressures, which one
weighs heavier? Ansolabehere, Snyder and Stewart (2001) examined this question in the
context of candidate positioning and found, “congressional candidates have primarily
espoused the ideology associated with the national party, moderating very little to
accommodate local ideological conditions” (136). In other words, candidates position
themselves closer to the national party than their own individual districts. In this instance.
party wins over district. However, if members of the House stray too far away from
district preference in favor of the party line, they face lower vote shares (Canes-Wrone,
Brady, and Cogan (2002). Moreover, another dimension to this argument that has
received much less attention in the literature on representative decision-making is that of
the Median Voter Theorem. Downs (1957) told us that at the national level, in order to
garner as many votes as possible, candidates should converge to a common point in the

middle. In addition, Canes-Wrone, Brady, and Cogan (2002) found that representatives

who voted moderately were more successful at the polls. Between the party, the district,
and the nation, where should a candidate in a US. House election place themselves to
keep all actors happy? The answer I provide -— between all three. Members need to take
into account all three.

In this paper, I argue that in order for members of the House to ensure seat safety,
they should balance all three aspects. The party, the district, and the nation do exercise
more or less influence across elections, but overall, members will balance all three in
order to protect their electoral fortunes. Representatives do not have to sacrifice the party
for the voters, both at the district and national level, or vice versa. They can consciously
balance it all in order to ensure electoral fortunes. With reelection as the proximate goal
of members of Congress (Mayhew 1974), securing ones seat is essential to gaining
seniority, building a reputation with the voters, and eventually running for higher office.
While literature has separately looked at the influence of party (Cox and McCubbins
1993, 1999; Aldrich 1995; Jacobson 1999; Alrdich and Rohde 1999, 2001; Aldrich,
Berger, and Rohde 2002; Sinclair 2002) and constituency (Miller and Stokes 1963;
Jackson 1971; Kingdon 1973; Mayhew 1974; Fiorina 1974; Fenno 1978; Page et al.
1984; Wright 1989; Jackson and King 1989; Arnold 1990; Bartels 1991; Bianco 1994)
over congressional roll-call behavior, none has looked at them coupled with national
constituency pressures. Because progressive ambition can affect House member
behavior (Hibbing 1986), we should expect members to consider present national
preferences in preparation for relying on larger a constituency for future election to
higher office. Candidates must balance the idea of the national Median Voter Theorem

with party and district in order to safeguard current and future electoral fortunes.

In this study, I examine incumbent positioning during the 103'd-108‘h Congresses
(1994-2004). Using roll-call data, I look at the impact of the nation, the party, and
districts on the voting behavior of members of the House of Representatives. Previous
literature has told us that all three are important, but which one do the representatives
themselves think is the most important? Do they balance them, and if so, how? Building
upon the findings of Ansolabehere, Snyder, and Stewart (2001), I add a discussion of the
importance of national constituency when it comes to legislative voting. Members are
aware of the possibility of punishment from all sides, and in order to secure reelection
now and in the future, will not give the voters or the party any reason to turn to another
candidate. The remainder of the paper is organized as follows: The first section outlines
the literature on congressional decision-making and position taking. The second section
discusses the data and measures used in this study. Next, I discuss the results, and finally
I offer conclusions and comments on future research.

Party

Members of the House face party pressure for several reasons: representatives
feel compelled to vote with the party to improve the party reputation (Cox and
McCubbins 1993), create rules that benefit the party (Sinclair 1998), provide more power
to the leadership through party homogeneity (Aldrich and Rohde 1999, 2001), and
advance the policy and electoral goals of the majority party (Kiewiet and McCubbins
1991, Rohde 1991, Cox and McCubbins 1993, Rohde and Aldrich 1995, Sinclair 1995,
Aldrich, Berger, and Rohde 2002, Sinclair 2002). Because the House is a majoritarian
institution in which the party in power has almost complete institutional control, the

single goal of the majority party is to stay in the majority, while the single goal of the

minority party is to gain the majority. Therefore, members of the House should always
vote the party line. The following is a graphical representation of the relationship
between congressional voting and district preference if members of the House were to

always vote the party line.1

[Figure l About Here]

Members who do not toe the party line in order to further the policy and electoral
causes of the party may face some sort of punishment. Since the party leadership
controls committee assignments (Cooper and Brady 1981), rank-and-file members know
that party loyalty is the best way to ensure being assigned to their desired committees.
The party leadership also can impact the diffusion of pork to congressional districts
(Krutz 2001), so party loyalty help keeps pork coming into a representative’s district.
Finally, and most severely, members who continually buck the party line run the risk of
losing party support in the primary. For example, The Republican Party abandoned
Representative Joe Schwarz (R-MI) in favor of Tim Wallberg in the 2006 Republican
primary. Schwarz was considered a more moderate Republican, and after only one term,
the Republican Party decided to throw their support behind a more conservative
candidate. As a result, Schwarz lost his seat. The Republican Party discarded Schwarz
for a candidate that would more closely toe the party line, and that shift in support lead to
Schwarz’s defeat. Members of the House could face the same fate if they consistently

voted against the party.

 

' The y-axis represents House member behavior (0,1) where 0 is the most conservative and l is the most
liberal. The x-axis represents district preference as Republican presidential vote in the district (0,1), where
0 is the most liberal district and l is the most conservative.

District
While party exerts a significant force on congressional decision-making, it is

tempered by constituency constraints (Jacobson 1999). There is an interactive
relationship between party-line voting and constituency preference. Therefore, party
pressure must be coupled with district preference. Many scholars have argued that
constituency influence does motivate legislative behavior (Jackson 1971; Kingdon 1973;
Mayhew 1974; Fiorina 1974; Fenno 1978; Page et a1. 1984; Wright 1989; Jackson and
King 1989; Arnold 1990; Bartels 1991; Bianco 1994). According to these narratives, in
order to secure reelection, members of the House should vote according to district
preference. The following is a graphical representation of the relationship between
congressional floor behavior and district preference if representatives always voted with

their district.

[Figure 2 About Here]

While the party can exercise punishment over members, the most devastating
punishment comes from voters. The best way to punish a member of Congress is simply
to vote for the challenger (Erikson and Wright 1980; Page, et al. 1984; Stimson,
MacKuen, and Erikson 1995; Ansolabehere, Snyder, and Stewart 2001; Canes-Wrone,
Brady, and Cogan 2002). Empirical evidence showed that House members vote closely
with the preferences of their district in order to avoid losing electoral support (Miller and
Stokes 1963; Wright 1981; Jackson and King 1989; Kingdon 1989; Arnold 1990; Bartels
1991; Jacobson 1992; Overby et al. 1992), and do believe that their behavior on the floor

has an impact on their chances of winning reelection (Matthews and Stimson 1975;

Aldrich 1995; Jacobson 1996; Ansolabehere, Snyder, and Stewart 2001; Canes—Wrone,
Brady, and Cogan 2002). Members of the House must balance party and constituency in
order to have the best possible chance to retain their seat.
Progressive Ambition

While the national constituency usually is discussed in terms of presidential and
even senatorial elections, it is a consideration for members of the House if they want to
run for higher office in the future. Rohde (1979) described progressive ambition as a
politician who holds an office and attempts to gain another regarded as more attractive.
Almost all members of the House hold this ambition. The expected value of running for
an office is the function of the probability of winning, the value of the office, and the cost
of running (Rohde 1979). A member only will run if the expected utility is positive. Part
of the probability of winning is the way in which the larger constituency views the
candidate. A member of the House with his or her sights set on higher office must take
into consideration the preferences of the larger constituency from almost the very
beginning of their tenure. Lublin (1994) argued that challengers act strategically when
deciding to enter based on local and national forces. Incumbents must consider those
national forces when voting on the House floor in preparation for running for a higher
office. Strategically, Downs (1957) told us that the best way to encompass as many
voters as possible is to moderate. Previous research posited that members of the House
who moderate are less likely to be punished at the polls (Schoenberger 1969; Erikson
1971; Johannes and McAdams 1981; Erikson and Wright 1993; Brady et a1. 1996;
Jacobson 1996; Canes-Wrone, Brady, Cogan 2002). Therefore, when progressive

ambition is at play, members should place themselves directly in the middle of the

national spectrum. The following is a graphical relationship between congressional floor

behavior and district preference if a member always followed national preferences.

[Figure 3 About Here]

Combining Pressures

The best way to ensure reelection in the present and future is to balance district,
party, and national preferences. Sinclair (2001) found that members of the House vote in
a manner that balances party, policy, and district goals. Crespin, Gold, and Rohde (2006)
also argued that multiple factors such as party and district have independent and
significant effects on House floor voting. Members of the House are aware the party
(Cooper and Brady 1981) and the voters (Kingdon 1989) are constantly monitoring their
voting behavior, and therefore, the best way to satisfy all actors involved is to balance the
preferences of those actors equally. The following is a graphical representation of the
relationship between congressional floor behavior and district preference if members of

the House balance all three.

[Figure 4 About Here]

Figure 4 is a hybrid of the previous three figures. The slopes of the lines in Figure
4 merely are an equal combination of the slopes of Figures 1, 2, and 3. In the case of
Figure 4, we see representatives giving consideration to voting the party line, voting the

district line, and moderating in order to appease as many national voters as possible.

Striking a Balance
The theoretical expectations that result from this literature are straightforward.

Members of the House want to be reelected both for their current seat and in the future
for a higher seat. In the face of party pressure, district pressure, and national forces, the
best way for representatives to secure their electoral fortunes is to balance these often
times competing and conflicting priorities. Punishment can come from all sides involved
if at any time one of those sides feels neglected. Parties will abandon representatives
either by revoking rewards or backing another candidate. Additionally detrimental is the
possibility that voters will cease to cast votes for House incumbents running for their
current seat or a higher seat who have not furthered their interests. Rather than
sacrificing one priority for another, representatives can appease all actors by weighing
party, district, and national goals. Therefore, in a multivariate context, the party, the
district, and the nation should all have an independent and significant effect on House
members’ voting behavior.
Data

To examine these theoretical expectations, I concentrate on the 103rd — 108’h
Congresses (1994-2004). During this time of Republican control, we saw a significant
increase in both the rate at which incumbents were getting reelected and the margin at
which they were winning (Abramson, Aldrich, and Rohde 2002). Therefore, incumbents
were keeping both the voters and the party happy. This time period is ideal in order to
determine the best strategy for representatives to continue to receive the support of the
party and the voters. What can we learn from the successes of the representatives from

the 103rd - 108th Congresses? How were they balancing party, district, and nation and

securing solid reelection? The answer I provide is that they were balancing all three
factors — party, district, and nation — equally.
Measures
District

I use Republican presidential vote measured as the two-party vote in the district
from the 1994-2004 elections as the measure of district preference.2 Congressional
scholarship has long used presidential vote in the district as a proxy for district
3

preference.

House Member
I use first dimension Dw-NOMINATE scores from the 103rd — 108th Congresses

(1994-2004) as a measure of overall legislator behavior on a liberal-conservative scale.
The scale goes from -1 to +1 (Liberal to Conservative) with extreme partisans at the
poles. For the purposes of linearity, I rescale the scores [0,1]. I also control for
representative partisanship by coding Democrats as 0 and Republicans as l.4

Nation
Because I am employing OLS regression, I will measure national preferences

using the constant. The constant is the predicted value of the dependent variable when all
other variables are 0. So in the model where a representative only follows the national
preference, the coefficient on party and district are 0, and therefore, the constant becomes
our indicator of national preference.

Results
In this section, I plot the data and discuss the empirical expectations. Next, I

present a series of OLS regressions and discuss the theoretical implications of my

 

2 I use the most recent two-party Republican presidential vote in the district as the measure of district
preference for each election.

Erikson and Wright 1989, 1993, 1997; Brady et al. 1996, Canes-Wrone, Brady, and Cogan 2002; Crespin,
Gold, and Rohde 2006.
‘ Bernard Sanders (I-VT) is coded as a Democrat.

findings. I expect the coefficients on the three predictors of House roll-call behavior
(district, party, and nation) to be positive and statistically significant.
How Are Members Voting?

According to the mixed model, and referring back to Figure 4, representatives
should strategically position themselves between party pressure, district preference, and
overall national preferences. In other words, they should combine voting the party line,
voting district preference, and moderating to garner as much of the national preference as
possible. The outcome, as represented in Figure 4, is one that appeases each individual
actor in play. Upon plotting the actual data from 1994-2004 against Figure 4, we see that

members are balancing according to my theoretical expectations.

[Figure 5 About Here]5

Figure 5 illustrates that members of the House are consciously taking into
consideration which actors they must cater to in order to secure present and future
electoral fortunes. Representatives were able to balance voters and party during this time
period when Republicans were cracking the party whip, and the nation was moving in a
more conservative direction. While Ansolabehere, Snyder, and Stewart (2001) found that
the district is abandoned in favor of the national party, it appears that representatives truly
do not have to pick and choose between the two. Members of the House do not have to

sacrifice district preference in order to show party loyalty. Moreover, representatives do

 

5 District preference is represented by Republican presidential vote in the district, and representative
behavior is represented by the l” Dimension DW-NOMINATE scores rescaled [0,1].

10

not have to run the risk of losing a pork project in order to appease the voters in either the
nation or the individual district.
Estimating the Model

Representatives in the House during the time period of this study are striking the
appropriate balance. When estimating a model of roll-call voting, what are the
expectations of the coefficients on the explanatory variables under the partisan, national
Downsian, district Downsian, and mixed models? Table 1 shows the expected
coefficients for party of the representative, Republican presidential vote in the district,

and the national preference (the constant) for all four models.

[Table 1 About Here]

In the partisan model, since representatives cast roll-call votes based solely on
where the party line is, the coefficient on party of the representative should be one, while
all other coefficients should be zero. In the district Downsian model, since members of
the House are following the median district preference, the coefficient on Republican
presidential vote in the district should be one, while all others should be zero. Looking
at the national Downsian model, the coefficient on the constant should be 0.50. This
represents a perfectly moderate member who is voting right in the middle in order to cast
the net as wide as possible. The predictive model, (the mixed model), provides
coefficient estimates that are one-third of each of the other three models. Theoretically, I
argue that members of the House balance party, district, and nation in order to secure
current and future seats of government. Balancing all three provides for coefficient

estimates that are one-third of the estimates of the other models. Therefore, in the mixed

11

model the coefficients on party of the representative and Republican presidential vote in
the district should be roughly 0.33, and the coefficient on the constant (national
preferences) should be roughly 0.17.

Value of the Mixed Model

Table 2 reports the results of an OLS regression with the 1St Dimension DW-
NOMINATE scores of all incumbents rescaled [0,1] as the dependent variable. The
independent variables are party of the representative, Republican presidential vote in the
district, and national voter preferences as measured by the constant. The results tell us a
few important things. First, holding all else constant, Republican representatives are
more likely to cast more conservative roll-calls. Second, the higher the Republican
presidential vote in the district, the more conservative a representative votes. The
relationship between party, district, and roll-call votes are as expected. The model has an
R2 of 0.82, telling us that party, district, and national influences explain 82% of the
variation of representative roll-call behavior.

Most importantly, these results illustrate that as predicted by the mixed model,
party, district, and the nation each exert a positive, independent, and statistically
significant effect in representative floor behavior. The coefficient on party of the
representative (0.31) is almost identical to the expected estimate (0.33). The coefficient
on district preference (0.45) is a bit higher than predicted by the mixed model (0.33).
Nonetheless, this is not a surprise since voters in the district exercise the most immediate
and damaging electoral threat over members of the House. If a member had to choose
between losing a committee assignment or losing his or her seat, the choice is simple — go

with the district to avoid the electoral threat. In regard to national preferences. the

12

coefficient on the constant (0.14) is very close to the estimate provided by the mixed
model (0.17). Because progressive ambition is held by most members of the House, the
national constituency is in their minds right along with their party and their districts. The
national constituency helps temper the extremity of voting the party line as well as voting

with the popular passions of individual districts.

[Table 2 About Here]

To add to the predictive value of the mixed model, Figure 6 plots the individual
party regression lines from the results in Table 2 against the plotted data from Figure 5.
As the figure shows, while the regression lines may be a bit steeper, the regression results
follow the estimates from the mixed model almost perfectly. This also is a graphical
representation of the slightly stronger relationship between district preference and roll-
call behavior. The lines are steeper because the districts are pulling their representatives
in their direction than the direction of the party. This is exactly the opposite of what
Ansolabehere, Snyder, and Stewart (2001) argue. During the time period they study
(1976-1996), members may have abandoned their districts in favor of the national party,

but since then, members have been more responsive to the district than the party.

[Figure 6 About Here]

In order to further analyze the slightly stronger effect of district preference and
roll-call voting, in Table 3 I run a bivariate OLS regression of Republican presidential

vote in the district on the rescaled DW-NOMINATE to examine the predictive value of

13

adding party of the representative to the model. The starkest difference is the coefficient
on the national preferences (constant). In the model without party, the relationship
between national preferences and roll-call behavior is no longer statistically significant.
Perhaps most importantly, the R2 drops dramatically from the 0.82 in the mixed model to
0.53. While the results in Table 2 illustrate that district preference is marginally more
important than party, not including it in a model of roll-call voting dramatically decreases
its predictive value. Party is one of the most important determinants of congressional

roll-call voting coupled with district and national forces.

[Table 3 About Here]

Conclusion and Discussion
Former Speaker Gingrich was correct when he discussed the importance of party

and constituency. Building, keeping, or trying to win back the majority in the House of
Representatives requires party loyalty from all members. Nonetheless, district
preferences and pressures temper this allegiance. Voters expect that their elected
representatives will reflect their preferences in their floor behavior, even in the face of
party pressure. Parties expect that their rank-and-file members will reflect their
preferences in their floor behavior, even in the face of district pressure. Ansolabehere,
Snyder, and Stewart (2001) found that in the face of these two pressures, district
preference is abandoned for the national party. What I have presented tells a different
story. Representatives can balance the two conflicting priorities in a manner that keeps
both happy. In fact, my empirical results show that representatives are choosing to

follow district preference more than party loyalty. Even during a time when party loyalty

14

was more important than ever, (1994-2004), both Republican and Democratic members
of the House still balanced party and district.

However, there is another dimension of this juggling act that had been previously
unexamined. The Median Voter Theorem argued that at the national level, candidates
should converge to a common middle point to gain as many votes as possible. Why
would members of the House have to take the national constituency into consideration if
they are only running in their specific districts? Progressive ambition, held by almost all
members of the House, makes it extremely important to begin voting early on in a
manner that is attractive to as many voters in the national constituency as possible.
Moreover, casting the net as wide as possible by moderating cannot come at the expense
of appeasing the extreme voting the party requires and the sometimes varying voting the
districts want. All three aspects must be in a representative’s voting calculus. This
mixed model predicts representative roll-call behavior between 1994-2004 extremely
well. While the empirical results show that representatives do favor district over party,
withholding party in a model predicting roll-call behavior dramatically decreases its
predictive value. Members of the House of Representatives know they are being
monitored and face punishment from all sides; therefore, they consciously consider all
the sides when casting a roll-call vote.

Examining the different forms of punishment from the party, the district, and the
nation is the next step in analyzing the factors that influence representative roll-call
behavior. We would expect members of the House who neglect the party to receive less
pork, not receive preferred committee assignments, and even lose party support in the

primary. Additionally, we could see the entrance of a quality challenger in the primary

15

and the general election as a form of district electoral punishment. Loss of vote share and
even loss of reelection are the most extreme forms of electoral punishment, and if a
representative is out of step with district preference for too long, this can occur. Finally,
those members of the House who do run for higher office, and have not considered the
national constituency may have a more difficult time getting elected than those
incumbents who took into account national preferences. Examining punishment is the
next step in appreciating the equilibrium that members of the House must maintain in
order to secure current and future electoral fortunes. This delicate balance between
conflicting and competing forces impacts all aspects of congressional politics. From roll-
call voting, to committees, to elections, representatives must juggle it all in order to

achieve the ultimate goal — retaining or gaining a seat.

16

Tables

Table l: Expectations for Models of Roll-Call Influence

 

 

Partisan District National Mixed
Republican Presidential Vote 0 l O 0.33
Party of the Representative 1 0 0 0.33
Constant 0 0 0.50 0.17

 

 

 

 

 

 

 

Table 2: Party, District, and National Forces on Roll-Call Behavior

 

 

 

 

 

 

1994-2004
Party of the Representative 0.3 l ***
(0.005)
Republican Presidential Vote 045*"
(0.016)
Constant 0.14* "' *
(0.007)
N 2587
F 5890.1 1*"
Root MSE 0.09
R2 0.82
Dependent variable —- 1sf Dimension DW-NOMINATE score,
rescaled to [0,1].
*** p < 0.001

(Robust standard errors in parentheses)

Table 3: District Preference and Roll-Call Behavior (without Party)

 

 

1994-2004

Republican Presidential Vote 1.14* It 1:

(0.02)
Constant _0.03

(0.01)
N 25 87
F 2867.36***
Root MSE 0.15
R2 0.53

 

 

 

 

Dependent variable — 1St Dimension DW-NOMINATE score,
rescaled to [0,1].
*** p < 0.001

(Robust standard errors in parentheses)

l7

Figures

Figure]
0.
'—
co.
O
h—
.9
E o.
(D O
m
0)
>
’53
.9
C
3 v
93 o‘
D.
0
CK
N.
O
O.
O

 

 

 

 

 

Partlsan Model
Republicans
Democrats
I l j l l l
0.0 0.2 0.4 0.6 0.8 1.0

District Preference

18

 

Figure 2

 

 

 

 

Dlstrlct Downsian Model
‘3 4 /
/"
/'/
/,

f; - /
§ l//
E (Q 4 /,.,
<1) 0 /.
ED /
a) /
.> /
E /
C
a v /
9 o' T /
g /
05 //

/'
8 i /./
/
,/
,/
/
8- i '
I I I I I l
0.0 0.2 0.4 0.6 0.8 1.0

District Preference

l9

Figure 3

1.0

0.8

0.6

Representative Behavior
0.4

0.2

0.0

National Downsian Model

 

 

 

 

I I

0.4 0.6

District Preference

20

0.8

1.0

 

Figure 4

 

 

 

 

Mixed Model
0. _
Republicans

00 /

O. _
.§
E «a. _
(D O
m
a)
.2
is
E
8 v
9 o' ‘
O.
a)
o:

N. _

0

Democrats
Q
C)
I I I l I l
0.0 0.2 0.4 0.6 0.8 1.0

District Preference

21

Figure 5

Mixed Model

 

 

 

Republicans

Democrats

 

 

oe

_
wd

_ _
no to
$22.20; 28w 88> __8__om

0.0

1.0

0.8

0.6

0.4

0.2

0.0

Republican Presidential Vote

22

Figure 6

Mixed Model, Estimated

 

1.0
9

Republicans

 

0.8

0.6

Rollcall Voting Score (NOMINATE)
0.4

0.2

 

0.0

 

 

I I I I , l I

0.0 0.2 0.4 0.6 0.8 1.0

Republican Presidential Vote

23

Education v. Defense: Issue Salience and the 106th Congress

Introduction
In a survey in the late 1970’s of more than 150 members of the House of

Representatives, a majority of those surveyed agreed with the statement, “I seldom have
to sound out my constituents because I think so much like them that I know how to react
to almost any proposal” (Davidson and Oleszek 2006, p. 122). This implies that
members of the House know their districts so well that they will almost always behave in
Washington the way their district wants. However, such congruence, agreement between
district preference and legislator behavior is not automatic. Conflicting pressures and
priorities faced by members of the House, coupled with district changes, create a tenuous
relationship between the representative and the represented. Therefore, the question
becomes: Under what circumstances do members of the House best reflect the
preferences of their district? Miller and Stokes (1963) were the first to provide
quantitative evidence that there is partisan, ideological, and issue preference congruence
between congressional representatives and their voters. While subsequent studies have
discussed this relationship, there are some dynamics that have been unexplored due to
data limitations.

One such dynamic is the importance of issue salience on voters’ evaluation of
representatives. The 2000 National Annenberg Election Survey (N AES) asked the
question, “In your opinion, what is the most important problem facing our country
today?” The answer given most often was education. During this time, the Elementary
and Secondary Education Act of 1965 was due to be renewed, and the federal
government had been discussing not only renewing it, but revamping it. This issue was

on the minds of the voters and elite members of government. Conversely, according to

24

the NAES, the least important problem facing the nation during this same time period
was defense spending. Since the end of the Cold War, Americans had turned their
attention to domestic issues such as education, social welfare programs, and the
environment (National Annenberg Election Survey). Defense issues were not at the
forefront of the public conscience or the public debate because the country was not
involved in a war or international crisis. Because voters did not see issues of defense as
salient and did not form strong preferences, there would be no expectation that voters
would punish members of Congress for voting with the party or their own personal
ideology and preferences. However, since education was the most salient issue facing the
nation during this time, would voters hold an electoral threat over representatives?

There has been much research examining the theoretical influence of issue
salience on legislator behavior. While most scholars agree that the more salient the issue,
the more responsive a member of Congress should be to district preference, there have
been difficulties quantifying which issue or issues are the most salient to the district. The
main contribution of this study provides for an alleviation of such data concerns by
employing reliable constituency preference data. In this study, I utilize the 2000 National
Annenberg Election Study (NAES). Most research examining the relationship between
constituency opinion and representative behavior has used the National Election Study
(NES) to approximate district opinion (Page et a1. 1984; Hurley 1989; Bartels 1991; Box-
Steffensmeier et a1. 2003), or presidential vote in the district (Erikson 1971; Deckard
1976; Hurley and Kerr 2000; Ansolabehere, Snyder, and Stewart 2000, 2001). Using the
NES is not a reliable measure since there are so few respondents per congressional

district. There are excluded districts and those that are included have only a handful of

25

respondents at best. Using presidential vote in the district as a proxy for district
preference is problematic because it varies both by the distribution of district preferences
and the positions of the presidential candidates (Clinton 2006). Most importantly, these
two measures capture partisan preference not issue preferences. While the NES does ask
issue salience questions, there are so few respondents per congressional district and not
all districts are represented. However, the NAES, provides a substantial increase in
respondents per district and includes all districts, thereby increasing the reliability of a
measure of district issue salience. Using the NAES, I employ both open and closed-
ended policy questions posed to more than 79,000 respondents to assess district policy
preferences and aggregate the individual responses to determine district preferences. The
large amount of respondents per congressional district provides an unprecedented
estimate of district preference and issue salience and a reliable insight into whether those
preferences impact legislator behavior.

This paper analyzes the relationship between district issue preference, issue
salience, and representatives’ roll-call votes. Using the 2000 National Annenberg
Election Survey, I examine the extent to which members of the House of Representatives
in the 106’h Congress (1998-2000) voted with district preference on both salient and non-
salient issues. The paper is organized as follows: The first section discusses
congressional behavior and voter information, and congressional decision-making on all
issues and specifically on salient issues. The second section outlines the survey data and
measures used to examine the theoretical relationship between representative behavior

and district preference. The third section presents the results from the empirical analyses.

26

And finally, I discuss the implications of the results for congressional representation as
well as future research.
Congressional Decision Making

Since Miller and Stokes (1963), additional research has argued that constituency
influence does motivate legislative behavior (Jackson 1971; Kingdon 1973; Mayhew
1974; Fiorina 1974; Fenno 1978; Page et a1. 1984; Wright 1989; Jackson and King 1989;
Arnold 1990; Bartels 1991; Bianco 1994). Other scholars have stated that congressional
members simply follow their own personal ideologies (Bernstein and Anthony 1974; Kau
and Rubin 1979, 1993; Kalt and Zupan 1984; Peltzman 1984; Dougan and Munger 1989;
Bernstein 1989; Poole and Rosenthal 1991; Poole and Romer 1993). However, it is not
so simple. Literature also has argued that congressional members’ behavior is influenced
by a combination of multiple factors such as party pressure (Cox and McCubbins 1993;
Aldrich, Berger, and Rohde 2002; Sinclair 2002), personal preferences (Aranson and
Ordeshook 1972; Aldrich 1983, Wittman 1983; Calvert 1985; Francis, Kenny, Morton,
and Schmidt 1994; Crespin, Gold, and Rohde 2006), party activists (Aldrich 1995;
Fiorina 1999; Jacobson 2000; Layman and Carsey 2000; Aldrich and Rohde 2001), and
policy orientation (Kingdon 1973; Fenno 1973; Rohde 1991; Aldrich 1995; Aldrich and
Rohde 2000). Crespin, Gold, and Rohde (2006) found that all of the above factors play a
part in congressional member behavior, but the district median opinion has an
independent and significant impact on legislator behavior.

Knowing that there are multiple and conflicting priorities and pressures driving
congressional members’ behavior, the most effective way for constituents to keep their

member of the House representing their views when voting is simply to vote for the

27

challenger (Erikson and Wright 1980; Page, et al. 1984; Stimson, MacKuen, and Erikson
1995; Ansolabehere, Snyder, and Stewart 2001; Canes-Wrone, Brady, and Cogan 2002).
Since voters evaluate representatives based on their floor behavior, (and members know
this), then the electoral threat is both present and also keeps members of the House inline
with their districts’ preferences.

Voter Preferences, Salience, and Representative Behavior

The debate as to whether or not voters even know or understand issues and
candidates’ policy position has been ongoing since the 19603. While some research
argued that voters did not understand policies, issues, or the differences between political
candidates (Converse 1964; Popkin 1991; Sniderman, Brody, and Tetlock 1991) and that
voters were incapable of educating themselves in order to become informed on policies
and issues (Converse 1964; Sniderrnan 1993), more recent research has found that voters
can use cues or heuristics to educate themselves and make more informed judgments
regarding policy (Carmines and Kuklinski 1990; Grofman and Norrander 1990; Mondak
1993; Lupia 1994; Lupia and McCubbins 1998). With the presence of cues, citizens are
capable of learning in order to form preferences and express those preferences to elected
government officials (Druckrnan 2001).

Moreover, the representatives themselves must believe that the voters in their
district are paying attention to their actions in Washington. Kingdon (1989) argued that
members of Congress are aware that at least some voters in their district are informed and
monitoring legislative behavior and evaluate their members of the House accordingly.
Empirical evidence found that House members vote closely with the preferences of their

district in order to avoid losing electoral support (Miller and Stokes 1963; Wright 1981;

28

Jackson and King 1989; Kingdon 1989; Arnold 1990; Bartels 1991; Jacobson 1992;
Overby et al. 1992) and do believe that their behavior on the floor has an impact on their
chances of winning reelection (Matthews and Stimson 1975; Aldrich 1995; Jacobson
1996; Ansolabehere, Snyder, and Stewart 2001; Canes-Wrone, Brady, and Cogan 2002).
Therefore, representatives are concerned with congruence when reelection hinges upon it
because the higher the congruence, the more likely a representative is to get reelected.
According to this narrative, members of Congress are concerned with the electoral
impact of their roll-call behavior.

Furthermore, members of Congress do monitor constituency opinion in order to
foresee and evade problems of casting the “wrong” vote, especially when the specific
issue is salient to the district (Kingdon 1989). The degree to which issues matter is based
on the level of saliency (Berelson, Lazarsfeld, and McPhee 1954; Schattschneider 1960;
Kuklinski and Elling 1977; McCrone and Kuklinksi 1979; Alpert 1979; Page and Shapiro
1983; Wlezien 1995; Jacobs and Page 2005). The salience of an issue dictates whether a
representative will take on the role of trustee, delegate, or politico. The issues at the
forefront of the constituency’s mind will trigger representatives to act more as delegates
and follow the instructions of his or her constituency.6 Representatives are more likely to
vote with their district on a particular issue when it is salient (Hutchings 1998; Kollman
1998; Canes-Wrone 2001). Members' roll-call behavior regarding fairly salient issues
may have a larger electoral impact than votes on less salient issues (Canes-Wrone, Brady,
and Cogan 2002). When the weight of an issue increases, so should representative

responsiveness.

 

6 For a further discussion of issue saliency and representational role theory, see Kuklinski and Filing (1977)
and McCrone and Kuklinski (1979).

29

Drawing upon the Hill and Hurley (2003) issue classifications for the specific
issues in this study, we can categorize education as an easy cross-cutting issue since it is
uncomplicated and the parties held similar positions. These types of issues call for the
one-way linkage from the masses to elites (Hill and Hurley 2003). It is expected that
representatives would be responsive to district preference since members of Congress
who are unresponsive to their district’s opinion on a salient issue are more likely to be the
target of electoral punishment (Schattschneider 1960). Again using Hill and Hurley’s
classification, defense is a complex party-defining issue. Defense and foreign policy
issues are the most difficult for voters to understand, and the parties take distinct stances
that are integral to their platforms. This type of issue calls for a one-way linkage of elites
to the masses (Hill and Hurley 2003). In this case, we would not expect representative
responsiveness since the elites make the decisions.

Because of the NAES, we are able to obtain a cOnsistent measure of district issue
preference to first garner reliable insight into the most salient issue areas, and second,
determine if district issue preference on such issues has a significant and independent
effect on legislator behavior. The congruence between district preference on salient
issues and legislator behavior should be higher than district preference and non-salient
issues and legislator behavior because the more salient the issue, the higher the electoral
threat. In a multivariate context, when controlling for district partisanship, district issue
preference on salient issues should have an independent and significant effect on

legislator behavior, while district issue preference on non-salient issues should not.

30

Education and Defense Spending in the 106th Congress
The theoretical expectations that develop out of the previous research are

straightforward. When voters feel an issue is salient, voters do evaluate representatives
based on their floor behavior and expect their issue preferences to be reflected in roll-call
votes. When this is the case, we observe congruence. Conversely, when voters feel an
issue is not salient, voters do not evaluate representatives based on their floor behavior
and representatives have more freedom to act as a trustee. Therefore, holding all else
constant, because education spending is a salient issue, district preference on education
spending should have a significant and independent impact on legislator floor behavior.
On the contrary, since defense spending is a non-salient issue area, district preference on
defense spending should not have a significant and independent impact on legislator floor
behavior. Congruence is absolutely necessary when voters require it, but when it comes
to non—salient issue areas, representatives can act as trustees.
Data

To evaluate these theoretical expectations, I focus on the 106th Congress (1998-
2000) and the 2000 House of Representatives elections. This allows me to measure
district preferences using individual-level data from the 2000 National Annenberg
Election Survey (NAES). The survey employed nationwide random digit dialing
sampling frames executed between December 1999 and January 2001. As mentioned
previously, there were around 79,000 respondents with 20 (FL-23) to 227 (MI-1) per
congressional district. The respondents were asked a series of questions regarding the
2000 presidential and congressional elections as well as governmental policies and
demographic questions. Some of the questions included favorability of increasing or

decreasing taxes, Social Security and Medicare benefits, and program spending, as well

31

as demographic questions such as age, race, religion, and partisanship. For the purposes
of this study, I utilize the education and defense spending questions as well as partisan
ideology questions to obtain district ideology and district issue preferences on salient
(education) and non-salient (defense spending)issues. The NAES provides an
unprecedented estimate of district issue preference in order to examine the relationship
between district preference, issue salience, and congressional behavior.
Measures
District

To control for overall district ideology, I construct mean district ideology using
the ideology of each respondent in a congressional district as asked on the NAES. 7 To
measure district issue preferences on salient and non-salient issues, I use the education
and defense spending questions on the NAES to construct mean district education and
defense spending preferences. The education spending question asks, “Providing
national assistance to public elementary and secondary schools — should the federal
government spend more money on this, the same as now, less, or no money at all?”8 The
defense spending question asks, “Maintaining a strong military defense—should the
federal government spend more money on this, the same as now, less or no money at
all?” 9 I drop those who answered “Don’t Know” or “No Answer.”
House Member Behavior

1 use First Dimension DW-NOMINATE scores from the 106th Congress (1998-2000)

as a measure of overall legislator behavior on a liberal-conservative scale. The scale goes

 

7 The NAES asks, “Generally speaking, would you describe your political views as: very conservative,
conservative, moderate, liberal, or very liberal?” l= Very Liberal, 2=Liberal, 3=Moderate,
4=Conservative, 5=Very Conservative.

8 The question is scaled l to 4: 1=More, 2=Same, 3=Less, 4=Nothing

9 The question is scaled 1 to 4: 1=Nothing, 2=Less, 3=Same, and 4=More.

32

from -1 to +1 (Liberal to Conservative) with extreme partisans at the poles. Using the
Rohde/PIPC Roll Call Database, I search bills in the 106th Congress (1998-2000) to
construct a W-NOMINATE score for each representative based on how they voted on
education and defense spending.10 I study the 106th Congress because respondents from
the 2000 NAES evaluate their representatives based on their floor behavior between

1998-2000.

Results

In this section I present a series of regression models examining the impact of
district issue preference, both salient and non-salient, on issue-specific roll-calls as well
as on overall representative behavior. I expect district issue preference on salient issues
to have an independent and significant impact on legislative behavior, while district
preference on non-salient issue areas will not have a significant and independent impact.
Issue Salience and Legislator Behavior

In Table 1, I present OLS regression results with representatives’ education w-
NOMINATE scores from the 106th Congress (1998-2000) as the dependent variable. The
main explanatory variable is the mean district preference on education spending, while
the control variables are overall district ideology and party of the representative. As I
hypothesized, because education spending is a salient issue, mean district preference on
education spending exerts an independent and statistically significant effect on
representative roll-call voting. The coefficient on mean district education spending is
significant at (1 < .04 and positive. This illustrates that Republican representatives vote

more conservatively on education spending, while Democratic representatives vote more

 

'0 I used authorization and appropriations votes to construct the W-NOMINATE scores.

33

liberally on the same issue. This translates to Democrats more often voting to increase
education spending. In regard to model fit, the model accounts for about 75% of the
overall variation or representatives’ legislative behavior. In the case of the 106th
Congress, because education was a very salient issue during this time, representatives had
to toe the district line. As hypothesized, we see congruence between district preference
and legislator behavior because representatives know that if they stray outside the bounds

of district preference on salient issues, they may face electoral punishment.

[Table l About Here]

What about non-salient issues? Table 2 reports the results from a second OLS
regression focusing instead on defense spending. As expected, since defense spending
was the least salient issue to voters, district preference on defense spending does not exert
an independent or statistically significant impact on representatives’ defense spending
roll-calls. However, we also see that representatives are able to follow other pressures
such as party. For example, the coefficient on party of the representative is highly
significant at (1 < .0001, telling is that representatives chose to toe the party line as
opposed to the district line when voting on defense spending. Previous research
examining the impact of foreign policy preferences on legislator behavior have concluded
that because foreign policy is the least salient issue area, (Almond 1950; Boyd 1985;
Bumham 1985; Ladd 1985; Shanks and Miller 1985; Weisberg 1985; Light and Lake
1985; Abramson, Aldrich, Rohde 1987), representatives are free to vote their own

ideology (McCormick 1985), party (Lindsay 1990, 1991; LeoGrande and Brenner 1993),

34

or personal policy preferences (Burgin 1994). In the case of the 106th Congress, defense
issues were the least salient, voters did not evaluate their representatives based on their
roll-call behavior on such issues, and in turn representatives voted the party. District
preference-legislator behavior congruence is not necessary if there is no electoral threat at

play, which is illustrated in this example.

[Table 2 About Here]

Additional Issues
While education was the most salient issue to voters, and defense was the least

salient, what about issues in between? Upon further examination of the NAES question,
“In your opinion, what is the most important problem facing our country today?” the
environment and taxes were issues that were barely mentioned. While the environment
has never been a salient issue, taxes traditionally have been at the top of the list.
However, during the time of this study, the economy was thriving, unemployment was
low, and tax rates were steady. While foreign policy is almost always the least salient
issue area, and we should expect there to be no relationship between district preference
and roll-call behavior on such issues, taxes are almost always salient. Issue salience is
not absolute. It is variable, contextual, and time dependent.

In order to illustrate the variability of issue salience, Table 3 reports a series of
regressions with the representative’s issue specific W-NOMINATE scores as the dependent
variable. The main explanatory variable is the mean district issue preference on the

corresponding W-NOMINATE score constructed from the issue specific questions on the

35

NAES.ll The control variables are mean district ideology and party of the representative.
What we see is that in both the environment and taxes models, district preference does
not exercise an independent and significant impact on legislator behavior. However, in
both models we again see the coefficient on party of the representative or < 0.001. The
environment and taxes were not salient issues, and since legislators are not being
evaluated based on their roll-call votes on issues that are not salient, they can act as
trustees and vote based on other priorities such as party. Representatives are electorally
safer as long as they are responding to and congruent with district preference on the most

salient issue. It is not about which issue it is, it is how salient it is.

[Table 3 About Here]

Issue Salience and Overall Behavior
The previous results have show that when an issue is the most salient to voters. it

affects the way in which representatives vote on those particular issues. Conversely,
when issues are non-salient, members of the House have more freedom to act as a trustee
and vote according to other pressures and priorities aside from the district. But what
about overall legislator behavior? Does district preference on salient issues impact
overall legislator behavior? In order to maintain alignment with district preference, or
remain congruent, members must balance the multiple and conflicting pressures and

priorities including district preference. Members of the House must take into account

 

” The environment question states, “Protecting the environment and natural resources—~should the federal
government do more about this, the same as now, less or nothing at all?” and is coded: 1=More, 2=Same,
3=Less, 4=Nothing. The taxes questions states, “The amount of money Americans pay in taxes—is this an
extremely serious problem, serious, not too serious or not a problem at all?” and is coded: 4= Extremely serious
problem, 3=Serious problem, 2=Not too serious, l= Not a problem at all.

36

partisan, ideological, and issue specific district preference when casting all roll-calls. For
example, if the voters in a district want the federal government to spend more money on
education, the representative would need to vote to increase education funding, which
affects votes on appropriations and the budget. The weight a member of the House
places on issue specific district preferences varies based on the level of issue salience.
Therefore, we would expect district preference on salient issues to have an independent
and significant impact on overall legislator behavior, while non-salient issue preferences
should not.

Expanding the analysis to all four issues: education, environment, taxes, and
defense, I run another OLS regression to examine the effect of district issue preference on
overall legislator behavior. I measure the dependent variable as the 15‘ Dimension DW-
NOMINATE score of each member from the 106th Congress (1998-2000). The main
explanatory variables are the issue specific mean district preferences, and I control for
party of the representative and mean district ideology. As expected, the coefficient on
mean district education spending is significant at 0t < 0.02, while none of the other issue
specific measures are statistically significant. These results tell us that district issue
specific preferences only impact overall legislator behavior when the issue preferences
are on salient issues. Representatives know that they may be electorally vulnerable if
they are incongruent with district preference when it comes to salient issues. The fear of
losing ones seat is enough to keep a member of the House in line, and not only affects
issue specific roll-calls, but also carries over to all floor votes. Legislators know that
while they are faced with pressures and priorities other than district preference, the most

severe punishment comes from the district: voting for another candidate.

37

[Table 4 About Here]

Conclusion and Discussion

In this paper I examined the relationship between district issue preference, issue
salience, and legislator behavior. My results demonstrate that district issue preference
only exerts a significant and independent impact on representative roll-call behavior
when it is in regard to salient issues. While previous literature had studied this question,
data problems made it extremely difficult to quantify issue salience and obtain a true
grasp on district preference. Employing the 2000 NAES provided a reliable measure of
district issue preference and issue salience to examine the impact of issue salience on
representative behavior. Using presidential vote in the district or the NES as proxies for
district issue preference has been problematic, and therefore any conclusions drawn were
suspect. The NAES has alleviated the sample size and scope problems in order to
provide a clear estimate of district issue preference, and therefore, an answer to the
question of district preference-legislator behavior congruence. Because issue voting does
occur on salient issues, and therefore impacts House member reelection, there is
congruence between district issue preference and representative behavior. On the other
hand, there is no expectation of congruence when dealing with non-salient issues since
voters do not form strong preferences and hold the electoral threat over their
representatives’ heads. Acting as a trustee in regard to voting on individual issues only
works if the issue is non-salient. Salience, on the other hand, calls for a delegate.

Issue salience is variable and contextual. At any given time an issue like taxes

that is stereotypically salient could be of little or no concern to voters. Conversely, an

38

issue like education that has historically been a state and local issue could be pushed to
the forefront and become the most important problem facing the nation. Since September
11, 2001, almost nothing has been more important to the electorate than foreign policy
and defense issues. Securing our borders and our troop placement in the Middle East are
in their forefront of the minds of most voters. The post-9/11 environment is the ideal
situation to conduct further research on the relationship between issue salience and
legislative behavior. The 2006 midterm election highlighted that the War in Iraq was
absolutely the most important issue, and some representatives that alienated their voters
on this issue lost their seats. The 2004 National Annenberg Election Survey has recently
been released. Using this data, further analysis can be conducted comparing the pre and
post-9/11 electorate, representatives and issues to obtain an accurate understanding of the
importance of issue salience on district preference and legislative behavior. What we
should see is that district foreign policy preference does demand representative
responsiveness when the issues are salient. We witnessed electoral punishment in the
2006 midterm, and by using the both the 2000 and 2004 NAES to track the movement of
district preference on foreign policy, we can observe how the salience of issues has
changed.

The next step in assessing the linkage between districts and legislators is the
electoral impact on non-responsive or out of step members. If there is congruence
between district issue preference on salient issues and representative floor behavior, then
there should be some sort of punishment for out-of-step members. Whether this
punishment is in the form of a representative receiving lower vote margins over time, or

eventually losing reelection, voters will inflict some form of electoral punishment if they

39

feel their interests are not being met. The core of the district-representative relationship
is a somewhat simple exchange — districts feel represented, and legislators secure

reelection.

40

Tables

 

 

 

 

 

 

 

 

 

Table 1: Education Spendin . and Legislative Behavior
Constant 0.25
(0.16)
Mean District Education 011*
Preference (0.05)
Party of the Representative 0.45*
(0.01)
Mean District Ideology 0.30*
(0.04)
N 425
F—statistic 407.62*
R2 0.75
Root MSE 0.15

 

 

 

 

Dependent Variable — lSt Dimension Education Spending W-Nominate

* Significant atp < 0.04

(Robust standard errors in parentheses)

Table 2: Defense Spending

and Legislative Behavior

 

 

 

 

 

 

 

 

 

 

Constant 0.03
(0.36)
Mean District Defense 0.06
Preference (0.08)
Party of the Representative 0.56*
(0.02)
Mean District Ideology 0.25*
(0.07)
N 431
F -statistic 364.27*
R2 0.71
Root MSE 0.196

 

 

 

Dependent Variable — 1St Dimension Defense Spending W-Nominate

*Significant atp < 0.001

(Robust standard errors in parentheses)

41

Taxes

Constant

0.31
431
F-statistic 1697.34*

0.92
Root MSE 0.142

V — Issue
* Significant atp < 0.01

Constant

Mean

Party the

Mean District

N 424

F-statistic 3 76.77*
0.69

Root MSE

 

l

(Robust standard errors in parentheses)

42

Table 4: Issue Preference, Salience, and Overall Legislator Behavior

 

 

 

 

 

 

 

 

 

 

 

 

Constant 0.26
(0.48)
Mean Education 019*
(0.08)
Mean Defense 0.06
(0.11)
Mean Environment 0.08
(0.07)
Mean Taxes 0.02
(0.08)
Party of the 0.69*
Representative (0.02)
Mean District 040*
Ideology (0.08)
N 432
F-statistic 358.07*
R2 0.80
Root MSE 0.196

 

 

 

 

Dependent Variable —1St Dimension DW-Nominate
* Significant at p < 0.03
(Robust standard errors in parentheses)

43

Extremity and Incongruence: The Impact of Party and District Preference on House
Elections, 1994-2000

“A congressman has two constituencies — he has his constituents at home. And his
colleagues here in the House. To serve his constituents at home, he must also serve his
colleagues here in the House " — Speaker Samuel Rayburn (Dulaney and Phillips 1 978,
pp. 263-264)

Introduction
The dual nature of Congress, one part lawmaking body and one part

representative assembly that depends on the voters, presents members of Congress with a
juggling act. This is especially true for members of the House of Representatives since
the House is “intended to be the more representative element of the government”
(Davidson and Oleszek 2006, p. 12). The House was designed to answer to the people,
and because its members are elected every two years, this accountability to the people is a
constant balancing act between lawmaking and representation. Consequently, members
of the House are faced with this question: When do I vote with the district, when can I
vote outside the bounds of district opinion, and under what circumstances do I need to
worry about losing reelection? The most comprehensive study of this question told us
that the more a member of the House voted with his or her party, defined as extremity,
the more likely he or she would see a drop in vote share regardless of district preference
(Canes-Wrone, Brady, and Cogan 2002). In other words, in order to avoid electoral
punishment, representatives should become more moderate even if district preference is
not moderate. The question that arises from the answer the authors provides is: What if
the district is extreme? Should a representative moderate anyway? The answer I
provide: No.

In taking on this widely accepted argument, I combine voter information and

voter behavior literature, such as elite cues and directional voting, with congressional

44

behavior literature and develop the concept of incongruence. The degree that House
member behavior corresponds to district preference, or congruence, affects the likelihood
of both the entrance of a quality challenger and a decrease in vote share. The more
congruence, the less likely an incumbent will draw a quality challenger, and the less
likely voters will cast their ballots for any challenger. House members who vote with the
party will not be automatically punished because they are extreme, but rather they will be
punished if those votes do not fall within the bounds of district preference. Conversely, if
an incumbent is incongruent, the more likely he or she will draw a quality challenger and
see a drop in his or her vote share. Representatives can be incongruent, or not fall within
the bounds of district preference, and suffer a loss of vote share because they are too
moderate. In fact, directional theory (Rabinowitz and Macdonald 1989) and policy
balancing (Kedar 2005) suggest that voters prefer more extreme candidates.

While I provide another take on the Canes-Wrone, Brady, and Cogan (2002)
argument, I also provide additional empirical insight that may change the way in which
we model the relationship between district preference, legislative behavior, and vote
share. Because challenger quality and relative campaign spending are post-causal
treatment variables, including these factors in a model of electoral punishment may
underestimate the true causal affect (Gelman et. al 2003). Therefore, it may be
inappropriate to include such variables in models of electoral punishment.

This paper analyzes how the voting behavior of members of the House of
Representatives affects the two phases of electoral punishment: the entrance of a quality
challenger and a decrease in vote share. I examine the voting records of members of the

House of Representatives in the 103rd - 106th (1994-2000) Congresses and compare those

45

votes to district preference in order to access extremity, incongruence, quality
challengers, and dropping vote shares. First, I test how extremity and incongruence
affect the likelihood of drawing a quality challenger. The second test focuses on
decreasing vote share for members that are incongruent with district preference. Finally,
I discuss the consequences of the results as well as directions for fiiture congressional
research.

Voter Information and House Member Behavior

The need to balance representation and lawmaking only exists if voters evaluate
members of Congress based on floor behavior. Previous research has suggested that
voters are ignorant about representatives’ policy voting (Converse 1964; Popkin 1991;
Sniderman, Brody, and Tetlock 1991). This would allow members of Congress to vote as
they pleased since their constituents are not evaluating them based on their policy
positions. There would be no electoral threat present.12 And while Downs (1957) stated
that obtaining information is too costly for voters, this cost can be somewhat alleviated
by the presence of cues (Carmines and Kuklinski 1990; Grofman and Norrander 1990;
Mondak 1993; Lupia 1994; Lupia and McCubbins 1998, Druckman 2001).

Even if voters can obtain information on the behavior of their representatives
using cues, representatives have to believe that they are being evaluated based on that
behavior. Much congressional research has argued that members of the House do believe
that roll-call voting affects their chances of reelection (Matthews and Stimson 1975;
Aldrich 1995; Jacobson 1996; Ansolabehere, Snyder, and Stewart 2001; Canes-Wrone,

Brady, and Cogan 2002).

 

’2 See Erikson and Wright (1980); Page, et al. (1984); Stimson, MacKuen, and Erikson (1995);
Ansolabehere, Snyder, and Stewart (2001); and Canes-Wrone, Brady, and Cogan (2002).

46

Since voters evaluate representatives based on their floor behavior, and members
know this, then the electoral threat is both present and keeps members of the House inline
with their districts’ preferences. Members must consider this representation in the face of
multiple and conflicting lawmaking priorities and demands such as party pressure (Cox
and McCubbins 1993; Aldrich, Berger, and Rohde 2002; Sinclair 2002), personal
preferences (Aranson and Ordeshook 1972; Aldrich 71983, Wittman 1983; Calvert 1985;
Francis, Kenny, Morton, and Schmidt 1994; Crespin, Gold, and Rohde 2006), party
activists (Aldrich 1995; F iorina 1999; Jacobson 2000; Layman and Carsey 2000; Aldrich
and Rohde 2001), and policy orientation (Kingdon 1973; Fenno 1973; Rohde 1991;
Aldrich 1995; Aldrich and Rohde 2000). However, when does voter preference affect for
whom they cast their ballot?

Previous Research on Electoral Punishment V

Research examining electoral punishment based on legislative floor behavior has
been mixed. Some studies focusing on vote share found that safe members or
incumbents in general need not worry about the electoral impact of legislative voting
(Bernstein 1989; Gaines and Nokken 1999). While other literature has found empirical
evidence that holding district ideology constant, incumbents’ vote shares were lower
when they increased the extent to which they voted with the extreme of their party
(Schoenberger 1969; Erikson 1971; Johannes and McAdams 1981; Erikson and Wright
1993; Brady et al. 1996; Jacobson 1996; Canes-Wrone, Brady, Cogan 2002). In other
words, when members of Congress stray away from the median positions of their

districts, there was a decrease in electoral support.

47

More specifically, Canes-Wrone, Brady, and Cogan (2002) found that holding all
else constant, representatives who consistently voted the party-line, or what they
identified as extreme, were more likely to experience a drop in vote share. This
extremity is detrimental regardless of district preference because representatives should

always moderate. Figure 1 is a graphical representation of their argument.

[Figure 1 About Here]

The authors’ analysis is somewhat problematic for a few reasons. First, they do
not take into account district preference. Canes-Wrone, Brady, and Cogan argued that
holding district preference constant, House incumbents should vote moderately because
extreme members are punished at the polls. We know that district preference does
impact legislative behavior and subsequently impacts electoral fortunes. While Canes-
Wrone, Brady, and Cogan (2002) title their article “Out of Step, Out of Office,” there is
no discussion of the role of district preference. The authors argue that even if a district
prefers a more extreme incumbent, that incumbent should still moderate in order to avoid
electoral punishment for being too extreme.

Second, the authors include both challenger quality and a measure of relative
spending in their model of electoral punishment. While it is widely accepted in the
literature to incorporate both challenger quality13 and challenger/incumbent spending‘4 in
any model of electoral punishment, it may be inappropriate to incorporate such variables

because they may create post-treatment bias. Controlling for the effects of challenger

 

'3 Jacobson and Kemell 1983; Bianco 1984; Bond, Covington and Fleisher 1985; Jacobson 1989, 1990.
'4 Jacobson 1978, 1980, 1985; Banks and Kiewiet 1989; Green and Krasno 1990; Abramowitz 1991;
Epstein and Zemsky 1995; Cox and Katz 1996; Erikson and Palfrey 1998.

48

quality and campaign spending may underestimate the true causal affect (Gelman et. a1
2003)
“For instance, in the incumbency advantage example, what if we were to include
a control variable for campaign spending, perhaps the logarithm of the number of
dollars spent by the incumbent candidate's party in the election?. . .F or the
purposes of estimating the incumbency advantage with a regression, however,
total campaign spending should not be included, because much spending occurs
after the decision of the incumbent whether to run for reelection” (Gelman et. a1
2003, pp. 368-369)
Roll call votes affect vote share indirectly by encouraging the entrance of a quality
challenger and greater relative challenger spending. Both challenger quality and
campaign spending is associated with predictions that incumbents have made about
possible vote share. The dependent variable, vote share, ends up predicting the
independent variables — in this case challenger quality and campaign spending. While it
is theoretically correct that both challenger quality and campaign spending play a large
role in election returns, both variables are accounted for in the measure of vote share
itself. Moreover, using such controls in a model of vote share causes endogeneity and
post-treatment. For this reason, I exclude challenger quality and spending variables from
the model of vote share.
Finally, Canes-Wrone, Brady, and Cogan (2002) use Americans for Democratic
Action (ADA) scores as their measure of legislative voting. However, research has
argued that interest group scores exhibit a substantial amount of stretching over time
(Groseclose, Levitt, Snyder 1999). In addition, different interest group ratings select
different votes for consideration (Brunell et al. 1999). There is not much overlap in votes

selected by different interest groups. Using a measure such as Poole and Rosenthal’s

NOMINATE provides the most systematic approach to measuring roll-call behavior.

49

Since the authors use ADA scores, they are measuring extremity by national
standards. If these scores are extreme by national standards, they do not hurt a
representative when his or her district is locally extreme. But if the scores are moderate
by national standards, and the district is extreme, it does hurt the incumbent. Members of
the House need to represent their district, not the nation, which reinforces the fact that
they do not take into account district preference. When district preference is extreme, it
does not serve a representative well to moderate, as Canes-Wrone, Brady, and Cogan
(2002) argued, because the representative would not be inline with district preference.
The only time a representative should moderate is when district preference calls for it.

As opposed to voters preferring moderation, other literature argued that voters
actually prefer extreme candidates. Directional theory has suggested, “In the case of the
majority supporting a particular policy direction, a candidate who takes strong and
aggressive stands in favor of that policy is likely to be advantaged against a centrist
candidate” (Rabinowitz and Macdonald 1989, p. 110). In fact, candidates that take more
extreme positions are more likely to win House races (Achen 1978; Powell 1982).
Therefore, candidates that converge to the median voter are less successful than those
who take positions somewhat more extreme. If this is the case, moderating neither serves
the district’s best interest nor helps the representative’s chances of reelection.

But why would voters prefer extreme candidates? The success of more extreme
candidates may stem from a voter’s need to counterbalance “watered down” policies by
voting for a Candidate that is somewhat more extreme than the voter herself (Kedar

2005). Since most voters prefer more extreme candidates, district preference calls for

50

more extreme incumbents, and said incumbents will vote more extreme because it

corresponds to district preference. This concept is what I refer to as congruence.

[Figure 2 About Here]

While Kedar discussed policy balancing in a parliamentarian context where the
parties are battling for control of the government, policy balancing is appropriate to apply
to the American context during times of divided government. Under divided
government, the legislative and executive branches are engaged in bitter partisan battling
over policy outcomes. This lack of consensus among the branches creates policy
outcomes that may be more moderate than what the voters prefer due to partisan
compromising. Under these circumstances, voters will support candidates that take more
extreme policy stances in order to overcompensate for watered down compromised
policies. If this line of research is correct, being extreme actually should increase an
incumbent’s vote share as opposed to decreasing it as Canes-Wrone, Brady, and Cogan
(2002) argued. Therefore, extremity should not account for the entrance of a quality
challenger and dropping vote share, and in fact should account for an increase in vote
share.

Extremity, Incongruence, and House Elections

The theoretical expectations that develop from the previous research are
straightforward. Voters evaluate representatives based on their floor behavior and will
exercise the electoral threat if they do not feel their preferences are being reflected by
their representative. As opposed to centrist candidates, voters prefer candidates that take

more extreme stances. Therefore, I expect that representatives that are extreme will be

51

less likely draw a quality challenger and will see an increase in their vote share. I expect
that representatives that are incongruent are more likely to draw quality challengers and
see a decrease in vote share. If extremity affects incumbents positively, and it is
incongruence that affects incumbents negatively, then this provides a new perception of
the relationship between voter preference and legislator behavior.
Data

To examine these theoretical expectations, I focus on the 103’d-106‘h (1994-2000)
Congresses. Since I claim that voters prefer extreme candidates and will not punish
representatives for voting with the party, there is no better time period to test the affects
of extreme polarization. There were 73 Republican freshmen elected in 1994 from
heavily Republican districts as well as competitive districts with Democratic incumbents
(Crespin, Gold, and Rohde 2006). The so-called “Republican Revolution” has endured,
and as F enno (1997) posited, the polarization of the House has become increasingly bitter
while members consistently vote their party lines. If Canes-Wrone, Brady, and Cogan
(2002) are correct, and incumbents are punished for being extreme, then we should have
seen a drop in the House incumbency advantage since 1994. In fact, the percent of House
incumbents reelected increased from 90.2% in 1994 to 97.8% in 2000 (Abramson,
Aldrich, and Rohde 2002).
Measures
District

I use Republican presidential vote measured as the two-party vote in the district

from the 1994-2000 elections as the measure of district preference.’5 Congressional

scholarship has long used presidential vote in the district as a proxy for district

 

'5 I use the most recent two-party Republican presidential vote in the district as the measure of district
preference for each election.

52

preference.16 While there has been criticism of using presidential vote in the district as a
measure of district partisanship,l7 the 2000 National Annenberg Election Survey has shed
new light on the debate. The NAES conducts multiple rolling cross section studies from
December 1999 to the beginning of January 2001. Overall, the survey had approximately
79,000 respondents with anywhere from 20 (F L-23) to 227 (MI-1) respondents per
district. The respondents were asked partisanship, ideology, and issue questions as well
as candidate preference questions. In order to analyze how presidential vote in the
district compares to a more direct measure of district partisanship, I constructed the mean
partisanship from the 2000 NAES and correlated it with the 2000 Republican presidential

vote in the district. The two measures are correlated at 0.85.

[Figure 3 About Here]

Both the correlation and Figure 4 convey that presidential vote in the district is an
appropriate proxy for district partisanship.

House Member and Challenger
I use first dimension Dwnommm from the 104th —106th Congresses (1994-2000)

as a measure of overall legislator behavior on a liberal-conservative scale. The scale goes
from -1 to +1 (Liberal to Conservative) with extreme partisans at the poles. I also
1 18

control for representative partisanship by coding Democrats as 0 and Republicans as

Challenger quality is coded 1 if the candidate has held elective office and 0 if not.

 

’6 Erikson and Wright 1989, 1993, 1997; Brady et a1. 1996, Canes-Wrone, Brady, and Cogan 2002;
Crespin, Gold, and Rohde 2006.

’7 See Clinton 2006.

'8 Bernard Sanders (I-VT) is coded as a Democrat.

53

Incongruence and Extremity
The main explanatory variables are extremity and incongruence. Extremity is

measured as the absolute value of the first dimension DW-NOMINATE score. Incongruence
is constructed by a bivariate regression of DW-NOMINATE on the Republican presidential
vote in the district. Next, I predict the residuals from the regression. Because I am using
residual estimates from a regression as an explanatory variable, I bootstrap the standard
errors in all models. With nonparametric bootstrapping, the obtained data is treated as if
it is a precise reflection of the entire population, and then draw many bootstrapped
samples by repeated sampling with replacement (Efron and Tibshirani 1993). Once the
OLS regression computes the bootstrapped standard errors, the absolute value of the
residuals reflects the degree of incongruence.19 What I expect to see is that the higher the
residuals for each district, the more likely an incumbent is to draw a quality challenger
and the lower the incumbent’s vote share should be. In other words, the more variation
around the regression, the more incongruent the representative is with his or her district,
and therefore the more likely they are to draw a quality challenger and see a decrease in
vote share in subsequent elections.
Results

In the first section I present a series of Probit models examining the impact of
incongruence and extremity on the entrance of quality challengers in House races. I
expect that more extreme incumbents are less likely to draw quality challengers, while
incongruent incumbents are more likely to draw quality challengers. After discussing my
findings, I move on to analyzing the second phase of electoral punishment: decrease in

vote share. In this section, I present a series of OLS regression models analyzing the

 

'9 For each model I conducted 2000 replications.

54

impact of incongruence and extremity on vote share. I expect that more extreme
members will experience an increase in vote share, while incongruent members will
experience a decrease in vote share.

Phase 1: Enter the Quality Challenger

In Table 1, I pool the elections (1994-2000) and run three Probit models: one with
only extremity, one with only incongruence, and one with both incongruence and
extremity.20 The dependent variable is whether or not there was a quality challenger
(1=yes, 0=no). I also control for party of the representative in each model. It is clear
from Table 1 that being extreme is beneficial to representatives since it decreases the
likelihood of drawing a quality challenger. In the first model with no measure of
behavior relative to district preference, extremity is both negative and statistically
significant. As I predicted, the more extreme a member is, the less likely he or she is to
draw a quality challenger. In the second model, my prediction that being incongruent
will increase the chances of drawing a quality challenger also holds. The third model,
with extremity and incongruence, shows that when taking into account representative-
district match and degree of extremity, incongruence entices quality challengers and
extremity deters them.

There are a few important conclusions to draw from these results. The model
with extremity only has the lowest Wald Chi-Square estimate (15.29) of all the models.
While extremity does exert a negative and independent effect on the entrance of quality
challengers, including a measure of incongruence strengthens the effect of extremity.

Substantively, this illustrates that extremity is not the most important determinant of

 

2° Since I pool the elections, I run fixed models for both Congress and district. Models show that neither
district type nor individual Congress account for a significant amount of model variance.

55

whether or not a House incumbent draws a quality challenger. Extremity is most
influential in deterring a quality challenger when it is coupled with incongruence.

These findings support my theoretical argument but are antithetical to the logic set
forth by Canes-Wrone, Brady, and Cogan (2002). If in fact being extreme was
detrimental to representatives, then we should see quality challengers enter races with
extreme members. Instead, what we see is that because voters prefer extreme candidates,
House members who consistently vote the party line are rewarded. On the other hand,
incumbents who are incongruent are punished. Extremity is most damaging to an

incumbent when in conjunction with incongruence.

[Table 1 About Here]

In order to further parcel out the effects of extremity and incongruence, Table 2
reports the predicted probabilities of drawing a quality challenger in two categories:
incumbents who are incongruent because they have moderate DW-NOMINATE scores in
extreme districts, and incumbents who are incongruent because they have extreme DW-
NOMINATE scores in moderate districts. This table provides an illustration of the
probability of an incumbent drawing a quality challenger holding party of the
representative at its mean and varying the degree of extremity and incongruence.2| The
results in Table 2 show that the probability of incongruent members drawing quality
challengers is much less when they are too extreme (0.19) than when they are too

moderate (0.41). Again, as policy balancing and directional theory states, being extreme

 

2' Since party of the representative is coded 1 for Republicans and 0 for Democrats, 1 set it 0.50 for the
mean. Extreme members are 0.70 and moderate members are 0.50 on the absolute vale of the NOMINATE
scale.

56

is better than being too moderate. Extreme members in moderate districts have a lower
probability of drawing a quality challenger, illustrating that even if a member is
incongruent with district preference, being too extreme is beneficial to seat safety.
Overall, the most vulnerable incumbents are moderates in extreme districts. Incumbents
who are not extreme enough face the biggest threat of a quality challenger, while extreme
members’ seats remain safer. Therefore, the Canes-Wrone, Brady, and Cogan (2002)
argument that moderate members are safer than extreme members ceases to be accurate.
Moderating does not help an incumbent, but rather it is detrimental. Now that I have
examined the first stage of electoral punishment, I will discuss decreasing vote shares.

[Table 2 About Here]

Phase 2: Decrease in Vote Share
In Table 3, I pool the elections (1994-2000), and run three OLS regression

models: one with extremity, one with incongruence, and one with both extremity and
incongruence.22 The dependent variable is the incumbent share of the two-party vote,
and the main explanatory variables are extremity and incongruence. I control for party of
the representative, and in the models with extremity I control for district partisanship
using Republican presidential vote in the district.23 As mentioned previously, I do not
use challenger quality or any form of campaign spending because they create a post-
treatment bias. Post-treatment variable bias may well be the largest overlooked
component of bias in estimating causal effects in political science (King, Keohane, and

Verba, 1994). According to King and Zeng (2001), to avoid such bias, either a model

 

22 Since I pool the elections, I run fixed models for both Congress and district. Models show that neither
district type nor individual Congress account for a significant amount of model variance.

23 I do not include a measure of district partisanship in the model with only incongruence because it is
accounted for in the measure itself.

57

must not contain post-treatment control variables, or included post-treatment control
variables must not vary with the dependent variable. While excluding certain control
variables may cause omitted variable bias, including them will cause a post-treatment
bias. In the case of a regression model that predicts vote share, challenger quality and
relative campaign spending do vary with the dependent variable since predictions of vote
share will have an effect on both the quality of a challenger and incumbent spending.

Canes-Wrone, Brady, and Cogan (2002) includes such control variables in their
models of vote share, and therefore underestimated the true causal effect of their
regression model of vote share. In recreating the authors’ model, I found that when
excluding challenger quality and relative campaign spending, the coefficient on extremity
became positive and statistically significant.24 In other words, when correctly specifying
the model, extreme members actually increase their vote share as opposed to decreasing
it.

As illustrated in Table 3, in all three models, extremity is both positive and
statistically significant, while incongruence is both negative and statistically significant.
All else equal, incongruent members will decrease their vote share while extreme
members will increase their vote share. Again, extreme members are only in electoral
danger if they are extreme and incongruent. Similar to the models of quality challengers,
the model with only extremity has the lowest Wald Chi-Square (71.97).

In the model with both extremity and incongruence, holding all else equal,
moving two standard deviations “up” in incongruity, or about 0.15 up or down on the
NOMINATE scale, lowers incumbent vote share by about 4 points. Along the same lines,

holding all else equal, moving two standard deviations to the extremes on the NOMINATE

 

24 See appendix for the full model and results.

58

scale increase incumbent vote share by almost 3 points. In the models with just extremity
and just incongruence, this effect is not as substantial since the coefficients are smaller
than in the model with both explanatory variables. Again, we must look at extremity
coupled with incongruence to capture the full effect of both concepts. As I predicted,
because voter prefer extreme candidates, incumbents who consistently vote the party line
are rewarded at the polls. Moreover, incumbents who are incongruent will be punished at

the polls.

[Table 3 About Here]

These results coincide with the incumbent election returns during this time period.
As stated, the post-Republican takeover Congresses were extremely polarized, yet the
incumbency rate increased. Even more convincing than the increasing incumbency
advantage is the consistent increase in incumbent vote share since 1994. From 1994-
2000, House incumbent vote share increased from 62.8% to 65.1% (Abramson, Aldrich,
and Rohde 2003). In a time when both incumbents and challengers became more
extreme, Canes-Wrone, Brady, and Cogan (2002) would tell us that incumbent vote share
should have decreased since voters prefer moderate candidates. Not only has the rate at
which incumbents are reelected increased, but they have been winning by larger margins.
Overall, being extreme is not only acceptable, but also welcomed by voters.

In order to further examine the effects of extremity and incongruence, Table 4
reports the predicted values of vote share in two categories: incumbents who are

incongruent because they have moderate DW-NOMINATE scores in extreme districts, and

59

incumbents who are incongruent because they have extreme DW-NOMINATE scores in
moderate districts. This table provides an illustration of incumbent percent of vote share
holding party of the representative at its mean and varying the degree of extremity and
incongruence.25 This shows incumbents who are moderate in extreme districts, thereby
incongruent, have the lowest percent of vote share (57%); while incumbents are
incongruent due to extremity have the highest vote percentage (67%). If an incumbent is
going to be incongruent, it is best to be too extreme rather than too moderate.

Extreme members, even when incongruent, are electorally more successful than
moderate members, providing evidence that voters do prefer extreme representatives as
opposed to moderate ones. These findings, which coincide with my theoretical
expectations, are in direct opposition to those of Canes-Wrone, Brady, and Cogan (2002).
The contrary findings are mostly from the authors overlooking the importance of

legislator behavior in relation to district preference.

[Table 4 About Here]

What about partisan differences? Which party would we expect to be more
incongruent with district preference? During the course of the Congresses in question,
the Republican party median increased from 0.395 to 0.435 on the NOMINATE scale.
Moreover, between 1994 and 2000, the average Republican presidential vote in districts
with Republican incumbents increased from about 50% to about 51%. Republicans were

mostly staying congruent. Meanwhile, the Democratic party median increased from

 

25 Since party of the representative is coded 1 for Republicans and 0 for Democrats, 1 set it 0.50 for the
mean. Extreme members are 0.70 and moderate members are 0.50 on the absolute vale of the NOMINATE
scale.

60

—0.33 to —O.403 even though the average Republican presidential vote in district with
Democratic incumbents increased from about 40% to about 42%. Democrats were
polarizing and becoming incongruent with district preference, thus helping to explain the
election returns during the second half of the 19905. The incongruent Republican
incumbents were not as incongruent as the Democratic incumbents, and therefore fared
better at the polls. In order to empirically test this, I run another OLS regression, but this
time I include a measure of incongruence that separates the representatives by party. In
other words, in Table 5, I use the signed residuals along with the absolute value of the

residuals to see specifically which party was rewarded at the polls.

[Table 5 About Here]

Table 5 corroborates the results in Table 3 in that the coefficient on extremity is
both positive and statistically significant, and the coefficient on incongruence is negative
and statistically significant. This reinforces the concept that extremity increases vote
share and incongruence decreases vote share. In regard to vote share, incongruence, and
party, representatives that were more conservative that their districts experienced
increased vote share. The more conservative a representative, the better he or she fared at
the polls. Holding all else equal, moving two standard deviations “up” in incongruity, or
about 0.15 toward the more conservative end on the NOMINATE scale, increases incumbent
vote share by almost 1 point. Democrats were more incongruent than Republican
incumbents; and therefore the more liberal a representative, the more incongruent, and

the more likely to receive a decrease in vote share.

61

Conclusion and Discussion
The Republican takeover in 1994 activated an increased polarization between

Democrats and Republicans that still continues today. Extreme House incumbents and
challengers are the norm rather than the exception. While Canes-Wrone, Brady, and
Cogan (2002) argued that extremity was detrimental to electoral fortunes, it appears the
exact opposite is true. Both directional theory and policy balancing stated that voters
prefer extreme candidates, and past congressional research has concurred. Extremity,
when it corresponds to district preference, is beneficial to incumbents in two ways. First,
it deters quality challengers, and second, it increases vote share. Members who are
extreme are not automatically punished, as stated by Canes-Wrone, Brady, and Cogan
(2002), but rather they are rewarded for being in-line with district preference. The key is
remaining in-step with district preference. Incumbents should try to stay as congruent as
possible, and most of the time, this requires some extremity. Moderate members are
more likely to face quality challengers regardless of congruence and are the most likely
type of incumbent to face quality challengers when they are in extreme districts.
Extremity is harmful only when it is not what the district wants.

The second way in which extremity is advantageous is that it increases incumbent
vote share. Voters prefer to vote for candidates who are more extreme than they are; and
therefore more extreme incumbents will pick up more votes than moderate incumbents.
Canes-Wrone, Brady, and Cogan (2002) assume that incumbents and candidates need to
appeal to the median members of the district. In fact, moderate members are the most
vulnerable. Even if voters themselves are in the middle of the partisan continuum, they
cast their ballots for candidates at the ends of the continuum. Incumbents need not fear

voting the party line if it is congruent to district preference, which it often is.

62

Incumbents do not need to, and in fact should not moderate unless the district calls for it.
The threat is incongruence, not extremity. The results Canes-Wrone, Brady, and Cogan
(2002) reported did not account for one of most important determinants of legislator vote
choice — district preference. The relationship between floor behavior and election returns
is more complex than they present. What I have provided is a more nuance view between
the relationship of district preference and legislator floor behavior. During the time
period of this study, district preference highly influenced legislative floor behavior. Also,
most voters wanted representatives to vote the party line, which eased some of the
tension created by pressure from the congressional parties. The juggling act just got
easier.

The next step in assessing extremity, incongruence, and vote share is to examine
whether or not the relationship varies by issue area. Kedar (2005) argued that voters feel
policies are watered down due to partisan congressional bargaining and therefore prefer
candidates that take more extreme policy stands. The question arises as to whether or not
this preference varies by issue area and salience of the issue. While this may not have
been possible before due to data limitations, the 2000 National Annenberg Election
Survey taps into an unexploited area of voter preference. The fundamental reality,
nonetheless, remains the same — incumbents who follow the beat of their districts

drumming will keep their seats.

63

Tables

Table l: Extremity, Incongruence, and Quality Challengers 0994-2000)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Extremity Incongruence Extremity &
Incogruence
Constant -0.50* Constant -1 .24 "‘ Constant -0.84*
(0.10) (0.08) (0.11)
Extremity -0.93 * - - Extremity -1 .44‘
(0.24) (0.24)
- - Incongruence l.48* Incongruence 1 .89"
(0.20) (0.23)
Party of -0.03 Party of -0.04 Party of 0.04
Representative (0.07) Representative (0.07) Representative (0.07)
N 1728 N 1728 N 1728
Pseudo R2 0-01 Pseudo R2 0-03 Pseudo R2 0-05
Wald x2 1529* LR x2 53.50" Wald x2 86.18“

 

Dependent Variable — If there was a quality challenger in the election (0,1)
"' Significant at p < 0.001
(Bootstrapped standard errors in parentheses)

Table 2: Probability of Drawing A Quality Challenger (1994-2000)

 

Congruence/Incongruence

Probability of Drawing a Quality
Challenger with Confidence Intervals

 

 

district

 

 

Incongruent with a moderate DW- 0.41
NOMINATE score in an extreme (0.34, 0.49)
district

Incongruent with an extreme DW- 0.19
NOMIN ATE score in a moderate (0.14, 0.23)

 

 

Average probability across all four elections (1994-2000)

64

 

Table 3: Extremity, Incongruence, and Vote Share (1994-2000)

 

 

 

 

 

 

 

 

 

Extremity Incongruence Extremity &
Incongruence
Constant 69.02* Constant 73.94* Constant 70.75“
(2.03) (0.78) (1.86)
Extremity 12.65”“ - - Extremity 19.56”
(2.48) (2-44)
- - Incongruence -24.03* Incongruence -27.33*
(2.21) (2.28)
Republican -0.17* - - Republican -0.09"‘
Presidential Vote (0.04) Presidential Vote (0.04)
Party of the 2.36”“ Party of the -0.68 Party of the 0.19
Representative (0.85) Representative (0.72) Representative (0.95)
N 1535 N 1535 N 1535
Wald x2 71 .97* Wald X2 l 18.01 * Wald x2 23608"
R2 0.04 R2 0.07 R2 0.13

 

 

 

 

 

 

 

 

Dependent Variable — Incumbent Two-Party Percent of the Vote
* Significant at p < 0.01
(Bootstrapped standard errors in parentheses)

Table 4: Predicted Values of Vote Share (1994-2000)

 

 

 

Congruence/Incongruence Predicted Value of Vote Share (%)
With Confidence Intervals

Incongruent with a moderate DW- 57.0

NOMINATE score in an extreme (54.9, 59.0)

district

Incongruent with an extreme DW- 67.0

NOMINATE score in a moderate (65.3, 68.2)

district

 

 

 

 

Average probability across all four elections (1994-2000)

65

Table 5: Extremity, Incongruence, and Vote Share (1994-2000)
with Party of the Representative

 

 

 

 

 

 

 

 

 

 

Extremity &
Incongruence
Constant 69.45**
(2.03)
Extremity 19.23**
(2.46)
Incongruence without Party -27.74**
(2.32)
Incongruence with Party 503*
(2.58)
Republican Presidential Vote -0.03
(0.05)
Party of the Representative -2.88
(1.87)
N 1535
Wald x 249.85”
R 0.13

 

 

 

 

Dependent Variable - Incumbent Two-Party Percent of the Vote
* Significant at p < 0.05

** Significant at p < 0.001

(Bootstrapped standard errors in parentheses)

66

Figures

Figure 1

Representative Behavior

0.7

0,6

0.5

0.4

03

National Downslan I Moderatlon Model

 

 

Equilibrium

 

 

0.0

I I I I

0.2 0.4 0.6 0.8

District Median

67

1.0

 

Figure 2

Representative Behavior

1.0

0.8

06

0.4

0.2

0.0

Extremism I Party Model

 

 

 

Equilibrium

Equilibrium

 

 

I I I I

0.0 0.2 0.4 06

District Median

68

08

1.0

 

Figure 3

District Partisanship and Republican Presidential Vote 2000

 

.4 .6 .8

Presidential Vote in the District

.2

 

publican
o

 

 

0

 

I

Re
4.1

.4 .5
Mean District Partisanship

69

Appendix

National Annenberg Election Survey Questions

Taxes a Problem:

The amount of money Americans pay in taxes — is this an extremely serious problem =4,
serious =3, not too serious =2, or not a problem at all =1 ?

Federal government should spend on military and national defense:
Maintaining a strong military defense — should the federal government spend more money
on this =4, the same as now =3, less =2, or no money at all =1?

Federal government should spend on spending on schools:

Providing national assistance to public elementary and secondary schools — should the
federal government spend more money on this =1, the same as now =2, less =3, or no
money at all =4?

Federal government should expend effort to protect the environment:
Protecting the environment and natural resources — should the federal government do more
about this = 1, the same as now =2, less =3, or nothing at all =4?

Excluded Districts and W-NOMINA TE Scores

Alaska and Hawaii were excluded. There were no W-NOMINATE scores generated
for the Illinois 14’“. There were no scores generated for education for the Arkansas 3’“,
California 32”“, Massachusetts 9’“, South Carolina 2““, Virginia 4’“, and Wyoming 1”. For
environment, there were no scores produced for the Arkansas 3'“, Florida 1”, Indiana 6’“,
Massachusetts 9’“, Oklahoma 1”, and South Carolina 2”“.

70

REFERENCES

Abramowitz, Alan I. 1988. “Explaining Senate Election Outcomes.” American Political
Science Review 82: 385-403.

Abramson, Paul R, John H. Aldrich, and David W. Rohde. 1987. Change and Continuity
in the 1984 Elections Rev. ed. Washington: Congressional Quarterly.

------- . 2002. Change and Continuity in the 2000 Elections. Washington: Congressional
Quarterly.

------- . 2003. Change and Continuity in the 2000 and 2002 Elections. Washington,
Congressional Quarterly.

Achen, Christopher H. 1978. “Measuring Representation.” American Journal of Political
Science 22: 475-510.

Aldrich, John H. 1983. “A Downsian Spatial Model with Party Activism.” The American
Political Science Review. 77: 974-990.

------- . 1995. Why Parties? The Origin and Transformation of Party Politics in
America. Chicago: Chicago Press.

Aldrich, John H. and David W. Rohde. 1999. "The Consequences of Party
Organizations in the House: Theory and Evidence on Conditional Party
Government." A paper presented at the Conference on the Congress and the
President in a Partisan Era, at Texas A&M University.

------- . 2000. “The Republican Revolution and the House Appropriations Committee.”
The Journal of Politics 62: 1-33.

------- . 2001. "The Logic of Conditional Party Government: Revisiting the Electoral

Connection." In Lawrence C. Dodd and Bruce I. Oppenheimer (eds.), Congress
Reconsidered 7th Edition. Washington, DC: CQ Press, pp. 269-292.

Aldrich H. John, Berger M. Mark and Rohde W. David. 2002. “The Historical
Variability on Conditional Party Government, 1997-1994.” In Brady W. David

and McCubbins D. Mathew (eds.), Party Process and Political Change in
Congress: New Perspectives on the History of Congress. California: Stanford
University Press, pp. 17-35.

Almond, Gabriel A. 1950. The American People and Foreign Policy. New York:
Harcourt, Brace.

71

Alpert, Eugene J. 1979. “A Reconceptualization of Representational Role Theory.”
Legislative Studies Quarterly 4: 587-603.

Ansolabehere, Stephen, James M. Snyder, and Charles Stewart III. 2000. “Old Voters,
New Voters, and the Personal Vote: Using Redistricting to Measure the
Incumbency Advantage.” American Journal of Political Science 44: 17-34.

------- . 2001. “Candidate Positioning in US. House Elections.” American Journal of
Political Science 45: 136-159.

Aranson, Peter H. and Peter C. Ordeshook. 1972. “Spatial Strategies for Sequential
Elections.” In Richard G. Niemi and Herbert F. Weisberg (eds.), Probability
Models of Collective Decision Making. Columbus, OH: Merrill Publishing.

Arnold, Douglas R. 1990. The Logic of Congressional Action. New Haven: Yale
University Press.

Banks, Jeffrey S. and D. Roderick Kiewiet. 1989. “Explaining Patterns of Candidate
Competition in Congressional Elections.” American Journal of Political Science
33: 997-1015.

Bartels, Larry M. 1991. “Constituency Opinion and Congressional Policy Making: The
Reagan Defense Buildup.” The American Political Science Review 85: 457-474.

------- . 2000. “Partisanship and Voting Behavior, 1952-1996.” American
Journal of Political Science 44: 35-30.

Berelson Bernard, Paul Lazarsfeld, and William McPhee. 1954. Voting. Chicago:
University of Chicago Press.

Bernstein, Robert A., and William Anthony. 1974. “The ABM Issue in the Senate, 1968-
1970: The Importance of Ideology.” American Political Science Review 68: 1198-
1206.

Bernstein, Robert A. 1989. Elections, Representation, and Congressional Voting
Behavior: The Myth of Constituency Control. Englewood Cliffs, NJ: Prentice-
Hall.

Bianco, William T. 1994. Trust, Representatives, and Constituents. Ann Arbor:
University of Michigan Press.

Bond, Jon R., Cary Covington, and Richard Fleisher. 1985. “Explaining Challenger
Quality in Congressional Elections.” The Journal of Politics 47: 510-529.

72

Box-Steffensmeier, Janet M., David C. Kimball, Scott R. Meinke and Katherine Tate.
2003. “The Effects of Political Representation on the Electoral Advantages of
House Incumbents.” Political Research Quarterly 56: 259-270.

Boyd, Richard W. 1985. “Electoral Change and the Floating Voter: The Reagan
Elections.” Presented at the Annual Meeting of the American Political Science
Association, New Orleans.

Brunell, Thomas L., William Koetzle, John Dinardo, Bernard Grofman, and Scott L.
Feld. 1999. “The R 2 = .93: Where Then Do They Differ? Comparing Liberal and
Conservative Interest Group Ratings.” Legislative Studies Quarterly 24: 87-101.

Burgin, Eileen. 1994. “Influences Shaping Members’ Decision Making: Congressional
Voting on the Persian Gulf War.” Political Behavior 16: 319-342.

Burnham, Walter Dean. 1989. “The Reagan Heritage.” In The Personal Vote:
Constituency Service and Electoral Independence. Cambridge: Harvard
University Press.

Calvert, Randall. 1985. "Robustness of the Multidimensional Voting Model: Candidate
Motivations, Uncertainty, and Convergence." American Journal of Political
Science 29: 69-95.

Canes-Wrone, Brandice. 2001. “The President’s Legislative Influence from Public
Appeals.” American Journal of Political Science 45: 313-329.

Canes-Wrone, Brandice, David W. Brady, and John F. Cogan. 2002. “Out of Step, Out
of Office: Electoral Accountability and House Members’ Voting.” American
Political Science Review 96: 127-140.

Carmines, Edward G., and James H. Kuklinski. 1990. “Incentives, Opportunities, and the
Logic of Public Opinion in American Political Representation.” In Information
and Democratic Processes, ed. John A. Ferejohn and James H. Kuklinski.
Urbana: University of Illinois Press.

Clinton, Joshua D. 2006. “Representation in Congress: Constituents and Roll Call in the
106’“ House.” The Journal of Politics 68: 397-409.

Converse, Philip E. 1964. “The Nature of Belief Systems in Mass Publics.” In Ideology
and Discontent, ed. David Apter. New York: Wiley.

Converse, Philip and Gregory Marcus. 1979. “Plus ca change. . .The New CPS Election
Study Panel.” American Political Science Review 73: 32-49.

73

C00per, Joseph, and David W. Brady. 1981. "Institutional Context and Leadership Style:
The House from Cannon to Rayburn. " American Political Science Review 75:
411-25.

Cox, Gary W. and Jonathan N. Katz. 1996. “Why Did the Incumbency Advantage in US.
House Elections Grow?” American Journal of Political Science 40: 478-497.

Cox W. Gary and Mathew D. McCubbins. 1993. Legislative Leviathan- Party
Government in the House. Los-Angeles: University of California Press.

------- . 1999. “Agenda Power in the US House of Representatives, 1877-1986.” In

Essays in the History of Congress, ed. David W. Brady and Mathew D.
McCubbins. Palo Alto: Stanford University Press.

Crespin, Michael H., Suzanne M. Gold, and David W. Rohde. 2006. “Ideology, Electoral
Incentives, and Congressional Politics: The Republican House Class of 1994.”
American Politics Research 34: 135-158.

Davidson, Roger H., and Walter J. Oleszek. 2006. Congress and Its Members 10’“
Edition. Washington, DC: Congressional Quarterly.

Deckard, Barbara Sinclair. 1976. “Electoral Marginality and Party Loyalty in House Roll
Call Voting.” American Journal of Political Science 20: 469-481.

Dougan, William R., and Michael C. Munger. 1989. “The Rationality of Ideology.”
Journal of Law and Economics 32: 119-142.

Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper.

Druckman, Jamie N. 2001. “The Implications of Framing Effects for Citizen
Competence.” Political Behavior 23: 225-256.

Efron B., and R. Tibshirani. 1993. An Introduction to Bootstrap. London: Chapman and
Hall.

Epstein, David and Peter Zemsky. 1995. “Money Talks: Deterring Quality Challengers in
Congressional Elections.” The American Political Science Review 89: 295-308.

Erikson, Robert S. 1971. “The Electoral Impact of Congressional Roll Call Voting.”
American Political Science Review 65: 1018-1032.

Erikson, Robert S. and Thomas R. Palfrey. 1998. “Campaign Spending and Incumbency:
An Alternative Simultaneous Equations Approach.” Journal of Politics 60: 355-
373.

Erikson, Robert S. and Gerald C. Wright, Jr. 1980. “Policy Representation of
Constituency Interests. Political Behavior 2: 91-106.

74

------- .1989 “Voters, Candidates, ad Issues in Congressional Elections.” In Congress
Reconsidered 4’“ Ed, eds. Lawrence C. Dodd and Bruce I. Oppenheimer.
Washington, D. C.: Congressional Quarterly Press.

------- ..1993 “Voters, Candidates, and Issues in Congressional Elections.” In Congress
Reconsidered 5’“ Edition, eds. Lawrence C. Dodd and BruceI. Oppenheimer.
Washington: Congressional Quarterly.

------- 1.997 “Voters, Candidates, ad Issues 1n Congressional Elections.” In Congress
Reconsidered 6’“ Edition, eds. Lawrence C. Dodd and BruceI. Oppenheimer.
Washington, D. C.: Congressional Quarterly Press.

F enno, F. Richard.l973. Congressmen in Committees. Boston: Little Brown.
------- . 1978. Homestyle: House Members in Their District. Boston: Little, Brown.

------- . 1997. Learning to Govern: An Institutional View of the 104’“ Congress.
Washington, DC: The Brookings Institution

Fiorina, Morris P. 1974. Representatives, Roll Calls, and Constituencies. Lexington,
MA: Lexington Books.

------- . 1999. "Whatever Happened to the Median Voter?" A paper delivered at the
Annual Meeting of the Midwest Political Science Association.

Francis, Wayne, Lawrence Kenny, Rebecca Morton and Amy Schmidt. 1994.

"Retrospective Voting and Political Mobility." American Journal of Political
Science 38: 999-1024.

Gaines, Brian J ., and Timothy P. Nokken. 1999. “The Presidential Shadow on Midterm
House Elections: Presidential Support, Presidential Agendas, and Seat Loss.”
Typescript, University of Illinois at Urbana-Champaign.

Gelman, Andrew, John B. Carlin, Hal S. Stern, and Donald B. Rubin. 2003. Bayesian
Data Analysis 2"“ Edition. Boca Raton: CRC Press.

Green, Donald P. and Jonathan S. Krasno. 1990. “Salvation for the Spendthrift
Incumbent: Reestimating the Effects of Campaign Spending in House Elections.”

American Journal of Political Science 34: 363-372.

Grofman, Bernard, and Barbara Norrander. 1990. “Efficient Use of Reference Group
Cues in a Single Dimension.” Public Choice 64:213-227.

75

Groseclose, Tim, Steven D. Levitt, and James M. Snyder, Jr. 1999. “Comparing Interest
Group Scores across Time and Chambers: Adjusted ADA Scores for the US.
Congress.” The American Political Science Review 93: 33-50.

Hibbing, John R. 1986. “Ambition in the House: Behavioral Consequences of Higher
Office Goals Among US. Representatives.” American Journal of Political
Science 30: 651-665.

Hill, Kim Quaile, and Patricia A. Hurley. 1999. “Dyadic Representation Reappraised.”
American Journal of Political Science 43: 109-137.

Hurley, Particia A. 1989. “Partisan Representation and the Failure of Realignment in the
19805.” American Journal of Political Science 33: 240-261.

Hurley, Patricia A. and Kim Quaile Hill. 2003. “Beyond the Demand-Input Model: A
Theory of Representational Linkages.” The Journal of Politics 65:304-326.

Hurley, Patricia A., and Brinck Kerr. 2000. “The Effects of Party Advantage on the
Partisan Support on New US. House Members.” Legislative Studies Quarterly
25: 53-73.

Hutchings, Vincent L. 1998. “Issue Salience and Support for Civil Rights Legislation
Among Southern Democrats.” Legislative Studies Quarterly 23: 521-544.

Jackson, John E. 1971. “Statistical Models of Senate Roll-Call Voting.” American
Political Science Review 65: 451-470.

------- . 1975. “Issues, Party Choices, and Presidential Votes.” American
Journal of Political Science 19: 161-185.

Jackson, John E., and David C. King. 1989. “Public Goods, Private Interests, and
Representation.” American Political Science Review 83: 1193-1264.

Jacobs, Lawrence R., and Benjamin I. Page. 2005. “Who Influences US. Foreign
Policy?” American Political Science Review 99: 107-123.

Jacobson, Gary C. 1978. “The Effects of Campaign Spending in House Elections: New
Evidence for Old Arguments.” American Journal of Political Science 72: 469-
491.

------- . 1980. Money in Congressional Elections. New Haven: Yale University Press.

------- . 1985. “Money and Votes Reconsidered: Congressional Elections, 1972-1982.”
Public Choice 7: 7-62.

------- . 1989. “Strategic Politicians and the Dynamics of US. House Elections, 1946-86.”
The American Political Science Review 83: 773-793.

76

------- . 1990. The Effects of Campaign Spending in House Elections: New Evidence for
Old Arguments.” American Journal of Political Science 34: 334-362.

------- . 1992. The Politics of Congressional Elections. New York: Harper
Collins.

------- . 1996. “The 1994 House Elections in Perspective.” Political Science Quarterly
111: 203-223.

------- . 1999. "Party Polarization in National Politics: The Electoral Connection." In Jon
R. Bond and Richard Fleisher (eds), Polarized Politics. Washington DC: CQ
Press: 9-30.

------- . 2000. "Party polarization in National Politics: The Electoral Connection." In Jon
Bond and Richard Fleisher (eds), Polarized Politics: Congress and the President
in a Partisan Era. Washington: CQ Press, pp. 9-30.

Jacobson Gary C. and Samuel Kemell. 1983. Strategic Choice in Congressional
Elections. New Haven: Yale University Press.

Johannes, John, and John McAdams. 1981. “The Congressional Incumbency Effect: Is It
Casework, Policy Compatibility, or Something Else? An Examination of the 1978
Election.” American Journal of Political Science 25: 512-524.

Kedar, Orit. 2005. “When Moderate Voters Prefer Extreme Parties: Policy Balancing in
Parliamentary Elections.” American Political Science Review 99: 185-199.

Kiewiet, Roderick D. and Mathew D. McCubbins. 1991. The Logic of Delegation.
Congressional Parties and the Appropriations Process. Chicago: The University

of Chicago Press.

King, Gary, Robert O. Keohane and Sidney Verba. 1994. Designing Social Inquiry:
Scientific Inference in Qualitative Research. Princeton University Press.

King, Gary, and Lanche Zeng. 2001. “How Factual is Your Counterfactual?”
Manuscript. Harvard University, Cambridge, MA.

Kingdon, John W. 1973. Congressmen 's Voting Decisions. New York: Harper and Row.

------- . 1989. Congressmen ’s Voting Decisions. 3rd ed. New York: Harper
and Row.

Kollman, Ken. 1998. Outside Lobbying: Public Opinion and Interest Group Strategies.
Princeton: Princeton University Press.

77

Krutz, Glen. 2001. "Tactical Maneuvering on Omnibus Bills in Congress." American
Journal of Political Science 45 (January): 210-233.

Kuklinski, James H. and Richard C. Elling. 1977. “Representational Role, Constituency
Opinion, and Legislative Roll-Call Behavior.” American Journal of Political
Science 21: 1135-147.

Ladd, Everett Carll. 1985. “On Mandates, Realignments, and the 1984 Presidential
Election.” Political Science Quarterly 100: 1-25.

Layman, Geoffrey C., and Thomas M. Carsey. 2000. “Ideological Realignment in
Contemporary American Politics: The Case of Party Activists.” Presented at the
Annual Meeting of the Midwest Political Science Association, Chicago, IL.

LeoGrande, William M. and Philip Brenner. 1993. “The House Divided: Ideological
Polarization over Aid to the Nicaraguan ‘Contras’.” Legislative Studies Quarterly
18: 105-136.

Light, Paul C. and Celinda Lake. 1985. “The Election: Candidates, Strategies and
Decision.” In The Elections ed. Michael Nelson. Washington: Congressional
Quarterly.

Lindsay, James M. 1990. “Parochialism, Policy, and Constituency Constraints:
Congressional Voting on Strategic Weapons Systems.” American Journal of
Political Science 34: 936-960.

------ . 1991. “Testing the Parochial Hypothesis: Congress and the Strategic Defense
Initiative.” Journal of Politics 53: 860-876.

Lublin, David I. 1994. "Quality, Not Quantity: Strategic Politicians in US. Senate
Elections, 1952-1990." Journal of Politics 56: 228-41.

Lupia, Arthur. 1994. “Shortcuts Versus Encyclopedias: Information and Voting
Behavior in California Insurance Reform Elections.” American Political Science
Review 88: 63-76.

Lupia, Arthur, and Mathew D. McCubbins. 1998. The Democratic Dilemma: Can
Citizens Learn What They Really Need to Know? New York: Cambridge
University Press.

Matthews, Donald R., and James A. Stimson. 1975. Yeas and Nays: Normal Decision-
Making in the US. House of Representatives. New York: Wiley.

Mayhew, David. 1974. Congress: The Electoral Connection. New Haven: Yale

78

McCormick, James M. 1985. “Congressional Voting on the Nuclear Freeze
Resolutions.” American Politics Quarterly 13: 122-136.

McCrone, Donald J. and Kuklinski, James H. 1979 “The Delegate Theory of
Representation.” American Journal of Political Science 23: 278-300.

Miller, Warren E. and Donald E. Stokes. 1963. “Constituency Influence in Congress.”
The American Political Science Review 57: 45-56.

Miller, Arthur H., and Warren E. Miller. 1975. “Issues, Candidates, and Partisan
Divisions in the 1972 Presidential Election.” British Journal of Political Science
5: 393-434.

Mondak, Jeffrey J. 1993. “Source Cues and Policy Approval: Dynamics of Public
Support for the Reagan Agenda.” American Journal of Political Science 37: 186-
212.

Overby, L. Marvin, Beth M Henschen, Michael H. Walsh, and Julie Strauss. 1992.
“Courting Constituents? An Analysis of the Senate Confirmation Vote on Justice

Clarence Thomas.” American Political Science Review 86: 997-1003.

Page, Benjamin 1., and Robert Y. Shapiro. 1983. “Effects of Public Opinion on Policy.”
American Political Science Review 77: 175-190.

Page, Benjamin 1., et al. 1984. “Constituency, Party, and Representation in Congress.”
Public Opinion Quarterly 48: 741-756.

Peltzman, Sam. 1984. “Constituent Interest and Congressional Voting.” Journal of Law
and Economics 27: 181-210.

Poole, Keith T., and Thomas Romer. 1993. “Ideology, ‘Shirking’ and Representation.”
Public Choice 77: 185-196.

Poole, Keith T., and Howard Rosenthal. 1991. “Patterns of Congressional Voting.”
American Journal of Political Science 35: 228-78.

Popkin Samuel. 1991. The Reasoning Voter: Communication and Persuasion in
Presidential Campaigns. Chicago and London: University of Chicago Press.

Powell, Lynda. 1982. “Issue Representation in Congress.” Journal of Politics 44: 65 8-
678.

Rabinowitz, George and Stuart Elaine Macdonald. 1989. “A Directional Theory of Issue
Voting.” American Political Science Review 83: 93-121.

79

Rayburn, Sam. 1978. Speak, Mr. Speaker, eds. H.G. Dulaney and Edward Hake
Phillips. Bonharn, Texas: Sam Rayburn Foundation.

Rohde, David W. 1979. “Risk Bearing and Progressive Ambition: The Case of Members
of the US House of Representatives.” American Journal of Political Science 23:
1-26.

------- . 1991. Parties and Leaders in the Postreform House. Chicago: IL University of
Chicago Press.

Rohde, David W. and John H. Aldrich. 1995. “Conditional Party Government Revisited:
The House GOP and the Committee System in the 104th Congress." Extensions
Of Remarks_(in the A.P.S.A. Legislative Studies Section Newsletter) 19: 5-7.

Romer, Daniel Kate Kenski, Paul Waldman, Christopher Adasiewicz, and Kathleen Hall
J amieson. Capturing Campaign Dynamics: The National Annenberg Election
Survey. 2000. New York: New York: Oxford University Press.

Schattschneider E. E. 1960. The Semisovereign People. New York: Holt, Rinehart, and
Winston.

Schoenberger, Robert A. 1969. “Campaign Strategy and Party Loyalty: The Electoral
Relevance of Candidate Decision-Making in the 1964 Congressional Elections.”
American Political Science Review 63: 515-520.

Shanks, Merrill J. and Warren E. Miller. 1985. “Policy Direction and Performance
Evaluation: Complementary Explanations of the Reagan Elections.” Presented at
the Annual Meeting of the American Political Science Association, New Orleans.

Sinclair, Barbara. 1995. "Change and Continuity in the Legislative Process: The US.
House and Senate from the 19708 to the 19908." A paper delivered at the Annual
Meeting of the American Political Science Association.

------- . 1998. "Do Parties Matter?" Center for the Study of Democracy. 98-107.

------- . 2001. “The New World of US. Senators.” In Congress Reconsidered
7th ed. , eds. Lawrence C. Dodd and Bruce I. Oppenheimer. Washington, DC:
CQ Press: 1-19.

------- . 2002. “Do Parties Matter?” In Brady W. David and McCubbins D.
Mathew,(eds.), “Party, Process and Political Change: New Perspectives on the
History of Congress.” California: Stanford University Press, pp. 36-63.

Sniderman, Paul M., Richard A. Brody, and Philip E. Tetlock. 1991. Reasoning and
Choice: Explorations in Political Psychology. Cambridge: Cambridge University
Press.

80

Sniderman, Paul M. 1993. “The New Look in Public Opinion Research.” In Political
Science: The State of the Discipline 11 ed. Ada W. Finifter. Washington, DC:
American Political Science Association.

Stimson, James A., Michael B. MacKuen, and Robert S. Erikson. 1995. “Dynamic
Representation.” The American Political Science Review 89: 543-565.

Weisberg, Herbert F. 1985. “The Electoral Kaleidoscope: Political Change in the
Polarizing Election of 1984.” Ohio State University, Typescript.

Wittman, Donald. 1983. “Candidate Motivation: A Synthesis of Alternative
Theories.” The American Political Science Review 77: 142-157.

Wlezien, Christopher. 1995. “The Public as Thermostat: Dynamics of Preferences for
Spending.” American Journal of Political Science 39: 981-1000.

Wright, J .D. 1981. “Public Opinion and Gun Control” A Comparison of Results from
Two Recent Surveys.” Annals of the Academy of Political and Social Science
455: 24-39.

Wright, Gerald C. 1989. “Policy Voting in the US. Senate: Who Is Represented?”
Legislative Studies Quarterly 14: 465-486.

81

  

ES

iiiiiii

9919

MICHIGA STATE U I

111111111111111111111l

3 1293