u..n--

{bk—9...: .; . 3

x.
a. it...
5.31;". .
m5.

“.3.

P
J's

3:.
S:
n‘

u, M.

338.... a
1.“. 3PM? .
5%.»qu 43...};
IIV... . uraniohwmul:

‘I...Ih¢ V.

..s.....&. ‘5... 3,3,5 ..

V:

II. H. 10?... .
V t! 9.3:.le
. t.
r.
.. Law
‘3 97de9qu
i... . RT. a...» ma. .. ,
‘29.)": . . . 3...}. .32?!an
; a: , . . £00....
fighhnfifi. c,
‘l.’{‘§.t I” ‘ -
Sioi-.‘I"I p; ' ~£ -
lizav 31.1.”..Ovldhra'd‘i 1-" at
in... .i :1 ”usalhr
. a. ”in." h. .1». .A...SOJJ«.th—l. as .
1.35.7? 1: ‘..w’.flwaafil§alflrnnn 9
3|

.1 (a... . 4‘1
.... 394.....11 t...
iii-El... . “H t.
. . I?
.
I
‘EIi:
I
6.9;..- .5 I
. i.— it??? t’vtl..:|§§.§

—2‘.“>¢N.‘II 1%” In.
w .3930

1051...!
‘l I. g"!ul".l’n
5.1%.». Juxhufihuuafii . -511"..th
II I: l3h..$§ .38 gut-Vaihklualii
It!!! 43% .Ilnv.....fln.u..
:Lnngn...1.mr .. {at
3..
gluin$>auiil3511i
92.17:”; “bunn-
ii:i..n n
.9531} ix“. 0... 11..
I .320 Ont.

.fi 1.

irmunuuhhsix 3.; a...
3..ka $.51. V
x .33“?

(3 ”Mn. ..

I;
3.1....
i: , ‘ ‘ 3.1(V

. ‘41 ,Il ,
(a: 15...; .
£333?
i , 4;
2:33.?
3 3%..
3. V i , 3
Wfiflww“! .1}

12' 7:3

If s u. , . .r
intuit. .. ,

Envy: h... 24“..“

£5.81

. .tathanmwie h.
. ’ "fi'fioflfio
5ogla '5‘. I

g»: i:
. :‘ISI hr
:5

.32 x. .l.

3.». 33.4.7133: $3.22!...ng
jut§fflwupi I
. . .i..oi\nfl.i\x..;..I.Y.?€..?._.).llao‘t

If trifle. .t... «it.
a at.) viii-fixtrilxxnummwflfl

13. I... {4.5;}.
:1 :iziitiii.)
. 3:3:
‘93....

hr?

ti.
I}.L.€.r¢hb1hit..n’Vu.-hlld

‘Sttr.¢31¢
i§ilol.‘

 

. IX .. (4..
11...)... l; .

 

THESIS

jg} (-

 

LiBRARY
Michigan State
University

 

 

 

This is to certify that the

dissertation entitled

Mixed Signals: Incentive Structures

and Party Strategies
in Mixed Electoral Sys

tems

presented by

Erik S. Herron

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in filiticalicience

 

@m

V

Major professor

 

8/1/00
Date_

MS U is an Affirmative Action/Equal Opportunity Institution 0- 12771

PLACE IN REI'URN 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

 

We? 4021102

 

 

 

 

 

 

 

 

 

 

 

 

 

111w W.“

 

MIXED SIGNALS: INCENTIVE STRUCTURES AND PARTY
STRATEGIES IN MIXED ELECTORAL SYSTEMS

By

Erik S. Herron

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Political Science

2000

MIXED SK?

2.

NC

0’16?”
:15.“
.v S!- S‘

.fhnqn
mr P U, len‘:

“"Mn

V

n\v

2,.
1N v
I\U .

ABSTRACT

MIXED SIGNALS: INCENTIVE STRUCTURES AND PARTY STRATEGIES IN
MIXED ELECTORAL SYSTEMS

By

Erik S. Herron

Institutional designers have recently shown a preference for mixed
electoral systems, combining proportional representation (PR) and single
member district systems (SMD) into a single election. Underlying this choice has
been a belief that Duverger’s Propositions would apply to the individual
components of the system. However, the unique features of mixed systems
undermine the application of standard analytical approaches.

The strategic incentives influencing party strategies are rooted in
contamination effects. Contamination is the interaction between PR and SMD
components of mixed systems that affects the decision-making of voters,
contributors, candidates and parties. While existing scholarship suggests that
political actors compartmentalize the PR and SMD sections, there is strong
evidence of interaction.

This dissertation assesses contamination effects with data from nine
mixed systems, but focuses on data from Lithuania, Russia and Ukraine. The
empirical analysis produces three major findings: the mechanical effect in mixed
systems is more forgiving to marginal parties than the mechanical effect in SMD;

parties tend to participate in the whole mixed system, contingent upon resources;

27:89.10 can
A3313??? :4;
UKIEI‘E den,
39.813; 5:73
“ETAP‘ A.

l

”‘2' p0,, -.
"9 ....e m

y
's

and SMD candidate placement positively affects local PR vote outcomes.
Additional qualitative data based on interviews with political actors in Russia and
Ukraine demonstrate that political parties recognize contamination effects and
develop strategies to take advantage of them. All of the evidence points to the
emergence of multiparty competition in the PR and SMD components, as well as

the entire mixed system, contradicting Duvergerian expectations.

Copyright by
Erik S. Herron
2000

For LeaMarie, my loving wife and patient editor

By Its Vt

.. 4. A‘
uFU‘JIIG-King W

355 of them by

The if

are 8.5

Last!
UHI'I

«C
9v

r

a .
.lufil

fl

'3' are;

ACKNOWLEDGMENTS

By its very nature, a dissertation is a collective enterprise. Although only
my name appears as author, I could not have completed this immense
undertaking without the help and guidance of many people. While I cannot thank
all of them by name without doubling the size of this volume, I want to give
special thanks to a few special people.

The first and most important person in the long list is my dear wife,
LeaMarie Bistak Herron. Without Lea’s love, patience, sense of humor,
innumerable sacrifices and skilled editing, this dissertation would not have been
completed. My family and Lea’s family have also played a major role
encouraging me as I worked both here and abroad. I especially want to
acknowledge my parents, Thomas and Yvonne, and grandmother Ivy, who raised
me to cherish education and pursue this final degree.

I have also had many mentors at Michigan State who have skillfully guided
me along the way. Brian Silver has always been ready to work with me on all of
my projects and has spent countless hours helping me to hone my professional
skills (his late night e-mails attest to his round-the-clock concern for my academic
development). He has been a great mentor throughout my tenure as a graduate
student. Dennis Patterson provided me with unquestioned support and counsel
from the moment I arrived at MSU. He first introduced me to the issue that has
become my dissertation. Mark Jones also encouraged my professional
development and sparked my interest in electoral systems. Bill Reed provided

invaluable assistance and helped me solve many methodological puzzles.

vi

Imus! e
or he intal ve
tie ideas m tr.
Dee‘sped 0::
5r: framertai :l

Asa Fir;“.er F

(I)-

I must extend special thanks to Misa Nishikawa, my friend and co-author
on the initial versions of our future “magnum opus” on mixed systems. Many of
the ideas in the dissertation were hatched in conversations with Misa as we
developed our c00perative and individual projects. Other scholars have been
instrumental in honing my ideas and my approach to the dissertation, including
Ada Finifter, Robert Lowry, and Tse-min Lin. The Department of Political
Science support staff, Karen Battin, Rhonda Burns, lris Dunn and Elaine
Eschtruth, also helped me in many ways.

I also received quite a bit of assistance beyond East Lansing’s borders.
Kathleen Bawn and Robert Moserk kindly offered their data to me, despite the fact
that it would be used to question the assumptions of their own work. Moshe
Haspel generously augmented the Russian data with material he acquired.
Timothy Frye and Eric Browne offered substantial support for this project and
continue to be advocates for my work.

Outside of academe, Paul Harris at the New Zealand Electoral
Commission responded to my requests for information about his country with
blinding speed. J. Ray Kennedy at IFES in Washington put me in touch with the
field offices in Russia and Ukraine that were instrumental in making my field work
successful. It is also imperative to thank the American Councils for International
Education under whose auspices I conducted field research in Ukraine and
Russia.

While I was in Kyiv, I had a large support network. Svetlana Napara

provided me with a home and warm kasha every morning; I really mean every

vii

matting I re

I'. .‘ 1
Wane. ind "
Vac-nine and "

at: Kr *'
wSLja Sh;

 

it???“ ‘
.0 .8] RC

CEP” ‘
_: I A

.9:- we
5.3.. at IFEt
\
M»
vb‘fix .
W’v CSNC'

M:%"-.'
,. diov

morning. I received assistance from many other people and organizations in
Ukraine, including: David Earl and the staff of IFES; Vladimir Pigenko, Ellie
Valentine and the staff of the Parliamentary Development Project; Evhen Lapin
and Kostya Shevchenko at Lab F 4; Thomas Garrett and Oksana Hasiuk at the
International Republican Institute (IRI); Svetlana Kravtsova at the Ukrainian
Central Electoral Commission; Olexander Lavrynovich and Igor Iosipenko.

In Russia, Olga Gulikova gave me a home as well as syrniki and olady. I
also received assistance from Bob Dahl, Christian Nadeau, Ksenia Tyurkova and
the staff at IFES; Dr. Lee Patterson at IRI; Nikolai Petrov at the Carnegie
Moscow Center; Viktor Sheynis; Vladimir and Natalya Ovcharenko and Valentin
Mikhailov.

While many names are not included in this list, their help is greatly

appreciated and not forgotten.

viii

LIST OF TABL
LIST OF FIGU

LIST OF ABEF

|
|
INTRODUCTIC L

CHAPTER 1
INSTITUTION;
ELECTORAL 5
Sector. 5
he I
he I
Che" ;
An LITTLE; ;
Instr.
IUQKI
Rese;
Seal ;'
COHCIus-O,

i
I
I
I
I

CHAPTER 2
WIEQRETICAL ,
[Ogle of c, .
Party Site: 3
ESEargh L
RESe.v
Case 3
MPCP;'
InveSf.
ECho g

 

TABLE OF CONTENTS

LIST OF TABLES

LIST OF FIGURES

LIST OF ABBREVIATIONS
INTRODUCTION

CHAPTER 1
INSTITUTIONAL DESIGN, POLITICAL PARTIES AND MIXED
ELECTORAL SYSTEMS
Election Rules and Political Parties
The Role of Political Parties
The Development of Party Systems
Challenges to Duverger’s Propositions
An Introduction to Mixed Systems
Institutional Variation in Mixed Systems
Institutional Designers and Mixed Systems
Research on Mixed Systems
Seat Distribution in Mixed Systems
Conclusions

CHAPTER 2
THEORETICAL APPROACH AND RESEARCH DESIGN
Logic of Contamination Effects
Party Strategy Under Mixed Electoral Rules
Research Design
Research Hypotheses
Case Selection and Data Sources
Mechanical Effect Hypothesis
Investment Effect Hypothesis
Echo Effect Hypothesis
Models 1a and 1b: Echo Effect Hypothesis with
Controls for Popularity
Model 2: Echo Effect Hypothesis with Controls
for Regionalism
Model 3: Echo Effect Hypothesis with Controls
. for Independents
Conclusions

ix

xi
xiii

xiv

13
15
15
20
25
31
31
36
46
51
55

58
61
65
71
71
73
74
76
79

83
85

86
87

CHAPTER 3 .
THE MECHAI.
Proportcr

CONCLdSi:

CHAPTER 4

THE lNVESTI.‘
Partlcapa;
The E.“e:'. I
Aftermate a
Canclus :'

CHAPTER 5
THE ECHO EFF
Em E‘.‘e:
Res;

Res-

ECDC “9:
Echo E‘ie:
C3723 defies

CONCLLSQ-fi

 

 

CONCLUSIONS
GC‘Li-Afone
St'a‘v‘igl: A:
CCDCLUS'CQ

Sump;
Exit-3":
COPS"

AP"ENDICES
AEC‘EndIX A
"C=9hd;x B

EP’JO’GRAPHY

CHAPTER 3

THE MECHANICAL EFFECT 89
Proportionality Profiles 101
Conclusions 1 14

CHAPTER 4

THE INVESTMENT EFFECT 116
Participation in PR and SMD 119
The Effect of Resources on Participation 124
Alternate Explanations for SMD Placement 130
Conclusions 140

CHAPTER 5

THE ECHO EFFECT 142
Echo Effect Hypothesis, Models 1a and 1b 143

Results for Model 1a 143
Results for Model 1b 147

Echo Effect Hypothesis, Model 2 - The Effect of Regionalism 164
Echo Effect Hypothesis, Model 3 — The Effect of Independent

Candidates 168
Conclusions 174
CHAPTER 6
CONCLUSIONS — PARTY STRATEGIES IN MIXED SYSTEMS 176
Go-lt—Alone Strategy — Ukraine’s Rukh Party 180
Strategic Alliance - Russia’s SP8 and Yabloko 182
Conclusions 185
Summary of Findings 185
Extensions of the Research 189
Concluding Remarks 192
APPENDICES 194
Appendix A - Data Sources 195
Appendix B - Additional Evidence for the Mechanical Effect 197
Appendix C — Additional Evidence for the Echo Effect 206

BIBLIOGRAPHY 215

Tat'e I, Ef‘eCf

Mled Systems

 

Tazie 8. Meet a

are 7 Mew,

'-
V‘!“

L535 ‘2 SMD I

L-‘uh. afl A,
~ ...a 13.;
V»;

T- I
3,719”. SMD F
«3081028.

U

 

 

 

C
‘A
e“: 4
7' e pS'ior,
V
\- . I
‘7 Pea.
; arr
‘A .
3:5 4:
V Per‘I‘OFr
\
=2:- 4A
3 penr".

LIST OF TABLES

Table 1: Effective Number of Parties for the SMD Component of

Mixed Systems, SMD and PR Systems 4
Table 2: Electoral Rules of Selected Mixed Systems 37
Table 3: Seat Distribution in the Ukrainian Rada (1998) 52

Table 4: Distribution of Seats in Mid and West Wales Region 56
Table 5: Hypothetical Distribution of Candidates and Votes 81

Table 6: Mechanical Effect for the Mixed System (Lithuania 1996
— Second Round) 92

Table 7: Mechanical Effect for the Mixed System (Russia 1993) 95
Table 8: Mechanical Effect for the Mixed System (Russia 1995) 96
Table 9: Mechanical Effect for the Mixed System (Russia 1999) 98
Table 10: Mechanical Effect for the Mixed System (Ukraine 1998) 100

Table 11: SMD Participation by Parties Registered in PR (Russia
1993, 1995 and 1999) 120

Table 12: SMD Participation by Parties Registered in PR
(Lithuania 1996) 122

Table 13: SMD Participation by Parties Registered in PR
(Ukraine 1998) 123

Table 14: Results for the Effect of Resources on SMD Placement 126

Table 15: Results for the Effects of Resources on SMD Placement

(Fixed Effects Models) 129
Table 16: Performance of Small Parties (Russia 1999) 132
Table 17: Performance of KT R (Russia 1999) 133
Table 18: Performance of the LFU (Lithuania 1996) 134
Table 19: Performance of YL (Lithuania 1996) 135

xi

T551920 Pe"
late 21 Per"
T352924. OLS
Tate 23 OLS
Ta:!e24 OLS

T321325 OLS

 

T. .
:atle 3C H'y’CA.
, U.

m.
-. n
Mistat

.Ion

Ewe I
‘ ., 31. Hits"
Jag/era.

~d,l0n

O

 

Table 20: Performance of WOM (Ukraine 1998) 137

Table 21: Performance of GDP (Ukraine 1998) 138
Table 22: OLS Results for Model 1a (Lithuania 1996) 144
Table 23: OLS Results for Model 1b (Russia 1995) 148
Table 24: OLS Results for Model 1b (Russia 1999) 151
Table 25: OLS Results for Models 1b and 2 (Ukraine 1998) 155
Table 26: OLS Results for Model 1b with Controls for Gubernatorial
Affiliation (Russia 1999) 163
Table 27: Mean PR Vote and Number of Candidates by Region
(Ukraine 1998) 166
Table 28: Number of Covert Partisans for Ukraine (1998) 170
Table 29: OLS Results for Model 3 (Ukraine 1998) 171
Table 30: Hypothetical Distribution of Votes and Seats with No
Cooperation 1 77
Table 31: Hypothetical Distribution of Votes and Seats with
Cooperation 178
Table 32: Nomination Patterns for SP8 and Yabloko in Moscow

and St. Petersburg Districts (1999) 184
Table 33: OLS Results from Germany (1953) 208
Table 34: OLS Results from New Zealand (1996) 210
Table 35: OLS Results for Japan (1996) 212
Table 36: OLS Results for Japan (1996) with POPULARITY 213

xii

 

Figure 1' A CI
nge 2, Pro;
Fzgs'e3 Pr:

Figure 4 Pen

Feces Prcp

Figure 7: Prcpr

"Th CCU/ ,
7v . V- ,‘5 C300

LIST OF FIGURES

Figure 1: A Classification of Mixed Systems

Figure 2: Proportionality Profile for Lithuania (1996)

Figure 3: Proportionality Profile for Russia (1993)

Figure 4: Proportionality Profile for Russia (1995)

Figure 5: Proportionality Profile for Russia (1999)

Figure 6: Proportionality Profile for Ukraine (1998)

Figure 7: Proportionality Profile for Lithuania, Russia and Ukraine

Figure 8: Proportionality Profile for Lithuania, Russia and Ukraine
with 95% Confidence Intervals

Figure 9: Proportionality Profile for Armenia
Figure 10: Proportionality Profile for Japan
Figure 11: Proportionality Profile for New Zealand
Figure 12: Proportionality Profile for Scotland
Figure 13: Proportionality Profile for Wales

Figure 14: Proportionality Profile for All Countries

xiii

49

104

107

108

109

110

112

113

198

199

201

202

203

205

 

LIST OF ABBREVIATIONS

FSU Former Soviet Union

PR Proportional Representation
SMD Single Member District
Political Parties in Germany

BP Bavarian Party

CDU Christian Democratic Union
DP Deutsche Party

FDP Free Democratic Party

SPD Social Democratic Party

2 Zentrum

Political Parties in Japan

 

DPJ Democratic Party of Japan

LDP Liberal Democratic Party

NF P New Frontier Party

NHP New Harbinger Party

SDPJ Social Democratic Party of Japan
Shin. Shinshakaito

Political Parties in Lithuania

CDU Christian Democratic Union

Econ Lithuanian Party of the Economy
LCU Lithuanian Center Union

LLU Lithuanian Liberal Union

LNU Lithuanian National Union

LL Party of Life’s Logic

Peasants Lithuanian Peasants’ Party

Poles Alliance Lithuanian Poles’ Electoral Alliance
Pol. Prisoners Lithuanian Union of Political Prisoners and Deportees
Socialists Lithuanian Socialist Party

Women New DemocracyNVomen’s Party
YL Lithuanian National Party “Young Lithuania”
Political Parties in New Zealand

CC Christian Coalition

MS McGillicuddy Serious Party

PG Progressive Greens

UNZ United New Zealand

xiv

 

 

 

 

 

Political Parties in Russia

AN-SF Bloc of Andrey Nikolayev and Svyatoslav Fyodorov

APR Agrarian Party of Russia

DN Spiritual Heritage

DVR Democratic Choice of Russia

KPRF Communist Party of the Russian Federation

KRO Congress of Russian Communities

KTR Communists — Workers of Russia — for the Soviet
Union

LDPR Liberal Democratic Party of Russia

NDR Our Home is Russia

OVR Fatherland — All Russia

RSP Russian Socialist Party

Rybkin Bloc of Ivan Rybkin

SPS Union of Right Forces

Political Parties in Ukraine

APU Agrarian Party of Ukraine

Bloc Dem Bloc of Democratic Parties

CDP Christian Democratic Party

KPU Communist Party of Ukraine

Nat F National Front

NDP People’s Democratic Party

PSP Progressive Socialist Party

RAZOM Party of Rights and Liberal Party

R&O Reform and Order

S&P Bloc Bloc of Socialists and Peasants

SDPU Social Democratic Party (United)

UNA Ukrainian National Assembly

Political Parties in Wales

Con Conservative Party

Green Green Party

Ind. Turner Independent Turner

Lab Labour Party

LD Liberal Democrat party

NLP Natural Law Party

PC Plaid Cymru

SocLab Socialist Labour Party

XV

 

DIME-'96
the two-

Duvergé
balls- 4
Dave";

he adv
logettef

Instead
(Sanon

Fordece

testes that er.

recesses Re:

I"!!!

me upon 2'

1?th
.c. dc not die

 

 

INTRODUCTION

Duverger’s Law: “[t]he simple-majority single-ballot system...favors
the two-party system" (Duverger 1954, 217)

Duverger’s Hypothesis: “[tjhe simple-majority system with second
ballot and proportional representation favors multipartyism”
(Duverger 1954, 239).

The advocates of plurality-PR hybrids believe that they are bringing

together the best of two worlds; but they are likely to obtain,

instead, a bastard-producing hybrid which combines their defects

(Sartori 1997a, 75).

For decades, political scientists have struggled to uncover universal
theories that explain the behavior of individuals and groups involved in political
processes. Rather than developing research programs that encompass and
improve upon previous analyses,1 scholars often pursue independent agendas
that do not directly contribute to the construction of general theories of politics.
This tendency is particularly prevalent in comparative politics where schools of
thought historically have been tied to researchers’ ideological preferences and
theories are transitory (Geddes 1991 ). There is at least one exception to this
process.

Riker (1982) indicates that the study of political party systems and the
interaction between institutions and parties has followed a “Lakatosian” path.
Since Maurice Duverger formalized his theories about the relationship between

institutional rules and the number of political parties, the study of party systems

has moved forward by incorporating earlier scholarship. While some critics

 

1 Imre Lakatos (1978) provides the seminal definition of a scientific research program.

 

suggest that :r
6513;831:002 ”
rcie In shaprr;

Duverge
|
crgmal some:
I
e‘ecrve In ex;
|
1959. Feeders-
|
mncepts to ve'
er: Sevetsora
Va'etzueia 132-
Ketashima 19;

..
”a“
v. y

~ as of Duve:

‘I.
”rein {‘ r- 5

"CI has.

 

suggest that institutions have little impact on the development of partisan
competition,2 most political scientists accept the idea that institutions play some
role in shaping the party system.

Duverger‘s work has spawned many efforts to refine and expand his
original concepts. Scholars have studied how Duverger’s Propositions are
effective in explaining party systems (Palfrey 1989; Taagepera and Shugart
1989; Feddersen, Sened and Wright 1990; Fey 1997) and have applied the
concepts to various countries (Lijphart 1984, 1994; Reed 1990, 1999; Ordeshook
and Shvetsova 1994; Wada 1995; Kawato 1996; Gebethner 1996; Siavelis and
Valenzuela 1996; Neto and Cox 1997; Chhibber and Kollman 1997; Cox 1997;
Kabashima 1998; Jones 1999; Chhibber and Norfuddin 2000). Although the
effects of Duverger’s Law and Hypothesis have been studied in various
institutional and cultural environments, the expansion of new electoral institutions
creates the opportunity to extend Duverger’s Propositions beyond traditional
settings.

I argue that mixed electoral systems - those with both proportional
representation (PR) and single-member district (SMD) components - present
strategic incentives to candidates and parties that differ from those found in
independent PR or SMD systems with plurality rules. These incentives should
influence the development of the party system in ways not addressed by

Duverger. This research challenges the extension of Duverger’s Propositions,

 

2 Epstein (1980) has criticized institutional approaches to the study of party system development.

2

 

 

that l w": I call '

bath ‘PPC'EiICE

Hill I: ivvr

I

Schoia'
systems miles I
fetter: lhcsl
o’esgrters =th I
|
fiaIe erred, Tr
|
C0920. have 58'
8933033 STOCE

VOTE 0,;
liter these n; .

53'3” SI‘Stem E

 

 

 

that I will call “Duvergerian expectations,” to hybrid systems and is important for
both theoretical and practical reasons.3

Scholarly research that has analyzed the two components of mixed
systems independently may be invalid if strategic incentives in mixed systems
differ from those in standard PR and SMD systems. In addition, institutional
designers who selected mixed systems under Duvergerian expectations may
have erred. The failure of mixed systems to produce the predicted outcomes
could have serious implications for the population’s trust in parliament and the
electoral process.4

Vote outcomes in mixed systems suggest that party systems emerging
under these rules diverge from Duvergerian expectations. The SMD portion of
the mixed system does not seem to have the anticipated reductive effect on the
party system. By comparing the effective number of parties in the SMD
component of mixed systems to the effective number of parties found in PR and
SMD systems, we see that the mixed system’s SMD portion generally exceeds

the effective number of parties in standard SMD systems (see Table 1).5

 

3 The dissertation thus challenges later scholars' interpretations of Duverger's work rather than
Duverger’s own conclusions. I use the term “Duvergerian expectations” to denote interpretations
of Duverger’s work and not Duverger’s own Propositions. Specifically, Duvergerian expectations
refer to the assumption that PR and SMD components of mixed systems are independent and
that Duverger's Propositions should thus apply to each part separately. I will use Duverger’s
Propositions when specifically referring to Duverger’s own work.

Ukrainian voters expressed their dissatisfaction with parliament’s inability to govern in a
referendum conducted on April 16, 2000. They authorized the president to redesign the
legislative branch in part due to fragmentation within the legislature that undermined its ability to
Qovem. This fragmentation is due, in part, to the mixed electoral rules used to elect

parliamentary deputies.
The effective number of parties is calculated using the Laakso-Taagepera index which is:

/ 2 where p is the proportion of votes received by the party in the district. For Armenia,

p,
Lithuania, Russia and Ukraine, I calculate the effective number of parties by taking the mean of
each district's effective number of parties in order to account for independent candidates.

Table 1: E‘

8110 Systems I

‘r
Terrey 1933-:

U 8. House 19'
N-C’way 1882. ‘|'
US Secate 1?
SCSI). Africa 132‘
Sieden 1887:
New Zealand ‘
Ll"- :ee‘ KINGS"
Canada 19:136.
De'rark 193:-
late 1951-71

 

Table 1: Effective Number of Parties for the SMD Component of Mixed
Systems, SMD and PR Systems6

 

 

 

SMD Systems # PR Systems # Mixed Systems #
Turkey 1950-57 2.1 Belgium 1848-94 2.1 Scotland 1999 3.8
US. House 1910-70 2.1 Austria 1923-70 2.4 Wales 1999 3.8
Norway 1882-1903 2.2 Austria 1971 -79 2.4 Japan 1996 3.9
US. Senate 1912-70 2.2 Greece 1928,33,35,52 2.6 New Zealand 1996 4.1
South Africa 1958-74 2.3 Australia 1919-80 2.9 Armenia 1999 4.2
Sweden 1887-1905 2.3 Greece 1951, 56-77 3.0 Russia 1999 4.6
New Zealand 1946-84 2.4 Ireland 1923-77 3.1 Russia 1993 4.7
United Kingdom 1900-77 2.5 Turkey 1961-77 3.1 Lithuania 1996 5.6
Canada 1949-80 3.1 Japan 1928-80 3.3 Russia 1995 5.9
Denmark 1901 -1 3 3.8 Sweden 1952-68 3.4 Ukraine 1998 6.0
India 1951-71 4.3 Belgium 1919-77 3.5

Italy 1946-76 3.5

Luxembourg 1919-79 3.5

Sweden 1908-49 3.5

Sweden 1970-76 3.5

Iceland 1959-74 3.7

Norway 1930-49 3.8

Norway 1953-77 3.8

Venezuela 1947-73 4.0

France 1945-46 4.5

Israel 1959-77 4.5

Greece 4.8

1926,32,36,46,50

Netherlands 1937-52 5.1

Switzerland 1919-75 5.3

Finland 1906-83 5.4

Netherlands 1937-52 5.4

Netherlands 1918-33 5.8

Germany 1920-33 6.0

Estonia 1919-32 6.6
Mean 2.7 Mean 3.9 Mean 4.7
findard Deviation 0.7 Standard Deviation 1.2 Standard Deviation 0.9

 

 

 

 

 

6 The designation “all“ in columns 2, 4 and 6 refers to the effective number of elective parties.
Sources: Asahi Senkyo Taikan; BBC; Central Electoral Commission of Armenia; Central Electoral
C0mmission of the Russian Federation; Central Electoral Commission of Ukraine; Electoral
Commission of Lithuania; Taagepera and Shugart (1989).

4

 

l'Ihle Dev/ET;

 

wmpettiiOI‘l lr
prs-duced a t.
he 'pure" Sl.’
xntrast. the r:
Systems iS 5.2
alerege hum:
”ended more

he Slit com

I

T)

sates In seit-r

 

The PR
hargrrta! partze
sari es Comps;
Tasy small {,3

L" 39 Omar my

While Duverger’s Law indicates that SMD systems should produce two-party
competition in districts over time, the SMD component of mixed systems has
produced a higher number of effective parties than “pure” SMD systems. Among
the “pure” SMD systems, the mean number of effective parties is 2.7. By
contrast, the mean number of effective parties among post-communist mixed
systems is 5.2. If we include mixed systems outside the former Soviet Union, the
average number of effective parties is 4.7. While the sample size is too small to
conduct more rigorous statistical analysis of all cases, the number of parties in
the SMD component of mixed systems tends to exceed the effective number of
parties in self-contained SMD systems.

The PR threshold in mixed systems also does not seem to restrict
marginal parties from participating in elections. In Russia’s 1995 elections, 43
parties competed in the PR section. This number was reduced to 26 in 1999, but
many small parties are active in both PR and SMD. While the number of parties
in the other mixed systems analyzed has not exceeded Russia’s 1995
contingent, a substantial contingent of both competitive and marginal parties
have committed to electoral competition.

The large number of parties in many mixed systems could be due, in part,
to the newness of the electoral systems; not enough time has passed for the
“psychological effect” to reduce the size of party competition.7 There is another
possibility as well. Perhaps the combination of PR and SMD rules undermines
the properties that produce party systems based on Duvergerian expectations. If

mixed systems present political actors with different incentives, political parties

and candidate

QTYI’ITOI'ImenI

Is :ased on a:

‘cnahon of pg
9933'), The ;
ETCOL'ages CC '

and voters to

r‘
k.

Pl"
I'rVTC'Ie .m'r‘t-v-
1.1“" L

I”

.ETESIS

 

and candidates should adopt different strategies to be successful In this
environment. How do optimal strategies differ in PR, SMD and mixed systems?

In a PR system, small parties have a greater potential for gaining seats in
the legislature if the threshold for representation is low. Because seat allocation
is based on aggregate performance, a party does not need to win a majority or
plurality of votes in any local electoral district to gain legislative representation.
This encourages small parties to participate in elections and facilitates the
formation of parties based around specific concerns (i.e., regions, ethnicity,
gender). The ability of small parties to gain access to the legislature further
encourages contributors to support these parties, candidates to align with them,
and voters to cast their ballots for them. For these reasons, PR systems tend to
promote multipartism and allow political parties to form around more specific
interests.

In addition, PR generally encourages greater party discipline. Although
open-list PR has been shown to discourage compliance with party norms
(Shugart 1994; Ames 1995a, 1995b), closed-list PR is more common.8 In
closed-list PR, the party itself controls the construction of the party list.9
Candidates who have shown party loyalty are more likely to gain access to the
list and to obtain a higher rank. Failure to follow the expectations of national-

level leaders can lead to a low rank or exclusion from the list. In a PR system,

 

:3 I address the psychological effect in detail in Chapter 1.
Among the mixed systems under analysis in the dissertation, only Lithuania uses open-list PR.
Seat allocation within a PR system is determined, in part, by the party list. After the number of
seats that a party is to receive is determined by application of the electoral formula, seats are
assigned based on the list. A party rank-orders its candidates from first to the end of the list. If a
party receives fifteen seats, for example, candidates ranked one through fifteen would receive
seats in parliament. Seat allocation will be addressed in greater detail in Chapter 2.

6

exclusion fror
from gamg :
control over c
natonaffy-fc-ct

manS
en a mapréty
le accsrhpl sh
'eccgnrze that
cistrid. they a
Scarces o‘ “L

III,,'

»

'eccgnzzed as

Catheters lf‘,

 

 

 

 

 

 

exclusion from the list or a very low ranking effectively eliminates the candidate
from gaining access to the legislature. If a party exerts a greater degree of
control over candidates, it converts local concerns and pork barrel politics to a
nationally-focused agenda.

In an SMD system, by contrast, the best way to gain representation is to
win a majority of the votes in a single electoral district. The most efficient manner
to accomplish this task is to form local electoral alliances. When voters
recognize that a candidate or party will not finish among the top two in the
district, they are more likely to throw their support behind a potential winner.
Sources of financial support will also erode over time if the party is not
recognized as a legitimate local contender. Through this process, the number of
competitors in SMD races should pare down to two.

This process does not guarantee two major parties competing nationally.
Rather, party competition will eventually approach two effective candidates in any
district. Party aggregation at the national level is a separate concern.
Nevertheless, the strategic incentives provided to parties in an SMD system
generally favor larger political parties and undermine small alliances.

While larger electoral blocs are generally more successful in SMD
systems, the degree of party loyalty is also lower. Candidates strive to satisfy the
needs of local constituencies to win votes in the local race because each
candidate must win his or her own district to gain representation. Thus, the
effectiveness of the party at the national-level is less relevant for local

performance. Policy positions that stray from official platforms are thus more

 

common In 87
oo..ntry like t'
It systems W
with the party
geoerafly pres
LID systems
The err

new and esta:

o7 PR and SI.’

 

Tee Integrate

"w‘fiidSng 5:3"

‘09 meed SVSf~
B5“ Ia I
“Ci-Tar strate-
I COME."
Si'afegrc anen.

I“. e
Coma?)

~ InailOr‘

‘ “an-es. M
E‘s-4
”Mal TUE-s

 

common in SMD systems than in PR systems. This is particularly true in a
country like the United States, where ballot access is determined by primaries.
In systems where ballot access is determined by the party, the degree of discord
with the party platform should be less measurable. Nevertheless, parties
generally present a more cohesive platform and vision in PR systems than in
SMD systems.

The emergence of mixed electoral systems as a popular choice for both
new and established democracies raises a new question: Does the combination
of PR and SMD into a single election affect the strategies of political parties?
The integration of electoral rules with contradictory incentives may send
confusing signals to political parties. Alternatively, parties could recognize that
the mixed system provides its own unique set of incentives and that certain
electoral strategies are more effective in this environment.

I contend that the latter approach is correct. Moreover, I will show that the
strategic incentives in mixed systems are rooted in contamination effects.
Contamination is the interaction between PR and SMD components of mixed
systems that underlies the decision-making of voters, contributors, candidates
and parties. Mixed systems are not a combination of two independent sets of
electoral rules. Rather, each component affects the other. The decisions of
political actors necessarily take the whole system into account. While existing
scholarship suggests that political actors compartmentalize the PR and SMD

sections, I will provide strong evidence of interaction.

 

T Ie e»
tentative y es
Ir an an ‘ys s
2030). i eat .
meme 'llS :

xIntaT’TlT 3M3"

tromtwo m"

 

control it ' SC”

Tr s d <-
twitt: de‘ 3 fr:
he come her
from thos . to;
roe-chants I e“

heChar‘I

.c IL e“
tc carfICIp rte l'

The sec

“knit.
.cy to pg 115,;

"A”

we ,;C:pate m I:

WI:

‘Dete In tr

I}.
9:293 W H
I5
Ill :39

 

The existence of contamination effects in mixed systems has been
tentatively established. I will expand upon the preliminary conclusions produced
in an analysis of Japanese and Russian election data (Herron and Nishikawa
2000). That work demonstrated the existence of contamination across the
components of the mixed system. While the evidence is convincing,
contamination is characterized as a single effect and the tests are limited to data
from two countries that have similar mixed systems. The research also fails to
control for some potential rival explanations.

This dissertation further develops the theoretical argument and evaluates
it with data from additional mixed systems. I assess three hypotheses to support
the contention that mixed systems produce a set of strategic incentives that differ
from those found in PR and SMD systems. The first hypothesis is that the
mechanical effect in mixed systems is more forgiving to marginal parties than the
mechanical effect in SMD systems. If this is true, small parties should continue
to participate in elections over time.

The second hypothesis builds off the first. Given that small parties are
likely to participate in elections, how will they participate? While parties could
participate in PR or SMD alone, the second hypothesis suggests that parties will
participate in the whole mixed system, contingent upon resources. That is,
parties will attempt to gain seats in both the PR and SMD sections of the mixed
system.

The final hypothesis evaluates whether or not a party’s placement of

candidates in SMD affects performance in PR. lf candidate placement affects

 

PR vote out:
SMD. Pl! titre
the PR 2 no S
Dorerge 'san -:

T re d:
LIIE'BiJTE on :
Part-es a Id pa

ESP-733’ es 1:

SISIQT’IS Is dag
I. .
“Gal/10p

.
. \ ,
__ — — — —

 

 

PR vote outcomes, parties will be further encouraged to place candidates in
SMD. All three hypotheses support the emergence of multiparty competition in
the PR and SMD components and in the whole system, contradicting
Duvergerian expectations.

The dissertation will be structured as follows. Chapter 1 reviews the
literature on party system development. In this chapter, I address why political
parties and party systems are important and specifically why institutional
approaches to party system development are the most fruitful. The chapter also
discusses mixed systems by providing a detailed description of existing mixed
systems, variation among the electoral rules within these systems, and how
seats are allocated in different types of mixed systems.

Chapter Two addresses the theoretical approach of the dissertation. I
provide a more formal definition of contamination and demonstrate the problems
of ignoring its existence. I also emphasize the failure of the existing literature to
adequately address mixed systems. Although much of the literature on mixed
systems is descriptive, some scholarly work uses mixed systems to test the
behavior of voters, political elites and parties. The dominant trend in this
literature is to ignore contamination effects, undermining our confidence in the
authors’ conclusions.

I also discuss the research design in Chapter Two. I will use district-level
election data to evaluate various forms of contamination. The analysis focuses
on data from Lithuania, Russia and Ukraine, although data from Armenia,

Germany, Japan, New Zealand, Scotland and Wales are used to augment the

10

anafysls anC
lelso prowde

In Che
Using a meat

demonstrate :

I’vestrnem E‘-

3')d ordInary

A, .-
.. eedoral pa
In Char
.Lamlna‘ron

 

' We 01 ch
dCSS Contra:

 

analysis and extend it beyond post-communist mixed systems.10 In this chapter,
I also provide information on the variables used in the analysis and their coding.

In Chapter Three, I analyze the mechanical effect in mixed systems.
Using a measure of disproportionality as well as a scatterplot smoother, l
demonstrate that the mechanical effect in mixed systems is more forgiving to
small parties than the mechanical effect in SMD systems (as defined by
Taagepera and Shugart 1989).

In Chapter Four, I address the first form of contamination, which I label the
Investment Effect. I employ descriptive statistics of party nomination patterns
and ordinary least squares (OLS) analysis to show that parties tend to participate
in both components of the mixed system and that their resources affect the level
of electoral participation.

In Chapter Five, l demonstrate the existence of another element of
contamination, which I call the Echo Effect. OLS analysis shows that the
placement of candidates in SMD affects the vote outcomes for parties in PR.
This type of contamination should encourage parties to participate in many SMD
races, contingent upon resources and the electoral environment.

In Chapter 6, I present qualitative evidence from research in Russia and
Ukraine that supports the existence of contamination effects. Not only do these
effects manifest themselves in quantitative analysis, but they are explicitly
understood by political actors in the region. International organizations have

advised parties to take advantage of the effects and some political parties have

 

 

‘0 These tests are included in the appendices.

11

C

. ‘
.
led 8 VV\

ll1
8
qual’
Iati‘

 

formed electoral strategies with contamination effects in mind. After presenting

the qualitative evidence, I summarize the findings and present conclusions.

12

 

INSTITUT

 

were faced 'A
.nsttstlonal r-

:Q‘ hi I
Lusan‘sne.d C

534911 to Tesf
IDSIliqf

fifties Of the n,

LC 953” Other 2

 

Chapter 1

INSTITUTIONAL DESIGN, POLITICAL PARTIES AND MIXED ELECTORAL
SYSTEMS

The final decade of the twentieth century witnessed an increased interest
in institutional design and its role in determining the environment for competitive
politics. Former communist regimes, emerging from authoritarian, one-party rule,
were faced with inventing institutions in a new era of independence. Interest in
institutional tinkering was not limited to newly independent states, however.
Established democracies also reevaluated the efficacy of their institutions and
sought to restructure some of them.

Institutions, the array of formal and informal arrangements that define the
“rules of the game” in politics, establish how different elements of the state relate
to each other and how individuals attain positions of authority (North 1990). They
determine, in part, who is in a position to make policy, what views are
represented at the highest levels of government, and what mechanisms are used
to resolve conflicts in society. Because institutions can exert such an important
influence on the outcome of political disputes, decisions to select or alter
institutions are not taken lightly.

Constitutional engineers face an almost unlimited combination of choices
that can lead to stable democratic governance: a presidential or parliamentary
system; a unitary or federal state; a centralized or decentralized state authority; a

unicameral or bicameral parliament; and a variety of electoral rules for local,

13

regsonal and
they are not
Afiert
control In Ea:
forced to take
representatrv
governments.
compromise t
existing syste
Corsetrr-g Cor

Deepne

i . ..
=EcSIP‘

we Instrtu
5‘3 primary mg
aSSDClated Wli
System Is not 9

5.,
l

 

regional and national offices. Institutional choices may seem wide in theory, but
they are not unlimited in practice.

After the initial period of euphoria following the ouster of Communist Party
control in Eastern Europe and the post-Soviet region, institutional designers were
forced to take into account the competing demands of the victorious opposition,
representatives of the failed regime, international organizations and foreign
governments. Institutional changes in established democracies also required
compromise between the advocates of reform and those who benefited from the
existing system. The development of new institutions is based on the desires of
competing constituencies (Bawn 1993; Remington and Smith 1996).

Despite these constraints, institutional designers have a large menu of
feasible institutional structures. To further complicate the environment, each of
the primary institutional choices mentioned above has lower-level options
associated with it. For instance, the choice of a presidential or parliamentary
system is not dichotomous; a collection of secondary, tertiary and other choices
must be made to define the rules of the new political system. The powers of the
president vis-a-vis other institutions and political actors must be established.
Decisions must be made about the term of office, the electoral formula, timing of
the election relative to other elections, and how candidates will be selected.
Each of these choices affects the development of political competition and each
institutional choice affects the others.

The collection of institutions that define the political system is influential

because it has a “genetic code that programs the kinds of bureaucracies,

14

leadersh:p ST

ultimately me

Important I L ,-

L r‘ s‘ I
IRSEILLIIOT‘IaI C

 

leadership structures, personnel arrangements, and other properties that will
ultimately make up its government” (Moe and Caldwell 1994, 192). Countries
rarely change their institutions dramatically, so the initial choice at every level is
important (Lijphart 1984; Shugart and Carey 1992; Jones 1995).11 Few
institutional choices doom democracy to failure, but some are more conducive to
democratic survival than others (Jones 1995). Although the total effect of each
choice cannot be completely understood independently, it is useful to analyze the

direct effects of each selection on political competition.

Election Rules and Political Parties

The selection of election rules for the legislature may seem to be a
relatively minor issue in the context of institutional design, but electoral
institutions play a critical role in structuring partisan competition. By establishing
the rules for political contests, electoral systems directly affect the strategic
behavior of political elites and voters. While sociological and historical factors
can also influence how competition develops, political variables are primary

determinants of the party system.

The Role of Political Parties
Political parties are organizations that not only are able to participate in
elections but choose to do so (Sartori 1976; Mainwaring and Scully 1995).

Despite pessimistic interpretations of the utility of political parties and

 

“ When established democracies change their institutional rules, the changes are usually limited
in scape and are more difficult to implement.

15

suggestion
in both est;
Sch:
democrat:
he org ns l
ilat Pa'izes
cadre of 92::
Vu’emer 1:...5
mil-lax p05

modernzzatll

the tomes Cl

 

suggestions that they are declining in importance, parties have critical functions
in both established and new democratic states.12

Scholars disagree about the most important contributions of parties to
democratic political life. Views of party functions are often linked to beliefs about
the origins of parties. Political scientists in the modernization school have argued
that parties emerged when public policy could no longer be made by a small
cadre of elites who were unconcerned with public opinion (LaPalombara and
Weiner 1966). The advent of mass suffrage, combined with the increasingly
complex post-Industrial Revolution world, required parties. Like much of the
modernization literature, however, this analysis emphasizes a vague notion of
the forces of history as the causal mechanism in the formation of parties. The
role of political actors is subsumed under the sweeping power of social and
historical change.

Other scholars focus on suffrage as a key element in the development of
parties. Parties link society and the political world by mediating society’s
cleavages. in this argument, the forum of conflict was longitudinally consistent in
Western democracies during the early twentieth century. The cleavages of the
19605 were virtually identical to those of forty years earlier (Lipset and Rokkan
1967). Thus, the substance of partisan competition was consistent. Worldwide
economic and cultural change in the 19608 challenged this notion of continuity,

however. The development of competitive party systems outside Western

 

'2 Many scholars have suggested that the importance of political parties has declined. Their
salience in the United States (Nie, Verba and Petrocik1976; Wattenberg 1990) and in other
countries has been questioned in the face of an increasingly disenchanted and independent
electorate (Dalton, Flanagan and Beck 1984).

16

 

Europe and
power of the
Defin

ccrtme {0 ;

 

Europe and North America also undermined the external validity and explanatory
power of the argument.

Definitions of parties as linkage structures that emerged due to suffrage
continue to permeate the literature. Epstein (1980) states that parties “were
needed to provide clarity and continuity for an enlarged public.” He emphasizes
social factors in the development of party systems and indicates that “the
different historical origin of parties helps to explain the differences in their
organizational type” (ibid., 99).

Prior to mass suffrage, there was little need to provide labels to help
identify candidates’ positions on the issues. However, after the extension of
suffrage, new, relatively uneducated voters required a simple means to
distinguish among competitors. Parties emerged to provide these cues to the
public. Once again, parties are viewed as a historic necessity rather than as a
product of ambitious political actors.

Epstein explicitly rejects institutional rules as an explanation for party
system development, preferring to explain party systems by their proximity to civil
service reforms and their relationship to social, cultural and geographic
cleavages. While he acknowledges that some institutions may affect party
systems (notably federalism and the parliamentary/presidential dichotomy),
Epstein denies the influence of electoral laws and rejects the basic tenets of
Duverger’s Propositions.

In contrast to scholars who emphasize social forces in the establishment

of political parties, researchers in the rational choice tradition argue that parties

17

 

 

emerged as
Parties are "'
politicians V
30:3:r
problems in
1934), in at".
for citizens i
pa'ticuiarly tr
between cam
spend gathe'
who are affil:
iBchnrton t.‘
atiiffy to gar
Candidates 1‘
W :t'mr
53’ iEduCing t
54159011 Of the

acrcss time C

emerged as tools for astute political actors to gain office and make policy.
Parties are not formed as a required step in political evolution, but because
politicians view them as useful for their own purposes: election and governance.

According to this argument, parties emerged to address collective action
problems in election campaigns and in the legislature (Schlesinger 1991; Aldrich
1994). In an election, political actors must overcome the perceived disincentives
for citizens to vote. Parties can mobilize citizens by reducing the costs of voting,
particularly those associated with collecting information. By providing a link
between candidates and an ideological label, parties reduce the time voters must
spend gathering data to make an informed decision about voting. Candidates
who are affiliated with parties also accrue benefits because of the instant
recognition the party label provides.13 Moreover, as a party demonstrates its
ability to gain office and make policy, it transfers credibility to affiliated
candidates.14

Within legislative assemblies, parties allow political actors to make policy
by reducing uncertainty. In the absence of parties, majorities may form in
support of individual policies, but these coalitions are not likely to be consistent
across time or issue. Political parties in the legislature provide consistency and
certainty, allowing representatives to achieve their preferred policy ends more

often than in the absence of these coalitions (Shepsle 1986). Because parties

_

13 The notion of parties as labels is rooted in the work of Downs (1957). This relationship has
been further explored by Kiewiet and McCubbins (1991), Cox and McCubbins (1993), Aldrich
1995) and others as an element in electoral and legislative behavior.
Conversely, as a party shows that it is incapable of winning office or providing a platform
responsive to the interests of the electorate, its credibility will decline. This will lead to
abandonment by voters, candidates and sources of financial support. See Aldrich's (1994)

18

generate 515
the benefits
membersh?

I
”per

whesaea

owever. if;
sebsumed L.’
problems V.
tins analysgs
gcrem. Thus

sectoral proé

 

 

generate stability and consistency, ambitious politicians are likely to form them if
the benefits of joining a party are greater than any costs associated with
membership. 1

If parties emerge as a response to changes in social structures and to
mediate political conflict, their primary functions are to provide citizens with a link
to the state and to information about the ideological beliefs of candidates.
However, if parties are tools for political actors, their communication functions are
subsumed under their role as instruments in overcoming collective action
problems. While parties may have multiple functions, the assumption underlying
this analysis is that a party’s main role is to help politicians win elections and
govern. Thus, a political organization is not a party unless it participates in the
electoral process.15

Candidates sometimes seek office without a party affiliation. This is
particularly true in countries with inchoate party systems. In some new
democracies, the costs of the party label may outweigh the benefits, prompting
candidates to participate as independents (Smyth 1997; Moser 1999b). In many
cases, however, institutional rules within the legislature encourage independents
to join parties once they gain seats in parliament. Access to committee

assignments and other perquisites are generally controlled by parties; unaffiliated

 

discussion of the decline of the Whig Party and emergence of the Republican Party in the United
States for an example of this process.

‘5 In many mixed systems, a distinction is drawn between parties that participate in elections and
those that emerge in the legislature. Parties that are in the legislature are labeled 'frakstii" or
factions in Russia and Ukraine. Thus, they are not parties by definition - only groups that
participate in the electoral process are called parties.

19

   

legzsiatzve p

from comp-e?

l
The Develo,

Wher

 

 

 

E‘Clher, The
$5er 19";
. ,U w

QC“; emanCe

 

legislators are handicapped in the policy-making process."3 While political actors
may choose to participate as independents in the election, pressures during the
legislative period should discourage large numbers of independent candidates

from competing in the long-run.

The Development of Party Systems

Wherever political parties emerge, they cooperate and compete with one
another. The “system of interactions resulting from inter-party competition”
(Sartori 1976, 44) constitutes the party system. Just as the development of
individual parties responds to the institutional features of the electoral and
governance systems, so too does the party system.

Political scientists have developed two general measures to define party
systems: the number of parties in the system and their ideological distance.
Initial forays into the classification of party systems focused on the number of
parties (Duverger 1954). Scholars subsequently expanded the analysis to the
competitive structure (Dahl 1966), the number and relative size of parties
(Blondel 1968), the number of parties and majority formation (Rokkan 1968) and
the number and ideological distance between parties (Sartori 1976). While
measures of party systems have varied, the number of parties in a system has
consistently played a central role.

The number of parties in a system is critical in determining who has

power. Giovanni Sartori (1976, 120) indicates that “it does matter how many are

 

‘6 Rules encouraging deputies to join parliamentary parties exist in Lithuania, Russia and Ukraine
(Haspel, Remington and Smith 1998; Clark, Holscher and Hyland 1999; D’Anieri 2000).

20

 

the partres

 

roughly. an
power is ire;
analyzing tr
can undersfs
and how mav
ccndrtrons cr

Maurlr

 

have their 0..

Ml‘. linked e,

the parties. For one thing, the number of parties immediately indicates, albeit
roughly, an important feature of the political system: the extent to which political
power is fragmented or non-fragmented, dispersed or concentrated.” By
analyzing the effects of institutions on the development of the party system, we
can understand how parties interact with one another, how power is distributed,
and how many and what types of political organizations thrive under the specific
conditions created by the country’s institutions.

Maurice Duverger formalized the direct link between electoral institutions
and the number of parties in the system. Duverger’s Propositions, however,
have their origins in the work of other scholars. As early as 1867, John Stuart
Mill linked electoral rules and the party system. Mill suggested that adoption of
proportional representation in Great Britain would undermine the two-party
system. Two years later, Henry Droop posited that two-party competition would
result from the use of a plurality rule for seat distribution in SMD races." These
conclusions were based on logic rather than empirical evidence. Nevertheless,
they show that a close connection between electoral rules and party systems
was recognized far earlier than Duverger’s more formal elaboration of his
Propositions.

Early scholars who supported the idea that electoral rules affect the nature
of partisan competition presaged a logic that is implicit in formal analyses
developed by modern political scientists. Plurality rules tend to lead to two-party
competition because the only way to be certain of victory in an SMD race is to

receive the majority of the votes in a district. When more than two candidates

21

 

compete. tr
reduced. T
and candid;

:ndwidual p;

hcwerer, tr»:

achzeve 11161

0'? predzction

 

Duverl
laws 10 pro:
'ep’esemazzz

”33W Sir‘:

compete, the likelihood that one candidate will garner a majority is substantially
reduced. This creates uncertainty about the outcome among voters, contributors
and candidates. If ambitious politicians group themselves into coalitions,
however, they can increase their probability of winning a majority. The notion of
individual politicians, weighing costs and benefits, and choosing cooperation to
achieve their ends is the basis of the rational choice approach and is at the heart
of predictions based on Duverger’s Propositions.

Duverger’s Law18 and Hypothesis assert that a plurality ballot system
tends to produce a two-party system at the district level and proportional
representation leads to multipartism. Duverger’s Law states that “[t]he simple-
majority single-ballot system...favors the two-party system” (Duverger 1954, 217)
and Duverger’s Hypothesis indicates that “[t]he simple-majority system with
second ballot and proportional representation favors multipartyism” (Duverger
1954, 239). Thus, Duverger’s Law and Hypothesis address the major forms of
electoral rules used at the time and their implications for party system
development.

The ballot systems’ effects take two forms. The mechanical effect is the
filtering of votes into legislative seats, that is, how votes are counted and
accumulated in order to allocate seats. Generally, small parties in an SMD
system with plurality rules are under-represented; they receive some degree of
support in the electorate, but win a plurality in few, if any, districts. Unless a

marginal party has a strong regional following and is thereby competitive in one

 

‘7 See Riker 1982 and Cox 1997 for a discussion of the origins of Duverger’s Propositions.

22

or a handft

 

praportrona
systems 95*

lnaF

afid distnou:
national leve

relatively hog

 

599,30” in CS
’EQ-Séture

531839 18p!
marginal p0;

eel-3mm Of

9337533th V

J‘s-fit

USlh

9 SMD \

or a handful of districts, it is unlikely to receive seats in parliament that are
proportionate to its support in the electorate. The mechanical effect in these
systems can be described as punishing to small parties.

In a PR system, smaller parties are more likely to be represented,
although the degree of proportionality depends on the rules used to count votes
and distribute seats. A small party needs only to overcome the threshold at the
national level to win a place in parliament.19 This can be achieved through
relatively homogenous support throughout the country or through concentrated
support in certain regions that are large enough to push the party into the
legislature. A party need not win a plurality of the vote in a specific region to
achieve representation. The mechanical effect in PR is thus less punishing to
marginal political parties because they are more likely to receive a proportionate
allocation of seats under this system than under SMD.

The mechanical effect interacts with the psychological effect. By
translating voter preferences into actual representation, the mechanical effect
produces the cues that parties use to decide whether or not to participate in
subsequent elections. The psychological effect suggests that voters in countries
using SMD with plurality rules will fail to support small parties in general because

they do not want to waste their votes. That is, voters will behave strategically.20

 

'8 Technically, Duverger never called Duverger's Law a law. He simply stated that it "approaches
most nearly...to a true sociological law.”

'9 In some cases, the distribution of seats in PR is aggregated at a regional level. In this case, a
plurality is also unnecessary - the party must simply overcome the threshold to receive seats
within the regional district. This facilitates representation for small parties with a regional focus by
increasing the likelihood that they will overcome the local threshold.

2° If a voter behaves strategically, she will not choose the most preferred candidate if that
candidate is unlikely to finish among the top two in an SMD race. Rather, the voter will select
between the most competitive candidates. The alternative is to vote sincerely and choose the

23

lndwrduals

 

—'_—

from non-CC
assoczate tr
win seats,

reseurces a

they are like

 

two Competil
aafidiclates a
5”laid- ever

sure“) We

to .
I' 0
”Users are l

RM? -
"’ 91" r.
”<er

Individuals and organizations with resources will also tend to withdraw support
from non-competitive candidates. In addition, candidates will be prone to
associate themselves with major parties to maximize the probability that they will
win seats. As the process is repeated over several elections, voters and
resources are expected to concentrate their support on the two parties that are
most likely to win the election. As support disappears for marginal candidates,
they are likely to remove themselves from electoral competition or join one of the
two competitive local parties. Because voters, sources of support and
candidates abandon poorly performing parties in SMD, the psychological effect
should eventually lead to two-candidate competition at the district level when
plurality rules are applied, as Duverger’s Law predicts.

In a PR system, the psychological effect is less likely to lead to the
abandonment of small parties. Because less competitive parties are able to gain
representation in PR systems (restricted only by the threshold or other barriers),
voters are less likely to abandon them. Constituencies with resources are also
more likely to support smaller parties because these parties may gain seats in
the legislature and positions in the cabinet. In addition, candidates are unlikely to
leave small parties if their ability to win a legislative seat is not hindered. In fact,
some candidates’ chances of winning a seat might be enhanced by affiliating with
a small party if they are able to obtain a high position on the party list relative to
the position they would obtain in another party. Because votes for small parties

are not wasted in a PR system, these parties are expected to receive support

 

most preferred candidate regardless of that individuals chances of winning the district seat (Cox
1997).

24

 

from voters

 

seeking pa

Will tend to

Challenges‘
|
Beta

953311010915.

validity of ti‘.

 

”e“! acce:

AD ea
“no" punish s
h .
”as been di!

demonSiiajE

from voters, contributions from donors and interest from ambitious politicians
seeking party affiliation. Consequently, Duverger’s Hypothesis suggests that PR

will tend to lead to a multiparty system.

Challenges to Duverger’s Propositions

Because Duverger’s Law and Hypothesis rest upon the mechanical and
psychological effects, challenges to both elements potentially undermine the
validity of the Propositions. While the existence of the mechanical effect is
widely accepted, a few researchers have challenged it.

An early challenge to the mechanical effect asserts that SMD systems do
not punish small parties more than PR systems (Grumm 1958). This conclusion
has been disproved by later work. Using cross-national evidence, Rae (1971)
demonstrates that the mechanical effect and electoral rules follow Duverger’s
expectations. Riker (1982) challenges Grumm’s evidence and indicates that
evidence supporting the mechanical effect is widespread. Further, Abramson et
al. (1995) show that the mechanical effect punishes third-party candidates in US
presidential elections. Other researchers have calculated proportionality profiles
and break-even points for different systems, showing that the mechanical effect
in SMD punishes marginal parties more than the mechanical effect in PR

(Taagepera and Laakso 1980; Taagepera and Shugart 1989).21 Thus, this

 

2‘ A proportionality profile is an array of all parties in a system on a figure in which the X axis
represents the aggregate proportion of votes received and the Y axis represents
disproportionality (called the Advantage Ratio). A horizontal line drawn through 1.00 on the Y
axis represents perfect proportionality; the point at which the proportion of seats and votes
received are equal. These measures provide insight into the mechanical effect and will be
addressed further in Chapters 2 and 3.

25

mailenge t.
refined.
Wh

have also t

 

nave focqul
scholars he
Fennalts‘
because the

15:3er the .

 

 

challenge to the existence of the mechanical effect has been convincingly
refuted.

While stronger attacks have been made on the psychological effect, they
have also been addressed. The major challenges to the psychological effect
have focused on the behaviors needed for it to be manifested. Specifically,
scholars have questioned the existence of strategic behavior. Ferejohn and
F iorina (1974) forcefully suggest that voters never engage in strategic behavior
because the regret that voters would feel for failing to cast the decisive vote in
favor of their most preferred choice outweighs the rationale for strategic voting.

Some empirical evidence has supported the idea that voters seldom
behave strategically or that the effect of strategic voting is small. Shively (1970)
tests the relationship between a party’s likelihood of winning and the change in
that party’s proportion of the vote in successive elections. His results do not
undermine the existence of the psychological effect, but suggest that its impact is
minimal. Riker (1982, 762), however, argues that Shively’s interpretation of
Duverger’s Law and analysis is flawed. Other comparative analysis has
evaluated strategic voting, but generally concludes that it exists (Spafford 1972;
Cain 1978; Reed 1999). Overall, challenges to the psychological effect do not
stand up to empirical scrutiny (Riker 1982; Abramson et al. 1995; Cox 1997;
Abramson, Aldrich and Rohde 1998).

Although we can be confident that at least some voters behave
strategically, do political actors behave strategically? That is, given the

institutional constraints predicted by Duverger's Law, will elite-level support for

26

tn:rd partiei
H991)anc
parties in S
d'stricfs.
Sch:
at the hand
natonai-le‘.

SystemshaW

3v represe
ulDan$n*
lew

 

 

third parties decline over time in SMD systems with plurality rules? Schlesinger
(1991) and Aldrich (1994) indicate that the strategic incentives presented to
parties in SMD should encourage them to coalesce into two parties in the
districts.

Scholars have debated whether the inertia toward two-party competition is
at the national or district level. The assumption that Duverger’s Law is a
national-level concept underlies the discovery of empirical anomalies where SMD
systems have not led to two-party competition (Rae 1971; Riker 1982;
Taagepera and Shugart 1989).22 Moreover, in reviewing his own theory,
Duverger suggests that exceptions to Duverger’s Law can be explained by
evaluating the ideological dispositions of the parties. Multiple competing parties
may represent slight variations of two general positions. A form of “bipolar
multipartism” that approximates a two-party system might emerge at the national
level (Duverger 1986).

The mechanisms at work in Duverger’s Law take place at the district level,
however. Political actors calculate the possibility of winning a single seat and
base decisions on expected local performance. Therefore, it is logical to assume
that Duverger’s Law is a district-level concept and that multipartism at the
national-level is not incompatible with the development of two-party competition
in the districts (Leys 1959; Sartori 1986; Cox 1997; Chhibber and Kollman

1998).23

 

22 All of these authors imply that party aggregation at the national level in India and Canada to
2more than two parties suggests that Duverger’s Law is malfunctioning in these countries.

In fact, there is variation in the local effective number of parties in India. In some states, the
effective number of parties is generally greater than three; in others, it is close to two as

27

In a
of casein.
not determ
rules. Vr’hr
recent ana'

arrangeme'

1o:
v.4

.m

Re.

.3

00h Influer
were Show
39 not clan;
invalid
Rece'
deergenan
P'CDOsmons

EXCjOre the If

D' “I
“~9Fger's p

 

 

In addition to these challenges, some researchers assert that the direction
of causality posited by Duverger’s Propositions is incorrect. Electoral rules do
not determine the party system; rather, the party system determines the electoral
rules. While early research erred in challenging the direction of causation,24
recent analysis has shown that existing institutional and competitive
arrangements structure institutional choices, including electoral rules (Bawn
1993; Remington and Smith 1996; Fish 1998). Thus, the party system could
both influence the choice of electoral rules and change as a result of these rules.
While showing that the selection of electoral rules is endogenous, these studies
do not claim that the basic assumptions of Duverger’s Law or Hypothesis are
invalid.

Recent research on the political effects of electoral systems challenge
Duvergerian expectations. The deterministic nature of the Duverger’s
Propositions and their reliance on institutions alone has prompted researchers to
explore the interactive effects of institutional features and social conditions.
Duverger’s Propositions imply that a two-party system will emerge under plurality
rules and a multiparty system will emerge under PR regardless of the political
and social cleavages in society. Thus, the United States would be expected to
evolve into a multiple-party system if PR were introduced and Italy would develop

two-party competition in the districts if SMD were selected.25 Critics of

 

Duverger’s Law predicts although India has a multiparty system at the national level (Chhibber
and Nooruddin 2000).

2" Riker (1982) points out that Grumm (1958) finds multipartism in countries that converted to PR.
However, they had dual-ballot systems prior to selecting PR. One would expect multipartism in
these states due to the electoral rules.

25 This is the core complaint of Epstein’s (1980) critique of institutional approaches to the study of
party systems as well as others who prefer sociological explanations.

28

 

institutional approaches assert that social forces better explain the party system
than institutional rules. According to proponents of sociological explanations, the
number of cleavages in society will ultimately determine the shape of the party
system. Thus, a system using proportional representation could have two parties
in competition and an SMD-plurality system could have multiple parties.26

A primary assumption of this critique is that salient social cleavages will
form political groups. However, we cannot be sure that this will be the case;
collective action problems may undermine the translation of social cleavages into
groups that take an active part in political life. In addition, it is unclear what
makes a cleavage salient. Some cleavages may be important enough in a
society to promote the development of a political party or parties; other cleavages
may not be as strong. It is also unclear how to discriminate between truly
salient cleavages and cleavages that are not important enough to influence
political competition.

Acknowledging that social forces play a role in the development of party
systems does not preclude the role of institutions. Likewise, focusing on
institutional explanations does not deny the role of society in the political process
(Lijphart 1994). Recent research has shown that both institutional and social
factors affect party system development. Ordeshook and Shvetsova (1994) and
Neto and Cox (1997) demonstrate that ethnic heterogeneity affects the number

of parties in a system, but that institutions also affect the evolution of competition.

 

26 This is not qualified by the term “at the district level" because those who challenge institutional
approaches generally interpret Duverger’s Law as a national-level concept.

29

inis work .

 

sociologice
In a I
features ir
rgue that

parliaments

ifiFOU’Jh the

presrdent‘s

a larger 00'

 

636320.?! FL]
15mg“ an
has been 5,]
lhfera

CtIOn t

I: H
e‘vClOTaj 8,4

 

 

This work underscores the influence of institutional variables as well as
sociological factors in party system development.

In addition to assessing interaction between institutional and sociological
features in determining how political competition will develop, some scholars
argue that different institutional features interact as well. Presidential,
parliamentary and local elections are thought to influence each other primarily
through their relative timing and electoral rules. Concurrent presidential and
parliamentary elections create a “coattail effect,” generally benefiting the
president’s party.27 Further, majority-runoff presidential electoral rules encourage
a larger number of parties than plurality rules (Jones 1999). Parliamentary
electoral rules, moreover, can influence competition in the presidential race
(Shugart and Carey 1992; Jones 1995; Cox 1997). Institutional contamination
has been shown to affect political competition, although research has focused on
interaction between levels of elections rather than interaction within a single
electoral system.

I provide another challenge to Duvergerian expectations by assessing the
validity of extending Duverger’s Propositions to a new type of electoral system.
Duverger did not foresee that PR and SMD rules would be combined into a
single election and his Propositions do not directly address the implications of
mixed electoral rules. The recent emergence of mixed systems on all of the
world’s populated continents provides political scientists with the opportunity to

evaluate how institutions affect the evolution of partisan competition under

 

27 See Calvert and Ferejohn (1983) for an example of research on coattail effects. It is also
addressed by Shugart and Carey (1992), Jones (1995) and others.

30

previously
work on ir‘.

pond-no

An lntrod U

Institution

about how .
Nowhere he
former Sov»
mixed elect:
designers tr
ciSi’Diayed tr

not

beenthe

 

previously untested conditions. The research also follows in the tradition of other
work on interaction between presidential, parliamentary and local elections by

providing the initial investigation of a different form of institutional contamination.

An Introduction to Mixed Systems
Institutional Variation in Mixed Systems

Institutional designers recently have expressed a preference for mixed
electoral rules to select national legislatures. Established democracies such as
New Zealand and Japan have adopted mixed systems in response to complaints
about how well the electoral system represented the preferences of voters.
Nowhere has the spread of mixed electoral systems been greater than in the
former Soviet Union. Nine of the fifteen Soviet successor states have chosen
mixed electoral rules for their parliaments.28 While scholars and institutional
designers thought that mixed electoral rules would produce party systems that
displayed the positive characteristics of the PR and SMD components, this has
not been the case.

Duverger’s Propositions assume that either an SMD or a PR system will
be selected. This is logical, given that Duverger first wrote about the institutional
influence on party systems around the time that Germany adopted the first long-
lasting mixed electoral system. The emergence of mixed systems in countries
with different cultural and historical environments allows us to assess the

incentives provided by mixed electoral systems and how political actors adapt to

31

 

these rnoe
Hypothesr

Pol
distinct ete

1996i Mas:

 

SMD and a
A n;
CCWponen;
‘G'mulas ca
”7 5’26 the
may differ, J
x'la'y by Its El
whether or ,
RC Single or
In a C
:‘stribmiOn
l

Ella-n

 

these incentives. Moreover, we can evaluate how Duverger’s Law and
Hypothesis apply to party systems under mixed electoral rules.

Political scientists generally consider elections that include two or more
distinct electoral formulas in a single election to be mixed (Blais and Massicotte
1996; Massicotte and Blais 1999). Usually, mixed systems include both PR and
SMD and are utilized to select national or local legislatures.29

A number of features can vary across mixed systems. Within the PR
component, seats can be compensatory or non-compensatory; a variety of
formulas can be used to distribute seats; a threshold can be utilized and can vary
in size; the list can be open or closed; the proportion of seats allocated to PR
may differ; and there can be one or many PR districts. The SMD component can
vary by its electoral formula; whether districts are geographic or corporatist and
whether or not a threshold of districts must be met for representation. There is
no single or ideal type of mixed system.

In a compensatory system, the distribution of seats in PR is linked to the
distribution of seats in SMD. Using a prescribed mechanism, PR seats are
allocated to compensate parties that perform well overall but fail to win a
proportionate number of seats in the SMD component. In an SMD system, a
party could enjoy the support of 49% of the population but fail to gain any seats

in the legislature if the distribution of support across districts is homogeneous.

 

28 These states are Armenia, Azerbaijan, Georgia, Kazakhstan, Kyrgyzstan, Lithuania, Russia,
Tajikistan and Ukraine. Although elections in some states, notably Kazakhstan, Kyrgyzstan and
Tajikistan, have not met international standards, they use mixed rules for parliament.

9 While some mixed systems fuse the PR and SMD votes, these systems provide voters and
parties with different incentives than separate ballots. For this reason, these systems will not be
included in this analysis.

32

That is. lff

51% in ali

 

 

While this
undererep'
mjxed sy’ST'

In a

bass of the

 

proportiona

l
mechanism
or larger pr
insulate pa

component

Man

That is, if the party receives 49% of the vote in all districts and a competitor gains
51% in all districts, the latter party would receive all of the seats in parliament.
While this is unlikely to occur in practice, it underscores how parties can become
under-represented in an SMD system with plurality rules. In a compensatory
mixed system, the PR component is designed to prevent this type of distortion.

In a non-compensatory system, PR seats are distributed solely on the
basis of the electoral formula and threshold. The seats are not used to promote
proportionality of the overall system, but are distributed by a separate
mechanism from the SMD component. A party could, therefore, garner a smaller
or larger proportion of seats than votes. The non-compensatory system does not
insulate parties from possible allocation “errors” of the mixed system’s SMD
component.

Many formulas can be used to distribute seats in PR and some provide
more proportionate outcomes than others. Blondel (1969) suggests that PR
formulae can be ranked in descending order of proportionality: STV,”0 Sainte-
Lague,31 d’Hondt”2 and Largest Remainders.33 Some scholars have disagreed

with aspects of Blondel’s classification. Lijphart (1986) reanalyzes the formulas

 

3° The Single Transferable Vote (STV) is used in Ireland and Malta. In this system, voters select
candidates, not party lists. While the results produced are generally equivalent to the LR formula,
it uses the Droop quota (total number of votes divided by the total number of seats plus one;
rounded up).

31 The Sainte-Lague formula is 9 highest average formula in which each party’s votes are
successively divided by a set of divisors. Seats are distributed based on the “averages” that
result. The divisors in the Sainte-Lague formula are 1, 3, 5, etc. The Modified Sainte-Lague uses
a different set of divisors (1.4, 3, 5, etc.).

32 Like the Sainte-Lague, the d’Hondt formula divides each party’s vote by a set of divisors (1, 2,
3, etc.) to distribute seats.

33 The Largest Remainders approach generally uses a Hare quota. A quota is first calculated that
is the total number of valid votes divided by the district magnitude. Each party’s votes are divided
by the quota and they receive the number of seats that correspond to the full quota (i.e. the whole
number). The remaining seats are given to those with the largest ”remainder.”

33

and finds
most pro;
Lagoe. Ir
Averages
is variato
The

COF'Tpéflth

 

 

 

 

and finds that the order of proportionality is different. According to Lijphart, the
most proportional is Largest Remainders, followed by STV, Modified Sainte-
Lague, Imperiali Largest Remainders,”4 d’Hondt, and finally Imperiali Highest
Averages.35 Any of these approaches to allocation could be selected and there
is variation in the formulas used in mixed systems.

The PR threshold is a device often used to reduce the number of parties in
competition. The threshold sets a minimal proportion of the vote that must be
obtained in PR for a party to gain legislative seats. Parties that do not surpass
the threshold in PR receive no seats through this portion of the ballot.
Thresholds can be introduced to dissuade extremely small parties from
participation, or to prevent radical parties from gaining representation. They
have also been promoted by existing political organizations to preserve their
advantages in new democracies (Moraski and Lowenberg 1999). The PR
threshold does not directly affect seat allocation in the SMD component of a non-
compensatory system, however.

Open-list and closed-list PR have been shown to influence the behavior of
candidates differently and either system could be applied within a mixed system.
Parties in PR must develop a rank-ordered list of candidates that identifies who
will receive seats in parliament if the party overcomes the electoral threshold. If
a party gains ten seats in parliament through PR, for example, the top ten

candidates on the party’s list receive those positions. These lists may be

 

3‘ Imperiali Largest Remainders is similar to LR, but the quota is lower. The quota is the total
votes divided by total seats plus 2.
35 Imperiali Highest Averages uses 1, 1.5, 2, 2.5, and so on as its divisors.

34

national-le-
a different

In c
on the PR

party's pla‘

l' -
and climb:

example, a

_ o
V3975 In;

 

 

national-level lists or regional-level; if they are regional, allocation of seats follows
a different formula based on the same principle.

In open-list PR, voters express their preferences regarding the candidates
on the PR list. Open-list PR has been shown to reduce the consistency in the
party’s platforms as candidates seek to distinguish themselves from each other
and climb the PR list (Ames 1995a; 1995b; Shugart 1994). Lithuania, for
example, allows parties to choose whether or not they will submit their lists to the
voters. In practice, most parties participate in the open list component and voters
sometimes change the order identified by the party in accordance with their
preferences. In closed-list PR, the parties themselves choose the order of
candidates and voters have no input on list construction. Russia and Ukraine
both employ closed-list PR.

Mixed systems, as indicated above, contain both PR and SMD
components in a single election. The number of seats allocated by each portion
of the ballot varies from country to country. This division could affect party,
candidate and voter behavior depending on whether the system emphasizes PR,
SMD or balances the two. The total number of PR and SMD districts also varies
cross-nationally.

Variation in the SMD component is also evident in the countries under
analysis. Many mixed systems use plurality rules for the SMD component.
Lithuania, however, uses a majority-runoff system. Majority-runoff rules are
generally expected to increase the number of parties in competition in

comparison to plurality rules (Jones 1999).

35

Wit
district in
Sowet Um
O'gafilZBIJ

Tate 2 m

Institution

MXE
because of
COPSt'tution
assumed tt
the PCSitive
3‘50 emergt

desgners‘ q

 

 

While the use of corporate districts is possible theoretically, all of the
districts in the countries analyzed herein use geographic districts. Although the
Soviet Union’s final parliament allocated seats both geographically and to
organizations, none of the successor states retained these corporatist seats.

Table 2 compares these features among the mixed systems under analysis.

Institutional Designers and Mixed Systems

Mixed systems are a popular choice among institutional designers
because of their perceived ability to facilitate compromise and solve problems of
constitutional engineering and domestic politics. While designers have generally
assumed that mixed systems produce the “best of both worlds” by synthesizing
the positive features of PR and SMD systems, unintended consequences have
also emerged. These have undermined the achievement of some of the
designers’ goals.

According to standard expectations, the PR component of the mixed
system should produce incentives for political parties to emerge and participate
in policy debates.36 This goal is particularly important for new democracies in
post-communist states where no tradition of independent political parties or
multiparty competition exists. PR should also provide a proportional distribution
of seats that encourages ideologically diverse parties to participate. The history
of one-party rule restricted individuals and organizations with dissenting views;

the PR component should facilitate wider debate on important issues.

36

 

Counhy

Annema

 

\

 

Gennany

Jacan If

i

Lthoanua t

 

 

 

 

New y
2633an

 

R‘JSSIa IN

 

SCOffand Y

”r...
ViceS

\L

 

 

 

 

 

 

 

'l‘hor Shl r—
‘irreah .
y. tr'df‘i‘ 3?:?
._;‘i ‘u
. y‘k
"3] gut-531 r

Table 2: Electoral Rules of Selected Mixed Systems

 

 

Country Comp. # of PR % Seats PR Threshold % Seats SMD Total #
Seats? Districts in PR” in SMD gormula of Seats

Armenia No 1 43% 5% 57% Pl. 131
(56) (75)

Germany Yes 1 50% 5% 50% Pl. 496
(248) (248)

Japan No 1 1 40% None 60% Pl. 500
(200) (300)

Lithuania No 1 50% 5% 50% M-R 141
(70) (71)

New Yes 1 46% 5% 54% Pl. 120

Zealand (55) (65)

Russia No 1 50% 5% 50% Pl. 450
(225) (225)

Scotland Yes 8 43% None 57% Pl. 129
(56) (73)

Ukraine No 1 50% 4% 50% Pl. 450
(225) (225)

Wales Yes 5 33% None 67% PI. 60
(20) (40)

 

 

36 Viktor Sheynis, the author of Russia’s electoral mles, indicated that PR was essential for party
formation, stating: “no proportional representation, no parties.” (White, Wyman and Oates 1997,

784).

37 The actual number of seats is in parentheses.

38 “PI." is plurality; 'M-R' is majority-runoff.

37

In a
interpretat
are more r

Because t:

behave or"

responsiv

 

deputjes. t

iessiation,

 

 

 

In addition, if PR and SMD systems are independent, as standard
interpretations suggest, PR should create a cadre of elected representatives who
are more responsive to national interests and are identified by a party label.
Because they do not represent local constituencies, PR legislators should
behave differently than their counterparts in SMD seats.39 In addition, their
responsiveness to party expectations and norms should be greater than SMD
deputies, facilitating the construction of coalitions in parliament that can pass
legislation.

The SMD component should contribute to a stable party system by
restricting the number of parties in competition. Whereas PR should encourage
parties to form, SMD should provide disincentives for a large number of parties.40
If SMD did not immediately restrict competition, it should at least reduce the
number of parties in competition over time. Legislators elected in SMD should be
more accountable to local constituents. In this way, both national-level and local-
level interests could be well represented in parliament.

These goals of constitutional engineering were not realized, however. PR
is not necessary for political parties to emerge. In fact, Ukraine used an SMD
system in its first post-Soviet legislative elections and parties participated even
though there was no PR. Further, SMD has not prevented parties from emerging
in other countries.

The PR component of the ballot has not always produced a proportional

seat allocation in parliament. In fact, SMD seats have been more proportionally

 

39 This assumption has also been evident in research about mixed systems. See Lancaster and
Patterson (1990) and Judge and llonszki (1995) for examples.

38

distributed than PR seats in some cases. This is particularly evident in the 1995
Russian Duma elections. While the overall mixed system is more proportional
than pure SMD systems, disproportionality has emerged from PR largely due to
the electoral threshold.

SMD does not seem to restrict the number of candidates in competition,
although there has been limited time for the psychological effect to influence
behavior. Nevertheless, as I will demonstrate in a later chapter, characteristics of
the mixed system undermine the reductive features of SMD.

Other research asserts that mixed systems have not created two “classes”
of legislators. Deputies occupying PR seats in mixed systems may not be more
responsive to party norms and national-level interests than those in SMD.
Further, SMD deputies may not represent local constituencies as expected.41
Unintended consequences of mixed systems have undermined the achievement
of constitutional designers’ objectives.

Domestic politics also played a role in the selection of mixed systems. In
many countries, institutional designers chose mixed systems as a compromise
between proponents of PR and SMD. Some political parties expected to perform
better under PR than SMD and insisted on the adoption of a PR system. Other
parties expected to benefit from SMD and advocated for a system based on local
districts. Because mixed systems combine both PR and SMD, they were
selected as a compromise between proponents of both systems (Elster, Offe and

Pruess 1998).

 

4° The electoral threshold was also expected to restrict the number of minor parties in parliament.

39

Political actors have also perceived mixed systems as conducive to
“hedging bets.” Because the environment for political competition in most post-
communist states was uncertain at the time new electoral rules were adopted,
politicians were concerned about selecting a system that would accord them the
maximum number of legislative seats possible, but not result in devastating
defeat if they miscalculated. In an uncertain environment, mixed electoral rules
appear to be an optimal system (especially non-compensatory mixed rules). If a
party expects to perform well in PR but fails to do so, it can still gain seats in
SMD and vice versa. The mixed system can insulate parties from poor
institutional choices.

Interpreting mixed systems as a way to “hedge bets” assumes that SMD
and PR do not interact. The failure to perform well in one component of the

system may affect the party’s ability to gain seats in the other component.

Instead of protecting themselves from the possibility of poor performance, parties

could magnify their poor performance in a mixed system.

While mixed systems were interpreted as a potential solution to problems

of constitutional engineering and domestic politics, the features of mixed systems

may create unintended outcomes. Institutional designers in Lithuania, Russia

and Ukraine seem to have operated under some of the above assumptions when

they adopted mixed electoral systems.
Lithuania was one of the first post-communist governments to select a

mixed system and the choice of this system reflected political compromise.

 

4‘ See Haspel, Remington and Smith (1998) for more information about the relationship between
seat mandate and legislative behavior in mixed systems.

40

Beginning in 1992, the Lithuanian Seimas was to consist of 141 legislators of
whom 71 were to be elected in SMD and 70 were to be chosen through PR. The
initial threshold in PR was set at 4%; it was raised to 5% for the 1996 elections.
Party coalitions were required to pass a 7% threshold to gain seats in the PR
component and there was a 2% threshold for parties representing ethnic
minorities. The latter provision was eliminated prior to the 1996 elections.42 The
SMD component uses a majority-runoff formula and the PR component is open-
list.

The designers of the electoral system, especially members of Sajudis,“
anticipated that the SMD component would reduce the number of parties in
competition. This did not occur in the 1992 or 1996 elections. A diverse set of
parties gained representation through the SMD component; fewer parties gained
seats in PR (Clark 1995; Krupavicius 1998). While this outcome has been
characterized as a paradox, such a characterization implies that each mixed
system component is independent of the other. This seems to have been the
assumption underlying the choices by institutional designers, but I will show that
it contradicts the actual incentives presented by the mixed electoral system.

As in Lithuania, institutional designers in Russia did not completely
understand the incentives that would be presented to political parties in a mixed
system. Proposals for redesigning the legislative electoral system circulated in

Russia as early as 1991. Two major plans were prepared at the Institute of

 

‘2 Coalitions are groups of smaller parties that join together as a bloc, but do not formally merge
into a single political party.

41

Soviet Government Structure and Legislation: one promoted a PR system, the
other promoted a mixed system (Vasilyev and Postnikov 1995).44 The
opportunity for implementation of electoral reform did not emerge until the
dissolution of the Soviet-era parliament in 1993. Because opposition groups had
been ousted from power in October 1993, pro-reform groups could design a
system to benefit themselves.

Some reformers advocated a pure SMD system because they thought that
the reform movement would perform well in the districts. The results of national
referenda suggested that reformers enjoyed support in the regions. Because
they expected SMD to encourage smaller factions to coalesce into larger, more
competitive parties, reformers thought that they could win a large number of
seats and dominate parliament. Proponents of SMD were confident that
reformers were popular and they would form an umbrella party that would
dominate a two-party system.

Supporters of PR were not convinced that reformers enjoyed the level of
popularity claimed by the advocates of SMD or that liberals would form a single
party. Moreover, they were concerned that conservative local bureaucrats and
Communist Party sympathizers would illegally undermine the success of
reformers and cause reformers to lose seats in many regions. Because
reformers seemed to be most popular in large urban areas, the advocates of PR

argued that they would experience electoral defeat in the countryside. PR would

 

‘3 Sajudis was the dominant political organization that drove efforts for Lithuanian independence.
It had the largest coalition in the pre-independence legislature, but subsequently declined in
popularity (Krupavicius 1998).

Soviet era-legislatures were elected through SMD with majority-runoff rules.

42

allow reformers to benefit from their overall support in society, resulting in
representation by a few smaller liberal parties in the Duma (Remington and
Smith 1996).

In late 1993, two major approaches were promoted as designs for the
legislature’s new electoral system. The first plan advocated retention of an SMD
system; the second promoted a mixed system. Although there were no credible
proposals for the introduction of a pure PR system in the country,45 some
reformers advocated for the inclusion of PR in the new system. Through a
process of elite-level bargaining between the executive branch and
representatives from the legislative branch, a mixed system was chosen that was
divided evenly between PR and SMD (each component was allocated 225 seats)
(Sheynis 1999). The threshold for representation in PR was set at 5%. The
system would allow reformers to hedge their bets and was expected to benefit
liberal parties overall.

As in Lithuania, Duvergerian expectations directed the choice of a mixed
system. Thus, designers anticipated that PR would provide incentives for parties
to participate, but SMD and the PR threshold would provide disincentives for
many small parties to form (White, Wyman and Kryshtanovskaya 1995). While
the author of the electoral rules acknowledged that there was likely to be

interaction between the PR and SMD components, he viewed the system as the

 

‘5 Personal conversation with Viktor Sheynis, deputy in the Duma from Yabloko. He was one of
the primary authors of the electoral rules and is an advocate of mixed systems.

43

best alternative to promote the development of parties while not encouraging too
many parties.46

The outcome of elections in Russia surprised the system’s designers. In
the 1993 elections, the Liberal Democratic Party of Russia (LDPR), a nationalist
group, performed best in the party list vote and received the most seats in the
Duma. The Communist Party of the Russian Federation (KPRF) received the
third highest seat total by performing reasonably well in both PR and SMD.
Independent candidates were also prominent, receiving 155 of 450 seats. While
pro-reform movements acquired many seats, they did not obtain the
parliamentary majority that they expected (Wyman, Miller, White and Heywood
1994). Although they controlled the system’s design, unintended consequences
of the system prevented reformers from dominating the legislature.

The establishment of Ukraine’s mixed system bears some similarity to the
process of institutional reform in Lithuania and Russia, but also diverges from
these models. The main author of the election law in Ukraine understood that
there would be substantial interaction between the components of the mixed
system that could affect the strategies adopted by parties and the number of
parties that would emerge in the Verkhovna Rada. Prior to composing the law,
he studied the theory and practice of institutional design as well as the
experience of other countries. Although no country was used as a specific model
for the Ukrainian system, it is quite similar to the Russian mixed system. The

number of legislators selected through PR and SMD are the same as in Russia.

 

46 Personal conversation with Wktor Sheynis. See the previous footnote for more information.

44

The primary difference is that the threshold for PR is 4% in Ukraine rather than
5%.47

The mixed system was selected in Ukraine as a reaction to the elections
of 1994, conducted under SMD with a majority-runoff formula. The electoral
rules were blamed for much of the fragmentation in parliament and the inability of
the Rada to perform as an effective legislative institution. The mixed system, it
was thought, would provide some cohesion among parties but would also allow
voters to select individual legislators to represent their districts’ interests.
Ukrainian President Leonid Kuchma sums up how the expectations of party
system development under a mixed system did not comply with reality:

I often ask myself if I didn’t make a mistake by agreeing to a mixed

electoral system for the last parliamentary elections... The logic of

the decision [was that]... elections on the party list would speed the

development of blocs and coalitions, bring greater clarity, direction

and simplicity to the party palette. It did not turn out this way.

Instead, it was the opposite - there was a fragmentation of parties

that was engineered by the electoral ambitions of their leadership.

(Kuchma 2000)
Kuchma’s complaints about the mixed system imply that he expected PR to
promote parties to emerge as stronger groups that were few in number.

Election outcomes under mixed systems suggest the application of
Duvergerian expectations to mixed electoral systems may not be appropriate.
Rather than promoting party development through PR, many parties were

excluded from seat allocation due to the threshold. By contrast, many small

parties gained representation through SMD districts that had large numbers of

 

‘7 Personal conversation with Olexander Lavrynovych, deputy in the Verkhovna Rada from Rukh.
He was the author of the electoral rules.

45

candidates in competition. Because election outcomes in mixed systems cannot
be predicted based on traditional approaches, and unanticipated consequences
appear, expectations of the system’s designers are often undermined. This could
potentially lead to dissatisfaction with the electoral rules, confusion about the
probable trajectory of the party system, and further tinkering with electoral

formulas.

Research on Mixed Systems

The general dearth of research on mixed systems is due, in part, to their
scarcity. While there have been historic instances of mixed electoral systems as
well as one long-standing example (Germany), the number of countries using
these rules only expanded substantially in the last decade. In mid-2000, over
thirty countries and additional sub-national units utilized mixed electoral rules to
select their legislatures (Massicotte and Blais 1999).48

In response to their relative newness and lack of institutional uniformity,
mixed systems have been regularly classified as anomalies. Mixed systems
were disregarded in studies that were conducted when few examples existed

(Rae 1967; Sartori 1968, 1986; Butler 1982; Bogdanor 1983; Rose 1983; Nohlen

 

‘8 The number of political units using mixed electoral rules is higher than Massicotte and Blais
indicate. Two Russian regions, Saratov and Tuva, use mixed rules for their local legislatures. In
Saratov, 33% of the parliamentary seats are allocated through PR. In Tuva, 15% of the seats are
selected in the PR component. The remaining seats in both regions are allocated through SMD
with plurality mles (Central Electoral Commission of the Russian Federation 1998; Slider 1996).
In addition, Kazakhstan, Kyrgyzstan and Tajikistan adopted mixed systems. While the
democratic nature of their elections may be called into question, all three countries use a mixed
system.

46

1984; Grofman and Lijphart 1986). Writing at the time of growth for mixed
systems, Lijphart (1994) classifies them as consensus-type systems.49

Shugart (in Budge et al. (1997, 234)) addresses mixed systems as a
phenomenon that can be distinguished from PR. By drawing a distinction
between compensatory and non-compensatory rules, Shugart further recognizes
that there is institutional variation among mixed systems. Nevertheless, Shugart
continues the pattern of identifying mixed systems by their perceived similarity to
PR or SMD rather than as a separate system.5° He indicates that mixed systems
with compensatory seats are a form of PR, while mixed systems with non-
compensatory seats are “softened” forms of SMD.

A recent classification of mixed systems explicitly recognizes that mixed
systems are not mere exceptions to more widely used PR and SMD formats.
Massicotte and Blais (1999) propose a typology for mixed systems, based on
historic and current manifestations of these electoral rules. The classification
scheme first divides mixed systems into compensatory and non—compensatory
seat allocation mechanisms. Within the compensatory (dependent) system,
there are two sub-types: correction and conditional. Under a correction
mechanism, PR seats are distributed to prevent disproportionality due to
allocation "errors” in the SMD component. Under a conditional system, seat

distribution mechanisms in PR depend on the results of the other portion of the

 

‘9 Lakeman and Lambert (1955) view mixed systems as a form of PR.

This tendency is also evident in the nomenclature. Compensatory mixed systems are often
called “Mixed Member Proportional” and non-compensatory systems are labeled 'Mixed Member
Majoritarian.” These labels imply that compensatory systems are influenced by the PR
component and non-compensatory systems are affected by SMD. Moreover, it implies that the
outcomes will conform with the expectations of Duverger’s Propositions. For these reasons,
other labels are used herein.

47

election. The authors note that few examples of the conditional system exist and
none are in operation currently.51

Among non-compensatory (independent) systems, there are three sub-
types: coexistence, superposition and fusion. In a system of coexistence, part of
a country uses PR while another part of the country uses SMD. Two electoral
systems are used, but they are separated geographically. In a superposition
system, both PR and SMD districts operate on the whole territory of the country.
In a fusion system, PR and SMD are combined within a single electoral district.
Figure 1 shows the distribution of current mixed systems at the national level for
countries that do not utilize a fused ballot.

Mixed systems were used as early as 1906 in Germany and 1915 in
Denmark (Elklit 1992), but the archetype for most modern mixed systems is
Germany’s post-World War II rules for election to the Bundestag. The founding
election of the Federal Republic of Germany in 1949 was held under a mixed
system in which 60% of the seats were allocated to SMD with plurality rules and
40% were allocated to PR. Voters could not split tickets, however because the
PR and SMD votes were fused. The next election in 1953 introduced two
modifications to the rules. First, votes were de-coupled; individuals could select
an SMD candidate from a different party than the party they supported in PR.
Second, a threshold for representation was introduced in both PR and SMD. To
receive seats in the party list component, parties were required to garner 5% of

the PR vote or win seats in three districts (Bawn 1993).

 

5‘ Massicotte and Blais identify supermixed systems as a unique form of mixed system, but
supennlxed systems are essentially compensatory. In a supermixed system, PR and SMD are

48

 

 

Compensatory

 

Conditional - None

 

 

Correction - Albania, Bolivia,

 

 

 

 

Non-Compensatory

 

> Germany, Italy, New Zealand,

Scotland, Tunisia, Venezuela, Wales

 

 

 

 

Superrnixed - Cameroon, Chad,
Hungary

 

 

 

Coexistence - France, Niger, Panama

 

 

 

 

 

p Fusion -None

 

 

Superposition - Andorra, Armenia,
Azerbaijan, Croatia, Georgia,
Guinea, Japan, Kazakhstan,

Kyrgyzstan, Lithuania, Russia,

 

 

Tajikistan, Ukraine

 

Figure 1: A Classification of Mixed Systems‘52

 

Ecombined in more than one way in a single election, but one of the mechanisms is compensatory.

2 Figure 1 is a modified version of Massicotte and Blais’ classification. The figure includes

countries that did not have mixed systems at the time the article was completed. It also classifies
supermixed systems as a form of compensatory system. It further excludes Ecuador, Mexico and

Philippines. Ecuador’s system is supermixed, but not compensatory. Mexico uses a fused vote,

truly mixed.

so its inclusion in the universe of mixed systems is questionable. The Philippines’ system is not

 

 

 

This version of the German compensatory mixed system has been in
place since 1953 with minor changes in the total number of seats. Because the
system has both PR and SMD, it has been the subject of numerous scholarly
studies of strategic voting (Barnes et al. 1962; Fisher 1973; Jesse 1988; Bawn
1993; Cox 1997). In these analyses, the difference between a party’s vote in PR
and a candidate’s vote in SMD is viewed as a measure of strategic voting. The
underlying assumption is that the PR and SMD components are independent and
can be analyzed separately.

Other scholars have also assumed no interaction, or contamination,
between the components of the mixed system (Lancaster and Patterson 1990;
Moser 1997, 1999a; Krupavicius 1998; Clark 1999; lshiyama 1999). In some
cases, the research design explicitly separates the two components in order to
test the effects of PR and SMD on the party system. In other cases, researchers
imply that interaction between PR and SMD is not important.

While acknowledging the possibility of contamination between the two
components of the mixed system in post-communist states, Moser (1997, 1999a)
suggests that contamination does not undermine direct comparisons of PR and
SMD in mixed systems. He concludes that election outcomes in new
democracies may not conform with Duvergerian expectations due to incomplete
information among voters and candidates as well as the inchoate nature of the
party system. The problems with the assumption of no contamination will be

further elaborated in Chapter 2.

50

Seat Distribution in Mixed Systems

Mixed systems introduce a complex set of rules that link individual votes to
seat allocation in the legislature. In a “pure” version of PR or SMD, the
connection between votes and seats is usually clear. Voters in PR are presented
the opportunity to vote for a single political party that best represents their policy
preferences. Votes are aggregated at the national or regional level and parties
receive a proportion of seats that reflects their support in the electorate,
contingent upon the electoral threshold and formula for allocation. In an SMD
system, voters also have a single choice. They select a candidate to represent
the district. The candidate that receives the highest vote total wins the seat.53
Seat allocation in these systems is simple compared to mixed systems.

Seat distribution in non-compensatory mixed systems with a single PR
district is the most straightforward among the different types of mixed systems.
Voters have the opportunity to cast two votes; one is for a party in PR and
another is for a candidate in SMD. The winner of each SMD race receives a
seat. In the PR component, each party that passes the threshold (if there is one)
receives the appropriate number of seats based on the formula. The three
systems that are the focus of this analysis use non-compensatory mixed rules.
Lithuania, Russia and Ukraine employ an LR-Hare formula to distribute PR seats.

The PR threshold in Ukraine is 4%; it is 5% in Lithuania and Russia.

 

53 In an SMD system with plurality rules, this is the case. In an SMD system with majority-runoff
rules, a candidate only wins the seat in the first round if she gains 50% +1 of the votes. If no
candidate wins in the first round, a second round is held between the top two candidates from the
first round. The candidate winning the plurality in the second round receives the district’s seat.

51

For example, eight parties in Ukraine passed the 4% barrier in the PR
component in 1998. They were thus eligible for seats in the PR section. To
determine how many seats each party received, a quota was calculated by
dividing the number of eligible votes (17,481,593) by the number of seats to be
distributed (225), yielding 77,696.54 Votes for parties that did not pass the
threshold were not considered in the seat distribution. Each party’s vote total
was divided by this quota, and the whole number that resulted determined the
number of “full quotas” the party received. This is the first step in the distribution
of seats.

Table 3 shows the vote in PR received by parties passing the 4% barrier,
their quotas and the total number of seats they received in the Rada. Based on
the initial seat distribution, 221 of the 225 seats were allocated to parties. The

last four seats were allocated based on the fractional component of the parties’

Table 3: Seat Distribution in the Ukrainian Rada (1998)

 

 

% PR # of Quotas # of Seats
KPU 24.65 84.31 84
Rukh 9.40 32.15 32
S&P Bloc 8.56 29.27 29
Greens 5.44 18.59 19
NDP 5.01 17.14 17
Gromada 4.68 15.99 16
PSP 4.05 13.84 14
SDPU 4.01 13.72 14

 

 

E

5‘ This is known as a Hare quota.

52

 

number .

 

largest r(
Socialzst
In none:
aggrega:

districts.

 

diSI'lCtS [
T?
CSUl‘trteg
.‘n .-
'VrmUia
W775 ll‘e
bETWEe

£191,851}

70/

1
I
, 1
m
V1
U, I.”

(if II
J

‘5

(1’
7.. (17
r _ _

l) (7
" ;.
4P
)

(fl
.7} .._

rt."
-o

number of quotas. The first seat was awarded to Gromada because it had the
largest remainder (0.99). The next seats were allocated to the Progressive
Socialist Party (PSP), the Social Democratic Party (United) and the Greens.

In non-compensatory systems with multiple PR districts, PR votes are not
aggregated at the national level. Rather, seats are distributed in regional PR
districts. Theoretically, parties could choose to participate in only some of the PR
districts in this type of system. Japan uses this form of seat distribution.55

The distribution of seats in the SMD component varies slightly among the
countries with non-compensatory systems. Russia and Ukraine utilize a plurality
formula. The candidate that receives the largest number of votes in the district
wins the seat. In Lithuania, a majority-runoff system is used. If no candidate
wins a majority (50% plus one vote) in the first round, a second round is held
between the top two candidates in the district. The candidate who wins a
plurality in the second round takes the district seat.

Mixed systems that use compensatory rules distribute seats in a more
complex fashion. Under the German system, voters also receive two votes. The
first vote is for a candidate in the district-level race. The second vote is for a
party in PR. Parties that receive more than 5% of the vote in PR receive seats in

the legislature. If a party is unable to pass the 5% threshold, but is able to win

 

55 Russia uses a hybrid of both national and regional lists to determine seat allocation within
parties in PR, but this is distinct from the use of regional lists to allocate seats to parties. In 1999,
parties could place eighteen candidates on their national PR lists. Other candidates were placed
on regional lists. The party’s overall seat allocation was not affected by this division. Rather, it
determined the identity of the candidate who would receive the seat (IFES 1999).

53

 

568‘.

cor:

isa

unit

the

:nF

.hl

Ii
In

.5? /

A i’

seats in three districts, it also receives representation in the party list
component.56

The German system strives to distribute seats proportionally. The PR vote
is aggregated at the regional (Land) level. Votes are pooled at the national level
unless the party specifies otherwise. The number of seats won by the party on
the first (SMD) ballot is subtracted from the total number of seats to be allocated
in PR for the following calculation. The number of votes received in PR is
multiplied by the number of seats in PR minus the number the party won in SMD.
This is divided by the sum of all votes in PR. The number that is obtained from
this calculation guides seat distribution (Federal Returning Officer 1991).

Each party receives the number of seats that match the whole numbers in
its index. For example, if a party received an index of 1.5, it would automatically
receive one mandate. After these seats are distributed, the remainders are used
to allocate remaining seats. If a party receives more seats through
constituencies than its proportion of the vote, these are not taken away. Rather,
additional seats are created to generate a more proportional distribution of the
vote for all parties in the Bundestag.

Compensatory systems can also allocate PR seats regionally rather than
nationally. For example, Wales has five PR districts. Votes are aggregated at
the regional level in each PR district. The d’Hondt formula is used to distribute
seats with one qualification. The number of seats won by the party in the SMD

component of the region is added to the denominator. This facilitates a more

 

56 If a party passes neither threshold, it is not represented in the Bundestag. However,
candidates that won the SMD seats are allowed to enter partiament as unaffiliated legislators.

54

proporti:

 

I

SMD 115
and heir
Tr;

region ir

totals. Ir

seats the

 

the hlgi‘fi
ICON) WO
One, its it
won the .
aim—t

Cl’r‘lru ( t

T.
ejeCIOia
Stab-gt.

pg ang-

m
."l
{1 '

proportionate distribution of seats. If a party wins a large number of seats in
SMD, its denominator is raised to a level that prevents it from winning PR seats
and being over-represented in parliament.

Table 4 shows the process of seat distribution in the Mid and West Wales
region in the 1999 elections. The region has four PR seats. In the first and
second columns, the parties are identified along with their regional PR vote
totals. In the first round, the parties’ vote totals are divided by the number of
seats they received in SMD plus one. The first seat is allocated to the party with
the highest total after dividing through by the divisor. The Conservative Party
(Con) won the first seat. In the second round, the Con’s divisor was raised by
one, its index was recalculated, and the next seat was allocated. Again, Con
won the seat. In the next round, Con’s total was adjusted and the seat was
allocated to the Liberal Democrat Party (LD). The final seat was given to Plaid
Cymru (PC). This process was repeated for each regional district.

These examples illustrate the complexity of seat distribution in mixed
electoral systems. Despite this variation in seat allocation methods, basic
strategic incentives should persist across mixed systems. Interaction between

PR and SMD should not be sensitive to these aspects of institutional variation.

Conclusions
This chapter places the basic questions addressed in the dissertation

within a broader scholarly context. The literature on institutional design has long

 

This status prevents them from gaining rights accorded to parties in the legislature.

55

 

 

.F

—fl\(|l.l|l‘

Table 4 - Distribution of Seats in Mid and West Wales Region

 

 

 

 

 

 

 

 

 

 

 

 

Party PR Vote Number of Round 1 Number of
SMD Seats PR Seats
Con 36622 0 36622 1
Green 771 8 0 771 8 0
Ind. Turner 1214 0 1214 0
Lab 53842 2 17947 0
LD 31683 2 1 0561 0
NLP 705 0! 705 0
PC 84554 4 1691 1 0
SocLab 3019 0| 3019 0
Party Round 2 Number of Round 3 Number of
PR Seats PR Seats
Con 1 831 1 1 61 04 0
Green 771 8 0 771 8 0
Ind. Turner 1214 0 1214 0
Lab 17947 0 1 7947 1
LD 1 0561 0 1 0561 0
NLP 705 0 705 0
PC 1691 1 0 1691 1 0
SocLab 3019 0 3019 0
Party Round 4 Number of
PR Seats
Con 61 04 0
Green 771 8 0
Ind. Turner 1214 0
Lab 8794 0
LD 10561 0
NLP 705 0
PC 1691 1 1
SocLab 301 9 0

 

 

56

 

 

have 59

msfrtuhc
0184168 \
The like:

Stra‘LEI" r r

I
:4 x

recognized the direct connection between the “rules of the game” and political
competition in democratic systems. I focus on electoral systems.

The link between electoral rules and the party system is well established
in the literature. Duverger's Propositions, although challenged and modified,
have served as basic guidelines for predicting party system evolution. Although
institutional designers have adopted Duvergerian expectations, the combination
of rules with contradictory incentives raises questions about this assumption.
The likelihood of interaction between PR and SMD suggests that alternate

strategic incentives might emerge, undermining Duvergerian expectations.

57

 

Chapter 2

THEORETICAL APPROACH AND RESEARCH DESIGN

Institutions play a central role in shaping partisan competition. I treat
institutional rules as exogenous factors that affect the development of the party
system. Although it is well established that existing power arrangements
influence the introduction of new institutional rules (Bawn 1993; Remington and
Smith 1996), political actors must know and understand the possible
consequences of the alternatives for endogeneity to be meaningful.

The rules that define the roles political actors play constitute a polity’s
institutions (North 1990). Institutions define the environment in which decisions
are made and affect the outcomes of those decisions. They limit the actors’
choices by identifying the parameters of accepted behavior. These boundaries
can be formal, defined bylaws or other official documents, or informal,
established by a culture of behavior generally accepted by participants.

By extension, institutions also influence the strategies of political actors.
Politicians who understand the rules and boundaries can maximize the
probability that they will be successful. In electoral politics, success Is measured
first and foremost by gaining elective office. Electoral institutions “weed out”
individuals or groups that do not understand the costs and benefits inherent in
the system.

When developing mixed systems, institutional designers had incomplete

information; the potential effects of the mixed system were not well understood.

58

 

\I'v'i‘ll

adva
Il'lSiii
were
HEW’C

syster

“Sf-3'?
inohoa
atf'ibut
IC the n
Election
lSExpeC
Vt
Dames p
Strategjc

1' .
he “mo-u

 

 

While political elites may have selected rules that they thought would provide
advantages vis-a-vis their opponents, they could not be certain how these new
institutions would affect the development of the political system because there
were few practical examples to study.57 The institutional rules chosen in many
new democracies in the 19905 differed from the few existing examples of mixed
systems.

Scholars have posited various reasons for the failure of many mixed
systems to produce a small number of parties. The large number of parties and
inchoate nature of party systems in countries using mixed rules has been
attributed to the unique qualities of initial elections. Before political actors adapt
to the new incentives of the system, the number of parties can be quite large. As
elections take place over time, the number of parties participating in the system
is expected to decline.58

While this argument may explain, in part, the relatively large number of
parties participating in elections in post-communist states, I contend that the
strategic incentives of the mixed system also contribute to multipartism. Despite
the unique qualities found in new democracies and incomplete information
among participants in the political system, incentives created by mixed electoral

rules also affect the development of party systems.

 

57 In addition, some designers did not analyze the existing systems before selecting mixed
electoral rules. Viktor Sheynis, the author of the electoral law in Russia, indicated that he and his
colleagues did not study the German system until one year after they adopted the mixed system
in Russia (personal conversation with Viktor Sheynis, November 30, 1999).

58 Turner (1993) contradicts this argument by asserting that initial post-authoritarian elections
cannot be characterized as more volatile than later elections.

59

Sl’Siem 14

TI"
SMD m P
the Other

 

As I explained in the previous chapter, the incentives for voters and
parties favor the creation of a two-party system at the district level in SMD and a
multiparty system in PR, according to Duverger. When they are combined into a
single system, two possibilities emerge. The incentives of PR and SMD may be
independent, with PR providing incentives for small parties to participate in
elections and SMD reducing their number in the districts. The underlying
assumption of this possibility is that decisions of voters and politicians in PR are
not affected by SMD and decisions in SMD are not affected by PR. Further,
individuals and organizations with resources will determine whom they will
support in SMD independent of the strength of the party in PR (and vice versa).
Lastly, this approach assumes that elite-level strategies in SMD (i.e., candidate
placement) will not be affected by concerns for the potential effect on PR.

In this model, either PR or SMD incentives could dominate the overall
system. If PR were to dominate, we would expect a multiparty system to
emerge. If SMD incentives were stronger, we would expect that a two-party
system would emerge over time in the districts.

The alternate possibility is that there is contamination between PR and
SMD in mixed systems. Contamination is one ballot component’s influence on
the other. Rather than acting independently, the incentives produced in PR and

SMD are combined and together affect the decisions of voters and political elites.

6O

Logic of Contamination Effects

The imprecise logic of the “no contamination” argument can be best
illustrated through an analogy. The suggestion that there is no contamination
across components of the mixed system is equivalent to the Independence of
Irrelevant Alternatives (IIA) assumption in multinomial logit models. If one were
attempting to model an individual voter’s choice for a party or candidate, the
dependent variable would be the party/candidate selected by the voter. In
multiparty systems, multinomial logit would provide an appropriate alternative for
modeling choice with more than two discrete, unordered options.59 However, the
IIA assumption in multinomial logit results in an imprecise model of the
underlying process.

In a standard IIA scenario, the ratio of the probability of choosing one
party to another party is not altered by the entry of a third party that is considered
to be indistinguishable from one of the participating parties. Assume that we have
a country in which there are four parties in competition (A-D) and that parties A
and B are liberal, and C and D are conservative. Further assume that the only
dimension of partisan conflict is the liberal/conservative dimension. In district 1,
parties A and C are in competition.‘50 The IIA assumption states that the entry of
party D (or party B) would not alter the ratio of probabilities of a voter selecting

party A or C. Thus:

 

5” See Whitten and Palmer (1996) for an example.
so This could either represent a single SMD race or a national PR competition. In other words,
the model fits both standard PR and SMD systems.

61

PIA |S[A,C] = PIA IS[A,C,D]
Pic|S[A,c] PiC|S[A,C,D]

where S is the choice set and the terms in brackets represent the elements of the
choice set. Under the IIA assumption, the ratio of probability of selecting A to C
is not changed by the entry of party D.

If, for instance, the probability that the ith voter will select party A is 50%

(PM = .50) and the probability that he will choose party C is 50% (Pic = .50), then

the ratio of probabilities is Pi = 1.0. If party D (a second conservative party)
iC

enters the race, two conservative parties would be competing. While the
probability of selecting a conservative party may remain .50, we would anticipate
that the probability of selecting party C would be altered, especially if parties C
and D are viewed as essentially identical. If the selection of the conservative
party were random, then the probability of selecting C would become .25 and the
probability of selecting D would become .25. The resulting ratio of probabilities
would also change, violating the IIA assumption (Alvarez and Nagler 1997, 13).
Suggesting that the process of selecting a party in PR is independent of
candidate selection in SMD in mixed systems presents a similar challenge to
logic. Assume that parties A-D participate in PR61 but do not place candidates in
all SMD districts. Assume again that parties A and B are considered to be
liberal, parties C and D are considered to be conservative, and that this is the

only salient dimension of partisan conflict. In district 1, candidates from party A

62

and p
from r:
select;

C0813?

where 5
district g
Olf‘ier WC
in PR d:

dlstrrct,

 

 

and party C compete. Under the assumption of no contamination, if a candidate
from party D enters competition in district 1, the relative probability of a voter
selecting between parties A and C in PR should not change. Under no

contamination, we would expect the following:

PIA", ISPR U SSMD[A,C] _ Pmm ISPR U SSMD[A,C,D]

Prom |SPR U SSMD[A,C] Prom ISPR U SSMD[A,c,o]

 

 

where Spa is the choice set within PR, SSMD is the choice set within the SMD
district and the terms in brackets represent the parties competing in SMD. In
other words, the ratio of the probability of the ith voter choosing party A to party C
in PR does not change when choices are added to the SMD district and/or PR
district.

If the process of choosing a party in PR and a candidate in SMD is viewed
as a single act, then the IIA/no contamination assumption is incorrect. In mixed
systems, the implicit assumption of no contamination is that the ratio of the
probability of choosing one party in PR to another party is not altered by the entry
of another candidate in the SMD component. Contamination effects suggest
that this assumption is inappropriate; the introduction of party D’s candidate into

SMD should alter the ratio of probabilities of selecting party A or C in PR. That is,

 

6‘ We will assume that there is one PR district for the entire country.

63

The ratio of
change whe
assumption
assumptions
The e
contaminatic
would expec
the PIObabilil
e"Dem conta
Party in PR if
advantage of

Arty mi

subjeq to bia.

PIA”, ISPR USSMD[A,c] PIAPRISPR U SSMD[A,C,D]

 

 

Prom ISPR U SSMDlA,C] PiCPRISPR U SSMD[A,C,D]

The ratio of probability of the it“ voter choosing party A to party C is likely to
change when choices are added to either the SMD or PR choice set. The
assumption of no contamination is thus analogous to a violation of IIA
assumptions.

The equivalence of the IIA assumption to the assumption of no
contamination can be extended beyond individual voter behavior. Just as we
would expect the entry of additional parties/candidates into competition to alter
the probability that voters would support a particular party/candidate, we would
expect contamination to alter elite behavior. If voters are more likely to support a
party in PR if the party fields a candidate in SMD, parties should strive to take
advantage of this incentive.

Any model that assumes IIA where the assumption is inappropriate is
subject to bias. Not only can parameter estimates can be incorrect, but the
actual decision-making process is improperly modeled. Similarly, assuming no
contamination in voter and elite decision-making leads to an inaccurate
representation of the processes at work. Just as the IIA assumption in
multinomial logit models is not logical, the assumption of no contamination is not
logical in mixed systems. Given that contamination is a logical outgrowth of

mixed systems, how would we expect party elites to behave?

64

Party Stratr
Prior
affect their I.
whether to p
decisions er
and technica
as well as th
wzth other lik
ilmust deteri
In a rr
PR and SMD
Candidates lr
Other Parties
Candidates, tc
system Shoot
the PR and S

interdepende

Denys DOIem

 

Party Strategy Under Mixed Electoral Rules

Prior to parliamentary elections, parties must make many decisions that
affect their likelihood of entering parliament. Among the most critical decisions is
whether to participate in the elections and if so, how they will participate. Entry
decisions are based on the party’s assessment of its material, financial, human
and technical resources, its ability to meet its electoral and non-electoral goals,
as well as the competitive environment and opportunities for coalition-building
with other like-minded organizations. Once the party has decided to participate,
it must determine how it will participate in the elections.

In a mixed system, parties must determine how they will participate in the
PR and SMD components of the ballot, how resources will be divided between
candidates in these components, and if strategic arrangements can be made with
other parties locally or nationally. The leadership must also prepare a slate of
candidates to participate in the party list and the district races. The mixed
system encourages parties to construct strategies that are conditioned on both
the PR and SMD components. These decisions are consequently
interdependent. They are influenced by three factors: the electoral rules, the
party’s potential resources, and external conditions.

All parties are subject to the same electoral rules. While the application of
these rules may be subject to bias that benefits some organizations at the

expense of others, the rules themselves are the same.62 Across mixed systems,

 

62 Arguably, powerful or well-connected political organizations may benefit from biased
interpretations of electoral rules at the national or local levels. For example, in the 1999 Russian
paniamentary elections, some regional governors used their influence with local electoral
committees to undermine the candidacy of parties competing against their preferred organization.

65

variation in
influence de
threshold in
influence wl
races. A hig
through the
participants
The t}
National PR
Opportunities
list IS Used, u
are mOre like
present the p
ImDI’OVeS a Ca
paws reg Ion
atom potent];

salley the par

Ballot S

 

variation in rules can influence party strategies. Thresholds in PR or SMD
influence decisions about the party list and candidate placement. A lower
threshold in PR increases the likelihood of gaining seats in parliament and can
influence who obtains a high ranking on the list and who participates in SMD
races. A higher threshold increases the risk that the party will not gain seats
through the PR component, encouraging expanded participation in SMD among
participants on the PR list.

The type of PR list also potentially alters the strategic decisions of parties.
National PR lists, regional PR lists or hybrid PR lists present different
opportunities for seat acquisition and should affect list construction. If a national
list is used, the highest ranked seats are the “safest” — candidates ranked higher
are more likely to receive a seat in the legislature. Regional or hybrid lists
present the party with a more complex environment. While higher list placement
improves a candidate’s opportunity of receiving a seat, this is conditioned on the
party’s regional performance. Parties must collect more detailed information
about potential regional performance in order to construct regional lists that will
satisfy the party’s candidate pool.

Ballot structure also influences party strategy. The type of information
about candidates printed on the ballot can alter placement decisions in PR or
SMD. The PR list can contain information about the party and information about
the top candidates on the list. The SMD ballot can include personal,

professional, party or other information. This information can influence where

 

In addition, the media arguably played an active role promoting some parties nationally and
locally. While the legal environment was the same for all parties during the campaign, the actual

66

candidate
specific di.
Wh.

party. A p;
resource. I
outcomes II
the election
another org
lnforr
decision abc
Placed in ~55
districts in wI
candidate ma
Daffy list. In 1

"ill;
MnGiless of

 

candidates are placed on the PR list and which candidates contest seats in
specific districts.

While the rules are constant in the system, resources vary from party to
party. A party’s popularity at the national and regional levels is a powerful
resource. Reliable data about national-level popularity helps parties predict vote
outcomes in PR and construct the PR list. If a party expects to perform poorly in
the election based on aggregate popularity data, it may choose to coalesce with
another organization.

Information about the regional distribution of popularity can inform the
decision about candidate placement in the districts. Popular candidates may be
placed in “safe” districts and young, untested politicians may be nominated in
districts in which the party expects to perform poorly.“3 Alternatively, a popular
candidate may be placed in an at-risk district, but given a high ranking on the
party list. In this way, the candidate may attract voters in the district, but is
guaranteed a legislative seat if the party crosses the electoral threshold. The
willingness of candidates to accept such a task will vary, but the mixed system
presents parties and candidates an opportunity to “hedge their bets” in a way that
is not available in standard PR or SMD systems.

The number and identity of prominent candidates under the party label is
another resource. Party “stars” can include prominent politicians, well-known
cultural figures or friends and relatives of prominent political actors. If these

individuals bring substantial financial resources to the party in addition to their

 

golitical environment sometimes favored one organization over its rivals.
This assertion assumes that the electoral rules do not include a residency requirement.

67

“star qualit
mSMD.k
campaign f
in a well-co
does not 56
Nevertheles
ranking to b
lApar
well-develop
Candidate pl
Organizamn
MOVGOver, lh
SMD ”Omina

a large ”Umbi

range of norm

Dames dO not

i
a .aige ”Umbe

Camdldage D00]

nc- .

The final

nefaqOrS als

I IF.“
Elicia, benef
' E

“star quality,” they may be able to influence placement decisions on the PR list or
in SMD. In some cases, these individuals may not wish to participate in a
campaign fully, but are willing to donate their image to the party. This may result
in a well—connected politician receiving a prominent position on the PR list that
does not seem justified by the individual’s history of loyalty to the organization.
Nevertheless, the party is willing to provide a prominent individual with a high
ranking to benefit from “star power” or from the individual’s financial investment.

A party’s regional organization is also a potential resource. Parties with
well-developed and financed local organizations may assert more influence over
candidate placement than organizations with weak regional affiliates. As local
organizations are more autonomous, they can influence list construction.
Moreover, the party’s central organization may have limited control over local
SMD nominations if it has powerful local affiliates.

The party’s cadre of potential candidates is another resource. Parties with
a large number of credible, experienced and popular politicians have a wider
range of nomination options. Particularly in new democracies, some political
parties do not have enough credible members interested in seeking office to field
a large number of candidates in either PR or SMD. The size of the credible
candidate pool thus becomes an important issue for parties and affects
nomination patterns in PR and SMD.

The financial support enjoyed by the party and the identity of its
benefactors also potentially influence decisions about candidate nominations.

Financial benefactors may require parties to provide spaces for particular

68

candidate
capable 0:
Par
itself. The
party. The
parties are
same elect:
Component
candidate p
their slates .-
SIgrlature (:0
environment
To rec
form Strategn
agree to dlwc
fielding a Slhg
redUce the “k!
"7 Sum.
COhSlItUte the l
i0d_ Wme

the Same f

Der

candidates on the PR or SMD slate. Also, well-financed parties are more
capable of participating in a larger number of SMD races.

Party resources are, to a certain degree, within the control of the party
itself. The external conditions, however, cannot be influenced directly by the
party. The most important external factor is the competitive environment - what
parties are competing in the election and to what degree they draw from the
same electoral base as the party in question. Further complicating this
component of the political environment is that parties must make decisions about
candidate placement simultaneously. Parties cannot wait for competitors to form
their slates and then react; efforts to gain access to the ballot often require
signature collection or resource gathering. Consequently, the competitive
environment is not fully known at the time candidate slates are presented.

To reduce uncertainty, parties that compete for the same electorate could
form strategic alliances that fall short of merger. For example, parties might
agree to divide some or all SMD districts in order to avoid splitting the vote. By
fielding a single candidate who appeals to a particular constituency, the parties
reduce the likelihood that votes will be split among candidates with similar views.

In sum, the rules, potential party resources and external environment
constitute the initial conditions faced by political organizations during the electoral
period. While the party resources vary, the rules and external environment are
the same for all. All of these conditions affect the nomination process and mixed
electoral rules influence the role that these factors play in strategic decision-

making.

69

featUres
strategfe
under‘h‘
electOral
included i
analyze in
competmol
from the ‘0'
The ‘
incenllves'
on the deve;
cadre 0f ideo
Parties may a
Particular Sho’
Party el
by the system-
with scholarly e
understood by P

manifest themse-

The organ.

This analysis emphasizes the behavior of political parties. The unique
features of mixed systems complicate the study of political parties and party
strategies. Much of the data from mixed systems comes from initial elections
under these electoral rules. Some scholars argue that first elections under a new
electoral system are highly problematic (Pedersen 1983) and should not be
included in cross-national analyses (Maguire 1983). Although others specifically
analyze initial post-authoritarian elections (Turner 1993), the structure of partisan
competition and the behavior of parties in first elections may substantially deviate
from the long-term equilibrium.

The behavior of parties may not solely respond to immediate institutional
incentives. Some groups may focus on short-term seat maximization and others
on the development of strong regional organizations, a healthy financial base, a
cadre of ideologically committed followers or some other long-term strategy.
Parties may also miscalculate and pursue strategies that are intended to fulfill
particular short- or long-term goals, but are unsuccessful.

Party elites may not fully understand the institutional incentives presented
by the system. Consequently, strategies in the initial election may not conform
with scholarly expectations because the system’s incentives are not yet well
understood by political actors. Over time, however, different behaviors will
manifest themselves.

The organizational capacity of parties may vary. Some parties may have
tighter control over ballot access than others; some may have strong local

organizations while others have weak local bases; and some parties may

70

inter?“one
ideolOQ’Cé
Parties to
analysis”
DeS
information
focusing] on
parties Und‘
data with qu
Russia and (

general strati

Research De
Research HYI

l evalua
effects and eve

upon each othe

intentionally field “covert” partisans who eschew the official party label but are
ideologically linked with the party. As a result, we would expect the capacity of
parties to respond to the system’s incentives to vary, complicating cross-national
analysis.

Despite these problems, the study of parties can provide valuable
information about organizational adaptation to institutional change. Rather than
focusing on the strategies of individual parties, I will emphasize the tendencies of
parties under different institutional constraints. I will complement the statistical
data with qualitative information gathered from discussions with party officials in
Russia and Ukraine. The core analysis of the dissertation, however, will focus on

general strategies rather than the specific strategies of any given party.

Research Design
Research Hypotheses

I evaluate three hypotheses to assess the credibility of contamination
effects and evaluate their impact on strategic incentives. The hypotheses build
upon each other to define the strategic incentives presented to political parties in
mixed electoral systems. The first hypothesis assesses the strength of the
mechanical effect in mixed systems. If mixed systems have a mechanical effect
that is less punishing to marginal parties than the mechanical effect in SMD
systems, we would expect continued participation by small parties.

The second hypothesis illustrates the incentives presented to parties for

participation in both components of the mixed system. Because parties can

71

choose to pa
have a choic
encourages
should be ex
Duverger's L
The tt
the SMD cor
limit the num
Performance
mixed Syster
Candidates u
multiparty Co
as a Whole,
I use c
that Vary In 9
rules, Data E
Assessmg thi

robustneSS 0

8'5
Eff

Q
~Cts are the

choose to participate in the PR, SMD or both components of the election, they
have a choice about how fully to participate in the system. If the mixed system
encourages participation in both components of the mixed system, multipartism
should be evidenced in the PR and SMD components, despite the predictions of
Duverger's Law.

The third hypothesis addresses how fully political parties will participate in
the SMD component. Resources and strategic alliances between parties may
limit the number of districts in which parties field candidates. However, if
performance in PR is positively affected by candidate placement in SMD, the
mixed system generates pressures for parties to maximize the number of
candidates they sponsor in the SMD component. This should also encourage
multiparty competition in both components of the mixed system and the system
as a whole.

I use data from national parliamentary elections in nine mixed systems
that vary in geography, degree of party system institutionalization and electoral
rules. Data availability prevents the inclusion of each system in every test.
Assessing the hypotheses in various systems, however, allows us to evaluate the
robustness of the findings. Contamination effects should be evident in all mixed
systems, regardless of the minor variations in institutional rules. If contamination
effects are manifested in all of the mixed systems in this study, we can be more

confident that they are a product of mixed electoral rules.

72

Case Selection and Data Sources

Inherent in the comparative enterprise is the need to choose the specific
cases for analysis; it is rarely possible to include the entire universe of all
possible cases. Consequently, comparative research is Open to bias emanating
from the manner in which cases are chosen for analysis.

A further complication to case selection is that mixed electoral systems
are not “rare events” cross-nationally, but they are rare longitudinally. In all but a
few countries, only a single election has been held under these rules. While
short-term effects theoretically can be analyzed cross-nationally, long-term
effects can be addressed in a limited number of cases. Data availability also
restricts case selection among mixed systems. While aggregate data are
available for many mixed systems, district level data are less accessible.64 In
addition, the diversity of languages complicates the collection and interpretation
of data.

While data from nine countries will be employed, the analysis will focus on
three post-communist states: Lithuania, Russia and Ukraine. The availability of
appropriate data in these states facilitates a consistent analysis across all
hypotheses. Data from six other political units, Armenia, Germany, Japan, New
Zealand, Scotland and Wales, will be presented in the appendices to confirm that

the incentive structures found in the non-compensatory systems of Lithuania,

 

6" The use of district-level election data is appropriate both theoretically and practically. This
analysis investigates strategic incentives provided to parties at the district-level. Duverger’s
Propositions are district-level concepts; using aggregate data to estimate the effects of a district
level process is inappropriate. Further, the employing district level data in the analysis does not
introduce a problem of ecological inference. The individual tendencies of voters are not assessed
in this analysis. Rather, the central tendencies of performance that guide the activities of political
actors are under investigation. The data and underlying mechanism are both district-level.

73

Russia and Ukraine are not limited to a particular sub-type of mixed systems or
to new democracies. The analysis seeks to provide a comprehensive picture of
strategic incentives and party behaviors in a few mixed systems, but additional

evidence will help to demonstrate whether the conclusions are generalizable.

Mechanical Effect Hypothesis

The first step in understanding party strategies under mixed electoral rules
is to determine how marginal parties should react to the system based on its
mechanical effect. Duverger’s Propositions provide different predictions about
the behavior of parties based, in part, on how the mechanical effect influences
the possibilities for small parties to gain legislative representation. In a PR
system, small parties are more likely to gain seats because performance in the
aggregate is emphasized; they are less likely to gain seats in SMD because
candidates must win a plurality in a single electoral district in order to win a seat.
How does the combination of these rules in mixed systems affect the electoral
fortunes of marginal political organizations?

The Mechanical Effect hypothesis states that the mechanical effect in
mixed systems is more forgiving to small parties than the mechanical
effect in SMD systems. If this hypothesis finds support in the data, the mixed
system produces an incentive structure that does not undermine the participation
of small parties. That is, the mechanical effect does not produce cues for small

parties to abandon participation. This should have an impact on the

74

psychological effect, reducing the likelihood that mixed systems will produce two-
party competition at the district level.

In order to assess this hypothesis, I first evaluate a measure of
disproportionality with data from Lithuania, Russia and Ukraine. The calculation
of disproportionality is straightforward - the proportion of seats gained by a party
is divided by the proportion of votes received by the party.65 The resulting index
provides some insight into the system’s mechanical effect. An index of 1.00
indicates that the party’s representation in parliament is perfectly proportional
because the percentage of seats gained is exactly equal to the percentage of
votes accrued. If parties that receive a relatively small amount of the vote are
represented proportionately, they should not be dissuaded from participating in
future elections.

The best way to represent the mechanical effect is derived from these
calculations of disproportionality. Proportionality profiles and break even points
allow us to view the mechanical effect on a two-dimensional figure. The
proportionality profile assesses the degree to which the mechanical effect
punishes small parties. Applying the scatterplot smoother loess (Hastie and
Tibshirani 1990), I will demonstrate that proportionality profiles of mixed systems
do not conform with the proportionality profiles typically associated with SMD

systems.66

 

65 This measure of disproportionality is labeled the Advantage Ratio by Taagepera and Shugart
(1989). It assesses the proportionality of seat allocation for each individual party. Other
measures of disproportionality, like the Least Squares Index of Disproportionality, calculate the
groportionality of seat allocation for the entire system.

I use standard proportionality profiles derived from empirical data by Taagepera and Shugart
(1989).

75

The analysis will show that the mechanical effect in mixed systems is less
punishing to marginal parties than the mechanical effect in SMD. Consequently,
the mixed system provides an incentive structure that does not undermine
participation by small parties or independents. While the total number of parties
in competition may be reduced somewhat over time, it should not mimic an SMD

system.

Investment Effect Hypothesis

The first form of contamination encourages parties to participate in both
components of the mixed system and can be called the Investment Effect. When
a political organization decides to participate in an electoral campaign in a mixed
system, it can support candidates in SMD only, in PR only, or in both PR and
SMD. The Investment Effect encourages parties to select the third option:
participation in both PR and SMD.

Theoretically, a party could participate only in the SMD portion of the
mixed system. However, pursuing this strategy would place the party at a
disadvantage vis-a-vis parties that also participate in PR. Some benefits
accorded to political parties often are not conferred to parties that fail to register
in PR. Participation in the SMD component alone does not facilitate seat
maximization by political parties.

The second approach is to participate in PR only. If parties aim to
maximize the number of legislative seats in parliament, this approach is also not

logical. Parties must make a substantial investment of resources to mount a

76

campaign in PR. Because the marginal cost of participating in some SMD races
should not be high, given that the party already has an administrative structure,
some fundraising capabilities, and some committed politicians campaigning for it
at the national level, parties can maximize seat acquisition by participating in
both PR and SMD.

Participation in SMD should be dictated, in part, by the ability of parties to
overcome costs of committing to electoral competition. Parties with a greater
ability to overcome the additional costs of participation in SMD should participate
to a greater degree than those with more limited resources. The Investment
Effect Hypothesis thus states that the decision of parties to participate in PR
will encourage some level of participation in SMD. If this is so, parties
participating in PR should place multiple candidates in SMD races. In
addition, parties with greater resources should participate more fully in SMD
than those with fewer resources.

The hypothesis is evaluated in two ways. First, I will present the SMD
candidate placement patterns of parties in Lithuania (1996), Russia (1993, 1995
and 1999) and Ukraine (1998) to demonstrate that parties committing to PR tend
to participate in both PR and SMD. Second, I will employ OLS regression
analysis to show that a party’s resources are directly related to SMD
participation. Parties that are more able to overcome resource restrictions are
likely to participate in a higher proportion of SMD races.

Data from the most recent parliamentary elections in Lithuania (1996),

Russia (1999) and Ukraine (1998) are employed to evaluate the second aspect

77

of the Investment Effect Hypothesis. The unit of analysis is political parties and
each party that participated in the PR component is included.67 The dependent
variable is the proportion of districts in which SMD candidates are placed and is
labeled SMDPER.

Three independent variables are included to reflect a party’s resource
base. While information about financial resources is not generally available, I
have developed proxies for other resources. The variable PRLIST measures the
party’s personnel resources. Parties that have more candidates on their PR list
have greater personnel resources and should be able to place candidates in a
higher proportion of districts than parties with a smaller contingent of capable
politicians. The variable is an interval-level measure of the proportion of total PR
seats covered by the party’s PR list.68

The explanatory variable PRVOTE measures the party’s level of support
in the electorate. Parties with greater levels of grassroots support should also
have more resources. While this variable is problematic temporally, it is the best
proxy for support in the electorate that could be developed for this analysis. It is
also interval level.

Lastly, the variable PREVELECT is a dummy variable measuring
participation in previous elections. Parties with a longer electoral history should
have better-developed local organizations than parties that are new and thus

should be more capable of overcoming additional costs of SMD participation.

 

67 In Lithuania, some parties participated in PR but did not submit their lists to the Electoral
Commission to participate in open list PR. These parties were omitted from the analysis.

68 Proportions, rather than raw numbers, are used because Lithuania’s partiament has fewer
seats allocated to PR than parliaments in Russia or Ukraine.

78

The variable is coded “1” if the party participated in the previous election and “0”

if the party did not participate. The model for this hypothesis is thus:

SMDPER = or + BIPRLIST + BzPRVOTE + BgPREVELECT + e

The Investment Effect suggests that a political party’s commitment to
participation in PR and its level of resources directly affect participation in SMD.
Furthermore, multipartism should be produced in the SMD component of mixed

systems.

Echo Effect Hypothesis

Whereas the Investment Effect asserts that resources dictate party
behavior, the Echo Effect focuses on indirect effects that encourage parties to
participate in as many SMD races as resources allow. The PR electoral benefits
that can accrue to parties that participate in SMD races further encourage
multipartism.

The Echo Effect describes an interaction that affects electoral outcomes
directly and party strategies indirectly. When a party sponsors a candidate in an
SMD race, the affiliation of a party with a local candidate could improve
performance in PR at the SMD district level. The candidate’s campaign would
raise awareness of the party locally, increasing the likelihood that the party would
receive votes in the PR component of the ballot. The Echo Effect Hypothesis

states that the placement of candidates in SMD will affect vote outcomes in

79

PR in corresponding districts. The positive effect of SMD placement on
parties’ PR performance should further encourage parties to maximize the
placement of candidates in SMD. The empirical test evaluates the effects of
SMD candidate placement on local PR performance and controls for rival
hypotheses.

There are two major rival explanations to the Echo Effect. The first is that
the direction of causality in the model is incorrect. SMD candidate placement
may not affect PR performance; rather, SMD placement could be a response to
evaluations of party popularity. That is, parties place SMD candidates where the
leadership knows that the party enjoys a high level of popularity. If the rival
explanation is correct, the statistical significance of the coefficient for SMD is an
artifact of careful candidate placement, not an Echo Effect.69

Another version of this rival explanation is that any significance attributed
to candidate placement is an artifact of cleavage structures in the country.
Specifically, variation in PR performance is better explained by regionalism than
by the Echo Effect. In a society with strong regional cleavages, the parameter
value for SMD placement may be statistically significant due to regionalism. I
use data from Ukraine, a country with established regional cleavages, to evaluate
this rival hypothesis.

Regionalism could undermine confidence in the existence of the Echo

Effect in the following way. Let us say that we have a country with four parties

 

69 This rival hypothesis could be rejected on logical grounds. The degree of electoral volatility in
post-Communist states undermines the argument that parties have complete district level
information about where they are popular. Nevertheless, the models include controls for this rival
explanation.

8O

that partici
strong in d

D has the I

Tr

that participate in PR (A, B, C and D) and six districts (1-6). Party D is regionally
strong in districts 1 and 2 and only runs SMD candidates in those districts. Party

D has the following distribution of candidates and PR performance:

Table 5: Hypothetical Distribution of Candidates and Votes

 

D 1 D 2 D 3 D 4 D 5 D 6
SMD 1 1 O O 0 0
PR Vote 50% 50% 10% 10% 10% 10%

 

 

 

In this case, we would find that the variable SMD is related to the
dependent variable PR Vote. However, we do not know if this is due to the Echo
Effect (candidate placement), the regional strength of Party D in districts 1 and 2,
or some combination of the two.

There is well-established literature about regionalism in Ukraine and the
effect of regionalism on politics in the post-Soviet period. While other countries
using mixed systems may display some elements of regional cleavages,
regionalism has not been a major force in their party systems.7o Regionalism has
been an important force in Ukrainian politics, however. During the 1994
presidential race, the eventual winner, Leonid Kuchma, catered to the East’s
preference for closer ties with Russia. His competitor, Leonid Kravchuk,
promoted distancing Ukraine from Russia. Kuchma won the East; Kravchuk the

West (Holdar 1995; Solchanyk 1995). In the 1998 parliamentary elections,

 

7° Russia, in particular, has experienced regional conflict. War in the Caucasus region and the
negotiation of various bilateral treaties between ethnic regions and Moscow underscores some of
Russia’s divisions (Ordeshook and Shvetsova 1995). However, these regional cleavages have
not yet been manifested in the party system. In contrast to Russia, regionalism in Ukraine is

81

certain pa

nafionahs

Assembly

Parties lik

fondness ;

2000)

Anc

contaminal
Electoral cc
the PFOport
Russian e15
aCtuauy “Co
i”dependen
We“? COntr
affiliation. bL
COnfidenCe n
Effeq

If the a

is we'l‘khowr

a statisjmauy

an
d 'OCai pen.

53%

certain parties constructed their messages to appeal to the West. Ukrainian
nationalist parties or blocs, such as the National Front or the Ukrainian National
Assembly, emphasized ethnicity and language to appeal to Western voters.
Parties like the Progressive Socialist Party, that view the Soviet period with great
fondness and nostalgia, targeted the East (Craumer and Clem 1999; Kubicek
2000)

Another issue that potentially undermines confidence in the existence of
contamination is the large number of independents that have participated in
electoral contests. Only Lithuania has a small number of unaffiliated candidates;
the proportion of independent candidates exceeds 35% in Ukraine and all three
Russian elections.71 It is very likely that some “independent” candidates are
actually “covert partisans” (Moser 1999b). That is, they are not truly
independent, but have a partisan affiliation that is masked in the coding. Local
voters, contributors and other candidates may be aware of the candidate’s true
affiliation, but it is not captured in the data. If covert partisanship undermines
confidence in the coding, it could also threaten the validity of the test for the Echo
Effect.

If the affiliation of “covert partisans” with the party that is sponsoring them
is well-known and substantially affects voter perceptions, we would expect to find
a statistically significant relationship between a measure of covert partisanship

and local performance in PR. By adding a variable that controls for covert

 

reflected in the party system and its effects have been addressed in scholarly works (Barrington
1997; Craumer and Clem 1999).

71 In Russia, 58% of the SMD candidates were independent in 1993; 35% in 1995 and 47% in
1999. In Ukraine's 1998 elections, 48% of the SMD candidates were unaffiliated.

82

partisans
attenuate
Th
analyzed
performan
model of ti
participate:
The ability 1
underminec
standpoint. i
COmpeted in

8temples of

MOdeIs 1a a]

MOdejs

for the first rive

”the party War
With the Party

5

934
dance 0f the

to .
W

Van . at » .
‘3th 0n thepan'Cl
Van;

C

partisans, we can assess the degree to which this impacts PR performance or
attenuates the Echo Effect.

The Echo Effect is evaluated on the level of political parties. Each party is
analyzed separately to gauge the precise effect of candidate placement on PR
performance. Because parties vary in size and relative performance, a pooled
model of the entire system would provide less useful results. All parties that
participated in at least 20% of the SMD districts are included in the analysis.72
The ability to assess the relationship between placement and performance is
undermined if parties run in very few, or virtually all, districts. From a practical
standpoint, this requirement tends to eliminate many of the “micro parties" that
competed in the newly independent states of the former Soviet Union. However,

examples of both small and large parties for each country remain in the analysis.

Models 1a and 1b: Echo Effect Hypothesis with Controls for Popularity

Models 1a and 1b evaluate the existence of the Echo Effect and control
for the first rival explanation. The dependent variable is performance in PR at the
SMD level and is labeled PRPER. SMDPL is a dummy variable that is coded “1”
if the party placed a candidate in the district and “0” if no candidate was affiliated
with the party. If the SMD coefficient is statistically significant, it demonstrates
that placement of candidates in SMD directly affects PR performance and is

evidence of the Echo Effect’s existence.

 

72 Parties that participated in over 80% of the districts were also eliminated to ensure some
variation on the variable SMDPL.

83

A
may hav
The van;
candidate

Cc
adequate
proxies. F
the district
Characteris
OfRussia, ,

to PErform i
that attract i
Parties are S
Calidldates i
The v
DOPUIanty. E
CENSUS an d E
are at the Obi

f ,. .
‘e’mollal Wilts
u ,;
fits matCh thé

al~
so mater, CUn

A control for incumbency is also included to account for candidates who
may have greater resources and are more established in their party and region.
The variable INCUMBENT is coded “1” if the candidate is an incumbent; “0” if the
candidate is not an incumbent.

Controls for popularity are also added the models. Because there are no
adequate measures of district-level popularity for the former Soviet states, I use
proxies. Parties may have limited information about their relative popularity at
the district level, but base some of their placement decisions on demographic
characteristics of the region. For example, agrarian parties (i.e., Agrarian Party
of Russia, Agrarian Party of Ukraine, Lithuanian Peasants’ Party) should expect
to perform well in rural areas and poorly in urban regions. Conversely, parties
that attract working class support should expect to perform well in cities. Liberal
parties are generally expected to appeal to more educated voters and may place
candidates in regions with a more educated population.

The variables URBAN and EDUCATION are added as proxies for
popularity. Both variables are based on data obtained from the 1989 Soviet
census and are aggregated at different levels. For Russia and Ukraine, the data
are at the oblast level. This unit is roughly comparable to a US state. There are
89 units in Russia, most of which are oblasts (provinces). There are also
territorial units called “krais” and “okrugs” which are included as well. The 89
units match the 89 regions in the current Russian system. The Ukrainian oblasts

also match current administrative divisions. There are 27 of these units in total.

84

map of
matche
geogra;
T

region th
of the reg
variables.
regional le

analysis. '

PRPE

Model lb In:

Russia and t

PRPER =(

Lithuanian data are a combination of regional- and city-level data. Using a
map of electoral districts obtained from the Lithuanian Electoral Commission, I
matched current administrative units with Soviet-era counterparts. These
geographic units conform with the 71 electoral districts in Lithuania.

The variable URBAN measures the proportion of the population in the
region that is made up of urban dwellers. EDUCATION assesses the proportion
of the region’s inhabitants that have a higher education. Both are interval level
variables. Because Lithuanian entries in the 1989 Soviet census did not include
regional level data for education, this control is excluded from the Lithuanian

analysis. This constitutes Model 1a:

PRPER = or + BISMDPL + leNCUMBENT + B3URBAN + 8 (Model 13)

Model 1b includes the variable EDUCATION and is evaluated with data from

Russia and Ukraine:

PRPER = or + BISMDPL + fileCUMBENT + B3URBAN + B4EDUCATION + a
(Model 1b)
Model 2: Echo Effect Hypothesis with Controls for Regionalism
Model 2 replicates Model 1’s assessment of the Echo Effect and adds
control variables for regionalism. The main difference in the country is between
the East and West (divided by the Dnieper River). However, the Crimean region

and Central region of the country have different characteristics that justify a

85

division of L
dummy varii
eliminates tl
base region

the model w

PRPEF

If the inclusi
COrltaminatir
and Social C
cleavages

variables. it

MOdeI 3: EC

MOde

SMD provide

~rirhe a hE
53ng a Ob 8y
Viv Ia

R n s

division of Ukraine into four regions. Regions are thus divided into four separate
dummy variables.73 By controlling for regionalism, we can see if it attenuates or
eliminates the effects of contamination. The Central region will be used as the
base region and its dummy will be excluded from the analysis. Consequently,

the model will be the following:

PRPER = or + BISMDPL + BZINCUMBENT + B3URBAN + B4EDUCATION
+ BsCRIMEA + B6EAST + B7WEST + a (Model 2)

If the inclusion of regional dummy variables attenuates the effect of
contamination (the significance of SMD), it will provide evidence that institutional
and social components may interact in countries with substantial regional
cleavages. If the effects of SMD remain significant after controlling for these
variables, it will show that institutional explanations are robust even in the

presence of strong regional cleavages.

Model 3: Echo Effect Hypothesis with Controls for Independents

Model 3 adds controls for covert partisanship. The Ukrainian ballot for
SMD provided voters with the candidate’s name, date of birth, occupation and
party membership. The ballot itself did not include party sponsorship, however.

Central Electoral Commission data from Ukraine include both the candidate’s

 

73 The Central region contains: Chemigiv, Cherkaska, Dnipropitrovsk, Kyiv City, Kyiv Region,
Kirovograd, Kherson, Mikolaevsk, Odessa, and Poltava oblasts. The Crimean region includes
Crimea and Sevastopol. The Eastern region includes Donetsk, Kharkiv, Lugansk, Sumska, and
Zaporizka oblasts. The Western region is made up of Chemivets, lvano-Frankivsk, Khmelnitsk,
Lviv, Rivnensk, TemOpil, Vinnitsa, Volinska, Zakarpatia, and Zhitomir oblasts.

86

party mei
an officia
political p
independi
Thr
partisan.
question 0
Party mem

in Which vc

PRPEj
”‘f

If Cc
paniCtPatiOr
is not, then ‘
Contaminatic

numbers or i.

party membership and the sponsorship. There were numerous cases in which
an officially independent candidate (labeled “bezpartiiniy”) was sponsored by a
political party. Such a candidate would be labeled on the ballot as an
independent.74
The variable COVPART is coded “1” if the district includes only a covert
partisan. In such a case, the voter would not see a candidate from the party in
question on the ballot itself. It is coded “0” if there is no covert partisan or if a

party member also ran in the district. This variable captures only those districts

in which voters find no one sponsored by the given party. The model is thus:

PRPER = or + BISMDPL + leNCUMBENT + B3URBAN + B4EDUCATION
+ BsCRIMEA + BsEAST + B7WEST + BsCOVPART + 8 (Model 3)

If COVPART is statistically significant and positive in sign, then the
participation of covert partisans influences the evaluation of the Echo Effect. If it
is not, then the existence of covert partisans has a negligible effect on
contamination. In the latter case, we can be confident that the existence of large

numbers of independents does not undermine conclusions about contamination.

Conclusions
The three main hypotheses provide a step-by-step approach to

determining the incentive structure of mixed systems and how party systems

 

7" The candidate’s decision to appear as independent on the ballot could be strategic. Through
party sponsorship, the candidate could gain access to resources. However, any negative
implications of association with the party could be avoided.

87

should develop based on these incentives. All three hypotheses provide
evidence for the continuation of multiparty competition in mixed systems.

The Mechanical Effect Hypothesis suggests that marginal parties are likely
to participate in mixed systems because the mechanical effect is more forgiving
to small parties than the mechanical effect in a “pure” SMD system. This
hypothesis will be evaluated in the next chapter.

The Investment Effect Hypothesis addresses the question: given that
marginal parties will participate in the mixed system, how will they choose to
participate? Parties are expected to participate in the system fully — competing
for seats in both the PR and SMD components. A constraint on participation in
SMD, according to the Investment Effect Hypothesis, is the party’s resources.
This hypothesis will be assessed in Chapter 4.

The Echo Effect Hypothesis evaluates a related issue: Are resources the
primary force guiding candidate placement decisions in SMD? The Echo Effect
Hypothesis suggests that parties can benefit in PR from participation in SMD,
encouraging parties to place SMD candidates in as many districts as possible.
The electoral environment, however, provides an additional constraint on

participation in SMD. This hypothesis will be analyzed in Chapter 5.

88

Chapter 3

THE MECHANICAL EFFECT

The mechanical effect (the translation of votes into seats) is one of the
main factors underlying the development of the party system. This effect occurs
in each election and unlike its counterpart, the psychological effect, no time
interval must elapse before its implications are known. The primary determinant
of the mechanical effect is the mechanism for the allocation of seats in
parliament — the electoral rules.

Because the mechanical effect demonstrates whether or not marginally
performing parties can gain a proportional seat allocation in the legislature, it
exerts an important impact on the psychological effect. If marginal parties can
gain representation in the legislature, the sources of support for these parties
are less likely to disappear in subsequent elections. Voters will continue to
support parties that can win, contributors will continue to fund these movements,
and prominent politicians are more likely to choose to affiliate with parties that
can gain office. Thus, the degree to which the mechanical effect punishes
marginal parties is a key to understanding the incentives that mixed systems
provide to political parties.

Parties take cues from the mechanical effect when assessing the
probability of success in an election. If the system punishes marginal parties by
under-representing them in parliament, parties may be encouraged to merge

with stronger organizations or leave electoral competition altogether. However,

89

if small parties are proportionately represented in the legislature, it provides a
signal that the system is conducive to seat acquisition by marginal parties.75
Over time, a system that is punishing to small parties is expected to experience
withdrawal by small parties and an overall reduction in the number of parties. A
relatively forgiving system should encourage participation by small parties.

SMD systems are considered to be more punishing to small parties than
PR systems. Because parties must gain a plurality vote in a district to win a
race, small parties are placed at a substantial disadvantage. A small party that
garners, for example, 10% of the votes in every district, would likely gain no
SMD seats. In a PR system, by contrast, a party receiving 10% of the votes
would receive approximately 10% of the seats. PR systems thus encourage
participation by marginal parties. How does the mechanical effect in mixed
systems affect marginal parties?

To assess the mechanical effect in the mixed system, the system should
be treated as an entity. Rather than calculating the mechanical effect for the PR
and SMD components separately, one should combine the results and
determine the effect of the entire system. One way to measure the mechanical
effect is to use a measure of proportionality.

Various measures can be used to assess a system’s proportionality. l
utilize the Advantage Ratio (A) which divides the number of seats obtained by

the number of votes for each party (Taagepera and Shugart 1989). This

 

75 It is important to note that parties may vary regarding their motivations to compete in elections.
Vlfinning a single seat, rather than a proportional allocation of seats, may satisfy some party
leaders. By contrast, a proportional seat allocation may not suffice for some parties that expect

90

measure is useful for two reasons. It allows us to evaluate the proportionality of
seat distribution for each party; other measures aggregate individual party
values into an overall index for the system. The Advantage Ratio is also used to
derive proportionality profiles, detailed later in the chapter.

If the value of A is near unity, then the party in question receives a
proportion of seats equivalent to its support in the electorate. A number
substantially above or below unity indicates that the party is over-represented or
under-represented in the legislature.

In each of the mixed systems under analysis, the mechanical effect is
less punishing than the mechanical effect in SMD. In Lithuania’s 1996 elections,
many parties received a proportion of seats roughly equivalent to their support in
the electorate (see Table 6).76 The party that received the most votes in 1996,
Homeland Union, is over-represented in Lithuania’s parliament (Seimas).
Although Homeland Union received 30% of the total vote, it received almost
50% of the seats. This is due primarily to the party’s ability to win many seats in
SMD. If results had been based on the first round of the elections alone, the
Homeland Union would have been even more highly over-represented. It won a
plurality in many districts in the first round of voting, but some of its candidates

failed to win the second round.

 

to be over-represented in parliament. Generally, however, parties are assumed to desire an
allocation of seats that corresponds with their support in the electorate.

76 Lithuanian results are somewhat problematic in calculating the mechanical effect. Because
the SMD component uses a Majority-Runoff system, seat allocation does not take place after the
first round. However, because only two parties advance to the second round, the total vote
cannot only be calculated with first round results. This analysis uses seat allocation from the
final round, but vote totals from the first round.

91

Table 6: Mechanical Effect for the Mixed System (Lithuania 1996 — Second

 

 

Round)77
Party % Votes % Seats” A’s
Homeland Union — Lithuanian Conservatives 30.0 49.6 1.66
Lithuanian Social Democratic Party 7.1 8.5 1.20
Lithuanian Center Union 7.7 9.2 1.19
Lithuanian Christian Democratic Party 11.9 11.3 0.96
Lithuanian National Union 3.0 2.8 0.95
Lithuanian Democratic Labor Party 10.6 8.5 0.81
Lithuanian Poles’ Electoral Alliance 3.0 2.1 0.72
Lithuanian Party of the Economy 1.6 0.7 0.43
Lithuanian Union of Political Prisoners and Deportees 1.7 0.7 0.42
Lithuanian Peasants’ Party 2.0 0.7 0.35
Lithuanian Liberal Union 2.3 0.7 0.31
Christian Democratic Union 2.4 0.7 0.30
New Democracy/Women’s Party 3.4 0.7 0.21
Lithuanian National Party “Young Lithuania” 4.4 0.7 0.16
Lithuanian Citizens’ Alliance 2.1 0.0 0.00
Union of Russians of Lithuania 1.3 0.0 0.00
Lithuanian Freedom Union 1.3 0.0 0.00
Lithuanian Freedom League 0.7 0.0 0.00
Lithuanian Union of Social Justice 0.7 0.0 0.00
Lithuanian Socialist Party 0.6 0.0 0.00
Lithuanian Republican Party 0.7 0.0 0.00
Party of National Progress 0.4 0.0 0.00
Lithuanian Party of Life’s Logic 0.3 0.0 0.00
Lithuanian People’s Party 0.2 0.0 0.00
Lithuanian Reform Party 0.0 0.0 0.00
Independent Movement "Elections 96" 0.5 0.0 0.00

 

 

 

77 Source: Electoral Commission of Lithuania. 1999. [Online] http://rc.lrs.lt/rinkimai/, and e—mail
with the Electoral Commission of Lithuania.

78 The seat percentage is based on the distribution that occurred after the second round of the
SMD component. The total percentage of votes received by parties in the system is based,
however, on the first round only in order to preserve the SMD vote totals received by parties that
did not pass through to the second round. Using data from only the first round does not produce
substantially different results.

79 The Advantage Ratio (A) is a measure of proportionality that is calculated by the dividing the
percentage of seats obtained by the percentage of votes obtained (T aagepera and Shugart
1989).

92

The Lithuanian Social Democratic Party (LSDP) and Lithuanian Center Union
(LCU) are also over-represented, but less so. Both parties received about 7% of
the total votes and 9% of the total seats in the legislature. Although slightly
under-represented, the Lithuanian Christian Democratic Party (LCDP) and
Lithuanian National Union (LNU) received a proportionate seat allocation.

The LCDP and LNU are notable because they achieved a proportionate
allocation of seats in different ways. The LCDP performed quite well in both
components of the system, receiving seats in PR and SMD. The LNU did not
pass the PR threshold, however, and only gained seats in SMD. The ability of
parties to gain a proportional allocation of seats in the system, even by failing to
pass the PR threshold, provides incentives for small parties to participate. While
thresholds are used to dissuade small parties from participating, they can gain
seats in parliament despite the threshold, as the LNU demonstrates.80

Results from Russia are similar to those from Lithuania. In 1993, seven
of the thirteen parties participating in the election were over-represented or
slightly under-represented in the legislature (see Table 7). Russia’s Choice, the
Agrarian Party of Russia (APR), Women of Russia and the Communist Party of
the Russian Federation (KPRF) were over-represented in the Duma. However,
the ratio of seats to votes for the Women of Russia and KPRF was almost

perfectly proportional. While they were under-represented, the Liberal

 

8° Rules in parliament regarding the formation of factions may favor winning seats in PR. Parties
that pass the threshold in the PR component can form parliamentary factions that have greater
access to committee assignments and other perks. While mechanisms exist for the formation of
factions by politicians that are not part of a party that passed the PR threshold, faction formation
is more difficult for these legislators. While there are additional incentives to pass the PR
threshold, a party’s ability to obtain a proportionate seat allocation only by winning SMD seats
should encourage small parties to participate in elections.

93

Democratic Party of Russia (LDPR), Yabloko and Dignity and Charity received
close to a proportional seat allocation.

While 1995’s Duma elections produced a larger number of over-
represented parties than the 1993 elections, the mechanical effect continued to
support participation by small political parties (see Table 8). Because only four
political parties passed the 5% threshold in PR in 1995, these parties were
substantially over-represented in the total system. The KPRF and Yabloko were
the main beneficiaries of the distortion. The KPRF received only 19% of the
overall vote but 35% of the seats. Yabloko received 5% of the vote, but 10% of
the seats. Our Home is Russia (NDR), Power to the People, 89 Regions of
Russia and the LDPR were also over-represented.

The tiny 89 Regions of Russia electoral bloc gained a single district seat
and was over-represented in the Duma. Just as in 1993, a small party was able
to gain representation in SMD, resulting in proportional seat acquisition in the
whole system. Two other parties were slightly under-represented. The Party of
Economic Freedom and APR received a proportionate share of seats.

The field narrowed in 1999 and the number of parties receiving a
proportional allocation of seats also declined (see Table 9). Five parties were
over-represented or slightly under-represented. Fatherland-All Russia (OVR)
and the KPRF were over-represented in the Duma, but the general level of over-
representation was not as extreme as in 1995. Medved and the Union of Right
Forces (SPS) were also over-represented. The Zhirinovskiy Bloc (the 1999

version of the LDPR) was slightly under-represented. Other parties were more

94

Table 7: Mechanical Effect for the Mixed System (Russia 1993)"1

 

 

Parties % Votes % Seats A

Russia's Choice 12.7 14.6 1.15
Agrarian Party of Russia 7.5 8.3 1.10
Women of Russia 5.0 5.2 1.04
Communist Party of the Russian Federation 9.0 9.2 1.03
Liberal Democratic Party of Russia 14.7 14.4 0.98
Yabloko 6.5 6.1 0.94
Dignity and Charity 0.7 0.7 0.91
Party of Russian Unity and Accord 5.3 4.7 0.89
Democratic Party of Russia 4.4 3.4 0.77
Civic Union 2.7 1.4 0.50
Russian Movement for Democratic Reforms 3.5 1.1 0.32
Future of Russia-New Names 1.1 0.2 0.20
Kedr 0.8 0.0 0.00

 

 

 

8‘ Source: Central Electoral Commission of the Russian Federation (1996); Rossiiskaya Gazeta
(1 996).

95

Table 8: Mechanical Effect for the Mixed System (Russia 1995)82

 

 

Party % Vote % Seats A

Communist Party of the Russian Federation 18.65 34.89 1.87
Yabloko 5.39 10.00 1 .85
Our Home is Russia 8.32 12.22 1.47
89 Regions of Russia 0.17 0.22 1.32
Liberal-Democratic Party of Russia 8.94 11.33 1.27
Power to the People! 1.77 2.00 1.13
Party of Economic Freedom 0.23 0.22 0.99
Agrarian Party of Russia 4.81 4.44 0.92
Bloc of Independents 0.36 0.22 0.62
Party of Russian Unity and Accord 0.38 0.22 0.59
Democratic Choice of Russia 3.44 2.00 0.58
Transformation of the Fatherland 0.40 0.22 0.56
Bloc of Ivan Rybkin 1.42 0.67 0.47
Common Cause 0.49 0.22 0.46
Forward, Russia! 1.73 0.67 0.39
Pamfilova-Gurov-Lysenko 1 .22 0.44 0.36
My Fatherland 0.66 0.22 0.33
Congress of Russian Communities 3.76 1.11 0.30
Bloc of Stanislav Govorukhin 0.86 0.22 0.26
Women of Russia 3.05 0.67 0.22
Union of Labor 1.28 0.22 0.17
Party of Worker Self-Government 2.47 0.22 0.09
Communists-Workers of Russia-for the USSR 3.40 0.22 0.07
Association of Russian Attorneys 0.26 0.00 0.00
Beer Lovers' Party 0.38 0.00 0.00
Christian-Democratic Union 0.23 0.00 0.00
Derzhava 1 .73 0.00 0.00
Duma-96 0.12 0.00 0.00
Electoral Iloc (various parties) 0.14 0.00 0.00
Electoral Bloc 2 (various parties) 0.28 0.00 0.00
Federal-Democratic Movement 0.09 0.00 0.00
For the Motherland! 0.29 0.00 0.00
Generation of the Frontier 0.06 0.00 0.00
lnterethnic Union 0.03 0.00 0.00
Kedr 0.93 0.00 0.00
National-Republican Party of Russia 0.28 0.00 0.00
People's Union 0.16 0.00 0.00

 

 

 

96

82 Source: Central Electoral Commission of the Russian Federation (1996).

Table 8 (cont’d)

 

Russian All-National Movement
Russian Muslim Movement "Nur"

Social Democrats

Stable Russia
Tikhonov-Tupolev-Tikhonov

Union of Housing Construction Workers

 

0.16
0.33
0.25
0.18
0.09
0.12

0.00
0.00
0.00
0.00
0.00
0.00

0.00
0.00
0.00
0.00
0.00
0.00

 

97

Table 9: Mechanical Effect for the Mixed System (Russia 1999)"3

 

 

Party % Votes % Seats A

Fatherland—All Russia 1 1.9 15.2 1.28
Communist Party of the Russian Federation 20.9 25.7 1.23
Medved 14.1 16.4 1.16
Union of Right Forces 6.4 6.6 1.03
Zhirinovsky Bloc (LDPR) 4.1 3.9 0.94
Yabloko 6.0 4.5 0.75
Our Home is Russia 2.2 1.6 0.73
Movement in Support of the Army 0.7 0.5 0.64
Russian National Union 0.8 0.5 0.57
Spiritual Heritage 0.6 0.2 0.41
Russian Socialist Party 0.7 0.2 0.34
Congress of Russian Communities 0.7 0.2 0.32
Bloc of Andrei Nikolayev and Svyatoslav Fyodorov 0.9 0.2 0.26
Party of Pensioners 1.5 0.2 0.15
All-Russian Party of the People 0.1 0.0 0.00
Communists—Workers of Russia—for the USSR 1.5 0.0 0.00
Conservative Party of Russia 0.1 0.0 0.00
For Citizens’ Dignity 0.5 0.0 0.00
Party of Peace and Unity 0.2 0.0 0.00
Peace, Labor, May 0.4 0.0 0.00
Russian Cause 0.2 0.0 0.00
Russian Party for the Defense of Women 0.4 0.0 0.00
Social Democrats 0.1 0.0 0.00
Socialist Party of Russia 0.1 0.0 0.00
Stalinist Bloc-for the Soviet Union 0.4 0.0 0.00
Women of Russia 1.4 0.0 0.00

 

 

 

83 Source: Rossiiskaya Gazeta, January 6, 2000 and [Online] http://www.rg.ru

98

substantially under-represented, but received seats in parliament. While
Yabloko’s total vote actually increased from 1995 to 1999, it received a less
proportional distribution of seats. NDR’s performance declined precipitously and
its seat acquisition decreased along with its loss of support in the electorate.
Although fewer parties were proportionally represented in the 1999 Duma than
in earlier elections, the mechanical effect remained conducive to the
participation of marginal parties.

As in Lithuania and Russia, Ukraine’s mechanical effect in 1998 did not
punish smaller parties to the benefit of large parties (see Table 10).84 The
People’s Democratic Party (NDP), Communist Party of Ukraine (KPU), and
Greens were over-represented in the Rada. Rukh and Gromada were also
over-represented, but less so. Menshe Sliv, the Progressive Socialist Party
(PSP) and the Social Democratic Party (United) (SDPU) were under-
represented, but only slightly. Nine parties were over-represented or slightly
under-represented. In sum, this evidence suggests that the mechanical effect
does not provide disincentives for small parties to participate in elections in a

mixed system.

 

8‘ In order to calculate the mechanical effect in Ukraine, parties that were part of an electoral
bloc in the PR component were combined with that party. Since a list of all of the parties in a
bloc appeared on the PR ballot, the combination of all members into a single party is logical.
Thus, the Socialist and Peasant Bloc includes the Socialist Party and Peasant Party. Working
Ukraine includes the Ukrainian Party of Justice and the Citizens’ Congress of Ukraine. National
Front includes the Ukrainian Republican Party, Ukrainian Conservative Republican Party and
Congress of Ukrainian Nationalists. RAZOM includes the Liberal Party and the Party of Rights.
Fonivard Ukraine includes the Ukrainian Christian Democratic Party and the Christian
Democratic Union. The Bloc of Democratic Parties includes the Democratic Party of Ukraine
and the Party of Economic Development. SLON includes the Inter-Regional Bloc of Reform and
the Constitutional Democratic Party. Menshe Sliv includes the Social National Party and

99

Table 10: Mechanical Effect for the Mixed System (Ukraine 1998)85

 

 

Party % Votes % Seats A
People's Democratic Party 4.9 6.3 1.29
Communist Party of Ukraine 20.3 26.1 1.29
Greens 3.4 4.2 1.23
Rukh 8.3 9.6 1.15
Gromada 4.4 4.5 1 .02
Menshe Sliv 0.2 0.2 0.96
Progressive Socialist Party 2.6 2.5 0.95
Social Democratic Party (United) 3.1 2.9 0.92
Socialists and Agrarian Bloc 7.0 6.3 0.90
Party of Regional Development 0.7 0.4 0.61
Agrarian Party of Ukraine 3.7 1.8 0.48
Soyuz 0.5 0.2 0.47
National Front 2.9 1 .1 0.39
Christian-Democratic Party 1.2 0.4 0.38
SLON 0.7 0.2 0.30
Reform and Order 2.6 0.7 0.26
Forward, Ukraine 1.2 0.4 0.18
Block of Democratic Parties 1.3 0.2 0.17
RAZOM 1.7 0.2 0.13
Working Ukraine 2.0 0.2 0.11
All Ukrainian Party of Women's Initiatives 0.4 0.0 0.00
All Ukrainian Workers' Party 0.5 0.0 0.00
European Choice of Ukraine 0.2 0.0 0.00
Party of Defenders of the Homeland 0.2 0.0 0.00
Party of Muslims 0.1 0.0 0.00
Party of National Economic Development 0.5 0.0 0.00
Party of Spiritual, Economic and Social 0.2 0.0 0.00
Progress

Republican Christian Party 0.5 0.0 0.00
Social Democratic Party 0.3 0.0 0.00
Ukrainian National Assembly 0.4 0.0 0.00

 

 

 

Government Sovereignty of Ukraine. European Choice of Ukraine includes the Liberal
Democratic Party of Ukraine and the Ukrainian Peasants’ Democratic Party.
85 Source: Central Electoral Commission of Ukraine (1998, 236-238).

100

It is important to note that a party receiving a proportional seat allocation
may not choose to participate in subsequent elections. For example, Dignity
and Charity received a small number of votes, but also a seat, in the Russian
Duma that was elected in 1993. However, it disappeared from the political
landscape after 1993. Even favorable institutional rules do not prevent financial
difficulties, personnel problems or political infighting from dooming parties that

perform well in elections.

Proportionality Profiles

Comparing the proportionality of seat allocation across parties is
instructive, but it does not allow us to measure the mechanical effect of the
mixed system against the SMD system. Because SMD with plurality rules is
less forgiving to small parties than PR, it is important to assess whether or not
the mixed system, which contains an SMD component, is more punishing to
marginal political parties. Proportionality profiles help make these direct
comparisons.

By juxtaposing the Advantage Ratio and the percentage of votes received
by parties on a figure, we can compare the mechanical effect in mixed systems
and SMD systems. While proportionality profiles were first outlined by Laakso
and Taagepera (1980), Taagepera and Shugart (1989) describe standard
profiles for PR and SMD systems based on cross-national empirical evidence.

In a proportionality profile, aggregate data from elections are arrayed on

a figure in which the Y axis represents proportionality (the Advantage Ratio) and

101

the X axis represents the proportion of votes received. The perfect
proportionality line is a horizontal line that intersects the Y axis at 1.00 (the
proportion of seats and the proportion of votes are equal). Every proportionality
profile has a break-even point, the point at which the average profile curve
intersects the perfect proportionality line. On average, parties accruing a
proportion of votes corresponding with the break-even point receive a
proportionate allocation of seats in the legislature.

Taagepera and Shugart (1989, 70) indicate that SMD systems typically
have a Very Late Rise profile in which the proportionality profile gradually rises
to a break-even point at approximately 40% on the X axis. The data continue in
an upward pattern and plateau at approximately 1.4 on the Y axis. In this
pattern, parties that receive a relatively small proportion of the vote tend to be
under-represented; parties that receive a moderate or large proportion of the
vote tend to be over-represented.86

The underlying question is thus: what is the best fitting functional form for
the data produced by mixed electoral systems? If the functional form of mixed
systems diverges from the functional form of SMD in a way that is more forgiving
to marginal parties, we can conclude that the mechanical effect of mixed
systems will encourage small parties to participate in elections.

Scatterplot smoothers are well suited to evaluate this hypothesis. With a

scatterplot smoother, such as loess, the precise form of the data is not assumed

 

‘36 It should be noted that comparing actual results to model results from Taagepera and Shugart
does not force the systems under analysis to conform to an unrealistic ideal. Rather than
representing an ideal, the proportionality profiles are derived from empirical evidence from many
countries.

102

to correspond to any particular shape. Rather, the data are used to generate
the best-fitting functional form. The test is non-parametric and allows the
researcher to conduct bivariate or multivariate analysis.

With loess, the smoother calculates the best-fitting line for a portion of the
data. This “windovV’ moves along the X-axis and the calculation of the best-
fitting line is repeated until the entire form is generated. The researcher can
determine the preferred size of this “window,” called the bandwidth (1). There is
no preferred bandwidth, although the fit becomes smoother as the bandwidth
increases. While there are many different types of scatterplot smoothers, loess
is widely employed because it combines some of the best features of smoothers
(Hastie and Tibshirani 1990). I applied loess to each of the countries under
analysis to generate proportionality profiles using the default bandwidth for
individual countries and a 0.75 bandwidth for the combined sample.87

The shape of Lithuania’s proportionality profile, based on loess, diverges
from the Very Late Rise profile, typical of SMD systems (see Figure 2).88 There
is a single outlier in these results, Homeland Union. This party performed
extremely well in the first round of the SMD component, garnering 30% of the
total vote. The remaining parties received less than 15% of the vote. The break-
even point is substantially lower than SMD because parties receiving a modest

proportion of the vote were proportionately represented in the Seimas. Based

 

87 Independent candidates were omitted from the analysis.

88 The Very Late Rise profile, associated with SMD systems, has been placed on each of the
figures and is labeled “Standard SMD.” This profile is an asymptotic result based on cross-
national and longitudinal data whereas individual profiles are from single countries. Thus, the
comparison of the standard SMD profile to mixed systems constitutes an evaluation of the
hypothesis rather than a formal test.

103

% Seats/% Votes

 

3.0

2.5 ‘

N
C
1

1.5—

.5
o
l

0.5 ‘

0.0 ‘

 

Lithuania

Standard SMD

 

 

—r

0

Figure 2: Proportionality Profile for Lithuania

104

60

on Lithuanian data, it appears that marginal parties can receive a proportional

allocation of seats in a mixed system. This should encourage small parties to

continue to participate in elections.
The Lithuanian results must be interpreted with care, however. The SMD

component of Lithuania’s mixed system is based on a majority-runoff system
not a plurality system like other mixed systems in the analysis. Although final
seat distributions were used to calculate disproportionality, votes are based only
on the first round.89

Russia’s mixed system proportionality profiles also diverge from the
standard SMD profile. Based on the loess analysis, data from the 1993 Russran
Duma elections appear to more closely approximate the pr0portionality profile of
PR systems rather than SMD systems (see Figure 3). The proportionality profile
rises quickly and then follows the perfect proportionality line as values increase
along the X axis. Like the Flat or Early Rise profiles (Taagepera and Shugart
1 989 70) that are typically associated with PR, the proportionality prof le for
Russia 3 1993 elections shows that parties receiving a relatively small pr0portion

of the vote are proportionally represented in the Duma. These results diverge

from SMD and promote multiparty competition.
Russia’s proportionality profile for 1995 also diverges from the Very Late

Rise profile (see Figure 4). Unlike the 1993 profile that followed the perfect

proportionality line, the profile in 1995 is close to a straight line. These results

 

\
L"il‘luania’ 5 use of a majority-runoff ballot for SMD complicates the proportionality profile

89
paalcuiation. Many seats were not allocated until the second round, but most parties did not
rt"lipate in runoff elections. Thus, I use vote outcomes from the first round and seat allocation

105

are due in large part to the disproportionate results in PR. Because only four
parties passed the threshold in the PR component and these parties divided all
225 seats among themselves, they were over-represented in the Duma.
Although there are differences between Russia’s 1993 and 1995 profiles, neither
approximates the SMD profile. Parties receiving a relatively small proportion of
the vote are able to gain a proportionate seat allocation.
The 1999 proportionality profile has an unusual shape but is similar to the
1 993 profile (see Figure 5). The profile has a large trough between 0 and 10 on
the X axis. This is due to the loess default bandwidth and distribution of the
data points. Despite the unusual shape, the break-even point for the profile is
far lower than the break-even point for the SMD system. Moreover, the profile
fol lows the perfect proportionality line as the values increase along the X axis.
Once again, the profile is shaped more like a PR proportionality profile than
S M D.
Ukraine’s proportionality profile is similar to the shape of the 1993 and
1 999 Russian profiles (see Figure 6). The break-even point is lower than in
SMD and many parties garnering a small proportion of the votes received a
pr Oportionate distribution of legislative seats. The proportionality profile for
Ukraine diverges from Russia in that it plateaus at a higher value along the Y

axis- Nevertheless, it is more forgiving to small parties than the standard SMD

VeW Late Rise profile.

 

\

fr
thoe' ' | the second round. In this way, I capture the votes received by all parties in the election and
aCtual seat distribution.

106

% Seats/% Votes

 

3.0

2.5 "

.N
O
1

1.5“

A
C
l

.0
on
r

0.0 ‘

 

 

 

 

 

Standard SMD
Russia 93
' ' o
O
O
T T l l l l
0 1 O 20 30 4O 50 60

% Votes

Figure 3: Proportionality Profile for Russia (1993)

107

% Seats/% Votes

3.0

2.5

N
o

1.5

.A
O

0.5

0.0

 

Russia 95

0
Standard SMD

 

 

 

 

 

1 0 20 30 40 50
% Votes

Figure 4: Proportionality Profile for Russia (1995)

108

60

%Seats/% Votes

 

3.0

 

 

 

 

2.5 T
2.0 T
Standard SMD
1.5 -< Russia 99
/\:y/0
1.0 T o
o O

O

O
0.5 T
0.0 T

I l I I T T

0 10 20 30 40 50 60

% Votes

Figure 5: Proportionality Profile for Russia (1999)

109

% Seats/%Votes

 

3.0

2.5 T

2.0 T

1.5T

1.0T

0.5 T

0.0 T

Ukraine

Standard SMD

 

 

 

 

 

20 30 40
% Votes

50

Figure 6: Proportionality Profile for Ukraine (1998)

110

60

All of the mixed systems analyzed here seem to differ from the Very Late
Rise profile identified by Taagepera and Shugart. Pooling the data from all five
elections provides a picture of the mechanical effect in mixed systems. Before
combining the data, I conducted a modified Chow Test for structural stability.
The test assesses if the intercept or slope of the individual cases differ
substantially from one another. If there is no structural change, the samples can

be combined (Gujarati 1995). The test is:

 

S

/2 n,- - 2k
where S1 is the Residual Sum of Squares (RSS) of the pooled data less the
RSS of the cross sections summed, 82 is the RSS of the cross sections
summed, k is the number of regressors, and in; is the number of cross sections.
The null hypothesis is that the cross sections can be combined (there is
structural stability). The calculated F statistic is 2.7; the critical F“ ,3) at 95%
confidence is 10.1. The null hypothesis cannot be rejected and the cross
sections can be pooled.

Figure 7 provides a proportionality profile for all of the previous elections
and compares it to the standard SMD profile. Like the individual proportionality
profiles, the combined profile has a break-even point that is lower than that of a
standard SMD system. There is a sizable cluster of slightly over-represented
parties and slightly under-represented parties between 0 and 20 on the X axis.

This suggests that parties receiving low to moderate support in the electorate

111

% Seats/% Votes

 

 

 

 

 

 

3.0
2.5 ‘
2 0 _ Lithuania, Russia and Ukraine
Standard SMD

1.5 -
1.0 ~
0.5 ~
0.0 ‘

T l I l l l

o 10 20 so 40 so so

% Votes

Figure 7: Proportionality Profile for Lithuania, Russia and Ukraine

112

% Seats/%Votes

 

 

 

 

 

 

 

N‘ “\I’

m I III II IIIII IIIIIHIIIHHHIIW iI sand,” SMD
"' ‘ I” I I 'II III III II' II I II III II III II. 'II * /\
o I" ‘ w

‘_' Tl ll

% Votes

Figure 8: Proportionality Profile for Lithuania, Russia and Ukraine with
95% Confidence Intervals

113

can indeed receive a proportionate seat allocation in mixed electoral systems.
Whereas the mechanical effect in SMD systems punishes small parties by
favoring parties that receive 40% or more of the vote, the mechanical effect in
mixed systems facilitates representation by marginal parties. This should
encourage small parties to participate in future elections. Figure 8 applies 95%
confidence intervals to the combined proportionality profile and shows that the
profile diverges from the Very Late Rise profile with a high level of confidence.
Proportionality profiles from other mixed systems further support the
notion that the mechanical effect in mixed electoral systems is more forgiving
than the mechanical effect in SMD systems. Results from Japan, New Zealand,
Scotland and Wales are quite similar to those found in Lithuania, Russia and
Ukraine (see Appendix B). Not only do these states have more institutionalized
party systems, but three of them use compensatory rules that link PR and SMD
seat acquisition. This suggests that the more forgiving mechanical effect in

mixed systems is not attenuated by institutional variation within mixed systems.

Conclusions

In sum, there is strong evidence supporting the Mechanical Effect
Hypothesis. Many parties in Lithuania, Russia and Ukraine were over-
represented or slightly under-represented in the elections under analysis. While
the mechanical effect “filtered out” some small parties, many marginal parties

gained proportionate representation in the legislature.

114

The proportionality profiles for each election further illustrated that the
mechanical effect encourages small parties to participate in elections. When
they were compared to the standard mechanical effect in SMD systems, the
break-even points were lower and the shape of the profile differed. The profiles
for the mixed systems were similar to those associated with PR, moreover,
suggesting that the incentives produced by the system may be similar to those
generated in PR.

The effect on party strategy should be clear. Although the mixed system
contains an element of SMD, this component, in concert with PR, does not
create the same incentive structure as a standard SMD system. If small parties
are not excluded from legislative representation due to the mechanical effect, it
is less likely that they will drop from competition. As voters recognize that small
parties can win seats in parliament, they are more likely to vote sincerely. As
contributors recognize that small parties can play an active role in government,
they are less likely to pull away resources. As candidates see that small parties
are viable engines for access to legislative positions, they are less likely to
abandon them for larger parties. In short, the less punishing mechanical effect

creates an environment that encourages multipartism.

115

Chapter 4

THE INVESTMENT EFFECT

The evaluation of the mechanical effect demonstrated that small parties
are likely to participate in mixed electoral systems because the mechanical
effect is more forgiving than SMD’s mechanical effect. The general tendency
should be for many parties to commit to competition in elections under mixed
rules. The next question is: how will parties choose to participate in the system?

Theoretically, a party could sponsor candidates in SMD but not register in
PR. However, parties that do not register in PR are generally accorded different
access to campaign finance and media access than those on the PR ballot.

In Lithuania’s 1996 Seimas elections, a registered party was allowed to
spend twenty times as much as an individual candidate on election activities
(Election Law 2000).90 Because the party’s spending limits were for the national
list, a candidate could benefit from the party’s national campaign and still mount
an individual campaign in the district, subject to the restrictions for candidates.
An unregistered party would not be permitted to spend the additional funds
restricted to party lists.

Registered parties also received additional benefits in advertising. While
all candidates and parties were allocated free advertising, parties received one

and one-half hours of national television and radio time. Individual candidates

 

9° Law on Elections to the Seimas, Section VII, Article 54.

116

were aiio
Once age
party’s ad
ln l
registered
both on th
organizatii
Candidate:
Individuals
COUId rece
individuals
Could also
IImes the r
times the r
adVemSms
OnI
RUSSIa'S 1
equal mm

addmofiaj

were allotted five minutes of television time and seven minutes of radio time.91
Once again, candidates affiliated with registered parties could benefit from the
party’s additional free air time.

In Russia’s 1999 elections, campaign finance regulations also favored
registered political parties over unregistered parties. Limitations were placed
both on the amount of money that could be received from individuals and
organizations as well as a maximum amount that could be spent.92 Individual
candidates could receive donations 100 times the minimum wage from
individuals and 2,000 times the minimum wage from organizations. Parties
could receive substantially larger donations — 150 times the minimum wage from
individuals and 20,000 times the minimum wage from organizations. Parties
could also spend far more than candidates. The limit for parties was 250,000
times the minimum wage while candidates could spend no more than 10,000
times the minimum wage (IF ES 1999).93 Affiliated candidates benefit from party
advertising and from both national and individual fundraising.

Only registered parties received free time on national stations94 in
Russia’s 1999 elections. Television and radio stations were required to provide
equal time to competing candidates/parties. Regional party lists were allocated
additional local air time. The amount of time given to SMD candidates was

multiplied by the number of districts in the region to determine how much free

 

9’ Law on Elections to the Seimas, Section VII, Article 51.
92 While limitations on the collection and expenditure of funds existed in practice in the 1999
campaign, conversations with two candidates (who requested anonymity) suggest that these
regulations were widely violated.
93 Law on the Election of Deputies of the State Duma of the Federal Assembly of the Russian
Federation, Chapter IX, Article 62.

These are networks with a clear signal to more than half of the residents of Russia.

117

time region
with a part)
PR Iist’s i0
unregisteri
process f0
television
WI
in Ukraine
Specify lir
of free teI
Candidati
SeDarate
tI’Iéiri ind.
parties 2
1999);?

advahte

time regional party lists received?"5 Thus, a candidate in an SMD race affiliated
with a party could benefit from the party’s national advertisements, the regional
PR list’s local advertisements, and his own time. Candidates affiliated with
unregistered parties could only use the time allocated to candidates. The
process for allocating free space in the print media was similar to radio and
television (IFES 1999).96

While the law is ambiguous about campaign finance and advertisements
in Ukraine, it seems to provide benefits to registered parties. The law does not
specify limitations on campaign finance, nor does it specify the minimum amount
of free television or radio advertising. The law, however, separates parties and
candidates when it outlines campaign finance, implying that they are treated
separately. Registered parties are allocated more free space in the print media
than individual candidates to publish their platforms, suggesting that registered
parties are likely to receive benefits not accorded to unregistered parties (NAU
1999).97 Thus, it is logical for political parties to register in PR in order to take
advantage of the additional benefits accorded to parties that participate in PR.

Just as a party could sponsor candidates in SMD alone, it could also
participate solely in PR. If parties aim to maximize the number of legislative
seats in parliament, this approach is not logical. Parties must make a
substantial investment of resources to mount a campaign in the PR component

of the elections. By committing to a national PR race, the party leadership is

 

95 Law on the Election of Deputies of the State Duma of the Federal Assembly of the Russian
Federation, Chapter VIII, Article 56.
96 Law on the Election of Deputies of the State Duma of the Federal Assembly of the Russian
Federation, Chapter VIII, Article 57.

118

presumat
fill its par
commits t
mSMDs
candidate
not be prI
Party imp
Candidate

Tr
Participa
is so! pa!
faces, Ir

inlly in S

PaniClp;

presumably confident that it has an adequate number of qualified politicians to
fill its party list and enough resources to advertise. Once the organization
commits to the national PR race, the marginal cost of running some candidates
in SMD should be low. Parties may not have adequate resources to sponsor
candidates in all SMD races, but the cost of participating in some races should
not be prohibitive. Furthermore, by taking part in SMD races as well as PR, the
party improves the odds of gaining legislative seats simply by fielding
candidates in more races.

The Investment Effect indicates that the decision of parties to
participate in PR will encourage some level of participation in SMD. If this
is so, parties participating in PR should place multiple candidates in SMD
races. In addition, parties with greater resources should participate more

fully in SMD than those with fewer resources.

Participation in PR and SMD

Table 11 shows the level of SMD participation for parties that registered
in PR during Russia’s 1993, 1995 and 1999 elections.” All thirteen parties in
1993 placed candidates in SMD. The Women of Russia party placed the fewest
SMD candidates (6) and Russia’s Choice placed the most (110). The pattern is

the same in 1995. All parties participating in PR placed at least two SMD

 

97 Law on the Election of Deputies to the Verkhovna Rada, Chapter 5, Article 34.
98 There were 225 districts in all three elections.

119

Table 11

PartI

'iWss
Russ
Agra
Civic
YablI
Pan)
Dem
Russ
Liber
ComI
Futur
Kedr
Digni
Worr
Our t
Cong
Forw
COmI
BlOc
POwe
UDlOI
Pamt
SOCIE
Pan)
Bldg
09':
My F
RUSS
BIQC
Stab
A830
Pan)
P80:
Tran:

D .
%

SOUrQe.
BJOSSIIska' Ger

‘1' mOCr-at.
The LDPRIC

Table 11: SMD Participation by Parties Registered in PR (Russia 1993,

 

 

1995 and 1999)
Party # of SMD
Candidates
1993 1995 1999
Russia’s Choice/Democratic Choice of 110 72 -
Russia100
Agrarian Party of Russia 74 87 -
Civic Union 73 - -
Yabloko 69 69 116
Party of Russian Unity and Accord 69 23 -
Democratic Party of Russia 62 - -
Russian Movement for Democratic Reforms 60 - -
Liberal-Democratic Party of Russia101 58 184 92
Communist Party of the Russian Federation 57 130 130
Future of Russia-New Names 36 - -
Kedr 23 19 -
Dignity and Charity 11 - -
Women of Russia 6 20 12
Our Home is Russia - 103 92
Congress of Russian Communities - 90 45
Forward, Russia! - 67 -
Communists-Workers of Russia-for the USSR - 64 18
Bloc of Ivan Rybkin - 64 -
Power to the People! - 41 -
Union of Labor - 40 -
Pamfilova-Gurov-Lysenko - 33 -
Social Democrats - 29
Party of Worker Self-Government - 27 -
Bloc of Stanislav Govorukhin - 25 -
Derzhava - 25 -
My Fatherland - 18 -
Russian All-National Movement - 18 -
Bloc of Independents - 17 -
Stable Russia - 17 -
Association of Russian Attorneys - 15 -
Party of Economic Freedom - 15 -
People's Union - 15 -
Transformation of the Fatherland - 15 -
Duma-96 - 14 -

 

 

 

99 Source: Central Electoral Commission of the Russian Federation (1995; 1996; 2000);
Rossiiskaya Gazeta (1996; 2000).

10° Democratic Choice of Russia (DVR) was the successor party to Russia’s Choice in 1995.
‘0’ The LDPR ran as the Zhirinovskiy Bloc in 1999 due to problems with registration.

120

Table 11 (cont’d)

 

Christian-Democratic Union

Beer Lovers' Party

Federal-Democratic Movement

For the Motherland!

lnterethnic Union

Union of Housing Construction Workers
Generation of the Frontier
Tikhonov-Tupolev-Tikhonov

Bloc of Dzhuna

National-Republican Party of Russia
Spiritual Heritage

Fatherland - All Russia

Bloc of Andrei Nikolayev and Svyatoslav
Fyodorov

Union of Right Forces

Russian Socialist Party

Medved

Party of Pensioners

Russian National Union

Movement in Support of the Army

For Citizens’ Dignity

Russian Cause

Stalinist Bloc - for the Soviet Union
Socialist Party of Russia

Party of Peace and Unity

All-Russian Party of the People
Conservative Party of Russia

Russian Party for the Defense of Women

 

13
11
11

imcootcnouoj

JAN
coo-souxtowo

 

121

Table 12 :

Party

FIE
Rept
Lithu
Lithu
Hom
Lithu
Lithu
Litht
New
Litht
Litht
Litht
Dep
Litht
Lithi
Lith
Chr
Lith
The
Lit!"
Litt
Litt
Th
Liti
Lit
\

//7

Table 12: SMD Participation by Parties Registered in PR (Lithuania 1996)102

 

 

Party # of SMD
Candidates

Lithuanian Social Democratic Party 71
Republican Party 68
Lithuanian Democratic Labor Party 67
Lithuanian Christian Democratic Party 67
Homeland Union - Lithuanian Conservatives 67
Lithuanian Centre Union 53
Lithuanian National Union 52
Lithuanian Party of Economy 45
New DemocracyNVomen's Party 44
Lithuanian Liberal Union 35
Lithuanian Peasants' Party 34
Lithuanian Union of Political Prisoners and 33
Deportees

Lithuanian Poles’ Electoral Action 22
Lithuanian Party of Life's Logic 22
Lithuanian Socialist Party 20
Christian Democratic Union 17
Lithuanian National Party "Young Lithuania" 16
The Party of National Progress 15
Lithuanian Union of Social Justice 11
Lithuanian People's Party 11
Lithuanian Freedom League 11
The Union of the Russians of Lithuania 10
Lithuanian Citizens' Alliance 10
Lithuanian Freedom Union 9

 

 

 

‘02 Source: Electoral Commission of Lithuania (1999).

122

Table 13:

Party

 

Rukr
Gron
Com-
Natic
Soci:
Soci;
RAZ
Gree
PeOI
Bloc
Agra
Refc
Prot
Chri
EUl’l
For

Table 13: SMD Participation by Parties Registered in PR (Ukraine 1998)103

 

 

 

Party # of SMD
Candidates’o"
Rukh 186
Gromada 181
Communist Party of Ukraine 174
National Front 164
Socialist and Agrarian Bloc 129
Social Democratic Party (United) 128
RAZOM 1 16
Greens 1 15
People’s Democratic Party 108
Bloc of Democratic Parties 103
Agrarian Party of Ukraine 85
Reform and Order 83
Progressive Socialist Party 74
Christian Democratic Party 66
European Choice of Ukraine 59
Forward, Ukraine 53
Ukrainian National Assembly 52
Republican Christian Party 49
Working Ukraine 48
Social-Democratic Party 42
Party of Regional Development 42
All Ukrainian Workers’ Party 40
SLON 37
Party of the Defenders of the Homeland 29
Menshe Sliv 23
Party of Spiritual, Economic and Social 18
Progress
Soyuz 12
Party of Muslims 11
Party of National Economic Development 10
All Ukrainian Party of Women’s Initiatives 5

 

 

‘03 Source: Central Electoral Commission of Ukraine (1998).

1°“ The measure includes candidates from related parties that were a part of the electoral bloc
and those who were sponsored by the party. The variable SMDPL is calculated in a slightly
different manner, reflecting the peculiarities of the Ukrainian ballot. Therefore, the number of
SMD candidates on this table and in the later analysis differ slightly. I include party sponsorship,
rather than membership, because the question I evaluate here is how parties nominate
candidates in SMD. Because they control sponsorship and indirectly control how members will
behave, sponsorship is the better measure here. In other analyses, however, membership is
more important because it is the factor that appears on the ballot.

123

candida
candida
(184). (
placed t

T
1996 ca
thuani
Party‘s '.
Women
Rukh s;
Particip;

F
typical“
mixed 5
SMD, as
Daniel;
8830C“

perpet,

The E

I

a /
l l
’l

(

Tm

hi

candidates. The National-Republican Party fielded the lowest number of SMD
candidates (2) and the Liberal Democratic Party of Russia placed the most
(184). Of the 26 parties in 1999, three fielded no SMD candidates; the KPRF
placed the most SMD candidates (130).

The pattern is repeated in other states. All political parties in Lithuania’s
1996 campaign sponsored multiple SMD candidates, ranging from the
Lithuanian Freedom League’s 9 candidates to the Lithuanian Social Democratic
Party’s 71 candidates (see Table 12).105 In Ukraine, the All-Ukrainian Party of
Women’s Initiatives placed the fewest SMD candidates (5) (see Table 13).
Rukh sponsored the largest SMD contingent (186).106 Once again, all parties
participating in PR also ran candidates in SMD.

Parties participating in the PR component of mixed systems have
typically fielded multiple SMD candidates. Because the mechanical effect of
mixed systems is not as punishing to small parties as the mechanical effect in
SMD, as established in Chapter 3, marginal parties should continue to
participate in elections. Committing to competition in PR seems to be
associated with placing multiple SMD candidates, a tendency that should

perpetuate multipartism.

The Effect of Resources on Participation
While the empirical pattern of participation in both components of the

mixed system is clear, what factors lead some parties to participate in just a few

 

105 There were 71 SMD districts in Lithuania's 1996 elections.
‘06 There were 225 SMD districts in Ukraine’s 1998 elections.

124

districts
suggest
SMD rac
assertIOI
election:
T
has plac
measure
P€fsonn
Paris/’8 t
eqUIpDE
grassro
ShOuld
PREVE
Parties

SITU ctu

districts and others to run many candidates in SMD? The Investment Effect
suggests that the ability to overcome the marginal costs of placing candidates in
SMD races should, in part, influence how fully a party participates in SMD. This
assertion can be evaluated with data from the most recent parliamentary
elections in Lithuania, Russia and Ukraine.

The dependent variable is the proportion of SMD districts in which a party
has placed candidates.‘°7 Three independent variables represent different
measures of a party’s resources. The variable PRLIST assesses the party’s
personnel resources by measuring the proportion of PR seats covered by the
party’s PR list. A party with greater personnel resources should be better
equipped to overcome the costs of SMD participation. PRVOTE measures
grassroots support for the party. A party with greater support in the electorate
should receive more financial support from the population. The dummy variable
PREVELECT represents parties that participated in the previous election.
Parties with a longer electoral history presumably have a stronger organizational
structure and a greater resource base.

The results of the OLS analysis in Table 14 show that a party’s ability to
overcome the costs of SMD participation should also affect its level of
participation in the SMD component of the mixed system (see the column
labeled Model 1). The coefficients for PRLIST and PREVELECT are statistically
significant and positive in sign. This indicates that parties with greater personnel

resources are likely to field a larger contingent of SMD candidates than those

 

‘07 For Ukraine, this is the proportion of districts in which candidates were sponsored by the
party, regardless of a candidate’s party membership.

125

Table 14: Effect of Resources on SMD Placement

 

 

Model 1 Model 2 Model 3
Variable ,6 and SE g and SE ,6 and SE
Constant -4.434 -62.016*** -.619
(4.013) (13.598) (6.542)
,6, PR List 512*“ 374*
(067) (198)
,6; PR Vote .453 1.048** .368
(.407) (.413) (.424)
,6; Previous Election 8.783“ 9.246“ 8.509”
(4.404) (4.818) (4.434)
,64 Logged PR List 22.147***
(3.603)
,6; Squared PR List .001
(.001)
Adjusted R2 .641 .571 .639
N 72 72 72

 

 

Dependent Variable: Proportion of SMD races in which candidates were placed.
Standard errors are in parentheses.

Note: All parties participating in the most recent elections in Lithuania, Russia
and Ukraine were included in the analysis. Some parties in Lithuania did not
submit PR lists and were excluded.

* ps.10
** ps.05
*** ps.01

126

with few
shows tl
candida
affect S!
is not sta
existing
N
PRLIST
SMD mg
a loQgec
T
(see Ta.
Ihe .05 I
HOWSVE
Indicate
mOre Ct
l
and PF
MOreQ.
Table
by ad;

reIaiic

with fewer personnel resources. The statistical significance of PREVELECT
shows that parties that are more established in society nominate more
candidates in SMD. The resources associated with stronger party organization
affect SMD placement strategies. The coefficient evaluating grassroots support
is not statistically significant, indicating that this is less important than the party’s
existing resources when determining how fully the party will participate in SMD.

Models 2 and 3 evaluate a potential non-linear relationship between
PRLIST and SMDPER. As party personnel resources increase, placement in
SMD may not increase at the same rate. To control for this possibility, I include
a logged version of PRLIST in Model 2 and squared PRLIST in Model 3.

The natural log of PRLIST is statistically significant and positive in sign
(see Table 14, Model 2). In addition, the coefficient for PRVOTE is significant at
the .05 level and PREVELECT is significant at the .10 level in Model 2.
However, based on adjusted R2, Model 2’s overall fit is lower than Model 1. This
indicates that despite the significance of logged PRLIST, Model 1 presents a
more complete picture of the effect of resources on candidate placement.

While the inclusion of squared PRLIST mitigates the effects of PRLIST
and PREVELECT in Model 3, the coefficients remain significant at the .10 level.
Moreover, the parameter of squared PRLIST is not statistically significant (see
Table 14, Model 3). The fit of Model 3 is also lower than Model 1, as measured
by adjusted R2. Thus, the linear approach seems to best capture the

relationship between resources and SMD placement.

127

A
function:
may be c
12 and 1
proportic
it is impc

M
While Pl
Russia c

reflects I

Although Models 2 and 3 do not undermine confidence in the linear
functional form, Model 1 does not fully control for other possible effects. There
may be country-specific factors that influence candidate placement. Tables 11,
12 and 13 show that major parties in Russia, on average, participate in a smaller
proportion of SMD races than their counterparts in Ukraine and Lithuania. Thus,
it is important to add controls for these influences.

Model 4 includes country dummy variables for Russia and Ukraine.
While PRLIST and PREVELECT remain significant and positive in sign, the
Russia country dummy is significant and negative (see Table 15, Model 4). This
reflects the lower level of SMD participation found among Russian parties.

Model 5 adds interactive variables for PRLIST and the country dummies.
The fit of this model improves over previous models, with an adjusted R2 of .81.
While PRLIST and PREVELECT persist as significant predictors of SMD
placement, RUSSlA*PRLlST is significant and negative in sign (see Table 15,
Model 5). The total effect of PRLlST is thus lower in Russia than its coefficient
suggests. The PRLIST coefficient value of .575 is attenuated by the -.291
coefficient of RUSSIA*PRL|ST, generating a total effect of .283 for Russia. The
total effect remains significant and positive, indicating that greater personnel
resources lead to fuller participation in SMD for all three countries. Even when
controls are added for country-specific factors, the effects of personnel

resources and organizational capacity remain.

128

Table ‘

De

St

anda
time. .
and UI
and We

1”

Table 15: Effect of Resources on SMD Placement (Fixed Effects Models)

 

Model 4 Model 5
Variable ,8 and SE ,6 and SE
Constant 1 .945 -3.648
(5.380) (6.641)

[3, PR List .511*** .575***
(.056) (.074)

pg PR Vote .453 .457
(.327) (.306)

,6; Previous Election 8.895“ 9250*“
(3.708) (3.457)

,84 Russia -19.474*** -.021
(4.341) (8.306)
,6; Ukraine 1.612 .905
(4.499) (7.863)
,66 Russia *PR List -.292***
(.102)
[37 Ukraine *PR List .054
(.099)
Adjusted R2 .775 .805
N 72 72

 

 

Dependent Variable: Proportion of SMD races in which candidates were placed.
Standard errors are in parentheses.

Note: All parties participating in the most recent elections in Lithuania, Russia
and Ukraine were included. Some parties in Lithuania did not submit PR lists
and were excluded. Lithuania is the excluded case in Models 4 and 5.

* ps.10
** p305
*** ps.01

129

Altema
L

Many fa
margins
and are
well- knt
this is St
party's a
determir
E
explanai
mixed 5)
that part
SDODSQr
”Wasn(

wefind:

Alternate Explanations for SMD Placement

Large and small parties tend to participate fully in the mixed system.
Many factors could motivate this behavior, not only a party’s resources. Some
marginal parties may place candidates in districts where they enjoy local support
and are likely to win the seat. Alternatively, they could have a contingent of
well- known local personalities who are strong contenders for the district seat. If
this is so, resources alone would not dictate placement strategies in SMD. The
party’s assessment of its likelihood of winning the seat could play a role in
determining where it sponsors local SMD candidates.

Examples from Lithuania, Russia and Ukraine suggest that this alternate
explanation does not account for parties’ tendencies to participate fully in the
mixed system. If we examine the performance of candidates in SMD by parties
that participated in fewer than ten SMD races, we find that parties did not tend to
sponsor highly competitive candidates and they did not tend to focus placement
on a single region. In addition, if we select a moderate party from each country,
we find similar results.108

In Russia’s 1999 Duma elections, three small parties placed more than
one candidate, but fewer than ten candidates, in SMD (see Table 16). The
Social Democrats (SD) had five affiliated candidates in SMD. The candidates
were not concentrated in a single geographic constituency, suggesting that the
party does not enjoy support in a specific location. The best-performing

candidate received 2.4% of the local vote and finished in ninth place. Thus, it

130

does m
were lik
better ir
E

(SPR) rt
perform;
were no
receive:
Party do
district, )
received
Tl

fOUnd w,
Candldat
the vote.
Camdidat
ability to
I01163! PR
K
KTR als<
organiza

N

‘ Set
4 .e
faneq to :1)

does not seem likely that the party sponsored candidates in SMD that it thought
were likely winners in the districts. In three of five districts, the party performed
better in the local PR vote than at the national level.

Each candidate participating in SMD for the Socialist Party of Russia
(SPR) received a proportion of the vote exceeding the party’s aggregate PR
performance. Nevertheless, none of the candidates were competitive and they
were not placed in a single geographic area. The most successful candidate
received 3.1% of the local vote and finished sixth in the district. Once again, this
party does not seem to have placed likely winners in the districts. In each
district, however, the party received a higher proportion of the PR vote than it
received in the aggregate.

The Stalinist Bloc — for the Soviet Union (STAL) repeated the pattern
found with the Social Democrats and Socialist Party of Russia. The top
candidate for this party finished in third place in his district, receiving 8.41% of
the vote. While this candidate was a potential competitor, the remaining
candidates were not. Candidate placement does not seem to be based on the
ability to win a district race. in all but one case, the party performed better in the
local PR race than it did at the national level.

KTR is an example of a medium-sized party in Russia’s 1999 elections.
KT R also participated in 1995, showing that it should have greater
organizational capacity than the small parties in the analysis. However, it failed

to pass the threshold in both 1995 and 1999. Table 17 shows the relative

 

‘08 I selected a moderate party for each country that participated in the previous election, but
failed to cross the threshold by the smallest margin. Because very small parties may not be

131

Table 16: Performance of Small Parties (Russia 1999)‘°°

 

 

Party District % Vote Candidate % Vote % Vote PR
SMo"° Finished’" PR (Local) aggregate)
so 104 0.82 10 (16) 0.09 0.08
so 173 1.38 13 (18) 0.08 0.08
so 193 0.53 11 (11) 0.07 0.08
so 197 2.38 9 (12) 0.12 0.08
so 211 0.60 9 (9) 0.09 0.08
SPR 18 0.55 10 (14) 0.17 0.09
SPR 39 1.07 7 (8) 0.11 0.09
SPR 52 2.71 10 (15) 0.16 0.09
SPR 99 0.52 15 (17) 0.11 0.09
SPR 110 3.09 6 (19) 0.15 0.09
SPR 184 1.84 7 (10) 0.12 0.09
STAL 33 3.42 5 (10) 1.34 0.61
STAL 100 1.86 10 (20) 0.86 0.61
STAL 106 1.64 9 (11) 0.98 0.61
STAL 108 0.83 16 (17) 0.52 0.61
STAL 149 2.13 11 (16) 0.97 0.61
STAL 156 8.41 3 (6) 1.42 0.61
STAL 176 1.64 9 (13) 0.91 0.61

 

 

 

serious political parties, including a moderate party increases confidence in the results.
“’9 Source: Rossiiskaya Gazeta (2000).

11° Includes “against all.”
"1 The value in parentheses is the number of candidates in the given district.

132

Table 17: Performance of KTR (Russia 1999)

 

 

Party District % Vote Candidate % Vote % Vote PR
SMD112 Finished” PR (Local) (Aggregate)
KTR 7 8.69 4 (5) 1.97 2.22
KTR 28 6.12 5 (13) 2.27 2.22
KTR 66 3.14 9 (14) 1.96 2.22
KTR 92 2.79 6 (13) 2.32 2.22
KTR 108 6.03 7 (17) 1.58 2.22
KTR 118 1.60 9 (10) 3.13 2.22
KTR 119 1.00 11 (12) 3.29 2.22
KTR 120 1.52 9 (12) 2.61 2.22
KTR 126 1.15 10 (16) 2.50 2.22
KTR 132 2.47 9 (13) 3.35 2.22
KTR 156 4.34 6 (6) 3.02 2.22
KTR 159 2.61 7 (9) 3.58 2.22
KTR 166 10.51 4 (10) 2.85 2.22
KTR 173 4.78 6 (18) 2.55 2.22
KTR 178 20.69 2 (7) 12.25 2.22
KT R 194 2.02 8 (13) 0.76 2.22
KTR 212 9.89 4 (14) 1.78 2.22
KTR 213 2.69 7 (13) 1.37 2.22

 

 

 

"2 Includes “against all.”
"3 The value in parentheses is the number of candidates in the given district.

133

perform
parties.
vote. OI
in their c
finished
party aIs

performa

Party
W
LFU
LFU
LFU
LFU
LFU
LFU
LFU
\

Tr
Freedom
N009 Of .
receide
With 240A

performance of KTR’s candidates in eighteen districts. As with the small

parties, KTR tends to perform better in the SMD vote than in the aggregate PR
vote. Only five of eighteen candidates accrued a smaller proportion of the vote
in their districts than the party gained at the national level. The best candidate
finished in second place, but many KT R candidates were not competitive. The
party also tended to perform better locally than nationally. The party’s local PR

performance outstripped its national PR performance in twelve districts.

Table 18: Performance of the LFU (Lithuania 1996)114

 

 

Party District % Vote Candidate % Vote % Vote PR
SMD Finished115 PR (Local) (Aggregate)
LFU 11 7.68 5 (15) 8.08 1.49
LFU 12 24.44 2 (12) 10.78 1.49
LFU 13 5.00 6 (13) 6.01 1.49
LFU 14 6.31 5 (16) 5.96 1.49
LFU 16 6.24 4 (16) 6.87 1.49
LFU 17 6.90 5 (13) 6.95 1.49
LFU 65 2.73 11 (12) 2.51 1.49
LFU 66 3.38 7 (10) 3.19 1.49

 

 

This pattern is not limited to Russia. In Lithuania, the Lithuanian
Freedom Union (LFU) placed nine candidates in SMD races (see Table 18).
None of the candidates received a lower proportion of the vote than the party
received in the aggregate. In one district, the candidate finished in second place
with 24% of the vote, allowing him to move to the second round. He did not win

the seat, however, leaving the LFU with no seats in the Seimas.

 

"4 Source: Electoral Commission of Lithuania (1999).

"5 The value in parentheses is the number of candidates in the given district.

134

The Va

Table 19: Performance of YL (Lithuania 1996)

 

 

Party District % Vote Candidate % Vote % Vote PR
SMD Finished” PR (Local) (Aggregate)
YL 5 3.54 9 (16) 2.77 3.81
YL 7 2.51 8 (16) 2.17 3.81
YL 11 8.64 2 (15) 7.32 3.81
YL 15 23.45 2 (11) 9.83 3.81
YL 16 8.24 3 (16) 7.43 3.81
YL 17 9.44 3 (13) 7.72 3.81
YL 18 5.81 6 (15) 6.78 3.81
YL 22 8.02 5 (11) 4.47 3.81
YL 28 3.33 10 (15) 4.17 3.81
YL 29 3.06 8 (12) 3.90 3.81
YL 41 4.34 7 (12) 6.18 3.81
YL 45 6.22 5 (15) 7.30 3.81
YL 59 4.10 7 (15) 4.87 3.81
YL 60 5.72 8 (13) 5.63 3.81
YL 65 7.17 6 (12) 6.32 3.81
YL 66 9.36 2 (10) 7.74 3.81

 

 

 

135

”6 The value in parentheses is the number of candidates in the given district.

 

in any c
SMD be
candida
11-18 a
that the
Russia.
nanonal
L
Lfihuan)
Table 1:
party at
one was
in the di:
Well in p
the 'OCal
TI
Cartdidat
recewed
performs
all hot Or
fifteen Ca

Deny did

With this exception, the LFU did not place highly competitive candidates
in any of the district races, suggesting that it did not solely sponsor candidates in
SMD because it thought the candidates would be competitive. All of the
candidates were concentrated in a single geographic area, however. Districts
11-18 are in the city of Kaunas; 65 and 66 surround Kaunas. Thus, it seems
that the party placed SMD candidates with the region in mind. Just as in
Russia, the party received a higher proportion of the local PR vote than the
national PR vote.

Lithuania’s moderately sized party, the Lithuanian National Party “Young
Lithuania” (YL) continues the pattern that is found with the small parties (see
Table 19). In twelve of sixteen districts, the candidate performed better than the
party at the national level. Three candidates finished in second place, but only
one was victorious in the second round. YL placed some competitive candidates
in the districts, but most were not likely to win seats. The party also performed
well in PR in districts where it nominated SMD candidates. In fourteen districts,
the local PR vote was greater than the party’s national PR vote.

The All-Ukrainian Party of Women’s Initiatives (WOM) also sponsored
candidates that received a higher local proportion of the vote than the party
received in the aggregate, but it did not win any seats (see Table 20). The party
performed better in the PR race at the local level than it did in the aggregate in
all but one case. The best performing candidate finished third in a district with
fifteen candidates. As with other small parties in Russia and Lithuania, this

party did not place candidates that were highly competitive in the districts.

136

Table 20: Performance of WOM (Ukraine 1998)117

 

 

Party District % Vote Candidate % Vote % Vote PR
SMD"8 Finishedm PR (Local) (Aggrgqate)
WOM 29 6.40 6(18) 1.18 0.58
WOM 42 1.24 10(20) 0.66 0.58
WOM 46 0.73 15(18) 0.66 0.58
WOM 47 5.92 3(15) 0.92 0.58
WOM 117 0.92 11(21) 0.38 0.58

 

 

Three of the five candidates ran in Donetsk Oblast, however, suggesting that
regional connections or resources may guide placement decisions.

The Christian Democratic Party of Ukraine (CDP) had many SMD
candidates on the ballot. In 39 of 69 districts, the candidate performed as well
as or better than the party at the national level. The competitiveness of GDP
candidates varied widely. While two candidates won their races, there were
also many non-competitive candidates. The party performed better in PR at the
local level than at the national level in 39 districts. Although local candidate and
PR performance exceeded national performance in the same number of
districts, the districts did not always correspond. While the relationship between
local and national performance is less consistent for the CDP than all previous
parties, the patterns generally persist — parties tend to perform better at the local
level where SMD candidates appear on the ballot.

Marginal parties and the medium sized parties in all three countries did
not tend to place highly competitive candidates in district races, undermining the

assertion that candidate placement strategies of marginal parties focus primarily

 

"7 Source: Central Electoral Commission of Ukraine (1998).

"8 Includes “against all.”

137

Table 21: Performance of the CDP (Ukraine 1998)‘2°

 

 

Party District % Vote Candidate % Vote % Vote PR
SMDm Finished122 PR (Local) (Aggregate)
CDP 1 0.82 15 (22) 0.43 1.30
CDP 2 1.36 10 (14) 0.40 1.30
CDP 3 0.68 12 (15) 0.36 1.30
CDP 4 1.73 8 (16) 0.52 1.30
CDP 5 0.66 11 (11) 0.80 1.30
CDP 6 0.72 14 (16) 0.66 1.30
CDP 7 1.04 11 (14) 0.79 1.30
CDP 8 0.47 17 (19) 0.43 1.30
CDP 9 1.96 7 (17) 1.23 1.30
CDP 10 0.42 15 (17) 0.37 1.30
CDP 19 7.12 6 (12) 2.49 1.30
CDP 20 1.49 9 (12) 2.26 1.30
CDP 23 2.46 9 (12) 3.34 1.30
CDP 27 1.26 15 (22) 0.59 1.30
CDP 33 1.75 8 (12) 1.41 1.30
CDP 34 1.39 14 (21) 0.95 1.30
CDP 53 0.35 18 (19) 0.56 1.30
CDP 68 20.40 1 (21) 17.68 1.30
CDP 72 2.09 10 (16) 2.02 1.30
CDP 78 0.54 18 (18) 1.05 1.30
CDP 81 1.19 12 (16) 0.58 1.30
CDP 84 3.83 8 (18) 0.84 1.30
CDP 86 2.12 13 (21) 1.48 1.30
CDP 88 10.29 4 (18) 1.87 1.30
CDP 90 0.44 20 (25) 0.93 1.30
CDP 91 1.30 13 (15) 1.31 1.30
CDP 92 0.92 6 (11) 1.63 1.30
CDP 93 1.30 13 (18) 1.43 1.30
CDP 94 1.97 12 (20) 1.58 1.30
CDP 95 1.92 10 (26) 1.80 1.30
CDP 96 2.36 11 (21) 1.55 1.30
CDP 114 1.14 13 (23) 0.71 1.30
CDP 115 1.32 11 (15) 0.77 1.30
CDP 117 0.86 13 (21) 0.76 1.30

 

 

 

"9 The value in parentheses is the number of candidates in the given district.

12° Includes candidates that were members of the party and appeared on the ballot as affiliated.
Because of this, the number of sponsored candidates in Table 13 differs slightly from this table.
121 Includes ”against all.”

122 The value in parentheses is the number of candidates in the given district.

138

Table 21 (cont’d)

 

CDP 118 7.16 5 (15) 1.78 1.30
CDP 119 1.02 11 (14) 1.47 1.30
CDP 121 0.61 10 (10) 1.75 1.30
CDP 122 3.78 9 (20) 1.80 1.30
CDP 123 1.54 11 (16) 2.03 1.30
CDP 124123 9.43 3 (12) 1.46 1.30
CDP 126 1.66 8 (18) 3.08 1.30
CDP 138 2.29 7 (17) 1.17 1.30
CDP 139 4.23 5 (19) 1.77 1.30
CDP 144 1.61 9 (27) 1.07 1.30
CDP 145 0.99 17 (22) 1.03 1.30
CDP 150 0.54 18 (26) 1.35 1.30
CDP 153 2.20 10 (22) 2.28 1.30
CDP 154 0.35 22 (23) 1.98 1.30
CDP 155 3.29 6 (14) 3.52 1.30
CDP 156 0.83 16 (22) 3.07 1.30
CDP 157 0.54 21 (21) 0.89 1.30
CDP 163 2.66 9 (17) 1.75 1.30
CDP 164 1.09 17 (24) 2.75 1.30
CDP 165 1.23 10 (17) 2.81 1.30
CDP 166 2.25 9 (17) 1.48 1.30
CDP 174 3.53 7 (21) 1.03 1.30
CDP 182 2.15 9 (19) 2.44 1.30
CDP 183 24.33 1 (22) 8.16 1.30
CDP 188 0.83 13 (20) 1.71 1.30
CDP 189 1.71 12 (16) 1.64 1.30
CDP 190 0.90 14 (16) 1.92 1.30
CDP 191 2.92 11 (18) 3.68 1.30
CDP 200 1.73 11 (18) 0.77 1.30
CDP 201 1.02 14 (19) 1.02 1.30
CDP 205 0.93 17 (19) 1.81 1.30
CDP 206 0.99 14 (21) 0.86 1.30
CDP 215 1.34 10 (24) 1.92 1.30
CDP 224 0.27 18 (19) 0.55 1.30
CDP 225 1.65 11 (19) 0.42 1.30

 

 

 

‘23 Two candidates appeared on the ballot as CDP candidates in the district. I report the results
for the best performing candidate.

139

0n probe
there doe
HC
pr0p0rti0.
aggregate
tended to
level. Th
placemer.

detail in ti

Conclusit

Thel
electOra) c
itSUQgesti
SMD- Poi
pamCipath
fewer reso

One
dictate We
are “RF-1y t:
Margina) p.
Were Door l

Winners in 1

 

on probable winners. While some parties clustered candidates geographically,
there does not seem to be a general pattern of placement based on geography.
However, SMD candidates for marginal parties received a higher

proportion of the vote in their respective districts than the party received in the
aggregate. In districts where these small parties placed candidates, the party
tended to receive a higher proportion of votes in PR than it did at the aggregate
level. This suggests that there may be an additional connection between
placement in SMD and performance in PR, 8 relationship that will be analyzed in

detail in the next chapter.

Conclusions

The Investment Effect predicts that parties that have committed to
electoral competition are likely to participate fully in the mixed system. Further,
it suggests that party resources partially dictate the degree of participation in
SMD. Political parties that are more able to overcome the marginal costs of
participation in SMD are likely to sponsor more candidates than those that have
fewer resources.

One possible alternative explanation to the argument that resources
dictate placement strategies in SMD is that parties place SMD candidates that
are likely to win. The evidence does not support this contention, however.
Marginal parties in all three countries generally placed candidates in SMD that
were poor performers. It is unlikely that they were considered to be probable

winners in the districts.

140

 

have a
candic

factors

the vol
raises
placen
Contan

Candid

Another alternate explanation is that parties place candidates where they
have a regional base of support. While some parties concentrated their
candidates in a particular region, this was not true for most parties. Other
factors may motivate parties in their placement decisions.

An additional finding, that parties tend to receive a higher proportion of
the vote locally than nationally in districts where they place SMD candidates,
raises the possibility that there is an additional effect influencing SMD
placement. In the next chapter, I will discuss the Echo Effect — a form of
contamination that encourages parties to participate in SMD, even if their

candidates have little prospect of winning seats in the districts.

141

Chapter 5

THE ECHO EFFECT

The prevalence of multiparty competition in mixed systems is not
surprising, given the relatively forgiving nature of the mechanical effect,
established in Chapter 3, and the Investment Effect, analyzed in Chapter 4. The
Echo Effect provides further incentives for multipartism in mixed systems.

A party’s sponsorship of a candidate in SMD races can improve vote
outcomes for the party in PR.124 Through advertising and active campaigning,
local SMD candidates would raise awareness of the party, increasing the
likelihood that the party would receive votes in PR. Even if the candidates are
not active campaigners, a party’s affiliation with a SMD candidate on the ballot
may encourage support for the party in PR by voters who otherwise would not
have voted for the party.125 Aggregated to the national level, this effect could
increase the number of seats that a party gains in parliament.

The name “Echo Effect” is derived from these direct and indirect effects.

The improvement of local PR outcomes based on SMD candidate placement is

 

124 In other words, affiliation with a candidate can have benefits whether or not the candidate is

popular or powerful. If the candidate is also well-known and well-connected to local political and
economic resources, the benefits accruing to the party could be even greater.

‘25 It Is possible that participation of some candidates may have a negative effect on PR
performance as well - a non-charismatic individual could hurt the party in PR. Also, some parties
might benefit from the lack of information about their actual platform - voters may be more likely
to support the image of the party rather than the substance. If the SMD candidate increases
knowledge about the party’s positions, it could damage the party’s performance. Despite these
possible scenarios, parties presumably field candidates that they expect will help their overall
performance. Moreover, the effect of increasing awareness of the party should generally be
positive.

142

the direct effect. The indirect aspect (or echo) is the strategic response of parties
to the potential electoral benefits.

The Echo Effect Hypothesis indicates that the placement of candidates
in SMD will affect vote outcomes in PR in corresponding districts. If the
effect is borne out by the evidence, the positive effect of SMD placement on PR
performance should further encourage parties to maximize the placement of
candidates in SMD. As I will show, the data support the existence of the Echo

Effect in all elections under analysis.

Echo Effect Hypothesis, Models 1a and 1b
Results for Model 1a

Lithuania’s 1996 parliamentary elections provide support for the Echo
Effect. The variable SMDPL measures placement across the districts (71 in
Lithuania). A statistically significant coefficient means that the placement of a
candidate in an SMD race directly affects vote outcomes in PR. In Model 1a,
controls are added for incumbency and for the party’s local popularity.

Of the eleven parties analyzed, the parameter value of SMDPL is
statistically significant and positive in sign in all but one instance (see Table 22).
The magnitude of the SMDPL’s effect also varies, as we would expect. Because
SMDPL’s coefficient represents the local bonus received by a party in PR from
placing a candidate in a district race, parties that received more support in the
elections should generally obtain higher coefficient values than parties that did

not perform as well.

143

Table 22: OLS Results for Model 1a (Lithuania 1996)

 

 

 

 

 

 

CDU Econ LCU LLU
Constant 3579*” .944 3889*” .345
(.545) (.645) (1.004) (.631)
,BISMDPL .243 1 .371 ** 2.999" 1 979*“
(607) (580) (928) (567)
lencumbent 3.138“ 6.316”
(1.123) (2.351)
flJUrban -.008 -.008 .036” .009
(.007) (.009) (.013) (.009)
Adjusted R2 .109 .049 .305 .175
N 71 71 71 71
Table 22 (cont’d)
LNU LL Peasants Poles Alliance
Constant 3.780*** .238 *** 3600*" 5.009”
(.401) (.033) (.454) (2.365)
,BISMDPL 2507*”r .201 *** 1 .170“ 10.845***
(.411) (.032) (.342) (2.217)
lencumbent __ _ 7.103” 7.856
(1.378) (8.711)
,63Urban -.005 -.001 -.028*** -.007**
(.005) (.000) (.005) (.032)
Adjusted R2 .337 .345 .560 .274
N 71 71 71 71

 

 

144

Table 22 (cont’d)

 

 

Pol. Socialists Women
Prisoners
Constant .851 .589*** 3.248***

(1.037) (.082) (.506)

,BISMDPL 1.391 ** .372*** 1.157“
(.454) (.078) (.377)

lencumbent 8.553” .055 __
(1.341) (.297)

,B3Urban .012 .011 .001

(024) (001) (006)
Adjusted R2 .314 .239 .100
N 71 71 71

 

Dependent Variable: Proportion of the vote in PR at the SMD level.
Standard Errors are in parentheses.

* ps.10
** p305
*** ps.01

145

For example, by placing candidates in SMD, the LCU accrued a mean
bonus of 3.0%. In the entire system, the LCU received 7.7% of the vote, so the
magnitude of its local bonus in PR is proportionate to its overall performance.126
This tendency carries over to very small parties in the analysis. The Party of
Life’s Logic received 0.3% of the total vote.127 At 0.2%, its local bonus for placing
a candidate was quite high. A small coefficient does not indicate that the bonus
itself was small; it is dependent on the relative size of the bonus in comparison to
the party’s overall performance. Small parties generally had small parameter
values for SMDPL; larger parties had larger ones.

The statistical significance of the coefficients for the control variables
varies. Five parties had no incumbents participating in the election. Of the
remaining six parties, INCUMBENT is statistically significant in four cases. The
coefficient value for INCUMBENT is higher than the coefficient for SMDPL in
each case, suggesting that incumbency provides a greater benefit for party
performance in PR than simply placing a candidate in the district.

The proxy for popularity, URBAN, is statistically significant in three of the
eleven cases and demonstrates the expected effect. The sign of the coefficient
is negative for the Lithuanian Peasants’ Party, indicating that the party performs
better in PR in rural areas. The coefficient of SMDPL is statistically significant in
these three cases, indicating that the inclusion of a proxy for popularity does not

undermine the Echo Effect in Lithuania.

 

‘26 The LCU received 8.2% of the vote in PR and 6.8% of the vote in SMD.
‘27 The Party of Life’s Logic received 0.24% of the vote in PR and 0.33% of the vote in SMD.

146

All cases were tested for collinearity and the influence of outliers.
Generally, the Variance Inflation Factor (VIF) was close to one, indicating that
collinearity should not undermine the efficiency of the parameters.128 Outliers
had a limited effect on the results. The removal of extreme outliers recovered the
significance of the constant in the case of the Lithuanian Party of the Economy,

but there were no substantial changes in other cases.129

Results for Model 1b

While the Lithuanian model controlled for urban/rural residence only, the
Russian and Ukrainian models also control for level of education. Just as the
proportion of urban dwellers in a region should provide a proxy for popularity, so
too should the composition of a region’s population by education. By adding this
variable, we raise our confidence that we have adequately controlled for the
alternative hypothesis.

Table 23 shows OLS results for ten Russian political parties in the 1995
Duma elections. The signs of the coefficients for SMDPL are positive, except in
the case of NDR. The parameter of SMDPL is significant at the .05 level in
seven of the models and at the .10 level in an additional model. The KPRF

received the largest average bonus for candidate placement, at 4.7%.

 

‘28 The VIF measures how a coefficient’s variance is inflated by collinearity. As collinearity
increases in the model, the VIF uses. The VIF measures how a coefficient’s variance is inflated
by collinearity. See Gujarati (1995, 328-329) for more information.

1 Outliers beyond three standard deviations from the mean were identified and removed from
the dataset. The data were then re-analyzed with each individual outlier removed and with all
outliers removed as a block.

147

Table 23: OLS Results for Model 1b (Russia 1995)

 

 

APR DVR Forward, KPRF KRO
Russia

Constant 13.652*** -3.240*** -.254 33.862*** 1.656“

(1.441) (.744) (.274) (2.874) (.717)
,3, .981 .924" .662*** 4.726*** 1.792***
SMDPL (.616) (.327) (.125) (1.194) (.295)
,3, 2.952" 2.987*** .742" 5.559" 1.595
incumbent (1.077) (.675) (.352) (2.343) (1.266)
p, -.140*** .003" .025*** -.170*** .017
Urban (.024) (.013) (.005) (.049) (.013)
p, -.002 .429*** .023 -.241 .064*
Education (.076) (.040) (.015) (.156) (.009)
Adjusted R2 .273 .608 .362 .210 .209
N 225 225 225 225 225

 

 

148

Table 23 (cont’d):

 

 

KTR LDPR NDR Rybkin Yabloko
Constant 8.337*** 11.125*** 8.122*** 1.552*** 4359*“
(.609) (1.753) (2.140) (.384) (1.011)
p, 1.180*** 1505* -.074 .587” 2.471***
SMDPL (.274) (.854) (.858) (.192) (.460)
'32 4595* .544 1.15*** 2112*
Incumbent (2.423) (3.122) (.325) (1.286)
p, -.032** .044 -.054 -.008 .007***
Urban (.011) (.028) (.036) (.007) (.018)
p, -.154*** -.649*** .571*** .003 .431***
Education (.033) (.088) (.115) (.021) (.055)
Adjusted R2 .282 .268 .116 .155 .576
N 225 225 225 225 225

 

ps.10
p305
p301

149

Dependent Variable: Proportion of the vote in PR at the SMD level.
Standard errors are in parentheses.

Among the control variables, INCUMBENT is statistically significant at the
.05 level in five of nine cases and at the .10 level in two cases.‘3° Just as in
Lithuania, the parameter value of INCUMBENT exceeds the value of SMDPL in
each case where INCUMBENT is statistically significant. This indicates that
incumbency status exerts a greater impact on local PR performance than
candidate placement.

The signs and statistical significance of URBAN and EDUCATION, proxies
for regional popularity, varied. The coefficient for URBAN was significant and
positive in sign for DVR, Forward, Russia, and Yabloko. It was significant and
negative for the APR, KPRF and KTR. Thus, pro-reform and pro-Western parties
tended to receive support in cities while communist and agrarian parties received
support in rural regions.

The parameter value of EDUCATION was significant and positive for
DVR, KRO, NDR and Yabloko. These parties received a higher proportion of the
PR vote in regions with a higher proportion of educated residents. It was
significant and negative for KTR and the LDPR, demonstrating that these parties
received less support where a region’s population was more highly educated.

Just as the results from the 1996 Seimas elections in Lithuania provided
support for the existence of the Echo Effect, so too do the results from Russia’s

131

1995 parliamentary elections. When controls are added for incumbency and

 

13° The variable could not be included in the KTR equation because the party had no incumbents
in 1995.

131 The results of diagnostic tests for collinearity and outliers showed that they did not exert a
substantial impact on the results.

150

Table 24: OLS Results for Russia (1999)

 

 

AN-SF DN KPRF KRO LDPR
Constant .224** .044*** 39.356*** -.538**'* 5.381***

(088) (.017) (1.206) (.172) (.725)
,3, .122*** .040*** 2152* .436*** .041
SMDPL (.038) (.007) (1.206) (.085) (.297)
[32 .462** .164*** 2.874** 2.403*** 1.716*
incumbent (.192) (.025) (1.284) (.511) (.897)
,3, .004*** .000 -.181*** .006** .048***
Urban (.002) (.000) (.043) (.003) (.013)
)3, -.001 .000 -.41 1*** .056*** -.255***
Education (.005) (.001) (.132) (.009) (.039)
Adjusted R2 .139 .322 .327 .424 .164
N 224132 224 224 224 224

 

 

 

‘32 The number of cases is 224 because no elections were held in Chechnya.

151

Table 24 (cont’d):

 

 

NDR OVR RSP SPS Yabloko
Constant 1 .537*** 7.932** .279*** -3.863*** -3.046***

(.496) (3.880) (.080) (1.356) (.901)
,3, .083 2.007 .117*** .928 1.494***
SMDPL (.216) (1.655) (.037) (.610) (.369)
p, .474 5.099 .679*** 2.840 .768
incumbent (.336) (4.000) (.1 10) (1.868) (.616)
p, -.006 -.213*** -.002 .181*** .098***
Urban (.009) (.068) (.001) (.024) (.015)
p, .026 1.753*** .007* -.096 094*
Education (.005) (.213) (.004) (.073) (.48)
Adjusted R2 .000 .291 .221 .306 .387
N 224 224 224 224 224

 

Dependent Variable: Proportion of the vote in PR at the SMD level.

Standard errors are in parentheses.
p s .10
p s .05

*** p301

152

popularity, the effect of SMD candidate placement remains in most of the cases.
This provides additional evidence for the Echo Effect.

Russia’s 1999 results also support the existence of the Echo Effect,
although they are less consistent than the 1995 analysis. The results in Table 24
show that the coefficient for SMDPL is statistically significant at .05 for five
models and at .10 in an additional model. The strength of the SMD effect is
highest for the KPRF, which received a 2.2% bonus, on average, by placing
candidates in SMD.

The coefficient for INCUMBENT was also significant at the .05 level in five
cases and at the .10 level in an additional case. Once again, the effect of
incumbency on PR vote outcomes is stronger in all of the cases in which the
parameter is statistically significant.

The coefficients for urban residence and education vary from party to
party. The parameter value of URBAN is statistically significant and positive in
sign for the Bloc of Andrey Nikolayev and Svyatoslav Fyodorov (AN-SF), KRO,
LDPR, SP8 and Yabloko. The relationship between urban residence and PR
performance for Yabloko remained the same in both 1995 and 1999; the party
tends to receive a higher proportion of PR votes as the proportion of the
population in the region is more urban. The coefficient is negative for the KPRF
and Fatherland - All Russia (OVR). This is not surprising given that the APR
dissolved during the 1999 election campaign and prominent members joined
forces with both the KPRF and OVR. Moreover, the sign of URBAN was

negative for the KPRF in 1995 and remained negative in 1999.

153

The coefficient for EDUCATION was positive and significant at the .05
level for KRO and OVR, and at the .10 level for the Russian Socialist Party (RSP)
and Yabloko. These results are consistent with those from 1995. The parameter
is significant and negative for the KPRF and LDPR. This is also consistent with
1995.

In sum, results from Russia support the Echo Effect Hypothesis. In both
1995 and 1999, the coefficient for SMDPL is significant in most cases. The
results are less consistent during the 1999 elections, a phenomenon that will be
addressed below.

The OLS analysis with Ukrainian data also supports the Echo Effect
Hypothesis, even in the presence of controls for incumbency and popularity (see
Table 25, columns labeled 1b). The variable SMDPL is statistically significant at
the .05 level in seven of thirteen cases and at the .10 level in an additional case.
Incumbency is statistically significant in ten of twelve cases and at the .10 level in
a single case.133

As in other countries, the bonus for candidate placement varies from party
to party. The larger parties, such as KPU, receive larger bonuses than more
marginal parties, such as UNA. The KPU bonus for candidate placement is 6.6%
compared to the UNA’s .51%. Thus, the KPU could expect to receive a 6.6%
bonus, on average, in PR in districts where it placed an SMD candidate; UNA

could expect .51%. The KPU received 20.3% of the total vote; UNA received

 

‘33 One of the parties had no incumbents in the 1998 elections.

154

 

 

 

 

 

mmm mmm mmm mum mmm mmm mmm cum 2
8mm. «2. ova. wt. #0. 80.. Pm. mos. am passes.
A341 663 E: 868 Lee:

iooom. icon. 33. £83 as
E: 803 3:3 883 Loom
S? can- moor men. as
ammo as; 88$ 833 88.5
48.- :33- out- So. as
$63 68.3 88c an“: ANNE 853 $83 $3 cosecscm
:18. icon. meo. tho. e8. .58. ac. moo. 4n
3:: :63 8:: £63 C83 :83 A83 8:: secs

£50.- £50.- moor immor 2.85.- £20.- 98.- £68.- an

:33 C :3 am: am: 8%. : Ev. : accesses

inc; tween 80. 80. £83 icons e.

863 :03 A83 am: 8e: 8?: can: as: dog

toes. the. tone. .mvm. m1. mm: mob. mmo. an
$63 ammo A83 803 as; 53 83”.: :03

£656 $23 New. famed £83 £83 £83 inc: E8628

m e m e m 8 m 8 Lone:

EoQ
ocoeco moo 69m 2%. End

 

 

some cases. u can be mace: 5.. ozone”. «Jo "mu e38.

155

 

 

 

 

 

 

mNN mNN mNN mNN mNN mNN mNN mNN 2
SN. N8. t e. 8 F. can. SN. 48. one. am ceases.
68.5 at; 83.5 8N9 5 one:
£ 55,1 Em. £33: £899- as
68.5 a 5.5 88.5 a a. 5 Loam
£o 5N Eb. cam. :05...” as
ASN.5 $8.5 $8.5 ENNNV 6685
5:. :o. 86.- £58: 5.
60 e .5 Go F .5 £85 53 EN e .5 so: 543 ASN.5 68856.5
£36. Nmor N8. m 5. £ :3. £6 5. £8N. e- £85 F- as
88.5 2&3 a :5 a 6.5 :85 $8.5 $8.5 :85 :25
88.5 aNNNv 85.5 33.5 834.5 82.5 385 88.5 accesses
SN .33 £2 e .6 £5 e .6 £83 £F {a tNNoe :33 as
88.5 Not 88.5 83.5 68.5 68.5 ace. 5 6.8. 5 udosm
Eb. £83 :62. :85 N 3. one. :82” £938 5.
:2. 5 GNN. 5 :35 ft 68. 5 as. 5 88.5 :8va
£68.? owNe £82 £53 NNme £5NN £858 86.- 5:65:60
N 2 N 2 N 2 N 2 Lace:
com 652 u 52 3% End

 

6.28. 3 can...

156

 

 

 

 

 

mNN mNN mNN mNN mNN mNN 2
SN. .35 one. 8N. 8m. 8N. am some}.
68. 5 863 Co: as:
£85m- £68N 93. st
:8. 5 68.5 AcNi seem
.mkN- 8N.- £83 as
63.5 A83 as: 85.5
£83. tmET me..- he
ASN.5 Cm: $8.5 $8.5 $8.5 $3.5 coco-56m
08.. 8? £EN. £30 £5:- £RN.- Q
$8.5 N35 8N3 $5.5 ES :55 :35
£30- £55- 08. £89 £86. £26. an
38.5 8R5 ASN.5 $8.5 68.5 88.5 Emcee-cs
Soc :33 £83 £83 £82” £83” an
885 El. 5 8:; EN: 5N5 63.5 .EQEm
mN: New. £5: £vaN EN. EN. as
some aNva 89. 5 act ENE :35
£8NNN £423 me:- £me.m 65.- £955 5988
N e N S N S Loco:
69m mem cam BEBE End

 

6.2.85 mN each

157

S.wau£ ”mode": Norman .-
.momo£c8ma 5 8m 22cm Eons—Em ._m>o_ 056 35 “m an. E 90> on. ,6 cos-.89.". 8385) E32800

 

 

 

 

m-NN mNN mNN mNN 2
Be. be. 68. 08. am 8532
:35 83 Lacs
£05. .8? S
83.5 it 58m
80.- m8. .6.
A8: :55 8E5
.8? 88.7 as
85.5 88.5 83 83.5 cosmoscm
£08. £N8. £83. £63. 8.
88.5 N03 ENS ES :85
30.- £e E.- £mNor £86.- 8
ASN.5 2 N5 ANNvé 68.5 568585
.58. £85 £858 £858 8
88.5 88.5 at; 83 .Eosa
£834. £88. oR. i8. e s.
$3.5 28.5 83. 5 68.5
3:. £58. £83 £83 59860
N 2 N 5 L86:
<2: :58 E8

 

6.2.8. 8 c.52-

158

.41 % of the total.134 Thus, the average bonus received by these parties is
proportionate to their overall performance in the election. Moreover, both large
and small parties show measurable effects.

The coefficient for URBAN was statistically significant in all but one case.
The coefficient values represent the effect of a one percent change in the urban
population on PR performance. For the KPU, an increase in 1% in the urban
population results in a .53% improvement in PR results at the district level.
URBAN was significant and positive for the KPU and RAZOM. It was significant
and negative for APU, Bloc of Democratic Parties (Bloc Dem), Christian
Democratic Party (CDP), Greens, National Front (Nat F), People’s Democratic
Party (NDP), Reform and Order (R&O), Socialist and Peasant Bloc (S&P Bloc),
Social Democratic Party (United) (SDPU) and UNA.

The parameter estimate of EDUCATION is statistically significant a the .05
level in eight cases. It is significant at the .10 level in an additional case. Again
using the KPU as an example, a one percent increase in the proportion of
population with a higher education results in a 1.5% reduction in the PR vote.
EDUCATION is significant at .05 and positive for CDP, Greens, Nat F, R&O,
SDPU and UNA. It is significant and negative for KPU and RAZOM.

In addition to assessing the effects of the explanatory variables, I
conducted diagnostic tests for collinearity and the influence of outliers. The
control variables presented some problems of collinearity, but this does not

introduce bias into the model. As expected, URBAN and EDUCATION are highly

 

‘34 The KPU received 24.65% of the PR vote; KPU candidates gained 13.38% of the SMD vote.
UNA garnered .40% of PR and UNA candidates received .38% of the SMD vote.

159

correlated; better educated individuals tend to be urban dwellers. While the VIF
reached as high as 4 for URBAN and EDUCATION for some parties, SMDPL
does not become significant for all parties even when URBAN and EDUCATION
are removed. The increase in inefficiency caused by the control variables does
not explain the failure of the Echo Effect to be manifested for each party.

Outliers also had a measurable effect on the results. Removing outliers
and repeating the analysis altered the outcomes in some cases. The p value for
SMDPL was raised to .32 after removing the outlying case in the CDP model.
While individually removing the outliers and testing the model did not affect the
results greatly for the Greens, removing the entire block of outliers raised the p
value of SMDPL to .11.

Removing outliers did not uniformly undermine the Echo Effect, however.
With the PSP model, the p value of URBAN became .02 after removing all
outliers, rendering that coefficient statistically significant. Removing all outliers in
the APU case also improved results. The variable SMDPL, which was not
significant with outliers included, became statistically significant when outliers
were removed (the p value became .03). However, the significance of
INCUMBENT was lost in the APU model when outliers were removed. The same
effect was found in the RAZOM model. Removing the block of outliers recovered
an effect of SMDPL, changing its p-value to .01. Likewise, the p-value of SMDPL
became .01 with all outliers removed for the Block of Democratic Parties.

Outliers have a measurable effect on the results in the Ukrainian case.

While it is instructive to remove outliers and reanalyze the data in order to assess

160

their effect, it is also important to note that outliers are real cases. That is, we
cannot simply remove them because they do not fit within our predictions.
Outliers also failed to affect the results in one direction; they did not uniformly
undermine or support contamination effects. In some models the effect was lost,
in others it was recovered. Overall, however, when outliers are removed,
evidence for the Echo Effect improves.

Evidence from additional mixed systems suggests that the results are
robust. Although the assessments of the Echo Effect in Germany, New Zealand
and Japan do not include the same variables as the analysis in this chapter, they
support the hypothesis in compensatory systems as well as established
democratic states with institutionalized party systems (see Appendix C for
details).

Russia’s 1999 and Ukraine’s 1998 elections are the least consistent in
demonstrating the existence of the Echo Effect. It is possible that the benefits
accruing to large parties diminish over time. As voters become more aware of
the party’s existence, the simple act of placing a candidate on the ballot may
have a less substantial effect on the PR vote. Because Russia has conducted
the most elections under the mixed system, the deterioration of the effect may
have begun to manifest itself in the 1999 elections.

Another possibility is that the influence of regional authorities could
undermine the Echo Effect. There have been claims of vote fraud in some

regions that affected the major parties (KPRF, LDPR, Medved, OVR, and SP8 in

161

particular).135 If regional governors indeed altered the results for these parties in
PR, the apparent effect of SMDPL on PR vote outcomes could be affected.
Alternatively, regional governors could use their administrative resources to favor
candidates that are affiliated with their parties. This could also affect the
relationship between SMDPL and local PR outcomes. In fact, if the governor’s
party affiliation is added to the equation, it significantly affects the PR
performance of the major parties.

The affiliations of regional governors were based on stated affiliations and
membership in leadership councils (McFauI, Petrov, Ryabov and Reisch 1999).
Only four political parties had close ties with regional authorities, limiting the
analysis. The KPRF has a strong regional base, stemming from its inheritance of
the Communist Party of the Soviet Union’s organizational capacity. While the
KPRF enjoys support in many regions of the country, it is particularly strong in
the so-called “Red Belt,” 8 set of regions along Russia’s southern tier.

The other three parties, NDR, OVR and Medved, have been considered
the “parties of power” with close ties to the central leadership. NDR, headed by
former Prime Minister Viktor Chernomyrdin, was the “party of power” in the 1995
elections. OVR, led by former Prime Minister Yevgeniy Primakov and Moscow
Mayor Yuriy Luzhkov, was the early favorite in the 1999 elections until it fell out
of President Yeltsin’s favor. Medved was the political bloc supported by Vladimir

Putin, Boris Yeltsin’s chosen successor for the presidential office. All three

 

‘35 Claims of fraud have circulated through Johnson's Russia List (a listserv). In addition, the

author has spoken with regional scholars who argue that there was some systematic fraud.

162

Table 26: OLS Results for Model 1b with Controls for Gubernatorial
Affiliation (Russia 1999)

 

KPRF MED NDR OVR
Constant 37.338*** 34.032W .945“ 1 5067*”
(2.494) (2.872) (.471) (3.599)

 

)3, 2.456“ 1.256 .000 1.438
SMDPL (1.150) (1.673) (.201) (1.482)
p, 2161* -2.721 .426 2.891
incumbent (1.232) (6.052) (.312) (3.598)
p, -.176*** -.007 -.000 -.320***
Urban (.041) (.050) (.008) (.062)
)3, -.343*** -.933*** .005 1.425***
Education (.127) (.151) (.024) (.196)

p, 6392*” 4.163” 1.763*** 12.193***
Governor (1.328) (1.982) (.293) (1.635)
Adjusted R2 .389 .256 .139 .433

N 224 224 224 224

 

 

Dependent Variable: Proportion of the vote in PR at the SMD level.
Standard errors are in parentheses.

* ps.10
** p505
*** p301

163

parties were considered, at one point, to be the party preferred by the central
leadership. Because of this, many regional governors chose to ally with these
groups to gain benefits from affiliation with the center. The unique position of
these parties allows us to code them based on gubernatorial support.

The results of the analysis are found on Table 26. While the statistical
significance of the parameter values for SMDPL do not change from the results
in Table 21, the parameter for the affiliation of governors is statistically significant
and positive in sign in all four cases. In other words, parties performed better in
regions where the governor was officially connected to that party. This suggests
that governors may have substantial influence over the electorate and electoral
process and that their authority affects vote outcomes. This could undermine the
connection between candidate placement in SMD and performance in PR. Thus,
the failure of SMDPL to display statistical significance for some parties in the
most recent Russian election may be due, in part, to the latter explanation. This

is not the only possible rival explanation for the results, however.

Echo Effect Hypothesis, Model 2 — The Effect of Regionalism

Extending the analysis to Ukraine raises an important question about the
robustness of the Echo Effect. Ukraine is a country with long-standing regional
divisions that could undermine our confidence in the existence of contamination
effects. The mechanism by which regionalism could be confounded with
contamination was outlined in detail in the Chapter 3. The argument is similar to

the alternative hypothesis about the effect of popularity: parties place candidates

164

in areas where they have regional bases of support. If the effects we attribute to
the Echo Effect actually reflect careful candidate placement based on geography,
they would be incorrectly interpreted as an institutional effect.

If parties with regional bases of support place candidates in regions
outside their strongholds, they may not receive the corresponding “bonus” in PR
vote outcomes. Table 27 shows that candidate placement and performance for
some parties varies substantially by region.136 For instance, the performance of
the National Front (Nat F) appears to depend heavily on the region. In western
oblasts, it received, on average, 7% of the local PR vote. It accrued less than
1% in all other regions. This disparity in performance is not limited to Nat F;
other parties also depend on support from particular regions. Given that
regionalism is entrenched in Ukraine’s political system, are contamination effects
robust when we control for regions?

The effect of SMDPL loses its statistical significance in two cases (PSP
and SDPU) when we control for regions (see Table 25, columns labeled Model
2). The coefficient value for SMD also drops in some cases (such as KPU and
R80), suggesting that regionalism may have an attenuating effect on the Echo
Effect.

The regional variables vary in terms of their significance. CRIMEA is
significant at the .05 level in four cases; EAST is significant in three; WEST is
significant in nine. The sign of the coefficients also varies. For the variable

WEST, the coefficient is positive for the APU, CDP, Nat F, R80 and UNA.

 

‘36 Candidate affiliations are based on party membership rather than party sponsorship.

165

Table 27: Mean PR Vote and Number of Candidates by Region (Ukraine

 

 

1998)137
Central Crimea East West
(86 SMD) (12 SMD) (64 SMD) (63 SMD)

APU 2.85 2.82 2.27 5.69
49 2 22 36

BIocDem 1.28 1.07 0.96 1.43
36 1 22 28

CDP 1.22 0.58 0.59 2.12
21 12 6 29

Greens 6.59 5.80 5.05 4.65
40 4 28 26

KPU 25.10 40.39 35.72 12.19
63 12 58 41

Nat F 0.94 0.20 0.33 7.04
56 3 34 48

NDP 4.71 5.14 4.42 5.84
52 5 33 35

PSP 3.76 1.41 7.53 1.67
25 0 33 7

RAZOM 1.33 0.63 4.01 1.19
30 10 37 27

R80 2.43 0.98 2.73 4.89
17 2 30 23

S&P 11.07 1.60 4.67 8.45
39 8 38 38

SDPU 3.90 1.94 2.67 6.09
40 0 32 27

UNA 0.35 0.09 0.18 0.70
18 0 10 22

 

 

 

‘37 The mean proportion of PR votes for the region is on top. The number of candidates in SMD
races is below. Number of candidates refers to candidates that appeared on the SMD ballot as
affiliated with the party.

166

The sign is negative for the Greens, KPU, PSP and the S&P Bloc. These results
are not surprising. The CDP, Nat F and UNA are parties that have promoted
Ukrainian nationalist positions. Thus, we would expect them to compete
successfully in the West. The KPU and PSP reflect ideologies associated with
the former communist system and are expected to perform poorly in the area
with the strongest manifestations of Ukrainian national pride.

The analysis was accompanied by diagnostic tests for collinearity and the
influence of outliers in all of the models. The problems of collinearity were similar
to those found in Model 1b. The variables URBAN and EDUCATION and
URBAN and CRIMEA were usually highly correlated. In the former case, we
would expect more educated individuals to live in cities. In the latter case, most
districts in the Crimean region are in urban areas. Nevertheless, the increase in
inefficiency due to collinearity did not exert a substantial impact on the results.

Removing outliers and repeating the analysis altered the results in some
cases. In the case of the Greens, removing the outliers raised the p-value of
SMDPL beyond the .05 level. This only occurred when all outliers were removed
as a block, however. Removing all outliers “recovered” the statistical significance
of SMDPL for APU, Bloc Dem and RAZOM, however.

In summary, including regional variables in Ukraine mitigates the effect of
contamination in some cases, but the Echo Effect is generally robust. While
sociological characteristics, specifically regionalism, seem to interact with the
institutional factors and attenuate the Echo Effect, the effects attributable to

institutions do not disappear.

167

Echo Effect Hypothesis, Model 3 — The Effect of Independent Candidates
Just as gubernatorial party affiliation affected PR vote outcomes in Russia
and regionalism exerted an impact on PR results in Ukraine, another prominent
factor in the elections could affect the relationship between SMD placement and
PR performance in these countries. The large number of independent
candidates who participated in elections could produce an artificial Echo Effect.
If voters, contributors and other political actors know the ideological preferences
of officially independent candidates, our measure of interaction may be
inadequate. Because we do not account for interactive effects among
candidates who are not officially affiliated with a political movement, but whose
“true” affiliation is known, we eliminate a large proportion of active candidates
from the analysis. If these covert partisans’ real affiliations are well-known
among the voters, contributors and other candidates, the data may be censored.
Data collected by the Ukrainian Central Electoral Commission facilitates a
analysis of the effect of covert partisans on PR outcomes. While the CEC
collected data about the membership status and sponsorship of candidates, only
party membership appears on the ballot. For example, a candidate might be a
member of the NDP but not be sponsored by the party on the ballot. That
candidate would appear on the ballot, however, as a member of the NDP.
Because parties cannot fully control the decisions of their members, there were
numerous instances in which the party ran an official candidate and another party
member also contested the seat. On the ballot, it appeared as if the party had

two candidates in the district.

168

In addition, a candidate without an official party affiliation (labeled
“‘bezpartiiniy”) could be nominated by a party on a local ballot. In this case, the
candidate would appear to be independent, but actually would be affiliated with a
political party. Because party membership alone was published on the ballot,
many covert partisans participated in the elections.138 Covert partisans are
defined as candidates with no party membership who are nominated by a party in
the SMD component of the system.

I included the variable COVPART to control for the role of nominal
independents who were sponsored by a party. These candidates’ ideological
leanings could be known to the voters. Table 28 shows the number of districts
with covert partisans for each party under analysis.

OLS results show that covert partisanship has a limited but important
impact on the Echo Effect (see Table 29). The variable COVPART is statistically
significant for only two political parties. In both cases, the inclusion of covert
partisanship renders the variable SMDPL statistically significant as well. For Bloc
Dem, the bonus received by having a candidate in SMD is 24%. Covert
partisans improve performance by 70%. A similar relationship is found with the
UNA. The coefficient for SMDPL is .48; the COVPART coefficient is .58. In
other words, covert partisans exert a stronger influence on PR performance than
party members in these instances. In the remaining cases, covert partisanship

has no measurable effect on the results.

 

‘38 To register in an SMD race, a candidate must either have been nominated by a party and
present proof of nomination or collect 900 signatures and obtain self-nominated status (“On the
Election of People's Deputies of Ukraine,” Article 25, as amended (NAU 1999)).

169

Removing extreme outliers from the analysis affects some parties. The
coefficient for SMDPL becomes significant for APU and NDP, but it is raised
above the .10 level for Greens. The statistical significance of coefficients for

some other parties change, but do not undermine my substantive conclusions.

Table 28: Number of Covert Partisans for Ukraine (1998)

 

 

Party Number of Covert Partisans
APU 16
Bloc Dem 16
CDP 7
Greens 29
KPU 7
Nat F 36
NDP 1 1
PSP 15
RAZOM 22
R80 24
S&P Bloc 19
SDPU 41
UNA 4

 

 

To summarize, by developing a measure of covert partisanship for
independents, we see that it has a slight influence on our assessment of the
Echo Effect. The results of the analysis do not undermine the existence of effect,
but enhance our confidence in it. When we control for covert partisanship, we
recover an effect that is lost in the existing coding scheme in two cases. This
suggests that some of our failures to find evidence of the Echo Effect may be due
to the influence of covert partisans and not a weakness of contamination effects.
The influence of covert partisanship potentially extends to Russia as well. The

instability in Russian results could also be due to the existence of covert

170

Table 29: OLS Results for Model 3 (Ukraine 1998)

 

 

APU Bloc Dem CDP Greens KPU
Constant 3.425” 1 .863*** .830 6990*“ 20.614***

(1.364) (.339) (.528) (.911) (3.985)
,3, .735 .243** .501 ** .640“ 3.961 **
SMDPL (.448) (.120) (.209) (.309) (1.580)
'32 5724*“ .048 3.153*** 3.996“
Incumbent (1 .449) (.425) (.886) (1 .689)
'33 -.033 -.016** -.005 -.007*** 232*"
Urban (.026) (.007) (.010) (.018) (.074)
[34 .076 .031 .045 .332*** -1.261***
Education (.084) (.022) (.033) (.058) (.243)
'35 .663 -.021 -1.00** -.851 14.359***
Crimea (.991) (.256) (.413) (.666) (2.733)
,6, .446 -.040 -.392 -.188 3.804“
East (.676) (.174) (.265) (.460) (1.928)
fl, 2668*“ -.038 853*” -2.057*** -12.032***
West (.661) (.171) (.257) (.445) (1.833)
[3,, .629 .701 *** .000 -.127 1.641
Cov Part (.853) (.213) (.486) (.460) (3.661)
Adjusted R2 .214 .088 .236 .235 .582
N 225 225 225 225 225

 

 

171

Table 29 (cont’d)

 

 

Nat F NDP PSP RAZOM R&O
Constant 1.511 7.673*** 11.988*** -.816 -1.156

(1.953) (.968) (1.754) (.822) (1.165)
,3, .522 708* 916 .252 1.610***
SMDPL (.789) (.365) (.665) (.286) (.452)
,3, 6.895*** 5.172*** 2.324 3.788*** 8.766***
incumbent (1.453) (1.317) (2.427) (1.028) (1.279)
)3, -.073*** -.059*** -.210*** .053*** -.01 1
Urban (.037) (.018) (.034) (.017) (.023)
p, .341*** .049 .505*** -.144*** .281***
Education (.121) (.067) (.108) (.054) (.077)
P: -.560 .560 -1.742 -.776 -1.766**
Crimea (1.450) (.806) (1.250) (.622) (.855)
,3, .946 .690 6.980*** 1.785*** -.053
East (.950) (.510) (.895) (.428) (.649)
p, 4.765*** .300 -4.466*** .548 3.011***
West (.948) (.483) (.866) (.407) (.602)
p, .022 -.783 .570 -.219 -.404
Cov Part (1.032) (.828) (1.235) (.474) (.703)
Adjusted R2 .353 .177 .344 .353 .378
N 225 225 225 225 225

 

172

Table 29 (cont’d)

 

 

S&P Bloc SDPU UNA
Constant 31 958*“ 3. 502“ . 144

(3.301 ) (1.423) (.139)
m 1.600 .719 .479***
SMDPL (1.148) (.537) (.063)
)3, 4.085 60.205*** 343*
incumbent (2.568) (3.431) (.198)
3, -.304*** -.075*** -.023
Urban (.064) (.028) (.003)
3., -.077 .466*** .022**
Education (.208) (.092) (.010)
3, -9.168*** -1.886* -.173*
Crimea (2.488) (1.099) (.103)
,66 -3.082* .344 -.110
East (1.678) (.729) (.074)
,3, -9.514*** 1347* .305***
West (1.605) (.705) (.070)
,3, 2.302 -.138 .577***
Cov Part (2.061) (.676) (.174)
Adjusted R2 .286 .627 .491
N 225 225 225

 

Dependent Variable: Proportion of the vote in PR at the SMD level.
The Central Region is the unit of comparison for the regional variables.
Standard Errors are in parentheses.

* ps.10
** ps.05
*** ps.01

173

partisans, although no comparable data exists to evaluate the hypothesis in

Russia at this time.

Conclusions

I have demonstrated the existence of an additional aspect of mixed
systems that should encourage multiparty competition. Candidate placement in
SMD affects vote outcomes in PR at the local level. This direct effect should
produce an indirect effect as well: political parties should strive to place
candidates in multiple SMD races in order to benefit from the “bonus” that can
accrue in PR.

In the empirical test, the variable SMDPL was found to be statistically
significant and positive in sign in every country and for a majority of the political
parties. Adding controls for rival hypotheses did not undermine confidence in the
existence of the Echo Effect. Incumbency was significant in many cases, but it
did not eliminate the effects of the variable SMDPL on PR vote outcomes. The
same was true for the proxies for popularity, education and urbanity.
Regionalism and covert partisanship had a more noticeable impact on the Echo
Effect. While these controls attenuated the Echo Effect in some cases, covert
partisanship was also shown to strengthen the results by recovering a lost effect
for two political parties in Ukraine.

Interaction between the PR and SMD components in mixed systems is at
the root of an incentive structure that encourages multiparty competition. While

the empirical evidence is clear, an important question remains: Do political

174

parties recognize these incentives and form their strategies based upon them?

This issue will be addressed in the final chapter.

175

   

Chapter 6

CONCLUSIONS — PARTY STRATEGIES IN MIXED SYSTEMS

The statistical analysis has established that interaction between the PR
and SMD components of mixed electoral systems substantially affects party
system development. While these effects have been shown to exist empirically,
are parties aware of them when they form district level campaign strategies?

Parties can adopt two main strategies to benefit from contamination
effects, both of which should contribute to multi-candidate races in the districts.
The first strategy focuses on the benefits accrued in PR through the Echo Effect.
By placing candidates in as many SMD races as resources allow, the party can
maximize the bonus received in PR. Moreover, it can win additional SMD seats
by participating in a greater proportion of SMD contests. This go-it-a/one
strategy is reasonable for parties that do not have competition for the same
subset of the electorate in district races or that believe they can win seats despite
the competition. If the party has a rival or rivals for the same constituency, an
alternate strategy may be more effective.

In the second strategy, resources are not the sole obstacle to participation
in SMD contests. Rather, the party may have at least one rival for the same
electoral constituency in the election. By running candidates in each district race,
the two parties could split the vote between each other, reducing the total
number of seats that each party obtains. Through a strategic alliance, however,

both parties could maximize the number of seats gained in the legislature.

176

The dilemma can be best illustrated with an example. In this hypothetical
case, the country has a non-compensatory mixed system and ten electoral
districts. There are three parties in competition. Parties A and B are
conservative, party C is liberal and the liberal-conservative cleavage is the only
dimension of partisan competition. When each party participates in all SMD
races in addition to the PR component, the system has the distribution of votes

and seats found in Table 30.

Table 30: Hypothetical Distribution of Votes and Seats with No
Cooperation

 

 

 

 

 

Party 01 02 D3 D4 D5 06 D7 DB 09 D10 Total
Seats
A PR 33 33 33 33 33 33 33 33 33 33 3
SMD 33 33 33 33 33 33 33 33 33 33 0
B PR 33 33 33 33 33 33 33 33 33 33 3
SMD 33 33 33 33 33 33 33 33 33 33 0
C PR 34 34 34 34 34 34 34 34 34 34 4
SMD 34 34 34 34 34 34 34 34 34 34 1 0

 

In the PR component, parties A and B receive three seats each and party
C receives four seats. Because parties A and B compete for the same electorate
(supporters of the conservative agenda), they split the local district vote. This is
not problematic in the PR portion because aggregate performance determines
seat allocation. The conservative parties obtain a majority of the PR seats.
However, because they split the vote in SMD races, the conservative parties

receive no SMD seats. Although conservatives enjoy the support of two-thirds of

177

the population, they receive a disproportionately low number of seats in the
legislature.

If parties A and B were to make a bargain and selectively withdraw
candidates from SMD competitions, the distribution of votes and seats could be
quite different (see Table 31). If Party A pulls its SMD candidates from Districts
1-5 and Party B agrees to withdraw its candidates from Districts 6-10, a strategic
agreement could result in the following distribution of votes and seats (if we
assume complete transferability of votes139 and that the Echo Effect is equal for

both parties”):

Table 31: Hypothetical Distribution of Votes and Seats with
Cooperation

 

Party D1 DZ D3 D4 D5 D6 D7 DB 09 D10 Total
Seats
A PR 23 23 23 23 23 43 43 43 43 43
SM D 0 0 0 0 0 66 66 66 66 66
8 PR 43 43 43 43 43 23 23 23 23 23
SMD 66 66 66 66 66 0 0 0 0 0

C PR 34 34 34 34 34 34 34 34 34 34
SMD 34 34 34 34 34 34 34 34 34 34

0-501030100

 

Although parties A and B lose some of their local PR vote by failing to run
candidates in five of the districts (due to the loss of the Echo Effect), they
substantially gain in the overall seat distribution. Instead of splitting the district
vote in SMD and losing all seats to party C, they consolidate the conservative

vote and split the SMD races between each other. While parties still gain from

 

‘39 If Party A does not run in a district, its supporters will choose the candidate from the other
conservative party, Party B, rather than voting for the liberal party, and vice versa.
14° The bonus received in PR by running a candidate in SMD is the same in each district.

178

running candidates in as many SMD races as possible, the benefits from
strategically withdrawing candidates outweigh the loss of votes in some local PR
districts. With this approach, parties can also engage in partial strategic
arrangements. This entails parties agreeing to cooperate in some regions of the
country, but competing against one another in other districts.

In choosing the optimal strategy, parties are faced with a high level of
uncertainty. The environment is generally more complex than the example
presented above. Often there are more than three parties in competition, parties
do not know their precise level of support in the districts, they cannot be sure that
voters will transfer allegiance between ideologically similar parties in cases of
strategic withdrawal, and they do not know the precise effect of withdrawal in
SMD on PR results. Therefore, the selection of either strategy includes an
element of risk.

Both strategies recognize the interactive effects of the mixed system. The
first strategy attempts to exploit the effect of SMD on PR despite the fact that it
could undermine SMD performance. The second strategy attempts to maximize
SMD performance and take as much advantage of the effect on PR as the party
can manage, constrained primarily by cooperative agreements. While these
strategies exist in theory, it is unclear how consciously parties choose their
strategies with contamination in mind. Because parties are often hesitant to
reveal their strategies, it is difficult to confirm whether or not they are aware of

contamination and how it fits into their calculations. However, based on

179

qualitative evidence, it is clear that both strategies have been utilized by parties

that are aware of contamination effects.

Go-It-Alone Strategy - Ukraine’s Rukh Party

Many political parties were aware of the potential benefits that could
accrue from contamination effects in Ukraine’s 1998 elections. lntemational
organizations that advised political parties urged like-minded parties to coalesce
so that resources could be pooled and parties could participate in as many SMD
races as possible. In addition, by fielding a single candidate in SMD races,
political movements with similar policy interests could avoid competing with one
another, splitting the vote in a district and losing the seat to a party with less
sympathetic policy aims.”1

Despite the efforts of advisory groups, coalition-building and resource-
pooling was not very successful in Ukraine. Two years before the elections,
Vyacheslav Chornovil, the leader of Rukh, indicated that the party would join
forces with other centrist parties (like the Liberal Party and People’s Democratic
Party) in the 1998 elections. This promise was uttered before the new rules were

in place and the incentive structure of the new system was clear.142

 

1“ Personal conversation with Fred Bradley, Advisor for the Parliamentary Development Project

in Ukraine on September 22, 1999. Mr. Bradley worked for the National Democratic Institute in
Ukraine at the time of the 1998 elections. NDI advised a number of democratic parties, including
Rukh, PDP, SDPU and the electoral blocs Reform and Order and Forward Ukraine. Mr. Bradley
indicated that the potential interaction between the components of the mixed system was
Bgesented to the parties. He suggested that it was logical that such interaction would take place.

The “Law on the Election of People's Deputies of Ukraine” was initially signed in September
1997. It was subsequently amended and challenged in the Constitutional Court. The law was
finally in place on March 25, 1998, four days before the actual election.

180

A member of Rukh authored the electoral rules that established a mixed
system in Ukraine and designed the rules with interactive effects in mind. The
author was aware that the simultaneous use of PR and SMD would generate
incentives like those described by the Echo and Investment Effects.143 Rather
than forming pre-electoral coalitions with similar parties, however, Rukh’s central
leadership decided to compete in as many SMD races as possible in order to
take advantage of the Echo Effect and to force smaller parties subsequently to
merge with Rukh under terms advantageous to the party. The party even
reneged on its earlier verbal agreement with other parties and mandated that
local organizations place candidates in as many districts as possible regardless
of the competition found in those districts (Tishchenko and Pikhovshek 1998,
282-287).144 Rukh nominated more candidates in SMD races than any other
pany.

Local Rukh organizations were more interested in coalition-building than
the national organization. The national leadership actively discouraged local
party organizations from cooperating with other parties that had similar policy
agendas. For example, Rukh’s organization in Cherkaska Oblast dutifully ran
candidates in most electoral districts as the national office demanded. This
created some difficulties in its relationships with other democratic-minded parties

and potentially damaged long-term cooperative possibilities in future elections

 

“3 Personal conversation with Olexander Lavrynovych- Deputy in the Verkhovna Rada from

Rukh, on October 11, 1999. He is also the author of the mixed system law for Ukraine.

1‘“ Rukh’s decision to run candidates in as many districts as possible was also addressed in a
personal conversation with Thomas Garrett, Director of the lntemational Republican Institute in
Ukraine. IRI advised parties during the 1998 campaign.

181

and the legislature.145 A similar breakdown in inter-party relationships occurred
in Lviv Oblast (Tishchenko and Pikhovshek 1998, 287).

By attempting to place candidates in as many districts as possible, Rukh
was acting in accordance with the system’s incentive structure and the go-it-
alone strategy. The proportion of Rukh-affiliated deputies elected to the Rada
increased from the previous election. While 4.8% of the deputies elected in 1994
were part of Rukh, 9.6% of the deputies elected in 1998 were members of Rukh
(Central Electoral Commission of Ukraine 1998). While we cannot know if
adopting another strategy would have further improved Rukh’s seat allocation,
Rukh’s strategy under the new mixed system improved its seat allocation

compared with the previous election.

Strategic Alliance - Russia’s SP8 and Yabloko

Just as major parties in Ukraine recognized potential benefits from
contamination effects, knowledge about the existence of these effects was
widespread among the largest parties in Russia.146 In the 1999 Duma elections,
some parties pursued the alternate strategy for candidate placement in SMD.
Rather than running candidates in as many districts as possible, some parties
chose to sacrifice potential Echo Effect benefits by making strategic

arrangements with other parties to potentially improve overall seat acquisition.

 

“5 Personal conversation with Thomas Garrett and Oksana Hasiuk of IRI-Ukraine on September
27, 1999.

“6 Personal conversation with Alexey Kuzmin, Research Director of the Institute of Humanities
and Political Studies on November 25, 1999. This position was also supported in conversations
with Nikolai Petrov, Scholar-in-Residence at the Carnegie Moscow Center on November 25,
1999, Boris Makarenko, Deputy General Director of the Center for Political Technologies on

182

The leadership of the Union of Right Forces (SPS) and Yabloko
recognized that they would share a similar electorate in most districts. While
they realized that the Echo Effect could improve PR performance, they further
calculated that placement of candidates in all districts was not feasible or always
advantageous. By placing candidates in all districts, financial and personnel
resources would be stretched to their limits and SPS and Yabloko candidates
would probably split the pro-reform vote. Consequently, the leadership decided
to cooperate in some regions by strategically placing candidates so that the two
parties would not directly compete.147

SP8 and Yabloko agreed to cooperate in Moscow and St. Petersburg.
Table 32 shows the distribution of SPS and Yabloko candidates in SMD races in
these cities. Districts 191-205 are in Moscow; districts 206-213 are in St.
Petersburg. In four cases, the parties withdrew candidates after registering them
for SMD races. Yabloko withdrew two candidates from Moscow; SPS pulled two
from St. Petersburg.148

While SPS and Yabloko candidates directly competed in seven districts,
they did not oppose each other in fifteen races. In some of the districts in which

they competed, one of the two parties fielded a candidate who was confident of

electoral victory; competition did not undermine the likelihood of winning the

 

November 25, 1999, and Lee Patterson, Director of the lntemational Republican Institute,
Moscow.

”7 Personal conversation with Viktor Sheynis, Deputy in the Duma from Yabloko and author of
Russia's mixed system on November 30, 1999. The strategic arrangement was also reported in
an on-Iine journal (polit.ru 1999). OVR and the KPRF also planned to strategically withdraw
candidates to improve their chances of winning district seats (Vandenko 1999). I do not know
their actual level of cooperation in the election, however.

"8 SPS withdrew a third candidate from St. Petersburg, but Yabloko did not have a registered
candidate in that district.

183

Table 32: Nomination Patterns for SPS and Yabloko in Moscow and
St. Petersburg Districts (1999)149

 

 

District SPS SPS Yabloko Yabloko Party
(Registered) (Actual) (Registered) (Actual) Finished
1 91 Yes Yes No No 2
1 92 Yes Yes No No 2
193 Yes Yes Yes N050 3
1 94 No No Yes Yes 3
1 95 No No Yes Yes 4
196"51 Yes Yes Yes Yes 4;3
1 97 No No Yes Yes 4
1 98 No No Yes Yes 2
1 99151 Yes Yes Yes Yes 6;3
200 Yes Yes Yes No150 3
201151 Yes Yes Yes Yes 3;1
202 Yes Yes No No 3
203 No No No No N/A
204 Yes Yes No No 2
..... 2 95 YesYesN°N°3
206151 Yes Yes Yes Yes 1 ;3
207151 Yes Yes Yes Yes 1;4
208"51 Yes Yes Yes Yes 32152
209 No No Yes Yes 1
210151 Yes Yes Yes Yes 4; 1 ‘53
21 1 Yes No"50 No No NIA
212 Yes No150 Yes Yes 1
21 3 Yes No150 Yes Yes 2

 

 

 

“9 Source: Rossiiskaya Gazeta, January 6, 2000 and [Online] http:l/www.rg.ru.
15° Change in status from registration to vote.

15‘ SPS and Yabloko competed against each other.

‘52 The SPS or Yabloko candidate may have been able to win with cooperation
‘53 Elections were invalidated.

184

 

district. In another case, cooperation could have yielded a victory, but both
parties failed to arrange a cooperative agreement (District 208).

Both SPS and Yabloko passed the 5% threshold in the PR component
and received 24 and 17 deputies, respectively. In addition, five SPS candidates
and four Yabloko candidates won SMD races. While Yabloko’s seat total
decreased from 1995 to 1999 (from 45 to 21), the total number of seats allocated

to pro-reform forces increased in the Duma.154

Conclusions
Summary of Findings

The analysis began with the underlying question: do mixed electoral rules
affect party system development in ways not predicted by Duvergerian
expectations? Maurice Duverger never considered how the combination of PR
and SMD that we call mixed systems would affect party system development.
His work, however, is consistently used to predict how party systems will evolve
based on the electoral rules.

Scholars influenced by Duverger have attempted to extend his work to
new electoral systems, but have erred in their interpretations of how electoral
institutions influence party system development under mixed rules. Based on
traditional interpretations of Duverger’s work, scholars and institutional designers
anticipated that mixed systems would behave in certain ways. Their

assumptions led to unanticipated consequences.

185

 

If the component parts of the mixed system were independent, PR should
distribute seats more proportionally than SMD. In fact, the PR component is
sometimes less proportional than SMD. The most egregious example of
disproportionate seat allocation in PR was in Russia’s 1995 elections when only
four parties cleared the 5% threshold, “wasting” many ballots. Further, standard
interpretations suggest that the threshold in PR should discourage parties that do
not have a reasonable chance of passing the threshold from contesting.
However, marginal parties participate in both PR and SMD in mixed systems.
SMD should also provide disincentives for multi-candidate competition in the
districts. Nevertheless, multi-candidate competition is common (see Table 1 in
the Introduction).

These unanticipated outcomes led to the empirical puzzle guiding the
dissertation: why have mixed systems failed to conform with Duvergerian
expectations? I proposed a possible solution: interaction between PR and SMD
undermines the incentive structure at the center of Duverger’s Propositions.
Contamination effects thus preclude the extension of Duvergerian expectations
to mixed systems and require us to rethink how we study countries using mixed
electoral rules.

I analyzed party systems because they show us, at least in part, who has
power in a political system. While there are many ways to assess party systems,
scholars consistently measure the number of parties in competition. Because

political parties and party leadership determine if and how they will participate in

 

15“ In June 2000, SPS and Yabloko signed an agreement to nominate a single slate of candidates
in the next Duma elections. While they did not agree to merge into a single party, they promised

186

elections, understanding the incentives that mixed systems provide to parties
allows us to predict how competition is likely to develop.

I first showed that marginal parties - parties that receive a relatively small
proportion of the vote — have incentives to participate in mixed systems. The
analysis in Chapter 3 showed that the mechanical effect of mixed systems
encourages participation by marginal parties. The proportionality profiles of
mixed systems were distinct from the standard SMD proportionality profile and do
not punish marginal parties as strongly as the mechanical effect in SMD.

Next, I demonstrated that parties committing to participation in mixed
systems tend to participate in both PR and SMD. The empirical evidence in
Chapter 4 demonstrated that parties tend to nominate candidates in both
components of the mixed system and that resources dictate, in part, how fully
parties participate in SMD. It provided evidence for the Investment Effect, which
contends that parties should compete in both PR and SMD because the benefits
of full participation in the mixed system outweigh the benefits of participation in
one component.

I further illustrated that there are electoral incentives for parties to run
candidates in as many SMD races as possible. The analysis in Chapter 5
provided evidence that parties can receive more votes in PR by participating in
SMD. The Echo Effect should further encourage parties to nominate candidates
in as many SMD races as possible, contingent upon resources and strategic

arrangements with other parties.

 

to cooperate even more closely in the next elections.

187

All of these incentives produced by the mixed system are based on the
interaction between PR and SMD and contradict Duvergerian expectations. In
this chapter, I provided evidence that not only do contamination effects exist in
theory, but political parties in the region form their strategies with the effects in
mind. Ukraine’s Rukh Party and Russia’s SPS and Yabloko understood that
optimal strategies in mixed systems were a response to the entire electoral
system, not only the PR or SMD components. They selected strategies that
would take advantage of the strategic incentives present under mixed electoral
rules.

Adaptation has not been the only response to the mixed system’s unique
incentive structure; some political actors have advocated change to electoral
rules with more familiar and predictable consequences. Russian President
Vladimir Putin has not directly attacked the mixed system, but has expressed
displeasure with the current party system and distribution of power.155 Putin has
shown his interest in instituting reforms to assert control over the regions156 and
may also pursue institutional changes to promote a two-party system. Ukrainian
President Leonid Kuchma, by contrast, has assailed the mixed system,
implicating it as the cause for increased fragmentation and uncertainty in the
Ukrainian parliament (Kuchma 2000). He sponsored a successful April 2000
referendum authorizing change in parliamentary structure and the electoral rules.

Political leaders have recognized the power of electoral rules in shaping partisan

 

‘55 Various news organizations have noted President Putin’s reported preference for a two-party

system (RIA Novosti 2000; Federal News Service 2000).
1 Putin instituted administrative reforms and proposed substantial changes in the Upper House’s
electoral rules. These changes were designed to limit the influence of regional governors.

188

competition and have realized that mixed systems have not delivered what

institutional designers promised.

Extensions of the Research

Although I focused on post-communist mixed systems that use non-
compensatory rules to distribute seats, the results are generalizable beyond
these countries. As I indicated in Chapter 1, there is substantial variation in
electoral rules among mixed systems that could affect strategies that parties
adopt. The analysis in the appendix provides preliminary evidence that that the
findings extend to countries using different allocation rules. It also provides
evidence that the effects extend to institutionalized party systems as well as new
democracies.157 The dissertation focused in detail on Lithuania, Russia and
Ukraine, but the analysis in the appendix shows that the findings also apply to
Armenia, Germany, Japan, New Zealand, Scotland and Wales. Although it is
important to more systematically address variation among mixed rules and party
system types to increase confidence in the validity of the conclusions, the initial
findings suggest that the results can be extended beyond the cases in the
dissertation.

The research also makes contributions to political science theory and the
practice of institutional design. By showing that Duvergerian expectations do not
apply to mixed systems, the research suggests that existing studies of mixed

systems may not be valid. Moreover, it provides insight into the conditions

 

‘57 See Mainwaring and Scully (1995) for the criteria defining institutionalized and inchoate party

systems.

189

under which Duverger’s Propositions may not operate as expected. The
research also follows in the tradition of other work about the interaction of
institutional factors in party system development, such as work on “coattail
effects” and other aspects of executive-legislative electoral linkages (Shugart and
Carey 1992; Jones 1995; Cox 1997), and provides the first extension of these
principles to mixed systems.

At the practical level, the findings provide useful information to institutional
designers. Mixed systems have been a popular choice because of their
perceived ability to facilitate compromise and solve problems of constitutional
engineering and domestic politics. Because institutional designers, like political
scientists, have assumed that PR and SMD should operate independently in
mixed systems, this research helps explain why the goals of institutional design
have not been realized in mixed systems. Further, the analysis points to features
of mixed systems that designers may wish to understand when selecting
electoral institutions.

This dissertation also opens new areas of research based on the idea of
contamination. By viewing mixed electoral systems as an entity rather than as
distinct components, the research identifies properties influencing party system
development that are not present in standard PR and SMD systems. This
analysis, however, is only the first step in understanding the potential
consequences of mixed electoral rules for party systems.

Expanding the study both spatially and longitudinally will permit additional

tests of the mixed system’s effects on parties. Variation in institutional rules

190

across mixed systems could alter party strategies. By increasing the number of
countries studied, the effects of this variation could be better understood.
Further, it is unclear if the effects found here are attributable to the inchoate party
systems and small number of elections that have taken place in these systems.
By collecting data over time, scholars can gain a better understanding of the
dynamic properties of the incentive structure and how parties adapt to the
incentives from election to election.

Investigating the strategies of political parties under mixed rules using
formal theory is another potential area of research. Parties may adopt different
strategies to take advantage of the rules over time and across systems.
Developing formal models of optimal election strategies that respond to variation
in the electoral rules will allow scholars to both understand and predict how
political parties behave in mixed systems.

By focusing on political parties, the dissertation has downplayed the
importance of individual voters. However, the strategic behavior of voters can
influence party decisions. Investigating how mixed systems, and institutional
variation among mixed systems, affects voter behavior will be another important
contribution to understanding how political competition develops under mixed
electoral rules.

This analysis has also ignored the potential influence of mixed systems on
legislative behavior. While studies of legislatures suggest that deputies elected
through PR are more likely to conform with party expectations than those elected

in SMD, this conventional wisdom probably does not apply to mixed systems

191

(see Haspel, Remington and Smith 1998). Because candidates can participate
in both PR and SMD races, it is likely that parties experience a diminished
capacity to command uniformity among affiliated candidates and legislators. This
could have serious implications for the ability to form legislative coalitions and
make policy and is thus an important question for future research.

Much work is yet to be done with data from mixed electoral systems.
Because the number of states using these systems continues to expand, there is
an increased need to understand the potential implications for party system

development.

Concluding Remarks

The rapid expansion of mixed systems in the 19905 no longer permits
them to be considered anomalies among the world’s electoral systems. Mixed
systems are used to elect legislatures at the local and national level in more than
thirty countries and are present on every inhabited continent. Mixed systems
have become a popular institutional choice in both new and long-standing
democracies because of their perceived advantages.

Under traditional assumptions of strategic incentives, mixed systems are a
perfect marriage of the benefits of PR and SMD systems. Mixed systems should
encourage parties to participate in elections as well as reduce the total number of
competitors to a manageable level. Unfortunately, the assumption that mixed
systems produce the “best of both worlds” by synthesizing the positive features

of PR and SMD systems is incorrect. As I demonstrated in the dissertation,

192

electoral rules have influenced party system development in unintended ways
due to contamination effects inherent in mixed systems. Mixed systems thus
seem more likely to spawn Sartori’s “bastard producing hybrids” (1997a, 75) than
the perfect balance of PR and SMD predicted by many institutional designers

and political scientists.

193

APPENDICES

194

APPENDIX A

DATA SOURCES

Data for the dissertation were obtained from many sources. The primary
countries under analysis are Lithuania, Russia and Ukraine. Lithuanian data
were compiled from the Lithuanian Election Commission web site and were
augmented with data forwarded by the Electoral Commission of Lithuania.
District-level results for PR and SMD are available as are biographies of the
candidates. These data were augmented by demographic information from the
Soviet census of 1989.

Data for the Russian Duma elections vary in quality. While data for PR
results at the SMD level are not available for 1993, they are available for 1995
and 1999. Russian election data were compiled from Central Electoral
Commission publications, Rossiiskaya Gazeta, and the 1989 Soviet census.
Ukrainian data are comparable to Russian data. They are compiled from similar
sources, including Central Electoral Commission publications and the 1989
Soviet Census.

Data from other mixed systems are used to complement the analysis of
Lithuania, Russia and Ukraine. Armenian electoral data were obtained from the
Central Electoral Commission web site and provide basic information on the
performance of parties in both portions of the system. While district results are
available for the SMD races, vote outcomes in PR are only accessible at the

aggregate level. In addition, information about candidates is restricted to their

195

party affiliation. Thus, Armenian results could be used for some cross-national
comparison, but they were inappropriate for inclusion when evaluating the
Investment and Echo Effects.

German data is more detailed than Armenian data and was obtained from
a scholar of German politics. District-level results of both PR and SMD
components from 1953 to the present are available, but incumbency information
and other pertinent data are not accessible for all elections. In addition, there are
minor problems with the quality of data. The raw election counts are not
available, only district level percentages. The degree to which these data have
been rounded is unclear, undermining their use in some tests. Because of data
quality issues, German data are used in the confirmatory analyses contained in
the appendices.

New Zealand data are also used in the appendices. These data were
obtained from the director of the New Zealand Electoral Commission. The data
include district-level results for PR and SMD races as well as information on the
characteristics of the candidates in SMD.

The primary source for Japanese data is Asahi Senkyo Taikan. Like the
above data, aggregate results are used in cross-national comparisons. More
detailed analysis of Japanese results is confined to the appendix. Data from
Scotland and Wales were obtained from the BBC. They include the regional PR

results and SMD results. PR results at the SMD level are not available.

196

APPENDIX B

ADDITIONAL EVIDENCE FOR THE MECHANICAL EFFECT

The evidence presented in Chapter 3 demonstrated that the mechanical
effect in mixed systems is more forgiving to marginal parties than the mechanical
effect in SMD systems. If we expand the analysis beyond the three countries in
the chapter, we find additional evidence that confirms the properties of the
mechanical effect in mixed systems.

Figure 9 shows the proportionality profile for Armenia. There is a single
outlier at 3.6 on the Y axis. The National Concord Party won a single seat in the
SMD component and did not participate in PR. Because it received a small
proportion of the vote in the total system but nevertheless received a seat, it is
over-represented in the legislature. The remaining parties are generally
clustered around unity on the Y-axis and all received less than thirty percent of
the total vote. Armenia’s proportionality profile shows that many parties received
proportional seat allocation in the legislature.

The Japanese proportionality profile conforms most closely to the Very
Late Rise profile among the countries under analysis (see Figure 10). Although
the shape is similar to the Very Late Rise profile, the break even point is lower on
the X axis. The Liberal Democratic Party and New Frontier Party drive the shape
of the profile. These parties received 36% and 28% of the votes, respectively,

and were over-represented. Most of the remaining parties received less than

197

% Seats/Vb Votes

 

P‘
o

.03
01
l

9’
o
l

N
01
l

.N
o
I

.3
U"
l

_s
O
I

.0
01
l

.0
o
l

 

Armenia

 

B
:-

Standard SMD

 

 

—

O

30
% Votes

Figure 9: Proportionality Profile for Armenia

198

60

% Seats/% Votes

 

3.0

2.5 ‘

N
o
l

.5
C
l

0.5 “

0.0 3

Standard SMD

Japan / /\

 

 

 

 

—

1 0 20 30 40 50 60
% Votes

Figure 10: Proportionality Profile for Japan

199

 

15% of the vote and were under-represented. The system is more forgiving to
small parties than SMD, but the functional form is similar to SMD.

New Zealand’s results diverge from a standard SMD system (see Figure
11). While a large number of parties received no seats in the legislature, those
that passed the barrier or won a SMD seat generally received a proportionate
number of seats. Act New Zealand was slightly over-represented, with a value of
1.35 on the Y-axis and United New Zealand was slightly underrepresented,
receiving a .57 on the Y-axis. The remaining parties that received seats received
within .10 of 1.00 on the Y-axis. Although small parties performed poorly in New
Zealand, those that received over 3% of the vote also received a proportionate
seat allocation in the legislature. This is well below the threshold of 5% for
representation in PR.

Scottish data further confirm expectations for the mixed system (see
Figure 12). Once again, we see that most parties are arrayed along the perfect
proportionality line. In addition, no party received over 40% of the total vote.
While the Greens are under-represented (with .43 on the Y axis), parties that
received seats generally received a proportionate number of seats.

As with all of the previous cases, Welsh election results also conform with
expectations (see Figure 13). The distribution of seats is generally proportionate
and small parties are not excluded from the legislature. All of the parties that
received seats in Wales were clustered along the perfect proportionality line and

the break-even point is far lower than we would expect in an SMD system.

200

 

% Seats/% Votes

3.0

2.5

.N
o

1.5

.3
O

0.5

0.0

 

 

Standard SMD

New Zealand /\
O

 

 

l l I l i F
0 1 0 20 30 40 50 60
% Votes

Figure 11: Proportionality Profile for New Zealand

201

% Seatsl% Votes

 

3.0

2.5 ‘

N
o
L

1.5“

1.0'1

.0
01
1

0.0 ‘

Scotland

Standard SMD

 

 

 

 

 

1 O 20 30 40
% Votes

50

Figure 12: Proportionality Profile for Scotland

202

60

% Seats/0A: Votes

3.0

2.5

N
o

1.5

.s
o

0.5

0.0

 

Standard SMD

 

 

 

 

Figure 13: Proportionality Profile for Wales

203

Wales 0
O O
l I r l l I
0 1 0 20 30 4O 50 60
% Votes

Figure 14 combines all of the countries in the analysis158 into a single
proportionality profile.“59 The profile crosses the perfect proportionality line under
10% on the X axis, diverging substantially from the Very Late Rise profile. Many
marginally performing parties receive a proportionate seat allocation in mixed
systems.

All of the evidence from individual countries as well as the combined
profile show that the mixed system’s mechanical effect is more forgiving to small
parties than a standard SMD system. Not only has the effect proved robust in
additional countries, but it is evidenced in fledgling and established democracies

as well as compensatory and non-compensatory systems.

 

‘58 That is, I combined the countries from the main analysis (Lithuania, Russia and Ukraine) with
those in the appendix (Armenia, Japan, New Zealand, Scotland and Wales) to generate a
ggoportionality profile for the entire sample.

Using the modified Chow Test, described in Chapter 4, I tested for structural stability. The
calculated F statistic is 1.94; the critical Fug) is 5.32. The null hypothesis of structural stability
cannot be rejected and the cross sections can be combined.

204

°/o Seats/%Votes

 

9
o

5”
01
l

9’
C
1

N
01
L

2.0 ‘

1.5A

1.0“

 

All Countries

0 O

 

Standard SMD

 

 

I

30
% Votes

40 5O 60

Figure 14: Proportionality Profile for All Countries

205

APPENDIX C

ADDITIONAL EVIDENCE FOR THE ECHO EFFECT

The evidence presented in Chapter 5 shows that the placement of
candidates in SMD positively affects PR vote outcomes at the local level. This
should encourage parties to place candidates in many SMD districts, promoting
multiparty competition. If we expand the analysis to other mixed systems, we

also find evidence of this effect.

Evidence from Germany

German data allow only a basic evaluation of the Echo Effect. Because
political parties quickly adapted to the mixed system and major political groups
fielded candidates in almost all districts even in the initial election, this analysis is
limited to the 1953 elections. Because this election was the first with separate
ballots under the mixed system, German parties arguably understood the
potential consequences of the mixed system as well as parties in the other
countries in this dissertation. Due to the lack of information about candidates
and districts in the elections, the model includes a single independent variable,

SMDPL.160 The model for Germany is:

 

16° The data available for the 1953 elections provide information about SMD performance as a
proportion of the vote received in a district. For the variable SMDPL, a district was coded "1' (a
candidate was placed in the district) if the party received more than zero in that SMD race. The

206

PRPER = a + BlsMDPL + s

The major German political parties (Christian Democratic Union (CDU),
Social Democratic Party (SPD) and Free Democratic Party (FDP)) cannot be
included in the analysis because they participated in almost all SMD contests.
There is adequate variation on the independent variable for four minor parties:
Bavarian Party (BP), Deutsche Partei (DP), DRP and Zentrum (Z).

The OLS results provide additional support for the Echo Effect Hypothesis.
The coefficient for SMDPL is statistically significant and positive in sign in three
of the four cases (see Table 33).

Among the parties under analysis, two (BP and 2) present a statistical
problem, however. In all districts where these parties failed to run a candidate
they received no PR votes.161 This substantially reduces variation on the
variables of interest.

The other two cases provide mixed results. Contamination is evident in
the DP results. The parameter estimate of SMDPL is statistically and
substantively significant for DP. The coefficient value of 8.4 for SMDPL indicates
that the party receives an 8.4% increase in the local PR vote in districts where it

fields an SMD candidate. SMDPL is not significant for DRP.

 

district was coded “0” if the party received no votes in the district. While the nature of the data
could introduce some coding error, the amount of error should be negligible.

‘61 It is possible that they received a very small proportion of the votes, but the data are rounded
to zero. This also creates the possibility that there were SMD candidates who received a very
small number of votes in the SMD component introducing some error due to coding into the

207

 

 

Table 33: OLS Results from Germany (1953)

 

BP DP DRP z
Constant 0.000 1.351 *** .970*** 0.000
(.156) (.284) (.150) (.151)

 

,8, 9364*“ 8.400*** .271 1569*”
SMDPL (.353) (.585) (.218) (.221)
Adjusted R2 .745 .460 .002 .170

N 242 242 242 242

 

Dependent Variable: Proportion of the vote in PR at the SMD level.
Standard Errors are in parentheses.

* ps.10
** ps.05
*** ps.01

I conducted additional diagnostic tests to assess the effect of outliers on
the analysis.162 There were many cases beyond three standard deviations from
the mean on the dependent variable for all four parties. Removing these cases
individually or as a block did not have a substantial impact on the results,

however.

Evidence from New Zealand
Data from New Zealand also allow us to perform a limited evaluation of

the Echo Effect. The model is:

 

analysis. The only data I currently have are district-level results. I have assumed that districts in
which a zero is recorded for SMD represents a district in which no SMD candidate participated.

208

 

 

PRPER = or + BISMDPL + leNC + 8

We can divide participation in the SMD component among parties in New
Zealand into two categories. In the first category are parties that participated in
almost all of the SMD contests. Six parties participated in virtually every
district.163 Among the remaining nineteen parties, only four participated in at
least 20% of the SMD contests. These are marginal parties in New Zealand
politics and only one of them gained a single seat in the legislature. While it is
preferable to have both major and minor parties in the analysis, lack of variation
on the independent variable only allows us to include minor parties.

The OLS results support the existence of the Echo Effect in New
Zealand’s mixed system. The coefficient for SMDPL was not statistically
significant in the Christian Coalition model, but it was significant in the other
cases. While the strength of the effect may seem substantively insignificant, we
must keep the relative size of the party in mind.

The bonus received by the McGillicuddy Serious Party for placing a
candidate in SMD is .08%. However, the party only received 29% of the total
PR vote. A bonus of .08% constitutes 28% of the total PR vote. This is a

substantial bonus relative to the party’s overall performance.

 

‘62 I did not conduct an analysis of collinearity for the German cases because there was only one
independent variable.

‘63 These parties are Act New Zealand, Alliance, Labour, Natural Law Party, New Zealand First,
and the New Zealand National Party.

209

 

 

Table 34: OLS Results from New Zealand (1996)

 

 

cc MS PG UNZ
Constant 4.197*** .219*** .198m .571***

(247) (020) (014) (058)
p, .010 .081*** .145*** .366"
SMDPL (.328) (.024) (.025) (.101)
A. 1393*"
INC (.169)
Adjusted R2 -.016 .145 .338 .883
N 85 85 85 85

 

 

Dependent variable is the PR vote at the district-level.
Standard Errors are in parentheses.

* (33.10
** p305
*** (33.01

The effect of the variable INCUMBENT could only be assessed in a single
case in New Zealand. Among the four parties in the analysis, only United New
Zealand had incumbent candidates in competition. The variable is statistically
significant and its effect on PR performance is greater than that of SMDPL.
However, after controlling for incumbency, the effect of SMDPL remains.

Diagnostic tests for the effects of collinearity and outliers revealed no
substantial impact on the OLS results. The results from New Zealand further

bolster our confidence in the existence and strength of the Echo Effect.

210

Evidence from Japan

The more detailed Japanese data permit a more sophisticated analysis.
Once again, the dependent variable is performance in PR at the SMD level.
There are three independent variables in the Japanese model. The first
independent variable is SMDPL. The second is INCUMBENT. The third
independent variable represents local popularity.

The variable POPULARITY is coded using data from Upper House
elections. The performance of parties in Upper House elections is coded at the
prefecture level. If parties indeed use their knowledge of regional popularity to
guide candidate placement, the effects of POPULARITY should sweep away the
results attributed to SMDPL. This model tests the existence of contamination
while controlling for incumbency and popularity.

The LDP was not included in the Japanese analysis because there was
inadequate variation in SMDPL (the LDP ran or recommended candidates in
each district). However, contamination effects were assessed for six other
parties. Table 35 shows the regression results for the initial model.

The results show that the coefficients for SMDPL and INCUMBENT are
statistically significant for all parties and the signs for all coefficients are positive.
While incumbency status affects performance in the PR component, candidate
placement also has an important impact on PR outcomes. Again, the coefficient
values represent the bonus received by parties in PR when the independent
variable is coded one. Thus, the Democratic Party of Japan (DPJ) received an

additional 7.9% of the PR vote in districts where it placed a candidate on the

211

Table 35: OLS Results for Japan (1996)

 

DPJ Jiyurengo NFP NHP SDPJ Shin.

 

Constant 11.859*** .732*** 22.166*** 1.285*** 6.129*** 1.832***
(.379) (.104) (.708) (.080) (.141) (.091)

fl, 7.9031m .802*** 5.737*** 3501*" 2.202*** 1.164***
SMDPL (.815) (.218) (.881) (.870) (.498) (.274)
[32 3.208*** 15.271*** 2.838*** 8.913*** 8.893*** 5.884***
Incumbent (.887) (1.198) (.750) (.781) (.581) (1.080)
Adjusted .471 .451 .238 .883 .505 .180
R

N 299 232 299 153 288 299

 

Dependent Variable: Proportion of the vote in PR at the SMD level.
Standard errors are in parentheses.

* ps.10
** ps.05
*** ps.01

SMD ballot. Moreover, it received a 3.2% bonus when the candidate was an
incumbent. While the coefficient for INCUMBENT was smaller than SMD for the
DPJ and New Frontier Party (NFP), it was larger for Jiyurengo, New Harbinger
Party (NHP), Social Democratic Party of Japan (SDPJ) and Shinshakaito (Shin).
These results show strong support for the Echo Effect.

When the variable POPULARITY is added, the effect of SMD and
INCUMBENT remains (see Table 36). The variable POPULARITY is a measure
of performance in Upper House elections and allows us to control for the main

rival hypothesis. Because of changes in the Japanese party system between the

212

Upper House elections in 1995 and Lower House elections in 1996, the variable
POPULARITY could only be assessed for two parties. Only the SDPJ and NF P
participated in both elections without division or merger.

Results for the SDPJ show that INCUMBENT still provides the largest
boost to PR performance (7.1% when a candidate is placed in SMD), but
candidate placement matters both statistically and substantively. Electoral
popularity also plays an important role, but provides a smaller bonus than the
other two factors. Results for the NFP are similar. Although the bonus for

incumbency is less than that for SMD candidate placement (2.0% versus 3.2%,

Table 36: OLS Results for Japan (1996) with POPULARITY

 

 

NFP SDPJ
Constant 4720*” .610*

(1.278) (.335)
,8, 3174*“ 2264*"
SMDPL (.687) (.347)
,62 1 .995*** 7052*"
Incumbent (.569) (.570)
,3; 643*” .331 ***
Popularity (.043) (.019)
AdjustedR2 .585 .758
N 299 286

 

 

Dependent Variable: Proportion of the vote in PR at the SMD level.
Standard errors are in parentheses.

* ps.10
** p305
*** p301

213

respectively), both remain statistically significant when the additional control for
popularity is added. The goodness-of-fit improves substantially for these models:
the adjusted R2 improves from .51 to .76 for SDJP and from .24 to .57 for NFP
when we add popularity.

The analysis of collinearity and outliers did not raise any problems. The
VIF was close to unity in all cases, indicating that there was only a small amount
of inefficiency introduced into the model due to collinearity. Removing the
outliers and reanalyzing the data did not substantially affect the results.

This evidence from Germany, New Zealand and Japan shows additional
support for the Echo Effect. Candidate placement in SMD affects PR
performance at the local level. In sum, there is evidence of the Echo Effect in
additional mixed systems, including compensatory systems and those with

institutionalized party systems.

214

BIBLIOGRAPHY

215

Bibliography

1989 Census of the USSR. 1995. CD-Rom Version 1.0. East View
Publications.

Abramson, Paul R., John H. Aldrich, Phil Paolino and David W. Rohde. 1995.
“Third-Party and Independent Candidates in American Politics: Wallace,
Anderson and Perot." Political Science Quarterly. 110:349-367.

Abramson, Paul R., John H. Aldrich and David W. Rohde. 1998. Change and
Continuity in the 1996 Elections. Washington, DC: CQ Press.

Aldrich, John. 1995. Why Parties? Chicago: University of Chicago Press.

Alvarez, R. Michael and Jonathan Nagler. 1997. “When Politics and Models
Collide: Estimating Models of Multiparty Elections.” Unpublished version of
forthcoming manuscript, American Journal of Political Science.

Ames, Barry. 1995a. “Electoral Rules, Constituency Pressures, and Pork Barrel:
Bases of Voting in the Brazilian Congress.” Journal of Politics. 57:324-343.

Ames, Barry. 1995b. “Electoral Strategy Under Open-List Proportional
Representation.” American Journal of Political Science. 39:406-433.

Asahi Shimbun Senkyo Honbu. 1997. Asahi Senkyo Taikan: Dai41kai
Shugiinsosenkyo, Dai17kai Sangiintsujyosenkyo. Tokyo: Asahi Shimbunsha.

BBC World Service. Scotland Elections ’99. [Online] Available
http:llnews1 .thdo.bbc.co.uklhi/englishlevents/
scotland_99lthe_scottish_parliamentlnewsidfl310000/310025.stm. May 28,
1999.

BBC World Service. Wales Elections ’99. [Online] Available
http:llnews1 .thdo.bbc.co.uklhi/english/events/
wales_99/newsl, May 27, 1999.

Barnes, Samuel H., Frank Grace, James K. Pollock, and Peter W. Sperlich.
1962. “The German Party System and the 1961 Federal Elections.”
American Political Science Review. 56:899-91 4.

Barrington, Lowell W. 1997. “The Geographic Component of Mass Attitudes in
Ukraine.” Post-Soviet Geography and Economics. 38:601-614.

216

Bawn, Kathleen. 1993. “The Logic of Institutional Preferences: German
Electoral Law as a Social Choice Outcome.” American Journal of Political
Science. 37:965-989.

Bilous, Artur. 1995. ll. Viborchisistemi: sviti Ukraina. [Part II. Electoral
Systems: the World and Ukraine]. Kyiv: lntemational Republican Institute.

Blais, André, and Louis Massicotte. 1996. “Electoral Systems.” In Lawrence
LeDuc, Richard G. Niemi and Pippa Norris (eds). Comparing Democracies:
Elections and Voting in Global Perspective Thousand Oaks, CA: Sage
Publications.

Blondel, Jean. 1968. “Party Systems and Patterns of Government in Western
Democracies.” Canadian Journal of Political Science. 1:180-203

Blondel, Jean. 1969. An Introduction to Comparative Government. London:
Weidenfield and Nicolson.

Bogdanor, V. 1983. “Introduction.” In V. Bogdanor and D. Butler (eds)
Democracy and Elections: Electoral Systems and their Consequences.
Cambridge: Cambridge University Press.

Brader, Ted and Joshua A. Tucker. 1998. “Congratulations, It’s a Party! The
Birth of Mass Political Parties in Russia, 1993-1996.” Paper presented at the
Midwest Political Science Association Conference.

Budge, Ian et al. 1997. The Politics of the New Europe. New York: Addison
Wesley Longman Limited.

Butler, D. 1982. “Electoral Systems.” In D. Butler, H. Penniman and Austin
Ranney (eds) Democracy at the Polls: A Comparative Study of Competitive
National Elections. Washington, DC: American Enterprise Institute.

Cain, B. E. 1978. “Strategic Voting in Britain.” American Journal of Political
Science. 22:639-655.

Calvert, Randall L. and John A. Ferejohn. 1983. “Coattail Voting in Recent
Presidential Elections.” American Political Science Review. 77:407-419.

Central Electoral Commission of Armenia. Election Results. [Online] Available
http:l/cec.freenet.am and http:l/www.elections.aml, June 9, 1999.

Central Electoral Commission of the Russian Federation. 1994. Rezul’taty

golosovaniya na vyborakh v Gosudarstvennuyu Dumu po odnomandatnym
izbiratel’nym okrugam. Mimeo.

217

Central Electoral Commission of the Russian Federation. 1995. Svedeniya o
zaregistrirovannykh kandidatov v deputaty Gosudarstvennoy Dum y
Federalnogo Sobraniya Rossiyskoy Federatsii po odnomandatnym
izbiratel’nym okrugam. Mimeo.

Central Electoral Commission of the Russian Federation. 1996. Vybory
deputatov Gosudarstvennoy Dumy. Moscow: Ves’ Mir.

Central Electoral Commission of the Russian Federation. 1998. Vybory v
zakonodateln ye (predstaviteln ye) organy gosydarstvennoi v/asti sub ’ektov
Rossiiskoi Federatsii 1995-1997. Moscow: Ves Mir.

Central Electoral Commission of the Russian Federation. 2000. Election Results
[Online]. http:l/www.fci.ru, January 20, 2000.

Central Electoral Commission of Ukraine. 1998. Parlament Ukraini: Vibori-98.
[The Parliament of Ukraine: Elections-98]. Volumes 1 and 2. Kyiv: Central
Electoral Commission.

Central Electoral Commission of Ukraine. 1999. Ukraina. Vibori-98: Dosvid,
prob/emi, perspektivi. [Ukraine Elections-98: Experience, Problems,
Perspectives] Kyiv: Central Electoral Commission.

Chhibber, Pradeep and Kenneth Kollman. 1998. “Party Aggregation and the
Number of Parties in India and the United States.” American Political Science
Review. 92:329-342.

Chhibber, Pradeep and Irfan Nooruddin. 2000. “Governing the Indian States:
How Parties [sic] Systems Count in India.” Paper presented at the Midwest
Political Science Association Conference.

Christensen, Raymond V. and Paul E. Johnson. 1995. “Toward a Context-Rich
Analysis of Electoral Systems: The Japanese Example.” American Journal of
Political Science. 39:575-598.

Clark, Terry D. 1995. “The Lithuanian Political Party System: A Case Study of
Democratic Consolidation.” East European Politics and Societies. 9:41 —62.

Clark, William A. 1999. “The Russian State Duma: 1993, 1995, 1999.” Problems
of Post-Communism. 4623-1 1.

Clark, Terry 0., Stacy J. Holscher and Lisa A. Hyland. 1999. “The LDLP Faction
in the Lithuanian Seimas, 1992-1996.” Nationalities Papers. 27:227-246.

Cox, Gary. 1997. Making Votes Count. New York: Cambridge University Press.

218

Cox, Gary and Matthew McCubbins. 1993. Legislative Leviathan. Berkeley:
University of California Press.

Craumer, Peter R. and James I. Clem. 1999. “Ukraine’s Emerging Electoral
Geography: A Regional Analysis of the 1998 Parliamentary Elections.” Post-
Soviet Geography and Economics. 40:1 -26.

D’Anieri, Paul. 2000. “Institutional Design and the Ineffectiveness of Ukrainian
and Russian Parliaments.” Paper presented at the Association for the Study
of Nationalities Conference.

Dahl, Robert (ed). 1966. Political Oppositions in Western Democracies. New
Haven, CT: Yale University Press.

Dalton, Russell J., Scott C. Flanagan and Paul Allen Beck (eds). 1984.
Electoral Change in Advanced Industrial Democracies. Princeton: Princeton
University Press.

Department of Press and Information of the Federal Republic of Germany. 1998.
Germaniya: Fakty. [Germany - Facts]. Hamburg: Societats Verlag.

Duverger, Maurice. 1954. Political Parties: Their Organization and Activity in the
Modern State. New York: Wiley.

Duverger, Maurice. 1986. “Duverger's Law: Thirty Years Later.” In Bernard
Grofman and Arend Lijphart (eds) Electoral Laws and their Political
Consequences. New York: Agathon Press.

Electoral Commission of Lithuania. Election Results. [Online]
http:llrc.Irs.It/rinkimail, December 9, 1998 and June 2, 1999.

Elklit, J. 1992. “The Best of Both Worlds? The Danish Electoral System 1915-20
in Comparative Perspective.” Electoral Studies. 11:189-205.

Elster, Jon, Claus Offe, and Ulrich K. Pruess. 1998. Institutional Design in Post-
communist Societies: Rebuilding the Ship at Sea. New York: Cambridge
University Press.

Epstein, Leon D. 1980. Political Parties in Western Democracies. New
Brunswick: Transaction Books.

Feddersen, Timothy J., Itai Sened and Stephen G. Wright. 1990. “Rational
Voting and Candidate Entry Under Plurality Rule.” American Journal of
Political Science. 34:1 005-1 016.

219

Federal News Service. 2000. “Remarks by Lilia Shevtsova, Carnegie Moscow
Center, at 8 Breakfast of the American Chamber of Commerce in Russia
(Aerostar Hotel, March 10, 2000” [Online] Available email:
davidjohnson@pop.erols.com, March 13, 2000.

Federal Returning Officer. 1991. The Election System in the Federal Republic of
Germany. Weisbaden: mimeo.

Ferejohn, John and Morris Fiorina. 1974. “The Paradox of Not Voting: A
Decision Theoretic Analysis.” American Political Science Review. 68:525-
536.

Fey, Mark. 1997. “Stability and Coordination in Duverger’s Law: A Formal Model
of Preelection Polls and Strategic Voting.” American Political Science
Review. 91 2135-147.

Fish, M. Steven. 1995. “The Advent of Multipartism in Russia, 1993-1995.”
Post-Soviet Affairs. 1 1 :340-383.

Fish, M. Steven. 1998. “Reversal and Erosion of Democratization in the
Postcommunist World.” Paper presented at the American Political Science
Association Conference.

Fisher, Steven L. 1973. “The Wasted Vote Thesis.” Comparative Politics.
52293-299.

Gebethner, Stanislaw. 1996. “Proportional Representation Versus Majoritarian
Systems: Free Elections and Political Parties in Poland, 1989-1991.” In
Arend Lijphart and Carlos H. Waisman, eds. Institutional Design in New
Democracies. Boulder: Westview Press.

Geddes, Barbara. 1991. “Paradigms and Sand Castles in Comparative Politics of
Developing Areas.” William Crotty (ed) Political Science: Looking To the
Future. Evanston: Northwestern University Press.

Grofman, Bernard, and Arend Lijphart (eds). 1984. Electoral Systems and Their
Political Consequences. New York: Agathon Press.

Grumm, J. 1958. “Theories of Electoral Systems.” Midwest Journal of Political
Science. 22357-376.

Gujarati, Damodar. 1995. Basic Econometrics. New York: McGraw-Hill, Inc.

Harris, Paul. NZ Election Results. Available e-mail: GET MAIL, DATE.

220

Haspel, Moshe, Thomas F. Remington and Steven S. Smith. 1998. “Electoral
Institutions and Party Cohesion in the Russian Duma.” Journal of Politics.
60:417-439.

Hastie, T. J. and R. J. Tibshirani. 1990. Generalized Additive Models. New
York: Chapman and Hall.

Herron, Erik S. and Misa Nishikawa. 2000. “Contamination Effects and the
Number of Political Parties in Mixed-Superposition Electoral Systems.”
Electoral Studies.

Institute of Postcommunist Societies. 1998. Pravo viboru: politiChni partii ta
viborChi bloki. [The Right to Vote: Political Parties and Electoral Blocs]. Kyiv:
Institute of Postcommunist Societies.

International Foundation of Election Systems. 1994. Vibori v Ukraini 1994 roku.
(The 1994 Ukrainian Elections]. Kyiv: IFES.

International Foundation of Election Systems. 1999. Electoral Legal Framework:
1999 Parliamentary Elections in Russia. Moscow: Human Rights Publishers.

lshiyama, John T. “Political Parties and Candidate Recruitment in Post-Soviet
Russian Politics.” Journal of Communist Studies and Transition Politics.
15:41-69.

Jesse, Eckhard. 1988. “Split-Voting in the Federal Republic of Germany: An
Analysis of the Federal Elections from 1953 to 1987.” Electoral Studies.
7:109-124.

Jones, Mark P. 1995. Electoral Laws and the Survival of Presidential
Democracies. Notre Dame: University of Notre Dame Press.

Jones, Mark P. 1999. “Electoral Laws and the Effective Number of Candidates
in Presidential Elections.” Journal of Politics. 61:171-184.

Judge, David and Gabriella llonszki. 1995. “Member-Constituency Linkages in
the Hungarian Parliament.” Legislative Studies Quarterly. 202161-176.

Kabashima, Ikuo. 1998. Seikenkotai To Yukensha No Taidohenyo. Tokyo:
Bokutakusha.

Kalmanson, A. A. and Yu. F. Zudinov. 1994. “Stanovlenie mnogopartiinosti v
postkommunisticheskoi Rossii.” [The Establishment of Multipartism in Post-
Communist Russia] In Yu. S. Novopashin (ed. ) Politicheskiye partii!
dvizheniya vostochnoi Yevropy. Moscow: Institute of Slavic and Balkan
Studies.

221

Kawato, Sadafumi. 1999. “Nidai Seitosei Ni Chikazuketa Shinsenkyoseido.”
Asahi Shimbun, October 11. 1996.

Kiewiet, D. Roderick and Matthew McCubbins. 1991. The Logic of Delegation.
Chicago: University of Chicago Press.

Kobayashi, Yoshiaki. 1991. Gendai Nihon No Senkyo. Tokyo, Japan: Tokyo
Daigaku Shuppankai.

Korgunyuk, Yu. G. 1999. Sovremennaya Rossiiskaya mnogopartiinost.
[Contemporary Russian Multipartism] Moscow: INDEM.

Krupavicius, Algis. 1998. “The Post-communist Transition and
lnstitutionalization of Lithuania’s Parties.” Political Studies. 46:465-491.

Kubicek, Paul. 2000. “Regional Polarisation in Ukraine: Public Opinion, Voting
and Legislative Behavior.” Europe-Asia Studies. 52:273-294.

Kuchma, Leonid. 2000. “About the Right and the Left.” [Online] Available
http:l/www.kuchma.org/pravlevlpravlev.html, June 12, 2000.

Lakatos, Imre. 1978. The Methodology of Scientific Research Programmes.
Cambridge: Cambridge University Press.

Lakeman, E and JD. Lambert. 1955. Voting in Democracies. London: Faber
and Faber.

Lancaster, Thomas D. and W. David Patterson. 1990. “Comparative Pork Barrel
Politics: Perceptions from the West German Bundestag.” Comparative
Political Studies. 22: 458-477.

LaPanmbara, Joseph and Myron Weiner. 1966. Political Parties and Political
Development. Princeton: Princeton University Press.

Law on Elections to the Seimas. [Online] Available: http:llwww.lrs.ltl, April 4,
2000.

Leys, Colin. 1959. “Models, Theories and the Theory of Political Parties.”
Political Studies. 7: 1 27-1 46.

Lijphart, Arend. 1984. Democracies: Patterns of Majoritarian and Consensus
Government in Twenty-One Countries. New Haven: Yale University Press.

222

Lijphart, Arend. 1986. “Degrees of Proportionality of Proportional Representation
Formulas.” In Bernard Grofman and Arend Lijphart Electoral Laws and Their
Political Consequences. New York: Agathon Press, Inc.

Lijphart, Arend. 1994. Electoral Systems and Party Systems: A Study of Twenty-
Seven Democracies, 1945—1990. New York: Oxford University Press.

Lijphart, Arend and Carlos H. Waisman (eds). 1996. Institutional Design in New
Democracies. Boulder: Westview Press.

Lipset, Seymour Martin and Stein Rokkan. 1967. Party Systems and Voter
Alignments: Cross-National Perspectives. New York: Free Press.

Maguire, M. 1983. “Is There Still Persistence? Electoral Change in Western
Europe, 1948-1979.” In Hans Daalder and Peter Mair (eds) Western
European Party Systems: Continuity and Change. London: Sage.

Mainwaring, Scott and Timothy R. Scully. 1995. Building Democratic
Institutions: Party Systems in Latin America. Stanford: Stanford University
Press.

Massicotte, Louis and André Blais. 1999. “Mixed Electoral Systems: A
Conceptual and Empirical Survey.” Electoral Studies. 18: 341 -366.

McFauI, Michael, Nikolai Petrov and Andrey Ryabov with Elizabeth Reisch (eds).
1999. Primer on Russia’s 1999 Duma Elections. Washington: Carnegie
Endowment for International Peace.

Miyake, Ichiro. 1989. Tohyo Kodo. Tokyo: Tokyo Daigaku Shuppankai.

Mizusaki, Tokifumi and Hiroki Mori. 1998. “Tokuhyo Data Kara Mita Heiritsusei
No Mekanizumu.” Senkyo Kenkyu 1998.

Moe, Terry M. and Michael Caldwell. 1994. “The Institutional Foundations of
Democratic Government: A Comparison of Presidential and Parliamentary
Systems.” Journal of Institutional and Theoretical Economics. 150:171-195.

Moraski, Byron and Gerhard Lowenberg. 1999. “The Effect of Electoral
Thresholds on the Revival of Former Communist Parties in East-Central
Europe.” Journal of Politics. 61 :151-170.

Moser, Robert G. 1997. “The Impact of Parliamentary Electoral Systems in
Russia.” Post-Soviet Affairs. 13:284-302.

Moser, Robert G. 1999a. “Electoral Systems and the Number of Parties in
Postcommunist States.” World Politics. 51: 359-384.

223

Moser, Robert G. 1999b. “Independents and Party Formation: Elite Partisanship
as an Intervening Variable in Russian Politics.” Comparative Politics.
31 :147-165.

Neto, Octavio Amorim and Gary W. Cox. 1997. “Electoral Institutions, Cleavage
Structures, and the Number of Parties.” American Journal of Political Science.
41 :149-1 74.

Nie, Norman, Sidney Verba and John Petrocik. 1976. The Changing American
Voter. Cambridge: Harvard University Press.

Nohlen, D. 1984. “Two Incompatible Principles of Representation.” In Arend
Lijphart and Bernard Grofman (eds) Choosing an Electoral System. New
York: Praeger.

NAU. 1999. NormativniAkti Ukraini. [CD-Rom]. Kyiv: NAU.

North, Douglass C. 1990. Institutions, Institutional Change and Economic
Performance. Cambridge: Cambridge University Press.

Ordeshook, Peter and Olga Shvetsova. 1994. “Ethnic Heterogeneity, District
Magnitude and the Number of Parties.” American Journal of Political
Science. 38:100-123.

Palfrey, Thomas. 1989. “A Mathematical Proof of Duverger’s Law.” In Peter C.
Ordeshook, ed., Models of Strategic Choice in Politics. Ann Arbor: University
of Michigan Press.

Pedersen, MN. 1983. “Changing Patterns of Electoral Volatility in Eur0pean
Party Systems; 1948-1977: Explorations in Explanation.” In Hans Daalder
and Peter Mair (eds) Western European Party Systems: Continuity and
Change. London: Sage.

Petrenko, Viktor, Olga Mitina and Ruth Brown. 1995. “The Semantic Space of
Russian Political Parties on a Federal and Regional Level.” Europe-Asia
Studies. 47:835-857.

Pikhovshchek, Vyacheslav, Svitlana Kononchuk, Yuliya Tishchenko and Vasil
Ganzha. 1999. Politichniy protsess v Ukraini: Suchasni tendentsii ta
inston'chniy kontekst. [The Political Process in Ukraine: Current Tendencies
and Historical Context] Kyiv: Agency “Ukraine.”

Polit.ru. Pravaya odnomandatnaya koalitsiya po-prezhnemu zhiznesposobna.

[The Right Single Member District Coalition is Viable as Before] [Online]
Available http:llwww.politm/documents/154869.html, December 16, 1999.

224

 

Rae

Ree

Rel

RI/

RiI<

Rc

Rt

 

 

Rae, Douglas W. 1971. The Political Consequences of Electoral Laws. New
Haven: Yale University Press.

Reed, Steven R. 1990. “Structure and Behavior: Extending Duverger’s Law to
the Japanese Case.” British Journal of Political Science. 20:335-356.

. 1999. “Strategic voting in the 1996 Japanese general election.”
Comparative Political Studies. 32:257-270.

Remington, Thomas F. and Steven S. Smith. 1996. “Political Goals, Institutional
Context, and the Choice of an Electoral System: The Russian Parliamentary
Election Law.” American Journal of Political Science. 40: 1253-1279.

RIA Novosti. 2000. “Kremlin Returns to Once-Malfunctioned [sic] Idea of
Bipartisanship to Put an End to Political Turmoil.” [Online] Available email:
davidjohnson@pop.eroIs.com, April 13, 2000.

Riker, William H. 1982. “The Two-Party System and Duverger’s Law: An Essay
on the History of Political Science.” American Political Science Review.
76:753-766.

. 1986. “Duverger’s Law Revisited.” In Bernard Grofman and Arend
Lijphart, eds. Electoral Laws and their Political Consequences. New York:
Agathon Press.

Rokkan, Stein. 1968. “The Growth and Structuring of Mass Politics in Smaller
European Democracies.” Comparative Studies in Society and History.
10:173-210.

Rose, Richard. 1983. “Elections and Electoral Systems: Choices and
Alternatives.” In V. Bogdanor and D. Butler (eds) Democracy and Elections:
Electoral Systems and their Consequences. Cambridge: Cambridge
University Press.

Rose, Richard, Evgeny Tikhomorov and William Mishler. 1997. “Understanding
Multi-Party Choice: The 1995 Duma Election.” Europe-Asia Studies. 49:799-
823.

Rossiiskaya Gazeta. January 17, 1996.

Rossiiskaya Gazeta. January 24, 1996.

Rossiiskaya Gazeta. January 6, 2000.

225

Rossiiskaya Gazeta. Registration of Candidates. [Online] http:llwww.rg.ru,
December 12, 1999.

Rossiiskiye regiony nakanunye vyborov-95. 1995. Moscow: Yuridicheskaya
Literatura.

Sartori, Giovanni. 1968. “Political Development and Political Engineering.” In
J.D. Montgomery and AC. Hirschman (eds) Public Policy. Cambridge:
Cambridge University Press.

. 1976. Parties and Party Systems. New York: Cambridge University
Press.

. 1986. “The Influence of Electoral Systems: Faulty Laws or Faulty
Method?” In Bernard Grofman and Arend Lijphart (eds). Electoral Systems
and Their Political Consequences. New York: Agathon Press.

. 1997a. Comparative Constitutional Engineering. New York: New York
University Press.

. 1997b. “Electoral Reform and Change in Party Systems.” Presented at
the International Forum for Democratic Studies. Washington, DC: National
Endowment for Democracy.

Schlesinger, Joseph A. 1991. Political Parties and the Winning of Office. Ann
Arbor: University of Michigan Press.

Shepsle, Kenneth A. 1986. “Institutional Equilibrium and Equilibrium
Institutions.” In Herbert F. Weisberg (ed.). Political Science: The Science of
Politics. New York: Agathon Press.

Sheynis, Viktor. 1999. Za chestnye vybory. [For Honest Elections] Moscow:
Center of Political and Economic Research.

Shively, W. Phillips. 1970. “The Elusive Psychological Factor: A Test for the
Impact of Electoral Systems on Voters” Behavior.” Comparative Politics.
32115-125.

Shugart, Matthew S. 1994. “Minorities Represented and Unrepresented.” In
Wilma Rule and Joseph Zimmerman (eds) Electoral Systems in Comparative
Perspective: Their Influence on Women and Minorities. Westport:
Greenwood.

Shugart, Matthew, and John M. Carey. 1992. Presidents and Assemblies:

Constitutional Design and Electoral Dynamics. Cambridge: Cambridge
University Press.

226

 

Sh

Si:

Sli

Sn

 

Shvetsova, Olga. 1999. “A Survey of Post-Communist Electoral Institutions:
1990-1998.” Electoral Studies. 18: 397-409.

Siavelis, Peter and Arturo Valenzuela. 1994. “Electoral Engineering and
Democratic Stability: The Legacy of Authoritarian Rule in Chile.” In Arend
Lijphart and Carlos H. Waisman, eds. Institutional Design in New
Democracies. Boulder: Westview Press.

Slider, Darrell. 1996. “Elections to Russia’s Regional Assemblies.” Post-Soviet
Affairs. 12:243-264.

Smyth, Regina. 1997. Building Democracy by Winning Votes? Candidates and
Parties in Transitional Russia. Dissertation: Duke University.

Spafford, D. 1972. “Electoral Systems and Voters’ Behavior: Comment and a
Further Test.” Comparative Politics. 52129-134.

Taagepera, Rein and Markku Laakso. 1980. “Proportionality Profiles of West
European Electoral Systems.” European Journal of Political Research.
8:423-446.

Taagepera, Rein, and Matthew Soberg Shugart. 1989. Seats and Votes: The
Seats and Determinants of Electoral Systems. New Haven, CT: Yale
University Press.

Tishchenko, Yu. and V. Pikhovshek. 1998. Vibori-98: Yak I kogo mi obirali.
[Elections-98: How and Who We Elected]. Kyiv: Agency “Ukraine.”

Turner, Arthur W. 1993. “Postauthoritarian Elections.” Comparative Political
Studies. 26:330-349.

Vandenko, Igor. 1999. “KPRF - OVR soyuz nerushimiy na printsipakh
vsevlastiya.” [The KPRF — OVR Union is Unbroken in the Principles of Full
Power] Novye Izvestiya. December 9, 1999.

Vasilyev, VI. and A.E. Postnikov. 1995. Vybory v Gosudarstvennuyu Dumu.
[Elections to the State Duma] Moscow: BEK.

Wada, Junichiro. 1995. “Shosenkyoku Hireidaihyo Heiritsusei Ni Kansuru
Geimuronteki Ichikousatsu.” Senkyo Kenkyu. No.10 1995.

Wattenberg, Martin P. 1990. The Decline of American Political Parties 1952-
1988. Cambridge: Harvard University Press.

227

White, Stephen, Richard Rose and Ian McAIlister. 1997. How Russia Votes.
Chatham: Chatham House Publishers.

White, Stephen, Matthew Wyman and Olga Kryshtanovskaya. 1995. “Parties
and Politics in Post-Communist Russia.” Communist and Post-Communist
Studies. 28:183-202.

White, Stephen, Matthew Wyman and Sarah Oates. 1997. “Parties and Voters
in the 1995 Russian Duma Elections." Europe-Asia Studies. 49:767-798.

Whitten, Guy D. and Harvey D. Palmer. 1996. “Heightening Comparativists’
Concern for Model Choice: Voting Behavior in Great Britain and the
Netherlands.” American Journal of Political Science. 40:231-260.

Wildavsky, Aaron. 1959. “A Methodological Critique of Duverger’s Political
Parties.” Journal of Politics. 21 :303-318.

Wyman, Matthew, Bill Miller, Stephen White and Paul Heywood. 1994. “The
Russian Elections of December 1993.” Electoral Studies. 13:254-271.

228

 

 

II