D O E S P R O T E S T M AT T E R ?
Parties’ Rhetorical Reactions to Protesters’ Claims in
Comparative Perspective
daniel bischof
Doctor of Philosophy (Ph.D.)
Department of Politics & International Relations
University of Leicester
01.2016
Daniel Bischof: Does Protest Matter?, Parties’ Rhetorical Reactions to Protesters’
c 01.2016
Claims in Comparative Perspective, ABSTRACT
In my PhD thesis I disentangle the rhetorical reactions of political parties to
public opinion and protest. Previous research on political responsiveness of
parties pre-eminently views the relation between public opinion polls and
party agendas as the key feature of responsiveness (Miller and Stokes 1963;
Soroka and Wlezien 2010; Stimson, Mackuen, and Erikson 1995; Burstein
1998; Adams et al. 2006; Ezrow 2010). Yet, taking to the street has become
an ever more important toolbox to articulate popular grievances. Social
movements have emerged throughout Western advanced democracies and
transformed the political landscape in Europe. Also new political parties
emerged from these social movements – such as Green parties and the
New Left (Kitschelt 1994; Kriesi et al. 1992). It is, therefore, surprising that
the link between political parties and protest has largely remained a lacuna in social movement studies and the literature on party competition.
My thesis is a first attempt to address this gap. I argue that besides public
opinion polls, political protest will affect party position taking. I hypothesise that growing protest leads to polarisation of party systems. While all
parties will increase their attention to the issue at stake during protest in
an effort to secure votes and/or office, they respond differently to protest
contingent on how their ideology relates to protesters’ demands. Furthermore, the success of protest depends on its support by the public at large. I
test my theoretical framework using a new and unique data-set containing
party positions on nuclear energy – revealed in interviews, press statements
and press conferences – of 67 parties across 12 Western Democracies. I run
time-series-cross-sectional models to test my theoretical arguments. Traditionally susceptible to responding to anti-nuclear protest, parties of the left
understand increased protest as a window of opportunity to influence policy debate in their favour, while right-wing parties perceive protest as a
threat to their ideological position on the usage of nuclear energy. Furthermore, I aim to understand in my last empirical chapter whether protest
also affects parties’ issue emphasis in manifestos. To this end, I use the
Comparative Manifesto Project data and protest data on 18 democracies
iii
across 15 years to estimate how parties adapted their issue emphasis to
postmaterialist issues. While I again find a significant influence of protest
on parties’ issue emphasis, the polarisation hypothesis does not find support in my last chapter. Finally, the instrumental variable models used in
this last chapter suggest that the causal direction runs from protest to parties’ position.
iv
P U B L I C AT I O N S
This thesis was written paper-based. All papers included in this thesis have
been presented at international conferences, some are under review at the
time of the viva and one paper has been published.
Publications from & during my PhD studies:
• Bischof, Daniel. 2015. “Towards a Renewal of the Niche Party Concept: Parties, Market Shares and Condensed Offers.” Party Politics:
forthcoming.
→ This paper was originally part of chapter 2 of this thesis. It was
also nominated for the best paper in comparative politics at the MPSA
2014 (Kellogg/Notre Dame paper award).
• Bischof, Daniel, and Simon Fink. 2015. “Repression as a DoubleEdged Sword: Resilient Monarchs, Repression and Revolution in the
Arab World.” Swiss Political Science Review 21(3): 377–95.
• Saalfeld, Thomas, and Daniel Bischof. 2013. “Minority-Ethnic MPs
and the Substantive Representation of Minority Interests in the House
of Commons, 2005-2011.” Parliamentary Affairs 66(2): 305–28.
Chapters under review at time of submission:
• Bischof, Daniel. “How Party Rhetoric after the Fukushima Accident
Reflects The Classical Left-Right Divide of Politics.” Environmental Politics: under review.
→ This is the descriptive part of Chapter 4.
• Bischof, Daniel. “Do Parties Walk like They Talk? Parties between
Rhetorical Positions and Substantive Policy Output.” Party Politics:
plan to submit in 02/2016.
→ This is the second (regression) part of Chapter 4.
• Bischof, Daniel.“Do Parties Adapt to Protest? Party Rhetoric on Nuclear Energy after the Fukushima accident.” American Journal of Politi-
v
cal Science: plan to submit in 03/2016.
→ This is chapter 5.
Furthermore, all empirical chapters in this thesis have been presented at
various international conferences:
• Chapter 4: Workshop at University of Leicester organized by ResponsiveGov project, Leicester (2014); ECPR General Conference, 2014 (Glasgow); EPOP, 2014 (Glasgow).
• Chapter 5: SPSA General Conference, 2014 (New Orleans); MPSA,
2015 (Chicago).
• Chapter 6: Workshop on Dynamics in Party Communication, 2015
(Vienna).
• Chapter 7: Workshop on Citizens Resilience in Times of Crises, 2015
(Florence).
Further papers under review written and submitted during my PhD studies:
• Senninger, Roman & Daniel Bischof. “Working in Unison - Political
Parties and Policy Issue Transfer in the Multi-Level Space.” Journal of
Politics: under review.
• Bischof, Daniel. “New Figure Schemes for Stata: plotplain & plottig.”
Stata Journal: under review.
vi
ACKNOWLEDGEMENTS
During the process of writing a thesis I met a lot of people. Thus, I hope
that everybody who expects to be named in the acknowledgements section
receive their fair share of attention.
I first have to acknowledge that this PhD thesis could have never been
written without funding from the European Research Council within the
ResponsiveGov project (ERC Grant 284277).
The first people to be acknowledged are those supervising me during
this three year period. I want to thank my supervisors Laura Morales and
Simona Guerra. Laura has taught me attention to detail, which is up until
today certainly not my strength and remains work in progress. She also
spotted each and every single issue coming with my theoretical approach,
my assumptions, and the conclusions I draw from my analysis. Simona
also helped to develop my results into a far stronger thesis than it would
have been without her. Without working for Thomas Saalfeld and his support during the application for the PhD position this thesis would also
not have been possible. Thank you Laura, Simona, and Thomas. Furthermore, Daniele Caramani put faith in me in offering me a longterm senior
researcher position even before I had submitted my PhD thesis. From the
very beginning of my appointment at his chair Daniele has been a calm,
understanding and supporting boss. Thank you as well, Daniele.
My co-authors were a further and very important source of knowledge
and support. Kaare Strøm was kind enough to support my application for
a visiting position at UC San Diego. Working with him has been a pleasure
throughout and hopefully we will continue our professional relationship as
well as the unfortunately unfrequent meetings for dinner. I also met Patrick
Dumont during my time at UCSD, who always came up with inspiring
ideas during meetings. Another special line needs to be dropped for my by
now permanent co-author Roman Senninger. Roman had to listen to a lot
of ups and downs during the process of writing my thesis, decisions about
my academic future, and methods considerations. I hope we will work on
vii
all the future projects we talked about during countless Skype meetings
across so many time zones.
Countless persons commented on the papers included in this thesis. During presentation at MPSA, ECPR, EPOP and other conferences colleagues
repeatedly listened to my thoughts on the impact of protest on political parties in Western democracies. Thereby, commentators and discussants gave
me wonderful and very helpful feedback. I tried to include most of their
comments and hopefully I did not miss to address any of the suggestions
made during my presentations. For their comments I specifically want to
thank Marco Giugni, Martin Hansen, Stefaan Walgrave, Sara Hobolt, Zac
Greene, Hanspeter Kriesi, Gijs Schumacher, Luca Bernardi, Maarja Lühiste,
Christine Arnold, Stuart Soroka, Prof. Hans Keman, Prof. Dick Katz, the
University of Leicester, the 3-PO cluster, the Department of Politics & International Relations, and especially Chris Wallace. I also want to thank
all participants of the ECPR summer school on parties, and especially the
organiser Prof. Ferdinand Müller-Rommel.
Special thanks are necessary for Markus Wagner and Thomas Meyer.
Both of them read more than one of the papers included in this thesis.
They both inspired me to re-think the concept of niche parties. Originally
this thesis was meant to be strongly linked to the niche party concept, but
like many other PhD theses this thesis more and more drifted away from
its original ideas. However, thanks to their inspiration I published a paper
in Party Politics which has had a profound impact on my young academic
career.
The final lines in the acknowledgements sections are the most important
ones. I want to dedicate these lines to the three most important people
in my life. First, my girlfriend Johanna Schönhöfer. Johanna has made this
thesis as it is possible. Her role in the process of my academic career cannot
be overestimated. She is not only a driving force of my train of thoughts,
but has also read each and every single line of this thesis and corrected so
many drafts that I will never be able to make it up to her. Besides these
rather professional facts she is simply gorgeous and astonishing.
Second, as many of my parents’ friends suggest, my parents had to go
through difficult times with me. I am extremely proud of my parents, their
dedication to each and everything they do in their lives, for not being boring snobs, for believing in me and especially for giving me a lot of freedom
viii
while growing up. While it is logically necessary that I would never have
been able to write this thesis without their existence, they are much more
to my life than this simple and logical conclusion. They always gave me reason to question the world, to challenge rules and to properly argue about a
fact before falling prey to simple conclusions. Georg is catalyser and faculty
of reason in a single person. Bärbel is certainly the emotional institution in
our family, but as such also the one person teaching us the importance of
emotions for our perceptions of the world over and over again. Georg: I
am sorry that football and I never were a successful combination. Bärbel: I
am overly proud of you, what you achieved during your life even though
people arranged obstacle after obstacle for you.
ix
CONTENTS
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvi
I
Parties and Protest
17
1
introduction
1.1 What is explained . .
1.2 How is it explained .
1.3 Synopsis of the thesis
1.4 Relevance . . . . . . .
.
.
.
.
18
20
21
23
24
.
.
.
.
.
25
25
28
35
37
41
2
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
rhetorical representation: the state of the art
2.1 The game of representation: the responsive politician .
2.2 What we know about responsiveness . . . . . . . . . . .
2.3 Parties in modern democracies . . . . . . . . . . . . . .
2.4 How protest matters for political parties . . . . . . . . .
2.5 Moderation of the protrusion of protest . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
44
II Measuring Rhetorical Responsiveness
3
data & methods
3.1 The ResponsiveGov data . . . . . . . . . . . . . . . . . .
3.2 Party selection . . . . . . . . . . . . . . . . . . . . . . . .
3.3 Using rhetorical statements to measure party positions
3.3.1 Recoding positions . . . . . . . . . . . . . . . . .
3.3.2 Parties’ rhetorical positions . . . . . . . . . . . .
3.4 Measuring the protrusion of protests . . . . . . . . . . .
3.5 Public opinion & controls . . . . . . . . . . . . . . . . .
x
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
45
45
48
49
51
53
55
59
contents
61
III Parties and Protest: An Empirical Analysis
4
5
6
7
is it worth the effort? the means and ends of parties’ rhetorical activities
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2 Parties between symbolism & policy-making . . . . . . . . . .
4.3 Rhetorical activities, manifesto positions & policy . . . . . . .
4.3.1 Linking parties’ rhetorical activities & policy outputs .
4.4 Modelling specifications . . . . . . . . . . . . . . . . . . . . . .
4.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.5.1 Congruence between rhetorical activities & policy outputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
who represents protesters – parties as crucial
keepers to the political arena
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . .
5.2 Protest, public opinion & party ideology . . . . . . .
5.3 How protest matters – parties between policy & vote
5.4 Modelling specifications . . . . . . . . . . . . . . . . .
5.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . .
80
84
gate.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
does timing matter? the electoral cycle & parties’
rhetorical reactions to public claims
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.2 The electoral cycle and responsiveness . . . . . . . . . . . . .
6.3 How timing matters . . . . . . . . . . . . . . . . . . . . . . . .
6.4 Model specifications & measuring the electoral cycle . . . .
6.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
does protest matter in the long-run?
alism to postmaterialism, 1981-1996
7.1 Introduction . . . . . . . . . . . . . . . . . .
7.2 Parties, postmaterialism & issue emphasis .
7.3 From materialism to postmaterialism . . .
62
62
64
66
70
72
74
.
.
.
.
.
.
86
86
87
90
94
97
106
.
.
.
.
.
.
108
108
109
111
113
115
121
from materi123
. . . . . . . . . . . 123
. . . . . . . . . . . 125
. . . . . . . . . . . 127
xi
contents
7.4
7.5
7.6
Data & methods . . . . . . . . . . . . . . . . . . . . . . . . . .
7.4.1 Modelling strategy . . . . . . . . . . . . . . . . . . . .
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7.5.1 The causal relationship between protest & party manifestos . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. 129
. 136
. 137
. 142
. 147
149
IV Parties and Protest: What have we learned?
8
conclusion
150
8.1 How protest matters . . . . . . . . . . . . . . . . . . . . . . . . 150
8.2 Contribution to the discipline . . . . . . . . . . . . . . . . . . . 153
8.3 Suggestions for further research . . . . . . . . . . . . . . . . . . 154
V Appendix
157
a appendix
a.1 Appendix chapter 3: . . . . . . . . . . . . . .
a.1.1 Re-coding actors’ positions . . . . . .
a.1.2 Codebook for party positions: . . . . .
a.1.3 MIP/MII questions used in the study
a.2 Appendix chapter 4: . . . . . . . . . . . . . .
a.3 Appendix chapter 5: . . . . . . . . . . . . . .
a.4 Appendix chapter 6: . . . . . . . . . . . . . .
a.5 Appendix chapter 7: . . . . . . . . . . . . . .
a.5.1 Imputation of postmaterialist index .
158
158
158
160
167
169
172
177
179
181
bibliography
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
186
xii
L I S T O F TA B L E S
Table 1
Table 2
Table 3
Table 4
Table 5
Table 6
Table 7
Table 8
Table 9
Table 10
Table 11
Table 12
Table 13
Table 14
Table 15
Table 16
Table 17
Table 18
Table 19
Rules for recodes of rhetorical positions . . . . . . . . 52
Party families & rhetorical activities . . . . . . . . . . 75
Relationship between rhetorical position & policy outputs: logistic regression models (standard b-coefficients) 81
The effect of protest on party agendas, regression results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
The moderation effect of the electoral cycle, regression results . . . . . . . . . . . . . . . . . . . . . . . . . 116
MARPOR and EPCD codes used . . . . . . . . . . . . 131
Postmaterialism vs. materialism, regression results . . 138
Testing causality between protest & party agendas,
instrumental variable models . . . . . . . . . . . . . . 146
Detailed rules for recodes of rhetorical positions . . . 159
Parties included in the Fukushima juncture . . . . . . 162
Factors favouring party rhetoric on nuclear energy . . 166
Overview of sources for MIP/MII questions . . . . . . 168
Effect of protest on amount of party rhetoric, negative binomial regression . . . . . . . . . . . . . . . . . 173
Robustness tests for polarisation hypothesis, regression results . . . . . . . . . . . . . . . . . . . . . . . . . 175
Congruence & the polarisation hypothesis, results of
GEE model . . . . . . . . . . . . . . . . . . . . . . . . . 176
Effect of electoral cycle on congruence, results of GEE
model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
Summary statistics (imputed dataset) . . . . . . . . . 179
Chain lengths of imputations . . . . . . . . . . . . . . 182
Overview of sources for postmaterialism index . . . . 184
xiii
List of Tables
Table 20
Robustness models postmaterialism vs. materialism,
regression results . . . . . . . . . . . . . . . . . . . . . 185
xiv
LIST OF FIGURES
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10
Figure 11
Figure 12
Figure 13
Figure 14
Figure 15
Figure 16
Figure 17
Figure 18
Figure 19
Figure 20
Linking congruence & responsiveness . . . . . . . . . 27
Overview of theoretical framework of the study . . . 41
Distribution of protest-index . . . . . . . . . . . . . . . 57
Relationship between parties’ rhetorical positions &
parties’ left-right placements . . . . . . . . . . . . . . . 77
Violinplot of rhetorical positions & manifesto positions 78
Movement on the nuclear issue in Spain & France . . 79
The relationship between policy outputs & party rhetoric,
logistic regression results across three models (standard b-coefficients) . . . . . . . . . . . . . . . . . . . . 82
Marginal effects plot of interaction between rhetorical activities & incumbency . . . . . . . . . . . . . . . 83
Marginal effect of polarisation & deflating hypothesis 101
Polarisation & deflating hypotheses, coefficient plot
of fully specified model . . . . . . . . . . . . . . . . . . 103
Marginal effect of polarisation & deflating hypotheses, results of GEE models . . . . . . . . . . . . . . . . 105
Marginal effect of interaction between protest, cycle
& mip . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Histogram of cycle variable . . . . . . . . . . . . . . . 119
Marginal effects of moderation through the electoral
cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Protest-index & postmaterialism across time . . . . . 134
Marginal effects of interaction terms with protest-index140
Diagnostics and descriptives of rhetorical position
measurement . . . . . . . . . . . . . . . . . . . . . . . . 169
Rhetorical position split by government status . . . . 170
Marginal effects plot of interaction between rhetorical activities & incumbency (controlling for manifesto positions) . . . . . . . . . . . . . . . . . . . . . . . 171
Densityplot of congruence . . . . . . . . . . . . . . . . 172
xv
List of Figures
Figure 21
Figure 22
Figure 23
Figure 24
Figure 25
Figure 26
Figure 27
Congruence & the polarisation hypothesis, results of
GEE model . . . . . . . . . . . . . . . . . . . . . . . . . 174
Margins at means of interaction between electoral
cycle & public opinion . . . . . . . . . . . . . . . . . . 177
Overview of lagged environmental protest-index, 19811996 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Overview of lagged pacifist protest-index, 1981-1996 . 180
Overview of lagged labor protest-index, 1981-1996 . . 180
Overview of lagged postmaterialism index (non-imputed
values), 1981-1996 . . . . . . . . . . . . . . . . . . . . . 181
Results of multiple imputation . . . . . . . . . . . . . 183
xvi
Part I
Parties and Protest
17
1
INTRODUCTION
uring the last decades protests have become an important way
for the public to express policy demands and grievances. For instance, students mobilised against the stagnant political systems
all over Europe during the “Protests of 1968”. In Germany, for example, students rebelled against the widespread hypocrisy of their elites, who were
largely reluctant to engage actively and transparently with their potential
enmeshment in the German Nazi regime. The same period also marked the
mobilisation climax of the US civil rights movement, with Congress passing
the Civil Rights Act (1964) and Voting Rights Act (1965) (Andrews 2001).
The 1980s saw ever more people mobilising against the peaceful usage of
nuclear energy all over Europe and the rise of the “New Left” and Green
parties (Kitschelt 1994; Kriesi et al. 1992). In the 1990s and 2000s large scale
protests arose against military interventions in Iraq and Afghanistan (Walgrave and Rucht 2003). These examples represent only a small and selective
sample of the major protests that have taken place over the last 70 years.
Some of these large scale protest mobilisations had an unmistakable influence on policies, while others did not result in policy change but did affect
the ways in which media and politicians discussed and thought about the
issues the protests raised.
In contrast to protests, electoral turnout and party membership have significantly decreased (Katz and Mair 1995); while especially Western democracies face strong opposition to policy decisions through protest, both the
increase of protest and the decrease of party membership, are said to be
results of public dissatisfaction with politicians’ decision making (Dalton
2008; Marien, Hooghe, and Quintelier 2010) and decreasing trust in political institutions (Norris 1999: 101-164). As early as the 1970s Inglehart
(1971, 1977) concluded from these simultaneous developments that a silent
revolution had taken place. The generations raised after the second world
war grew up in peace and an age of economic prosperity. Thus, due to the
satisfaction of these basic needs the public had altered its attitudinal prefer-
D
18
introduction
ences and was now seeking the satisfaction of postmaterialist values – such
as environmental protection, pacifism and issues of equality.
While social science has engaged with this attitudinal changes of the public, research on the impact of protest activities on policy and politics is still
scarce. Research on political responses to popular claims is constrained to
the nexus between public opinion and politics. Political actors are understood to respond to public opinion to win votes and secure their office.
This mechanism of dynamic representation is well understood and studied across a wide set of countries (Stimson, Mackuen, and Erikson 1995;
Wlezien 1995; Soroka and Wlezien 2010). However, the role of collective action in the political sphere seems well worth scrutinising, as protests might
affect political discourse and even policies (Koopmans and Statham 1999;
Giugni 2004). Some of the illustrative examples listed above support this
claim. Martin Luther King and his supporters succeeded in winning major
rights, including voting rights, for the African-American community. The
student revolutions in the 1960s led to major shifts in political systems all
over Europe. And, the German government decided not join the ‘Coalition of the Willing’ in going to war in Iraq in 2003, which is likely due to
pressure it felt from the large scale mobilised masses. Even in cases where
protests have not lead to policy change, they have often led to a public
debate about the potential options on the table.
Nonetheless, political science still largely neglects the possible impact of
protest events on politicians’ positions or policy decisions. While several
scholars argue that support by political parties matters for protests’ impact
on policies (Kitschelt 1986; Kriesi et al. 1992), there are only a handful of
studies that explicitly discuss political parties’ role as a mediators between
protesters and policy making (Müller-Rommel 1984). This neglect partly
stems from the subject of political science itself, whereby social movements
and protest attract sociologists’ interests more strongly. On the other hand,
sociology is rarely interested in the impact of political protest on political
parties, leaving this influence as a lacuna in the study of social movement
organisations and protest.
Yet, political science has invested a lot of time and effort in analysing the
relationship between public opinion polls and the reactions of the political
elite. Governments are understood to adapt their policies and agendas to
public preferences in efforts to seek re-election and public support (Stimson,
19
1.1 what is explained
Mackuen, and Erikson 1995; Mackuen, Erikson, and Stimson 1992; Wlezien
1995; Soroka and Wlezien 2010, 2004; Jennings and Wlezien 2015). Similar
theoretical arguments have been put forward for the relationship between
public opinion and political parties in general (Ezrow 2010; Adams et al.
2006; Adams, Haupt, and Stoll 2008; Adams 2012). Again, political parties are thought to react to public opinion shifts to secure votes and office
(Strøm 1990). However, in the last decades the frequency and turnout at
protest events has increased throughout Western democracies. This has led
scholars to articulate doubts that the conventional institutional channels of
representative democracies are capable of absorbing grievances articulated
in popular protest. Unfortunately both of these literature strands have not
looked at how political protest affects party positions, or if this influence
exists at all.
1.1
what is explained
My thesis is interested in whether protest influences the agendas of political parties, and if so, how political protest affects party rhetoric. In doing so, it attempts
to address the aforementioned academic gap which has been largely left
open by political science and sociology. I argue in my theoretical framework that demonstrators’ first goal is to seek the attention of external actors. This argument deviates from previous studies researching political
protest, which often assume that political protest seeks to change policies
or the status quo (Gillion 2012; Giugni 2004). While protest might seek to
change policy in the long-run, I argue that before seeking policy influence
protesters want to gain the attention of political decision-makers and the
media.
Political parties in turn only respond to protest if protest manages to
stand out above all of the other external stimuli political parties face. Protest
is often little more than random noise on the streets that follows unstable
waves of mobilisation. Generally demonstrators will only have an influence
on political parties if they manage to be organised in a larger group across
a longer period of time.
Parties can react to external stimuli in various ways and certainly the
most often studied response is policy change. However, I deviate from this
20
1.2 how is it explained
oft-employed approach and look specifically into the earlier, agenda-setting
stage of the policy cycle to try to understand if protest can influence it. I
argue that influencing the agenda is crucial for protesters. Generally it is
only if demonstrators manage to have an impact on this very early stage
of the policy cycle that they also influence policy drafting and making in
the longer run. In this thesis, I use the statements made by representatives
of political parties in press releases, press conferences, media interviews
and speeches to assess whether protests may impact this early stage of the
policy cycle.
In summary, this thesis raises four related research questions:
1. Does protest affect party rhetoric?
2. Are parties’ rhetorical activities cues for substantive policymaking?
3. Are there contingent factors moderating the relationship
between political protest and party rhetoric (e.g. party ideology; public opinion)?
4. Does protest have an effect on more binding party commitments such as issue emphasis in party manifestos?
Addressing these questions may shed more light on how parties talk
and signal policy positions between elections and how political discourse
develops across time and countries. Finally, I study how these discourse
developments are affected by external stimuli such as political protest and
public opinion.
1.2
how is it explained
The thesis employs a comparative approach, which allows a variety of institutional, ideological and country-specific factors to be controlled for. It
also allows other variations across countries and parties to be singled out
to provide insight into how divergent factors affect the relationship between protest and political parties. I mainly employ quantitative analyses
throughout all empirical chapters of the thesis.
The analyses use time-series-cross-sectional models across 17 advanced
democracies. To obtain measures of protest I use two different datasets. The
21
1.2 how is it explained
first part of the empirical analysis uses the data currently collected by the
ResponsiveGov project. This data not only provides information on protest
across 12 democracies, but also detailed information about how political
parties present their positions in the media. The last chapter employs the
data coded by the Comparative Manifesto Project (MARPOR) merged with
the European Protest and Coercion Dataset (EPCD).1 Thus, this final analysis serves as a sort of validation of the results found by the examination of
the ResponsiveGov data.
I focus on the issue of nuclear energy after the Fukushima meltdown
throughout most of the thesis. Nuclear energy is a very interesting case to
study the influence of protest for at least three reasons. First, as shown in
chapter 4, nuclear energy is a highly confrontational issue for political parties in the countries included in this study. Parties were outspoken about
the issue after the Fukushima accident and mostly adopt clear-cut positions for or against nuclear energy. Second, the confrontation on this issue
mirrors the general ideological divide of party competition. While parties
on the left of the ideological spectrum tend to reject the usage of nuclear
energy, parties to the right tend to support it. Therefore, one can argue that
the question of the usage of nuclear energy nowadays tends to reflect the
classical party ideological divide between left and right. Thus, the results of
this study might also speak to the general left-right divide of party competition and not only for the specific issue of nuclear energy. Third, in contrast
to economic issues, the salience of the issue varies hugely across countries.
For instance nuclear energy was salient in countries such as Germany, Italy
and Switzerland after the Fukushima meltdown, while it was not salient
in the UK, USA or Canada. Thus, the issue of nuclear energy also assures
that we can make meaningful arguments about how the salience of an issue
affects parties’ reactions to political protest.
The final analysis, using the MARPOR and EPCD datasets, looks into a
different set of issues. It examines ways in which protests on postmaterialist issues such as the military and the environment managed to shift the
salience parties attach to these issues in their manifestos.
1 The MARPOR dataset was formally abbreviated as CMP data.
22
1.3 synopsis of the thesis
1.3
synopsis of the thesis
Based on the general theoretical framework sketched above, this thesis
makes four distinct but strongly related claims which are discussed in four
empirical chapters. The claims are outlined below.
First, political protest matters. Throughout the thesis we will see that
political protest matters for party competition. In contrast to previous research on party responsiveness, I show that political parties do not only
adapt to public opinion polls, but also to political protest. Parties adapt
their agendas as a response to increased protest. This result is robust in
three empirical chapters (chapters 5; 6 & 7) and is also supported by the
two different data sources employed in these chapters.
Second, how parties rhetorically react to protest is contingent on three major factors. I argue that parties’ general ideology changes how they perceive
and react to protest. In the case of nuclear energy I show that parties to the
left are more likely to react in favour of anti-nuclear protests, while parties
to the right are more likely to reject them. Note that this is not a contradiction to the protest attention hypothesis: demonstrators successfully seek
the attention of political parties irrespective of their ideology. The reaction
of parties, however, is very different depending on their understanding of
protest as either a window of opportunity or a threat to the status quo. A
second obstacle for the influence of protest is the public in general. Parties reject protesters’ claims if the public at large also does. Parties tend
to support the opinion of the general public, instead of taking chances by
supporting the demands of protesters. These factors are the main concern
of chapter 5. A final factor moderating the reaction of political parties to
protest is the electoral cycle. I argue that due to vote-seeking strategies
parties become less likely to react in favour of protest the nearer the next
election is. Instead, parties then support the position of the public at large
and ignore protesters’ claims. The effect of the electoral cycle on parties’
reaction to protest is explored in chapter 6.
Third, the effect of political protest is not only visible in parties’ reactions
to them in the media but also in the issues they address in their manifestos.
Looking into a period of 15 years (1981-1996) across 17 democracies and 80
elections, I find that political parties adapt their issue portfolios to the rise
of postmaterialist values and postmaterialist protest. Chapter 7 looks into
23
1.4 relevance
this long-term effect of protest and provides an answer for the often raised
question of causality in political science nowadays. Using an instrumental
variable approach I show that political parties react to popular protest and
not the other way around.
1.4
relevance
Apart from addressing the academic gap left open by social movement
scholars and political science, this thesis adds to the broader questions
of whether we can expect politicians to answer to external stimuli and
whether the public has the possibility to affect politics through non-institutional channels.
We know that the public has a substantial influence through institutional
channels (elections; referenda) on politics and policy. Yet, if the public seeks
to challenge politics through unconventional channels without political actors reacting to it, the question arises as to what meaning such channels
actually have. Political protest is an important mechanism to influence politics, as I will show in the upcoming chapters. Thus, it is important that
people actively mobilise by taking to the streets.
The often paraphrased public opinion is not almighty, but limited in
scope and power. Public opinion polls do not speak for themselves and they
are not a mean of active participation in modern democracies. Respondents
reply to structured questionnaires via phone and pollsters summarise their
findings, which politicians might or might not read.
In contrast, political protest is a means of active participation that attempts to influence how democracy evolves and develops. People organise
themselves and their claims and bring them to the street as forcefully as
possible. If the people succeed in mobilising a critical mass then they will
be heard.
24
2
R H E T O R I C A L R E P R E S E N TAT I O N : T H E S TAT E O F T H E
A RT
2.1
the game of representation: the responsive politician
he term “political representation” and its studies belong to modern
societies. While Athenian democracy was a representative democracy, with people electing and selected leaders, the term “representation” did not exist (Vieira and Runciman 2008: 6).2 In ancient Greece
most citizens of a polis voted on most policy decisions.3 In contrast, modern
democratic nation states rely upon systems of delegation and accountability (Strøm 2000; Strøm, Müller, and Bergman 2003): Citizens understand
representatives as acting for them, with elections being the principal mechanism to assure their liaison to public preferences (Dahl 1971).
Ever since the establishment of modern democracies, observers and researchers of politics have been debating how representatives should balance the relationship between themselves and the public. On the one hand,
scholars such as James Madison (1787-88) have suggested that representatives should act as “delegates” and follow the preferences of their citizens.
On the other hand, the public’s will might be often uninformed (Zaller 1992:
6-39) and subject to strong fluctuation (Holsti 1996). In such occasions the
public might be better off in the long run with elected “trustees” following
their own guidelines to present the best possible policy for all represented.
As Burke (1774) famously reasoned in his speech to the electorate in Bristol:
T
2 Representation finds its origin in Latin: Originating from the verb “repraesentare” which
expresses (a) the necessity to do something immediately or (b) to present oneself to a
crowd. In Rome it was however used to talk about cash or debts and their immediate
payback. In contrast, politicians on the forum romanum re-presented themselves to the
public. Yet, this latter procedure can be largely understood as a promotional act. We have
to go back to Hobbes and Rousseau to find theoretical arguments about representation
that use the term in the sense of modern political science.
3 Note however, to vote one had to be an adult, male citizen who owned land. Slaves and
women were excluded from voting.
25
2.1 the game of representation: the responsive politician
“[. . . ] his unbiased opinion, his mature judgment, his enlightened conscience, he ought not to sacrifice to you, to any man, or to any set of
men living. These he does not derive from your pleasure; no, nor from
the law and the constitution. They are a trust from Providence, for
the abuse of which he is deeply answerable. Your representative owes
you, not his industry only, but his judgment; and he betrays, instead
of serving you, if he sacrifices it to your opinion.”
Burke stresses the need for representatives not to follow each and every public tamper, but rather to serve in the best long-term interests of
the people even though this might reject temporary popular mood swings.
Scholars nowadays tend to agree that democracies need to be responsive
to public demands to a certain degree. However, there is less agreement
about how to define the responsive politician. For instance, scholars often
use the terms congruence and responsiveness nearly interchangeably. Yet,
the relationship between congruence and responsiveness is more complex.
While responsiveness clearly calls for a reaction to an external stimulus,
congruence does not necessarily need to be preceded by a external stimuli,
such as public opinion shifts. Figure 1 illustrates this fact. The horizontal
axis symbolises time, while the vertical axis can be understood as the policy
position of the relevant actors. For instance higher positions could outline
the preferences for more public spending. Let P1 and P2 be two collective
representative actors – such as two competitive parties – and PO public
opinion. At time t−1 neither party P1 nor party P2 are congruent to public
opinion (PO). P1 appears to be somewhat closer to public opinion than P2
at t−1 . Between t−1 and t the public opinion gradually changes and moves
towards the position of P2 .
However, both parties – P1 and P2 – do not update their initial policy
preferences until t and they remain true to their original positions. Therefore, in t public opinion is not only fairly closer to P2 ’s position than to P1 ,
but it appears that public opinion and P2 are now congruent.
Between t and t+1 , party P1 shifts its policy position as well. Consequently in t+1 , P1 is responsive to public opinion and now represents the
will of the people. This definition of responsiveness is partly at odds with
existing research (Lax and Phillips 2009, 2012). However, it is closer to how
the public perceives parties and their policy making efforts, either reacting
in favour to their claims or being unresponsive. Furthermore this definition
26
2.1 the game of representation: the responsive politician
Figure 1: Linking congruence & responsiveness
Policy Position
P2
PO
P1
t −1
t
t +1
time
Source: Author’s own.
also finds support in definitions of responsiveness in medicine or computer
science. Medicine, for instance, understands responsiveness as the ability
of respond to a stimulus” (McGraw-Hill Concise Dictionary of Modern
Medicine, 2002).4 Using this example as an analogy for political science:
the public sends a policy signal and observes whether the political elite
accordingly adapts to this signal (is responsive) or not (is unresponsive).
Therefore, I understand responsiveness as a congruent adaption of a political actor to an external stimuli. However, a party holding a congruent
position to public opinion is not necessarily responsive, as P2 in the example above. Congruence is thus a necessary but not sufficient condition for
responsiveness. In contrast, parties that are congruent to public opinion do
not necessarily need to have been responsive beforehand.
However, political actors can perform responsiveness in various forms
and not only with ideological shifts. Originally strongly bound to political
4 The definition of responsiveness in computer science appears to be similar: “the response
of a software/system to complete a task which was given by a user within a certain time
interval” (Weik 2001: 1484).
27
2.2 what we know about responsiveness
actors’ positional adaptation to external stimuli, research also focused on a
range of other activities subject to responsiveness (for an excellent overview,
see: Eulau and Karps 1977). Allocation of goods and resources has been understood as responsive pork-barrelling in line with constituents’ material
interests (Eulau and Karps 1977: 245-246; Martin 2014; Samuels 2002; Fenno
1977).5 A now well developed experimental literature stresses the necessity
of service responsiveness – e.g. helping constituents with administrative
tasks (Grose, Malhotra, and Van Houweling 2014; Butler, Karpowitz, and
Pope 2012). Furthermore, governments’ adaptation to the issue priorities of
citizens has received increased attention in the literature on responsiveness
(Wlezien 1995; Stimson, Mackuen, and Erikson 1995; Hobolt and Klemmensen 2008; Bevan and Jennings 2014). Lately, this interest in the salience
politicians attach to different issues has stood in the limelight of the party
politics literature (Spoon and Klüver 2014). As outlined in the next sections,
this thesis addresses this last concept of responsiveness, attempting to enrich the literature on issue salience by linking it to positional shifts. Thus,
I propose a concept and measurement combining political parties’ salience
adaption, also taking their initial position into account.6
Notably, the “simple” congruence between citizens’ preferences and rulers’
actions appears insufficient for defining democratic responsiveness because
such a definition would include responsive and congruent autocrats as
well. Besides politicians’ behaviour and decisions, institutions within which
these actions take place need to assure their responsiveness (Powell 2004:
273-274).
2.2
what we know about responsiveness
Researchers standing on the shoulders of Miller and Stokes (1963) and the
Michigan school usually follow the logic of “rational anticipation”, mean5 Note that in this context the term responsiveness has a quite different normative connotation. While political actors are responsive to public claims and their voters, pork-barrelling
is certainly not a normatively desired form of responsiveness.
6 In this vein, responsiveness contains all four dimensions of Pitkin’s (1967) concept of representation: formalistic, symbolic, descriptive and substantive representation (Eulau and
Karps 1977).Thereby formalistic representation looks into institutional arrangements preceding political action. It looks at symbolic representation, the way a representative stands
for his/her constituents and descriptive representation, which observes the resampling of
voters’ characteristics; Substantive representation the activities of politicians.
28
2.2 what we know about responsiveness
ing elected politicians are subject to the pressure of satisfying constituents’
will due to the anticipation of electoral penalties if they fail to do so. They,
then, frequently adapt their positions to public mood swings in an effort
to secure their goals of being (re-)elected (Stimson, Mackuen, and Erikson
1995: 544-545). Building on Miller and Stokes’ results and methods, a rich
body of literature was published dealing with the congruence nexus in the
US (see e.g.: Adams 1997; Bishin 2000; Erikson 1978; Lax and Phillips 2012;
Page and Shapiro 1992; Soroka and Wlezien 2010; Stimson 1991). All of
these scholars partly differ in their empirical results, although they stick to
similar theoretical arguments in the sense that they all underline the causal
link between public opinion and politicians’ policy-making decisions. Furthermore, a couple of studies have been published dealing with the representation of public opinion in other countries (Barnes 1977; Converse
and Pierce 1986) and since the late 1990s, studies in comparative perspective have also been published (Miller et al. 1999; Wessels 1999; Hobolt and
Klemmensen 2008).
Political scientists largely share a common notion of representation in
the US. In most instances politicians are responsive to citizens’ demands
expressed in opinion polls (Hobolt and Klemmensen 2008: 311).7 Political
scientists seem to know that changes in the public mood are somehow
reflected in policy decisions, conditional on the size of the majority of public opinion on a particular issue and its salience (importance) (Page and
Shapiro 1983; Burstein 1999: 9-12). Furthermore, it seems to be true that
congruence is higher for strongly politicized issues than for contested ones
(Belchior 2013: 363; Mattila and Raunio 2006: 437-439; Dalton 1985).
Most of the outlined studies treated public opinion as a single entity,
which in most instances is understood as the mean voter opinion in public
opinion surveys. Notably, other public claims such as protests are unobserved. While there has been an extensive debate on whether “the mean,
median or something else” (Achen 1978: 476) provides the ideal formula
for capturing the masses’ opinion in polls, less effort has been made to
disentangle the public into its constituting pieces.8
7 Lax and Phillips (2012: 164) is one of a few offering a more pessimistic and nuanced view
of representation in the US.
8 Interestingly such efforts were undertaken on the representatives side by differentiating
between collective and dyadic representation (Weissberg 1978; Hill and Hurley 1999), but
29
2.2 what we know about responsiveness
Yet, people also articulate their opinions through collective actions. Since
the large scale civil rights and anti-Vietnam War social movements during
the 1960s and 1970s, people have become increasingly involved in demonstrations, boycotts, strikes and petitions (Inglehart 1977; Barnes et al. 1979).
This observation is accompanied by a major decline in traditional forms of
participation such as party membership (Katz and Mair 1995: 15) and voter
turnout. Since “unconventional” participation is becoming more conventional (Dalton 2008; Inglehart 2002; Marien, Hooghe, and Quintelier 2010),
the question on how protests affect policy, politics and polities should be
given increased attention in political science.
The relevance of protesters’ demands is becoming even more crucial considering that polls do not directly address politicians. Rather, politicians
need to be aware of polls themselves. In contrast, protest expresses direct
demands towards the political arena:
“Protest groups are groups of citizens who do not normally interact
with governmental officials, but who, under certain conditions [. . . ] organize on an informal, issue-specific basis to make demands on public
officials through pressure processes” (Schumaker 1975: 490).
Protest might be an essential information source about the public mood
and its swings on specific issues. Protest events are usually bound to a certain topic and participants often present policy expectations.9 In this regard
scholars might have underestimated the impact of the protesting minority
and overestimated the impact of the silent majority. While the silent majority responds with a certain answer in a survey without clear policy demands, the unquiet and mobilised minority mounts clear demands bound
to a certain issue (Lohmann 1993).
Yet, research linking protesters’ claims and political outcomes is scarce
(but see: Agnone 2007; Giugni, McAdam, and Tilly 1999; McCammon and
Campbell 2001; Soule and Olzak 2004) and the few existing studies present
mixed results (Burstein and Linton 2002: 398; Amenta et al. 2010). Tradiwe still mostly interpret the masses as a single and uniform actor (but, see: Bishin 2000;
Egan 2013, 2014).
9 This is especially true for the so called “new social movements”, as outlined by Kriesi et al.
(1995: xvii-xxii). These movements stressed demands for ecology, pacifism, equality of
women and the Third World as well as various others issues (Kriesi et al. 1995: xviii). A
few exceptions to this rule can be found. For instance the occupy Wall-Street movement
surely lacks a clear policy demand.
30
2.2 what we know about responsiveness
tionally scholars analysing protest activity largely deal with protests as a
dependent variable. They want to explain why people demonstrate and not
how demonstrations affect policy outcomes or the political sphere (see e.g.:
Gurr 2011; Opp and Kittel 2009; Van Aelst and Walgrave 2001; Tarrow 1994;
Kitschelt 1986; Muller and Opp 1986). Similar to other disciplines of political and social science, most studies focus on the American case (Gillion
2012; Baumgartner and Mahoney 2005; Amenta and Zylan 1991; Barkan
1984; Morris 1993; Burstein 1985). While some comparative approaches and
studies exist, researchers largely stress the necessity of looking into single
cases, limiting the possibility for generalisability (Kitschelt 1986; Kriesi et al.
1995; Giugni 2004; Kolb 2007). Thus, it is difficult, if not impossible, to generalise about the impact of protest using existing research (Uba 2009). Results partly suggest that comparable factors drive politicians’ reactions to
protest, similar to those already outlined for the impact of public opinion
on policy. Thus, results suggest a contingent effect of protest on policies
in cases of highly salient issues. Scholars aiming to link protest with the
political arena have tended to look at protests’ impact on policy decisions
(Giugni 2004; Agnone 2007).
However, as summarised above politicians have more means than just
policy change for being responsive. While governments can propose, vote
and enact policies (Bishin 2000; Page and Shapiro 1983), parties can change
their manifesto positions and politicians can visit their constituents and
directly respond to their claims. Politicians can increase their attention to
voters’ preferences in an effort to change the political agenda (Baumgartner
et al. 2009) and governments can influence actual policy outcomes (Soroka
and Wlezien 2010). Nevertheless, existing studies mainly focus on policymaking as the object of interest in measuring politicians’ responsiveness to
protest or public opinion.
Yet, parties’ influence on policies varies across political and environmental circumstances. While incumbent politicians can form and enact policies,
the opposition usually has to rely on the government’s will to enact policy
reforms. Above all, minority governments need to bring together a large
faction of the whole parliamentary chamber to enact their legislative agendas.
Therefore, the focus on policy output leaves a huge part of the representational puzzle untouched. Before policies are successfully implemented,
31
2.2 what we know about responsiveness
politicians need to recognise the necessity for policy change and subsequently formulate new policies to address public grievances. Furthermore,
politicians face serious institutional constraints, which may forestall any
change, when deciding which policies to implement. Therefore, public expenditure and governmental responsiveness depends on a range of factors
that are not influenced by governments, parties or even politicians. This
means that parties are confronted with serious restrictions that may impact
their capability of fulfilling policy promises or changing policies according
to the public mood. Certain constellations of veto-players (Tsebelis 1995) or
path dependence can make a change almost impossible (Pierson 2000; Rose
1984). Thus, parties or governments might be willing, and even show this
willingness publicly, to be responsive to public demands, but established
environmental circumstances that are not under their influence thwart their
capacity to be responsive. In this way, the idea of policy representation
partly neglects the frictions that politicians face when intending to change
policies (Jones and Baumgartner 2005; Baumgartner et al. 2009).
To address such criticism, some authors discuss other activities undertaken by representatives on behalf of the represented. The seminal work
in this regard is Fenno’s (1978) extensive study on representatives’ interactions with their constituencies.10 He concluded that it was the goal of Members of Congress to gain trust in their constituencies by interacting with
them and explaining their motivations for their political decisions (Fenno
1978: 56). Bianco (1994) picked up Fenno’s ideas and re-established his argument in a game-theoretical framework. Bianco finds that trust, what he
calls “voting leeway”, is given when constituents have a favourable opinion
of the representative despite their voting behaviour. In a similar vein both
authors suggest that representatives symbolic activities, such as not shying
away from direct interaction with constituents; listening to constituents’
grievances; and verbally replying to public concerns, have a positive effect
on their reputations and chances of (re-)election.
Despite several other studies that deal with the non-legislative actions of
politicians (Franklin 1993; Saalfeld 2011; Saalfeld and Bischof 2013; Searing
1994), the dominant strand of literature on substantive representation deals
with the policy-voter nexus as outlined above, while leaving rhetorical ac10 Fenno accompanied several Members of Congress during their direct interactions in their
constituencies for a couple of years.
32
2.2 what we know about responsiveness
tivities untouched. If discussed, rhetorical activities are often understood
as symbolic acts (Hinckley 1990). Thus in the American context, the President’s State of the Union address has found large appeal amongst scholars of Presidential responsiveness. On the one hand, presidential speeches
are understood as pure symbolism, containing little, if any, policy content
(Tulis 1987). In this vein presidential rhetoric does not share policy information but is used to unite the public behind the politicians’ policy endeavours. Presidents seek support from the public to put congress under
policy-delivering pressure in order to pass legislation, which helps presidents’ re-election goals (Canes-Wrone 2001b,a). They are said to share “policy symbols”, but not detailed policy plans (Cohen 1999: 35). Druckman
and Holmes (2004) showed that presidential rhetoric is successfully used
for priming, creating trust among the public (Bianco 1994; McGraw, Best,
and Timpone 1995) and altering the public’s perception of presidents’ issue
competence (Holian 2004).
On the other hand, it has been argued that speeches are used to set the
president’s or government’s policy agenda (Cohen 1995, 1999; Jennings and
John 2009). Rhetorical events are therefore understood as a step that precedes policy. The priorities and positions presented in speeches are likely
to result in substantive policy making. Bevan, John, and Jennings (2011) for
instance show that the Queen’s speech is a strong and positive predictor of
public policy-making in the UK.
Taking into account the fact that policy output may be influenced by
much more than protests or public opinion (e.g. veto-players) it seems more
fruitful to look at the agenda setting stage of the policy cycle to understand
the impact protest may have on politics. Protest might influence the agendasetting stage of politics (Soule and King 2006; Baumgartner and Mahoney
2005) and through this, in the long run, policies (Olzak and Soule 2009).
From this point of view protest today influences the agenda of tomorrow,
which subsequently affects policies of the day after tomorrow. The impact
of protest should decline along the policy cycle, due to institutions, vetoplayers and frictions that hinder policy change.
The public witnesses politics largely via the media and the exchange
of words in debates. The discourse between parties reflects the issues parties want to address and how they want to address them. This stage gives
protest the opportunity to matter by influencing which issues are picked
33
2.2 what we know about responsiveness
up by politicians, and gives researchers the opportunity to look into the
earliest stages of the policy cycle, which is the awareness and definition of
problems. Thus, politicians’ verbal statements in the media, in parliament,
in campaign meetings, party meetings and public gatherings, which are
here subsumed under the term “rhetorical responsiveness” should give a hint
as to which issues matter for parties and how they intend to address them.
I argue that parties’ verbal statements, like speeches, are costly signals
for politicians (Fearon 1997). Voters might punish or reward parties based
on the distance between their preferences and the information conveyed
via party rhetoric.11 As such politicians are not only believed to be judged
for their actual policy decisions, but also for the policy sets they outline as
feasible via their rhetoric. While it is true that party rhetoric rarely contains
extensive policy plans and descriptions (Hinckley 1990), it often forwards
clear cut positions on whether parties support or reject a certain policy goal
to the public. Parties communicate to the public the range of policy options
they consider as being feasible and which policy goals emerge from these
at the end.
There are several reasons that verbal statements should matter for parties and voters. First, the underlying assumption that actions further down
the policy cycle matter most for voters in representative democracies is
rarely challenged. However, as Bianco (1994) outlines, voters usually stick
to low cost information for future voting decisions. A large part of representatives’ work is devoted to generating trust amongst constituents, which
results in leeway in the case of unresponsive policy making (Bianco 1994).
Trust can be obtained by explaining policy decisions and positions to voters. Therefore, the ways in which politicians explain their policy positions
and decisions is essential for voters’ future political activities.
Second, the literature on political representation largely concentrates on
representatives’ actions and behaviour undertaken within institutional settings. Yet, it is unlikely that voters cast their votes based on these institutional actions. The mass of voters do not read through lengthy party programs, but form their opinions and voting decisions based of what the media presents them and what is reasonable to access. As such party rhetoric
11 In fact, message sending can only be cost-free in the case that the public does not listen
to the messages at all. However, it is fair to assume that not all voters will be uninterested
in seeking information about politicians’ positions. As soon as one voter listens to the
forwarded message, talk is no longer cheap.
34
2.3 parties in modern democracies
transferred via the media has an essential impact on voters’ perceptions of
their positions.
Third, besides fishing for voters’ trust, politicians might influence issue
salience by using verbal statements. As studies have shown, politicians use
speeches to prime certain issues (Druckman and Holmes 2004). Presidents
use their speeches to spotlight issues they think are capable of influencing
voters’ rank ordering of the importance of a certain issue. Thereby representatives are said to influence their electoral fortunes (Druckman and Holmes
2004: 763-768). Furthermore, statistics for the US case outline that American presidents and their campaign teams spend a huge amount of money
on rhetorical courses and consultation, which underlines the importance
politicians designate to rhetoric.
Finally, party rhetoric is not bound to a certain point in the electoral cycle, unlike manifestos, which are aligned to the campaign period. Parties’
interactions with citizens are mainly studied during campaign periods. The
most common object of analysis are party manifestos (Budge et al. 2001;
Klingemann et al. 2006), with researchers looking at how promises made
in them translate into policy making after elections (Budge and Hofferbert
1990; Artés 2013), or how parties’ positions, measured with manifesto data
or expert judgements, reflect the will of the people (Dalton 1985; Ezrow
2010). Yet, this leaves the actual interaction between parties and voters between elections untouched. The question remains as to how parties interact
with citizens between elections and how parties’ positions are articulated
to citizens apart from party manifestos. This costly meaning of rhetorical
activities is debated at length theoretically and empirically in chapter 4 of
this thesis.
2.3
parties in modern democracies
This study contributes to the existing literature on the influence of protest
by focusing on political parties, one of the most crucial players for shaping
protesters’ access opportunities to the political agenda in modern democracies. They decide which issues are relevant to be brought to and discussed
in politics and put into policies. As Lipsky (1968) put it: “The ‘problem of
the powerless’ in protest activity is to activate ‘third parties’ to enter the
35
2.3 parties in modern democracies
implicit or explicit bargaining arena in ways favourable to the protesters”
(Lipsky 1968: 1145, emphasis in original). Even though Lipsky made this
claim almost fifty years ago, in depth studies looking into parties’ responsiveness to protesters’ claims in comparative perspective do not exist. Thus,
while addressed theoretically (Kitschelt 1993, 2005), the linkage between
political parties and the activities of social movements is a lacuna in the
empirical literature on political protest.
Before outlining how parties matter in modern democracies, it is necessary to define what a party actually is and demarcate it from other political organisations, such as social movements. Scholars stress different party
characteristics in their conceptualisations and do not seem to agree upon
a definition of parties (see for a short summary: Meyer 2013: 15-16).12 By
discussing and acknowledging these facts Sartori came up with what he
calls a “minimal definition” of parties:
“A party is any political group identified by an official label that
presents at elections, and is capable of placing through elections (free
or nonfree), candidates for public office” (Sartori 1976: 63).
Sartori’s definition comprises all relevant features of parties: a) the organisational level as a group; b) the necessity to participate in elections and c)
the goal to place candidates into office. Therefore, I use Sartori’s definition
in my following argumentation.
After having defined political parties, one realises that modern democracies are characterised by party politics. Parties exist because they a) reduce
transaction costs for the public and b) help to solve collective action problems (Müller 2000: 309). Today most polities, whether democratic or not,
rely on political parties and as such parties should be perceived as crucial
for the implementation of protesters’ claims into the political arena. Parties
are the essential link calling political responsiveness into being in modern
democracies. Their duties and responsibilities might vary across polities
(Müller 2000), but their utility and necessity does not. The institutional
setting of Western democracies places parties at the forefront of political
12 Schattschneider (1942: 35) seems to focus on their goal to achieve power by defining parties as “an organized attempt to get to power”. In contrast, Schumpeter’s (2003), as an
economist, understanding of political party revolves around the idea of competition: “A
party is a group whose members propose to act in concert in the competitive struggle for
political power” (Schumpeter 2003: 283). Surprisingly neither Duverger (1954) nor Epstein
(1980) define parties in their seminal works on the latter.
36
2.4 how protest matters for political parties
decision-making. Parties decide on the supply of candidates in elections,
form governments, and decide on policies.
Parties might also be crucial when voters decide whom to vote for. In
many cases voters might not know their member of parliament (MP) by
name and thus use party affiliation as a cue to reflect on their preferences when casting a vote (Inglehart and Klingemann 1976; Cain, Ferejohn,
and Fiorina 1984). It is essentially in this stage in the chain of democracy
that parties hold a unique characteristic and duty. Most voters in western
democracies cast a vote on the party-level and not for a single candidate.
Even if voters cast a vote for a candidate or might be allowed to disturb
candidate lists, the party still decides which candidates run under their
party banner. Due to this distinct position and the meaning of parties for
modern democracies, I decided to focus my research on the impact protest
has on political parties.
2.4
how protest matters for political parties
In this thesis protesters are understood as attention seekers whose first
goal is not to achieve policy change, but rather to win the attention of
external actors, such as the media, political actors or the public in general.
Certainly protesters might have other goals as well, such as seeking policy
change in the long run. Other scholars have made comparable arguments
about protest, describing it as a resource, signal or source of information for
the political arena (Gillion 2012; Lohmann 1993; Lipsky 1968). Looking at
common tactics used by protesters, this assumption seems to be reasonable.
Protesters try to be as noisy as possible, seek to disrupt the daily routine
of people by marching on blocked streets and rely heavily on the usage
of symbols strongly rejecting a specific status quo, to seek the attention of
external actors that are not part of the protest activity.
However, protest is time and again little more than short-term noise. First
of all, protest often lacks the numbers to seek attention and is regularly undertaken by marginalised and minority groups. Thus, protests frequently
lack the capability to sway the critical mass to take to the street. This problem is well known in the literature of rational choice as “collective action
problem” (Olson 1965; Opp and Roehl 1990). However, to seek attention
37
2.4 how protest matters for political parties
protesters and their organisation need to overcome such problems to successfully seek the attention of external actors. Gaining political parties’ attention depends particularly on the size of the protest, as the numbers of
protesters can serve as a proxy for the amount of support protesters’ claims
find in the larger public (DeNardo 1985; Lohmann 1993). Thus, larger
protests suggest to political parties that a wider part of the public might
support protests’ claims. Power in numbers can help to legitimise protest
and its claims and research suggests that size might be the most crucial
means for protesters’ success or failure (Verhulst and Walgrave 2009; Earl,
Soule, and McCarthy 2003; Mueller 1997). For instance Earl, Soule, and McCarthy (2003) find that police presence at protest events is more likely the
higher the turn-out on the protesters’ side is, which suggests that the attention of the police as an external actor is heavily dependent on the number
of protesters.
The major difference between public opinion and protest is the disparity
in consistency of both concepts. Even though politicians might not be interested in the position of the public on each and every issue, and might also
ignore the public on issues which are not salient, they sense that public
opinion is always present as a latent dimension. They also sense that, to
a certain extent, they need to be responsive to public opinion, since otherwise they will likely be punished at the ballot box (Wlezien 1995; CanesWrone, Minozzi, and Reveley 2011; Woon 2012). In contrast, protest, even
on very specific issues, is very unstable in its duration and claims. Protest
can peak at times on a specific issue, but this mobilisation is often shortlived (Eisinger 1973). To successfully seek the attention of external actors,
protest needs to show durability across time. Thus, the more frequent and
persistent protests are, the more likely they are to attract the attention of
political actors.
Strongly related to durability are the organisations surrounding protesters.
The study of the influence of protest relies on the literature of social movement studies, which frequently emphasises the importance of organisational strength for the success of social movements (Zald and Ash 1966).
However, in contrast to the kinds of protests this thesis is interested in,
social movements are frequently surrounded and supported by an organisational structure of political protest. Such organisations help and allow
social movements to organise, structure and repeat their actions (Andrews
38
2.4 how protest matters for political parties
2001; Soule et al. 1999; Gamson 1990). Furthermore, irrespective of the organisational hierarchy, an organisation has a spokesperson (or many spokespersons) that is interviewed by the media or can approach the media to send
messages to supporters. For instance they can reach out to the public in an
effort to attract new supporters for upcoming protest events. Thus, a third
factor that positively impacts protesters’ capability to attract the attention
of external actors is their organisational strength.
Researchers who suggest that the organisational strength of social movements is important mostly also stress the importance of disruptiveness of
protest. The more radically and well organised a protesting group is, the
more likely it is to successfully impact politics (Gamson 1990; Lipsky 1968;
Piven and Cloward 1977). In the line of this argument the radicalisation
of protest helps to steer politicians’ attention to protesters’ claims. Recent
research on the effect of disruptiveness of protest on party agendas underpins the theoretical arguments put forward by Gamson (McAdam and Su
2002). For instance the protests surrounding the shooting of Michael Brown
on August 9, 2014 in Ferguson, USA, captured the most media attention
when they adopted extreme methods of mobilisation, such as burning cars,
damaging public property, police presence and wounded civilians.
In contrast to previous research, my theoretical argument does not want
to suggest that any of these factors outweighs any other. I suggest that all of
the named factors matter. Consequently I argue that the more of the named
characteristics (size; duration; organisation; disruptiveness) are given the
more likely protesters are to attract external attention and to penetrate the
agendas of external actors. Thus, the protrusion of protest increases with its
size, duration, organisational strength and disruptiveness.
As a consequence, I hypothesise that parties have an increased interest
in reacting to protest the higher its protrusion is. Political parties are more
likely to be forced to adapt their agendas in order to sufficiently address
demonstrators’ claims the more protrusion protest has. I argue that parties have at least three major reasons to rhetorically react to protesters’ demands.
First, the higher the protrusion of protest, the more likely it becomes that
it will not simply fade away with time. This suggests to political parties that
there is a need to address the issue at stake. Otherwise the mobilisation on
the issue might even increase in the future. The odds are high that ignoring
39
2.4 how protest matters for political parties
protesters in the first instance could incrementally increase the costs to not
have done so in the near future. Thus, political parties might actually be
forced not only to rhetorically react to protesters claims, but also to change
policy in an effort to secure votes or office (Strøm 1990; Müller and Strøm
1999).
Second, instead of leading to more mobilisation on the protesters side,
not reacting to protest with higher protrusion might reshape the overall
salience of an issue which is damaging for the reputation of political parties. Political parties are also problem solvers in the sense that they are
expected to solve certain social and societal crisis. Thus, in such a scenario
protest might still be successfully penetrating other external agendas, such
as the media agenda. This might lead the media to increase the coverage on
the topic of interest to protesters. Thus again, not reacting rhetorically to
protesters’ demands might not solve the pressure parties might feel when
faced with large scale, disruptive and durable protest.
Third, in the most disadvantageous scenario for political parties, ignored
protest with high protrusion could lead to public opinion shifts (Burstein
1979). Clearly, this is a scenario in which parties will face intensive costs
in the long run which could even reshape the environmental conditions
they compete in. Thus, new parties addressing the issue might arise and
challenge conventional politics successfully. A new challenger in the market will always result in costs for at least some established parties within
a political system (Bischof 2015). Thus, again, parties have an incentive to
undergo short-term costs by rhetorically adapting to protesters’ interests
instead of facing harsh long-term costs by losing votes or office.
In summary, protest with higher protrusion, meaning with large support, durability across time, a proper organisational structure and disruptiveness, can lead to increased costs if political parties do not adapt their
agenda accordingly. Due to these long-term costs, I suggest that political
parties become more likely to address the issue at stake with their rhetorical agenda:
H1(protest attention hypothesis) : The higher the protrusion of protest
the more likely political parties are to increase their attention to
the issue at stake during protests.
40
2.5 moderation of the protrusion of protest
This claim is the major hypothesis derived from the theoretical stipulations outlined above. The hypothesis provides the core theoretical structure
for the whole thesis. I also use it to develop further hypotheses that I test
in the upcoming chapters.
2.5
moderation of the protrusion of protest
While the protest attention hypothesis stipulated above is the main theoretical argument for all chapters to come, a range of other factors also
tend to influence parties’ rhetorical reactions to protesters’ claims. Figure 2
summarises the overall theoretical approach of my thesis which I discuss
briefly in the following section and extensively in the upcoming empirical
chapters.
Figure 2: Overview of theoretical framework of the study
causality
ideology
protrusion
of
protest
political
parties
rhetorical
reaction
public
opinion
electoral cycle
Source: Author’s own.
The protest attention hypothesis is represented in bold in figure 2. Demonstrators take to the streets, they successfully seek parties’ attention with
increased protrusion and then finally parties will adapt their rhetorical position to demonstrators’ claims. The major theoretical mechanism standing
behind this assumption is that parties might face higher costs in the long
run if they do not react to protest with high protrusion in the short-run.
41
2.5 moderation of the protrusion of protest
This argument is derived from the assumption that parties aim to secure or
expand their public support in an effort to secure votes and office (Strøm
1990; Harmel and Janda 1994).
Even though political parties might universally increase their attention to
protesters’ claims, they might not all react equally to protest. Originally parties were understood as vote-seekers, competing on a general left-right dimension and drifting towards a position on this dimension which was perceived to assure the highest vote gains in the upcoming elections (Downs
1957). More recent research suggests that parties’ general left-right ideology is not only an important factor for their vote-seeking incentives, but
also restrains political parties’ perceptions of their environment and their
understanding of shifting external conditions (Vliegenthart, Walgrave, and
Zicha 2013; Walgrave and Lefevere 2013; Jones and Baumgartner 2005). I
assume that such an effect could also be present for parties’ reactions to
protest. Thus, as outlined in figure 2 parties’ ideology could moderate how
they perceive and react to protest.
A rich body of literature proposes that the impact of protests depends
on favourable environmental circumstances, which it subsumes under the
term of “political opportunity structures” (see amongst others: Kitschelt
1986; Amenta, Carruthers, and Zylan 1992; Amenta, Dunleavy, and Bernstein 1994; McAdam 1982; Tilly 1978; Eisinger 1973).13 Studies tend to differentiate between “open” and “closed” structures and subsume countermovements, institutional arrangements and favourable interactions with public
opinion under the term opportunity structure (Kitschelt 1986; Meyer and
Staggenborg 1996; Giugni 2007; Agnone 2007). Re-evaluating the literature
review in this chapter, a major question is how public opinion as a major
external opportunity interacts with political protest. It seems that protest
might only have a sufficient impact on parties’ rhetorical positions if the
public at large supports the claims by protesters. Once parties perceive
an accordance between the masses’ and protesters’ demands a responsive
move across all parties is more likely. In this vein, protest can be seen as
the peak of the iceberg of grievances against a particular status quo. These
13 For an in depth overview of literature on opportunity structures see: Meyer (2004). Although political opportunity arguments have become the dominant theoretical force in
social movement studies, a common definition of what constitutes an opportunity is lacking (Meyer 2004: 134-135).
42
2.5 moderation of the protrusion of protest
two contingent factors, public opinion and parties’ ideology, are explored
in chapter 5.
Research on responsiveness by political elites stress that the timing of
stimuli for responsive activities is important. On election day the electorate
makes a decision to keep its previous representatives in office or to reshuffle them. Studies suggest that the temporarily closer the election day the
more are politicians’ activities under the limelight of public scrutiny. In
turn, politicians become more careful in their statements, but also more
likely to react in favour of public concerns in an effort to secure their (re)election.14 This mechanism of an electoral cycle is well-known in the American context (Kuklinski 1978; Tufte 1978; Cohen 1999; Canes-Wrone and
Shotts 2004; Rottinghaus 2008). Comparable mechanisms might also be at
stake for parties’ reactions to protesters’ claims. Chapter 6 looks into how
the timing of protest determines the reactions of political parties to them.
The first three empirical chapters (Chapters 4, 5 & 6) rely on the same
data coming from the ResponsiveGov project. They look into parties’ shortterm rhetorical reactions to protesters’ claims, while acknowledging that
protest might also have a long-term effect on the agendas of political parties. Chapter 7 uses data from the comparative manifesto project (MARPOR), the European Protest and Coercion dataset (EPCD) and several public opinion polls. The chapter examines the impact of the “silent revolution”
on party agendas (Inglehart 1971, 1977; Inglehart and Baker 2000). I analyse whether mobilisation on postmaterialist values had an impact on the
salience parties attach to postmaterialist values in their manifestos in 17
Western democracies (1981-1996). Thus, this final empirical chapter of the
thesis aims to deliver a triangulation of the results presented in the previous empirical chapters. Furthermore, it aims to partially address issues of
causality, suggesting that parties might not only react to protest, but instead
protesters might also react to party positions. I run instrumental variable
models to show that indeed parties react to protest and not the other way
around.
14 Note that Donald Trump in the current GOP race for the Presidential elections in 2016
seems to be an extreme and interesting counter example for this generalisation.
43
Part II
Measuring Rhetorical
Responsiveness
44
3
D ATA & M E T H O D S
efore turning our attention to the empirical research on the influence
of protests on party positions, a few words on case selection, data
and the measurements used in the following chapters are needed.
This chapter also describes how I conceptualise and measure parties’ rhetorical positions in the following three empirical chapters.
B
3.1
the responsivegov data
I use the ResponsiveGov data in the following three empirical chapters.
The last empirical chapter uses different data in an effort to triangulate the
results stemming from the chapters using the ResponiveGov data.15
To analyse whether party agendas react to protesters’ claims, longitudinal data of protest events and parties’ reactions to them are needed,
preferably reported on comparable ideological scales. The ResponsiveGov
data provide detailed information on protests and responses to them by
politicians. The ResponsiveGov project aims to find out the extent to which
“democratic governments [are] responsive to citizens’ demands and preferences between elections” (Morales 2014: 1). Data collection is thereby based
on pre-defined time periods, which are called “policy junctures”. Every juncture is bound to a certain policy issue. Thus, the project intends to code data
in relation to several issues.
From the policy junctures included in the ResponsiveGov data, I focus
on nuclear energy policy after the Fukushima catastrophe in 2011. Apart
from providing information on this event, the ResponsiveGov project will
eventually also provide information on up to five more issues.16 Given this,
the issue dimensions used for my study needed to be chosen carefully in
order to forestall biased conclusions due to specific attributes of certain is15 Chapter 7 has an extensive section on the data employed in the chapter.
16 The issues being: copyright policies; genetically modified crops; the financial crisis after
2007 and the Afghanistan war. At the time I was writing up my thesis the Fukushima
juncture was the only entirely finished juncture available.
45
3.1 the responsivegov data
sue dimensions (Geddes 1990). The reason for selecting the nuclear issue is
twofold. First, the issue used within this thesis needs to be conflicted along
party lines, meaning that parties need to have a public debate about the
matter. Parties must not only be outspoken about the issue, but also need
to take different positions on the matter. Otherwise, the hypothesis about
polarisation of party positions cannot be tested. Second, the issue needs
to achieve a meaningful salience on the party level and within the public.
Research shows that parties are less likely to be responsive to non-salient
issues and, thus, we might have very different expectations for such issues
than those of the assumptions put forward in the last chapter (Burstein
2014). Yet, the more important point is that we find enough mobilisation
on the protesters’ side to actually be able to link protest to parties’ positions.
Thus, in the case of the only other juncture that was available at the time of
my research, there was not enough mobilisation by protesters to be able to
reliably link protests with party positions. As we will see in the following
chapter (chapter 4), nuclear energy is a highly contested issue throughout
most Western democracies included in this thesis. Partly this is the case because after the Fukushima shock nuclear energy became a primary concern
in most Western democracies, but with very different consequences across
the sample of countries used in this study.
Following theoretical arguments by Soroka (2002b) nuclear energy can
be classified as a sensational issue. Sensational issues remain unobtrusive
to the public and thus are severely affected by external shocks or “focusing events” (Birkland 1998), such as the accident at the Fukushima nuclear plant.17 Somewhat at odds with Soroka’s classification salience varied largely across the countries included in this thesis. Due to this crosssectional variation the nuclear energy issue is an excellent case to look into
my theoretical arguments, as the variation helps to partly forestall selection
bias to highly salient topics. Previous literature on responsiveness and the
influence of protest tends to fall into the trap of generalising from results
based on highly salient issues (Soroka and Wlezien 2012; Burstein 2014:
5-18; Burstein and Linton 2002; Uba 2009: 437).
17 A focusing event is here defined as: “[. . . ] an event that is sudden; relatively uncommon;
can be reasonably defined as harmful or revealing the possibility of potentially greater
future harms; has harms that are concentrated in a particular geographical area or community of interest; and that is known to policy makers and the public simultaneously”
(Birkland 1998: 54).
46
3.1 the responsivegov data
The RepsonsiveGov project collects data by manual coding of the content of a country’s main newswire, legislative and parliamentary databases,
surveys and newspaper editorials. First, coders select the relevant news articles to be coded with an extensive keyword search. Second, coders extract any relevant event taking place during a certain policy juncture from
these newswires. Thus, the unit of coding is not a single publication of a
newswire, but all events reported within the publication relating to nuclear
energy. All acts included in the data involve demands, claims, declarations,
criticisms, or proposals related to the issue of nuclear energy. A wide range
of different types of events, ranging from speeches, acts, parliamentary debates and court rulings to protest events and public opinion polls are coded
(for a full list of events, see: Lühiste and Morales 2014). For example, in the
case of nuclear energy after the Fukushima juncture, Mona Sahlin (the actor), leader of the Socialdemokraterna party in Sweden (the organisation),
gave a speech (the event) on the 25.03.2011. She stated that nuclear energy
should not be pictured as the only solution for Swedish energy needs, but
instead green energy should be thought of as a viable alternative (the position). I will provide more detail about the positional coding in the following sections. The ResponsiveGov data provides a systematic coding of
Mona Sahlin’s statement, specifying a classification and date of the event,
the actor, actor’s organisation and the actor’s position. As such the ResponsiveGov data supply a detailed account of which actions parties, protesters
and the mobilised public undertook and which positions they reveal on
different policy issues across a total of 23 countries.18
The ResponsiveGov data on the Fukushima accident provides data for
USA, Belgium, Canada, France, Finland, Germany, Italy, Netherlands, Spain,
Sweden, Switzerland and the United Kingdom. At the time of the Fukushima
accident, all of these countries had either already used nuclear energy or
the parties in office had plans to phase it in again (Italy). The data collection
on the Fukushima juncture starts on the day it occurred (11.03.2011) and
ends two years later (31.03.2013). However, in case a government decided
to phase-out nuclear energy (Germany, Italy19 & Switzerland) or in case
general elections took place a minimum of half a year after the Fukushima
18 Reliability of coding is high with Krippendorf’s Alpha being 0.88 for the major variables
used in this thesis.
19 To be precise in the case of Italy the government withdrew from the plan to re-enter
nuclear energy.
47
3.2 party selection
accident (France, Spain & Netherlands), these events mark the end of the
coding period. The ResponsiveGov data contain any party activity reflecting on nuclear energy reported by a nations’ press agency, and the parliamentary or legislative database for all twelve countries included in the data
collection process.
Given that data collection ultimately ends once a government decides to
phase-out nuclear energy, protest mobilisation can hardly depend on these
policy decisions undertaken by governments. This provides the advantage
that protest mobilisation is not endogenous to parties’ policy decisions, but
pictures protesters’ dissatisfaction with the status quo.
3.2
party selection
The visibility of parties’ rhetorical positions in the media partly depends
on their relevance in Sartori’s (1976) sense, their incumbency status and the
time period of observation. Scholars have shown that parties’ size and office
positions are favourable factors for obtaining media attention (Hopmann
et al. 2010), with campaign periods representing more balanced coverage
by the media across all parties (Harris, Fury, and Lock 2006). Since the ResponsiveGov data mainly relies on manual coding of newswires stemming
from national press agencies, the selection of parties included in my study
takes parties’ ability to gain media attention into account. Even though this
study is not interested in the sheer amount of news coverage per party, the
selection of parties has to make sure that a substantial amount of access
to the media is guaranteed across all parties chosen for investigation. Otherwise, the results might be systematically biased by parties’ capacity to
penetrate the media agenda. Thus, parties included in this study have been
selected to assure that they fulfil Sartori’s (1976) ideas of party relevance,
as well as coalition and blackmailing potential.
As a general rule, parties that managed to secure at least five seats in
the national parliament and at least 5% of the national vote are included
in the analysis (For more details, please see: Bischof, Lühiste, and Morales
2013).20 Furthermore, parties needed to repeat this success at least once
20 Sceptical readers might suggest that parties should at least have made a certain number of
statements. However, such a selection would bias the sample. Following such a suggestion
would result in a selection bias on the dependent variable (Geddes 1990; Golder 2003: 434-
48
3.3 using rhetorical statements to measure party positions
during the period of interest of the ResponsiveGov project (1980 - 2013).21 I
also included all government parties during their time of office and parties
that helped to stabilise a minority government.22 This results in 67 parties
included in the following analysis (For an overview of these, see: Table 10
on page 165).
3.3
using rhetorical statements to measure party positions
First, an enormous amount of ready made data is already available to estimate ideological positions of political actors. First, roll-call votes can be
understood as the classical approach to estimate party positions (Poole and
Rosenthal 1985). While roll-call votes allow a fine-grained estimate of not
only party’s but also of individual MPs’ policy positions (Hug and Schulz
2007), in strongly disciplined party systems roll-call data frequently quantify conflict lines between government and opposition parties instead of
party positions (Debus 2009: 286-287). Given that, transferability from the
US case to parliamentary systems has proven to be difficult. Even though
studies show that there is a strong relation between MPs’ policy preferences and their voting behaviour in the EP (Gabel and Hix 2007), it is still
thinkable that other factors – party discipline; package deals – force MPs
to deviate from their policy preference when voting. Additionally, not all
votes in parliaments are recorded and not all issues make it to parliamentary floor as a bill. Hence, such measurements stem from biased samples
(Carrubba et al. 2006). Finally, voting in parliament should instead be understood as actual behaviour of politicians and not be conflated with MPs’
policy preferences or positions: roll-call votes should result to a large extent from MPs’ policy preferences, but are certainly not identical to their
preferences.
435; Heckman 1979). Excluding silent parties leads to a potential overestimation of the
effect of protest on party agendas because those parties not reacting to protest are not
represented in the sample.
21 While these rules pertain to most parties included in the following analysis, exceptions
have been made on several occasions to avoid ignoring country particularities (Please
see: Bischof, Lühiste, and Morales 2013). Table 10 on page 165 in the appendix gives an
overview of all included parties and reasons for their selection.
22 A minority government supporter is defined as: “Party in Support of Government are those
parties that are not represented at the ministerial level but which at the same time support
the investiture of that government” (Woldendorp, Keman, and Budge 2000: 15).
49
3.3 using rhetorical statements to measure party positions
Second, derived by asking experts or the public to place parties on a
given left-right scale, surveys have been used to estimate an aggregated
mean of parties’ positions (See for instance: Benoit and Laver 2007).23 The
most substantial disadvantages of surveys are: commonly low response
rates (For instance for Switzerland: Hug and Schulz 2007: 307); sparse observations across time, which veils party position shifts; and respondents
holding on to a single dimension of party competition which also accounts
for a high variation of interpretation of the dimensions included in this
single scale across country experts (Mair 2001: 17-18).24
Third, party positions can be assessed by investigating political texts,
with party manifestos as the most common source. The creation of such
data can either be undertaken by hand coding (Budge et al. 2001; Klingemann et al. 2006) or automated via software (Pennings and Keman 2002;
Laver, Benoit, and Garry 2003; Slapin and Proksch 2008). The main benefit of using party manifestos is their availability and comparability across
parties, countries and time. However, manifestos are only published immediately before elections. Therefore, meaningful policy position changes
between electoral campaigns, such as the German SPD’s Godesberger Programm (1959), remain unobservable to the researcher. Indeed, policy position changes that take place outside of the campaign period might be
reflected in the party manifesto for the next election as well, but this might
lead to a misinterpretation of the date and reasons behind such a policy
change.
In order to overcome such observational issues, parties’ policy reactions
in this study are measured by using their rhetorical statements as presented
in news agencies and coded by the ResponsiveGov project. By using such
data it is possible to generate detailed, issue bound positions of political
actors, be it single politicians or parties. However, since the ResponsiveGov
data is structured around certain policy issues and thereby ignores others,
a general party position on a left-right scale cannot be derived from the
data. While some scholars might spot this as a serious flaw, for my study
23 In case public opinion polls are used to compute party positions it is highly debatable
whether this actually quantifies party positions and not the public’s perception of party
position.
24 Furthermore, the well-known issues of any survey come into play here, namely: social
desirability; mistakes in retrospective placements etc. For a more detailed discussion of
the issues related to expert surveys, see: Mair (2001: 18-27).
50
3.3 using rhetorical statements to measure party positions
this assures the comparability of protesters’ claims, public opinion and how
political parties react by measuring all three on the same issue and scale.
As outlined in the next chapter, using parties’ rhetorical positions to
measure their attention to and position on an issue does not come without limitations. Firstly, generating such data is costly. Parties send out a
huge amount of messages via the media. All of these messages need to be
spotted, interpreted and coded by coders. Second, as discussed in the last
section, such measurements might be subject to media biases. Helbling and
Tresch (2011: 180) show that the selection of topics by the media leads to biased estimates derived from the media on how salient issues are for parties.
Yet, they also show that the actual content of party positions is reported
accurately and thus is valid for studying party position changes across
time by relying on media data (Helbling and Tresch 2011: 179-180). Since
this study is interested in parties’ positioning on a specific issue across
time and not their selective attention to an issue, using rhetorical positions
derived from manual coding of newswires appears to be a valid and generally fruitful approach for studying party responsiveness, as illustrated in
the upcoming chapters.
3.3.1
Recoding positions
Given that the ResponsiveGov project is interested in governments’ responsiveness, all positions coded within the project are recorded in relation to
the respective government’s initial policy position on the issue (For more information on the coding procedure, please see: Lühiste and Morales 2014).
In a first step, government positions are coded on a pre-defined five point
scale ranging from ‘very anti-nuclear’ to ‘very pro-nuclear’. This coding
is based on governments’ manifestos, coalition agreements or policy documents made prior to the start of the juncture.
In a second step, the positions coded during the data collection process
are compared to the original position of each government. Thus, coders
are asked to place any position made during an event on a five point scale,
ranging from ’(-2) an actor’s position is radically more anti-nuclear’ to ‘(2) an actor’s position is radically more pro-nuclear’ than the government’s initial policy
51
3.3 using rhetorical statements to measure party positions
position – with ‘(0) being the same position as the government’s initial position’.
Table 1: Rules for recodes of rhetorical positions
Initial government
position
-2
(Radically anti-nuclear)
-1
(slightly anti-nuclear)
1
(Slightly pro-nuclear)
2
(Radically pro-nuclear)
statements’ original
classification
-2
-1
0
1
2
-2
-1
0
1
2
-2
-1
0
1
2
-2
-1
0
1
2
position after
recode
anti-nuclear
anti-nuclear
anti-nuclear
anti-nuclear
pro-nuclear
anti-nuclear
anti-nuclear
anti-nuclear
pro-nuclear
pro-nuclear
anti-nuclear
anti-nuclear
pro-nuclear
pro-nuclear
pro-nuclear
anti-nuclear
pro-nuclear
pro-nuclear
pro-nuclear
pro-nuclear
Source: Author’s own, based on Lühiste et al. (2014: 12-16).
Since this study is interested in party rather than government positions,
the original coding scale is not comparable across countries for the purpose
of studying party positions due to varying initial government positions.
For example, events outlining a more anti-nuclear position than the government might still be in favour of nuclear energy depending on the government’s initial position in the country. Thus, the existing codes needed to
be recoded in order to assure comparability of rhetorical positions across
parties, countries, protests and public opinion. Therefore, I recoded all variables used from the ResponsiveGov project according to the rules outlined
in table 1 (see also: Lühiste et al. 2014: 12-16). For example in the case of
Italy, the Berlusconi government’s position prior to the Fukushima catas-
52
3.3 using rhetorical statements to measure party positions
trophe was to re-enter nuclear energy and construct new nuclear plants,
which was coded as ‘(2) very pro-nuclear’. Any position made during an
event which was classed between ‘-1’ and ‘2’ was then recoded into an activity supporting nuclear energy, while any action classified as ‘-2’ was recoded into an activity rejecting the use of nuclear energy. These rules were
used to recode party rhetoric, public opinion surveys and protest events.
The positions posted during protest events by political parties or public
opinion surveys are classified as being against nuclear energy and in favour
of nuclear energy, rather than on a five point scale relying on the initial
government position. This assures that all of these three variables can be
compared with each other.
3.3.2
Parties’ rhetorical positions
Since parties undertake a wide range of activities and not of all of these can
be subsumed as rhetorical, events which can be understood as such need to
be selected. In all upcoming analyses I decided to understand rhetoric as:
statements to the media (interviews); press conferences; any kind of speech
made during assemblies or party meetings; public letters, including tweets;
statements/speeches given during rallies and campaign events; party resolutions and declarations; parliamentary questions; statements given during
hearings or during any other event.25 All of these activities are undertaken
to persuade the public or to share information on a party’s position with
the public. None of these are events of substantial policy making themselves, however they might reflect on past policy decisions or present upcoming ones. Yet, none of these events directly result in legislative output.
They might describe an upcoming policy proposal or a past one, but cannot
be confound with substantial policy making itself.
Furthermore, I only included statements made by national-level politicians and the government. Neither statements by parties’ regional politicians, nor members of the European Parliament are included. Regional and
supranational politics differ substantially from national politics. Therefore
25 The attentive reader might suggest that party actors could present different positions via
twitter than in a press release, since attention to a press release by the media should be
considerably higher than to a tweet for most actors. However, recall that all these events
are coded from newswire releases. Thus, the statements are all equally important to the
media, as otherwise they would have not been recognised and reported by the media.
53
3.3 using rhetorical statements to measure party positions
party members of regional or supranational branches often deviate from
the national party line. Since this study is not interested in party factionalism between regional, national and supranational politics, these statements
have been excluded from the analysis.
As outlined above, all events are matched into two ideological categories,
one outlining support for nuclear energy and the other rejecting the usage
of nuclear energy. In a first step I counted the number of pro- and antinuclear rhetorical statements made by each party each month.26 In a second
step, all rhetorical activities were aggregated by party and months using the
following formula:
Rhetorical positionit = log(∑ pro nuclearit + 0.5) − log(∑ anti nuclearit + 0.5)
(1)
I subtracted the sum of “anti nuclear” rhetorical events from the sum of “pro
nuclear” rhetorical events for each party per month, comparable to measurements of parties’ left-right placements.27 Thus, values greater than zero
indicate rhetorical positions favouring nuclear energy, while values below
zero indicate positions against the use of nuclear energy. Just subtracting
pro- and anti-nuclear talk results in a highly skewed measurement, with
values on the extremes being heavily overrepresented. This might substantially hamper the validity of the results reported in the next section. To
address this issue, I use a log transformation to control for skewness, as
shown in equation (1).28 Such methods are well established and follow a
26 Ex ante there is no reason to stick to a monthly measurement. One might also think of a
weekly or daily aggregation period. Yet, the smaller the time periods, the more parties will
not talk about an issue at all. Thus, it is useful to find mid ranging time intervals which
ensure detailed coverage of party rhetoric without artificially increasing the zeros of the
measurement.
27 This means that zeros can have two meanings. Either the party did not talk about nuclear
energy at all, or the party is completely divided regarding the nuclear issue. The latter case
only occurred ten times out of 1165 possibilities. Running the upcoming regression analysis excluding these ten cases does not substantially change the results. The results were
also similar once I ran regressions with a separate a pro- and an anti-nuclear rhetoric variable. However, my proposed measurement eases interpretation of the regression results
and was therefore preferred to the other two outlined options for the following analysis.
28 log(0) is undefined. I chose to introduce ‘0.5’ into the equation which assures that in the
case a party did not talk at all, it is not undefined in the measurement but represents the
middle of the scale with zero.
54
3.4 measuring the protrusion of protests
more comparable logic than for instance log odds-ratios for bounded count
data.
In addition to controlling for skewness, the interpretation of a measurement based on a logged ratio is appealing, since through logging change
is not defined by the absolute difference between the counts of pro- and
anti-nuclear positions but by their ratio (Lowe et al. 2011: 130-132). In the
measurement outlined in equation 1 the marginal effect of a single piece of
rhetoric decreases with the number of statements already made publicly on
the issue of nuclear energy. This is a particularly interesting side-effect that
fits very well with psychological literature on how human beings interpret
written text and messages in general as has been described elsewhere in
detail (Lowe et al. 2011: 130-132).
Certainly several parties remained silent on nuclear energy at different
points of the analysis period. To estimate whether this was voluntary or
whether the media chose not to report their positions, I run a logistic regression with the dependent variable being ‘1’ in case a party did talk
about nuclear energy in a given month and ‘0’ otherwise. Following the
previous discussion about party size and incumbency affecting whether
or not the media absorbs party claims, I included party size (vote-share
and seat-share) and incumbency status as covariates (table 11 in the appendix). While more seats are a favourable and significant factor in gaining
media attention, the effect size is comparably small with the probability
of the media reporting the party position increasing by 1 %. Neither vote
share nor incumbency are significant factors for the media penetration. In
summary, it appears that the selection criteria for parties assure the comparability across parties and countries of the rhetorical position measurement
and that the sample does not suffer from a substantial media bias towards
larger and/or governing parties.
3.4
measuring the protrusion of protests
Data to measure protrusion of protest stem also from the data gathered by
the ResponsiveGov project. As discussed in the theoretical section, protests
come in different size and shapes. Instead of using the raw data as provided
55
3.4 measuring the protrusion of protests
by the ResponsiveGov project, I created an index measuring protrusion of
protest (protest-index) for each country and month.29
First, the characteristics for each protest are recoded in eight binary variables reflecting the relevant characteristics of protest. The variables are: (1)
more than 100 participants; (2) more than 1000 participants; (3) more than
10,000 participants; (4) duration of two or more days; (5) any organisational
support; (6) illegal protest; (7) violent protest; (8) one or more participants
wounded.30 Second, the binary variables are added for each protest event,
with each event receiving a score ranging from between 0 and 8. Next, I
aggregated the protrusion of all protest events for each country and month.
In case the ResponsiveGov data reported any form of counter-mobilisation,
these protests’ protrusion was subtracted from the protest-index.31 Two
protests in France supported the use of nuclear energy in an effort to secure jobs at a local nuclear power plant. Thus, they are defined as countermobilisation, while the remaining protests communicate an anti-nuclear position. In several instances the media interviewed politicians joining protest
events or politicians gave speeches during the protest event. In case a party
participated in a protest as an actor, I excluded the protest event due to potential issues of endogeneity. However, the results reported in the following
section are robust to the inclusion of all protest events, meaning that the
effects reported in the following chapters do not depend on this decision.
Finally, the protest-index is highly right-skewed. I logged the protest-index
to control for its right-skewness.32
Figure 3.4(a) reports the distribution of the protest-index for each protest
event coded within the ResponsiveGov data. As suggested earlier, most
29 Included are protests (including vigils), protest camps, any form of symbolic action, blockades and any form of occupation of land.
30 Using three binary variables to measure protest size, assures that protest size is weighted
higher than the remaining characteristics. This decision was made deliberately based on
arguments about the pre-eminence of protest size (Lohmann 1993; DeNardo 1985).
31 Note, however, that the results do not change if these protests are included. Some readers
might suggest that excluding the statements from the dependent variable would be the
best means for controlling for this issue. However, this would again result in a selection
bias on the dependent variable. Yet, even more important is the fact that it would be
unclear how to deal with protests in which political parties themselves were present as an
actor. The question is then raised as to whether other parties then take the protest more
seriously, or disregard it so as not to set a favourable stage for competitors. This itself is a
separate research question to be addressed.
32 To be precise I added 1 as a constant before logging the index: logged protest-index =
log(protest-index + 1). This was done to control for the fact that log(0) is undefined. Logging also makes the results more comprehensible (Gelman 2008).
56
3.4 measuring the protrusion of protests
Figure 3: Distribution of protest-index
−2 s.d.
−1 s.d.
mean
+1 s.d.
+2 s.d.
70
70
63
60
54
Frequency
50
40
30
20
16
10
3
2
0
−1
0
1
2
3
4
protestindex
(a) Histogram of aggregated characteristics for each
protest
(6,17.3]
(3.95,6]
(2.5,3.95]
[.4,2.5]
No data
(b) Average protest-index per country
Source: Author’s own.
Note: Figure (a) shows the distribution of characteristics for each protest coded within
the ResponsiveGov project. Figure (b) reports the average protest-index per country and
month. Figure b) omits USA (=1.0) and Canada (=2.2) to assure readability. Protest-index
is split into quantiles.
57
3.4 measuring the protrusion of protests
protests have low protrusion and a maximum of four out of the eight characteristics are fulfilled by three protest events. In summary, three quarters
(74 %) of all protest events coded in the ResponsiveGov data managed to
obtain at least one of the eight characteristics of protrusion. Figure 3.4(b)
shows the average protest-index for each country for the whole period of
observation. The map underpins observations shared in previous studies,
describing Germany and Italy as the countries that experienced the highest
public pressure by social movements to phase-out nuclear energy (Jahn and
Korolczuk 2012; Ramana 2013: 68-69). While France, Spain and Switzerland
also experienced significant upheaval against the usage of nuclear energy,
the remaining countries were largely spared from large-scale protests.
The characteristics measured in the protest-index have been carefully
chosen according to the results of previous research on the successful impact of protest. While the inclusion of protest size, duration and organisational support (1-5) appear to be theoretically straightforward based on
arguments found in previous contributions to the field of social movement
studies (Walgrave and Vliegenthart 2012; Soule and Olzak 2004; Lohmann
1993; DeNardo 1985), the characteristics of illegality, violence and wounded
participants deserve more attention (6-8).33 Since Gamson’s studies (1990),
scholars have repeatedly argued that the disruptiveness of protest increases
protestors’ chances of being heard and visible to the public, the media and
party politics. Recent research on the effect of disruptiveness of protest on
party agendas supports Gamson’s theoretical arguments (McAdam and Su
2002). Wounded people not only signal a certain amount of commitment to
the course of the protest, but more importantly focus the media’s attention
to the event. In case protesters are wounded during a protest in a democracy, questions tend to follow about how and why it happened. Examples
of the media and parties’ increased attention to a protest following the
wounding of protesters are numerous, for instance, wounded protesters after the shooting of Michael Brown in Ferguson (USA) (Alcindor and Bello
2014) or protesters harmed by police monitors during the protest against
the new railway station in Stuttgart (Germany) (Marquart 2010). After considering these illustrative examples, it seems to be appropriate to include
33 Illegal protest are explicitly framed by the media as such (e.g. lacking a protest permit),
occupations or blockades of land/buildings. If it is clear from the source that the protesters
initiated the violence, a protest event is categorised as violent.
58
3.5 public opinion & controls
injuries of protesters as another characteristic that should lead protests to
protrusively affect political parties’ agendas.
Like most protest event data, the ResponsiveGov data relies on codes
originating from the media. Even though media data still represents the
gold-standard in protest event coding, cross validation of different data-sets
shows considerable deviation depending on which media source was employed (Nam 2006). Information about the number of participants showed
particularly considerable variation. Thus, using binary variables to measure the presence of certain characteristics mitigate potential biases, e.g.
the variation of protest duration usually increases with its endurance, but
rarely contradicts the fact that a protest lasted longer than a single day.34
In contrast to previous research using comparable measures (Gillion
2012: 955-956; Gillion 2013), the protrusion of each component has been
tested by means of regressing each component separately with the amount
of statements each party published on nuclear energy.35 The results reveal
that the more of the eight characteristics that are ticked off, the more likely
a protest is to raise the parties’ level of attention. These results substantiate
the theoretical arguments outlined above and justify the decision to include
the eight components into a single aggregated index aiming to measure the
protrusion of protest.
3.5
public opinion & controls
Public opinion is measured using two variables. First, the position of public opinion is measured by subtracting the percentage of respondents who
disagree with the usage of nuclear energy from the percentage of respondents who favour it, and also comes from the ResponsiveGov data. Thus,
34 In case the ResponsiveGov data reported more than one informational resource for the
size of protests, I used the mean across all informational resources to decide whether a
protest mobilised more than 100 or 1000 participants. Even though there is high variation
across the different sources, in all instances the sources agreed whether a protest was
larger than 100 or 1000 participants.
35 Please note this is not an indirect selection of the dependent variable. The following analysis tries to link protest with party positions and not the sheer amount of talk. E.g. while
parties increase their attention to violent protests, they might still be more likely to reject
protesters’ claims. Factor analysis is not a suitable option for validating the measurement,
since a) the variables are measured as binaries and b) an empirical correlation between
each category would not substantiate the index, but rather show that most characteristics
go hand in hand with each other.
59
3.5 public opinion & controls
60
higher values on the public opinion scale indicate a more pro nuclear public
mood.36 Second, comparable to other studies, the number of respondents
who name environmental issues as the most important problem/issue in
each country is used to measure the salience of the issue of nuclear energy
(Jennings and Wlezien 2011).37 As suggested in the second set of assumptions, parties might only respond to issues which are salient, while ignoring
protest and public opinion on non-salient issues. This might mediate the
effect of the electoral cycle. Political parties have mainly framed the the usage of nuclear energy after the accident in Fukushima as an environmental
issue. Thus, using the number of people naming the environment as the
most important problem facing the nation should depict the salience of the
nuclear issue for the public well.
The following chapters use a range of controls. I include binary variables
for green parties and government parties. While green parties are expected
to support the anti-nuclear stance of protesters, government parties are suspected to speak out against protesters’ interests. Furthermore, I control for
party ideology by using the logged rile score stemming from the manifesto
data project (Lowe et al. 2011). Finally, parties might be influenced by their
country’s use of and or dependence on nuclear energy. Instead of calling
for reforms they might favour the more convenient current situation, which
does not force them to replace existing policies. I used the amount of energy produced by nuclear reactors in a given country to measure the status
quo. The data come from the World Nuclear Association and are produced
annually.38
36 Public opinion polls have been selected to either reflect a respondent’s position towards
a country’s status quo policy or her/his general opinion on the use of nuclear energy, and
come also from the ResponsiveGov data.
37 Data stem from the Eurobarometer for European countries; the Sorgenbarometer for
Switzerland; Gallup for the USA; and Focus Canada Report for Canada, and are measured on a yearly basis (with the exception of quarterly for the US). In contrast to the
remaining covariates, this measure varies on a yearly level. Thus, I also replaced the measure with a monthly salience measure using the amount of total newswire pieces coded
in the ResponsiveGov data (comparable to, also see for a validation of the measure: Lax
and Phillips 2012). The substantial interpretation of the results does not change if I use
this measure instead.
38 http://www.world-nuclear.org/info/Facts-and-Figures/Nuclear-generation-by-country/,
last checked: 07.01.2015.
Part III
Parties and Protest: An Empirical
Analysis
61
4
I S I T W O RT H T H E E F F O RT ? T H E M E A N S A N D E N D S O F
PA R T I E S ’ R H E T O R I C A L A C T I V I T I E S
“Politics is oftentimes little more than an endless exchange of public
conversations between those who control scarce resources and those who
wish to control them” (Hart 1987: 5).
4.1
introduction
s Hart (1987: 5) points out in the quote above, politicians reach the
public to an enormous extent through words. For instance: candidates duel on TV; parliamentarians fight with words during
debate sessions, which the public can observe; politicians answer question
from the press; send press releases; and hold speeches at all kinds of public
gatherings. Thus, the public might rightly have the impression that politics
is mainly a battle of words and not of substantive policy-making. Interestingly, these rhetorical activities have largely been neglected by research on
party politics which is mostly bound to campaign periods (Meyer and Wagner 2015). Since this thesis focuses primarily on such rhetorical events, this
chapter aims to understand the meaning of such events for political parties.
In this chapter I treat parties’ rhetorical positions mainly as a dependent
variable and try to validate the measurement I proposed in the last chapter.
Politicians’ rhetorical activities have mainly been studied by looking at
their inaugural addresses – such as the state of the union address given by
the American president (Cohen 1995; Canes-Wrone 2001b; Druckman and
Holmes 2004) or the Queen’s speech in the UK (Hobolt and Klemmensen
2008; Jennings, Bevan, and John 2011), which takes place at very distinctive
points in time and gains ample attention by the public and media alike.
In contrast, politicians’ daily rhetorical activities, press releases, statements
during interviews, public speeches, party meetings or parliamentary debates, have attracted less interest in the literature. Lately, several studies
A
62
4.1 introduction
have started to fill this gap by analysing parliamentary questions (Saalfeld
and Bischof 2013; Martin 2011; Russo and Wiberg 2010) and parties’ press
releases (Senninger and Wagner 2015; Hopmann et al. 2010; Brandenburg
2002). Rhetorical activities are often understood as symbolic acts and rarely
as a means of providing substantial policy frames to the public (For an
overview of the debate, see: Cohen 1999: 24-26). If not interpreted as the
latter, studies usually claimed theoretically that rhetorical activities can be
substantive but rarely delivered analysis showing that this is actually the
case. Rhetorical activities, thus, have been often described as “hollow gestures” (Gillion 2014: 2). However the link between parties’ rhetorical activities and their policy outputs still needs to be studied in more contexts in
order to be able to make stronger claims. Testing the congruence between
how parties’ talk and what they subsequently deliver in their policy outputs
(e.g. bills, legislative acts and laws) in parliaments however can not only
tell us how resilient parties’ rhetorical activities are, but also enhance our
knowledge of the relationship between concepts of symbolic and substantive representation (Pitkin 1967).
In this chapter I show that parties’ substantive attempts to influence their
country’s nuclear policy (e.g. via parliamentary motions, bills, legislative
acts and laws) reflect their policy positions, as shared through their rhetorical activities (via press releases, interviews, public speeches, party meetings or parliamentary debates). I show this by analysing ResponsiveGov
data about parties’ rhetorical activities and their policy outputs on nuclear
energy after the Fukushima nuclear meltdown. In order to do this, I proceed in a two-step manner in this chapter. First, I present theoretical arguments as to why rhetorical activities are likely to reflect parties’ manifesto
positions and ideology. Drawing on these theoretical arguments I derive hypotheses linking parties’ rhetorical activities to policy outputs. Second, by
means of face validity I present evidence that parties’ rhetorical activities
are indeed structured around their a) manifesto positions and b) party families’
traditional positions on nuclear energy. Then I link these rhetorical positions
to parties’ policy outputs. The final analysis shows that party rhetoric is
indeed congruent to policy outputs, even though this effect is mediated by
being in government. Incumbent parties are more likely to deliver policy
outputs in line with their rhetorical positions. This is because incumbent
63
4.2 parties between symbolism & policy-making
parties’ track records are tested more severely as citizens’ hold them accountable to deliver policies.
The following section presents the current debate on parties’ rhetorical
activities and their policy output. The third section outlines why rhetorical activities can be understood as substantive policy signals by parties.
The fourth section describes the methods used in the analysis, while the
fifth section shows the results of the pooled regression analysis. Section six
concludes.
4.2
parties between symbolism & policy-making
The ideological positions parties hold and compete on in modern democracies are one of the core interests of political science (Downs 1957; Budge
et al. 2001). The literature on party positioning is manifold and shows that
party positions matter for parties’ vote- (Adams et al. 2006), office- (Bäck,
Debus, and Dumont 2011) and policy-seeking incentives (Budge and Hofferbert 1990). Yet, party positions are measured mostly via extractions from
electoral programs (Budge et al. 2001) or via expert surveys (Castles and
Mair 1984). In the former case, party positions can only be measured during
campaign periods when manifestos are published. This assures large samples for comparisons across several countries and time periods, but leaves
party position shifts between elections untouched. Expert surveys, on the
other hand, depend on the availability of suitable experts and their subjective perception of changes in party position across time. In summary, even
though there are valid reasons for both approaches, and they have proved
reliable for a range of research purposes, they lack the capability of indicating party positions at more frequent time intervals, especially between
elections, as discussed in the last chapter.
Expert judgements and manifesto positions augment political science
knowledge about party behaviour. By using these measurements, scholars
show that incumbent parties’ policy positions taken from manifestos are
an excellent predictor for what they subsequently do while in government
in a variety of different institutional settings: for Spain (Artés 2013), the
Netherlands (Thomson 2001), the UK (Bara 2005; Hofferbert and Budge
1992), the USA (King et al. 1993), Greece (Kalogeropoulou 1989), Ireland
64
4.2 parties between symbolism & policy-making
(Costello and Thomson 2008) and in a comparative perspective (Mansergh
and Thomson 2007). As Budge and Hofferbert (1990: 112) show, studies report that impressively around 70 % of pledges lead on to substantive policies and a recent review confirms this observation (Pétry and Collette 2009).
Consequently, we know how substantive promises made in manifestos of
incumbent parties are.
In contrast, parties self-selected expression of their position on and attention to an issue on a daily basis – made in press releases, interviews, public
speeches, party meetings or parliamentary debates – have attracted less interest in the literature on party politics and representation (Gillion 2014:
5). If discussed, rhetorical activities are often understood as sheer symbolic
acts and as means of symbolic representation (Pitkin 1967: 92-111).
In the American context, the president’s State of the Union Address has
found large appeal amongst scholars of presidential responsiveness. On the
one hand, presidential speeches are seen as symbolism, containing little, if
any, policy content (Tulis 1987; Hinckley 1990). In this vein presidential
rhetorical activities do not share policy information, but are symbols for
increasing presidential popularity. Presidents seek support from the public
to put Congress under pressure to deliver policy, in order to pass legislation
which underpins their re-election goals (Ragsdale 1984; Brace and Hinckley
1993). Theoretically akin, but distinct in their arguments, studies showed
that rhetorical activities are successfully used for priming (Druckman and
Holmes 2004), creating trust among the public (Bianco 1994; McGraw, Best,
and Timpone 1995) and altering the public’s perception of the president’s
issue competence (Holian 2004).
Scholars have also suggested that speeches are used to set the president
or government’s policy agendas (Cohen 1995, 1999; Jennings and John 2009;
Bernardi 2014). Rhetorical activities are therefore understood as a preceding step to policies, with the priorities and positions presented in speeches
likely to result in substantive policy making. Bevan, John, and Jennings
(2011) for instance show that the Queen’s speech is a strong and positive
predictor of legislative outputs in the UK. Most studies showed that an increase in bills (Bevan, John, and Jennings 2011) or public spending (Budge
and Hofferbert 1990) on a certain issue was preceded by an increase of
attention on the issue in speeches or manifestos. Consequently, these studies did not differentiate among different types of attention, with all men-
65
4.3 rhetorical activities, manifesto positions & policy
tions of a policy issues being treated equally regardless of their ideological
frames and positions (Cohen 1999: 67).
Studies focusing on party rhetoric between elections and accounting for
positions shared by rhetorical activities are scarce (Helbling and Tresch
2011: 174). This is mainly because scholars measure party positions by
sticking to well established approaches, namely survey or manifesto-driven
techniques (Keman 2007: 77-78), which are bound time-wise either to elections in the case of manifestos or the period of data-collection in the case
of expert surveys. While political science has found substantial evidence
that mass media seems to have a profound impact on public perceptions of
parties (Schmitt-Beck 2003; Baum 2013), we know very little about the policy positions shared by politicians via the media (Helbling and Tresch 2011:
174-175; Kriesi et al. 2008) and even less about the relationship between
parties’ rhetorical activities and their policy output (Gillion 2014).
4.3
rhetorical activities, manifesto positions & policy
Using parties’ rhetorical activities to measure their positions comes with
several advantages to traditional techniques (for an extensive comparison
please see, Helbling and Tresch 2011: 174-177). First, parties’ rhetorical activities are a daily routine. Therefore, they can help us to fill the positional
blackbox between elections.39 Second, rhetorical activities allow an issuebased evaluation of party positions. Such issue-based measurements are
especially important once we try to estimate party congruence or responsiveness to, for example, public opinion (Lax and Phillips 2009: 368-369) or
protest, as I do in the upcoming chapters. Third, voters can be assumed
to largely rely on cost-low information to judge parties’ positional offerings. Voters can reasonably assess rhetorical activities via TV, newspapers,
campaign events, public speeches or public assemblies. In fact, rhetorical
activities in the media might even reach voters without them seeking such
information. Just alongside everyday activities people are often exposed to
party talk when listening to the radio, watching TV, reading newspapers, or
checking Twitter. This, however, is not the fact for party manifestos, which
39 Expert surveys cover this period as well, but scholars are not capable of disentangling the
bases on which experts evaluate parties. Especially once experts are asked to judge on
party positions retrospectively (Mair 2001: 24-25).
66
4.3 rhetorical activities, manifesto positions & policy
are often extensive documents, partly filled with complex jargon, and ignored by many voters (Bara 2005: 586 & 597). Consequently, nowadays the
public mainly keeps itself informed about politics via the media (Baum
2013: 442). Hence, leading party figures are likely to be cautious about the
positions they share with the public, in other words, voters, when talking.
Given these arguments about why parties’ rhetorical activities seem to
matter for voters’ perception of them, it is surprising how little scholars
of party politics know about parties’ rhetorical position taking. The main
argument on which I draw below, is that party rhetoric mainly reflects on
the traditional positions parties and party families stand for. Thus, I argue
that parties’ rhetorical activities should be similar across parties within the
same party family. Furthermore, they should be committed to past or future
manifesto positions, as is argued below.
Once parties’ rhetorical activities are understood as costly signals for
politicians by creating public audience costs (Fearon 1997), politicians are
unlikely to deviate from the party line. Voters listen to rhetorical activities presented in the media and judge parties based on the signals parties
send. Between elections parties do not draft party manifestos. However, the
political agenda is constantly in flux. Specifically between elections, party
positions might change in an effort to adapt to external stimuli. Thus, between elections rhetorical activities essentially mark out which set of policies parties regard as feasible and sensible and which options they disregard. Above all, rhetorical activities give parties the opportunity to frame
the debate on relevant issues or to prime one issue above the other (GreenPedersen and Stubager 2010: 665; Druckman and Holmes 2004). Voters also
expect parties to position themselves in the political space between elections, especially if public opinion shifts or focusing events draw the attention to an unresolved issue. Therefore, remaining silent results in loosing
the capability to influence how a topic is framed and therefore appears to
be risky business (Green-Pedersen and Mortensen 2010).
Once a larger fraction of a party is immune to party discipline and party
members send contrary policy cues, the overall party position blurs. However, position ambiguity is harmful for parties’ vote-seeking goals. Especially for salient issues, voters are likely to penalise parties without a clear
cut message (Dalton 1985: 294; Ferrara and Weishaupt 2004: 299). For that
reason maximising ambiguity is often not a sensible and profitable strategy
67
4.3 rhetorical activities, manifesto positions & policy
(Schedler 1998: 196). This should especially be the case for nuclear energy,
which became a highly salient topic in several countries included in this
study. Due to the Fukushima shock the public focused on what politicians
had to say about this issue in the public sphere, especially in Italy and Germany where the topic was already on the political agenda before the shock.
Party leaders need to make sure that members follow the party line and
party members themselves should have an interest to follow the party line,
too. Especially in times of strong party leadership accompanied by rising
mass media influences, leading figures of parties are likely to have an eye
on their cadre’s and member’s rhetorical activities (Poguntke and Webb
2005).40
However, changing environmental circumstances provide parties with
opportunities to update and change their believes (Harmel and Janda 1994;
Schedler 1998: 202-206; Pétry and Collette 2009: 66). Unpredictable events
– such as the Fukushima accident – can lead parties to rethink their policy
promises and update them in order to produce reliable positions the party
can follow through with policy. In those situations parties might be able
to renege their initial positions and start to forward new messages to voters, divergent to their initial policy promises laid out in party manifestos.
Doing so might in fact be the only possibility to not lie to the voter. Yet,
again parties should aim to keep the amount of external debate about their
policy shift low in an effort to not send an ambiguous policy message to
voters. Only when the party agreed on a position should members be more
outspoken about the positional change.
I firstly assume that parties’ rhetorical activities either echo past or anticipate future party manifesto positions. Since party positions might change
between elections, such changes are likely to be reflected in the rhetorical
cues parties share between elections. Switchers are therefore likely to not
talk in line with their bygone manifesto position, but rather anticipate the
upcoming one.
Secondly, I expect parties’ talk to be structured around party families.
Nuclear energy was largely neglected by political parties until the 1980s.
Yet, with Green parties fitting their party agendas and ideology around
40 However, lately it has been shown that parties might benefit from blurring their position
on some issues not part of their raison d’être, while sharpening them on issues which
mirror their core ideology (Rovny 2012b,a).
68
4.3 rhetorical activities, manifesto positions & policy
anti-nuclear and peace movements, nuclear energy has become an ever
more present issue across Western party systems. Thereby, nuclear energy
can be understood as a sensational issue which remains rather unobtrusive
to the public (Soroka 2002a; Walgrave, Soroka, and Nuytemans 2007; GreenPedersen and Stubager 2010). Though unobtrusive to the public in general,
external events or shocks cannot only lead to an incremental increase of
attention by political parties to sensational issues but also in the course to
party position shifts, as argued above.
This is also accurate for nuclear energy, which remained an issue mainly
debated by Green parties and environmental movements throughout the
1970s and 1980s. Yet, the Three Mile Island accident (1979) and the Chernobyl meltdown (1986) cast doubt on the safety of nuclear energy all over
the globe and increased the attention of the public, media and parties to the
topic. Several Social Democratic parties also started to question the safety
of nuclear energy, with Social Democratic parties in Sweden, Germany, Belgium, the Netherlands and Switzerland declaring themselves against the
building of new nuclear reactors even before the 1990s (Nohrstedt 2008: 264;
Koopmans and Duyvendak 1995: 239; Swyngedouw N.d.: 12; Aarts and Arentsen N.d.: 11; Kriesi N.d.: 19). Though, the Parti Socialiste (PS) in France
remained unperturbed by their family members’ switches at first, internal
debates became public on the nuclear issue in the early 1990s. Even though
the PS subsequently updated its ideology towards more anti-nuclear sentiments to attract the Green party (Les Verts) as a coalition partner under
Jospin, these positional shifts remained largely symbolic and did not result
in a renewed manifesto position on the matter (Brouard and Guinaudeau
2013: 19).
In contrast, Conservative and Christian Democratic parties largely kept
their pro-nuclear stances before the Fukushima disaster. Especially in Canada,
the US, and the UK there appears to be system wide pro nuclear agreement,
with Green parties never having been represented in the parliament.41 Consequently, within these party systems, mainstream parties never had to react to the challenges provided by niche parties such as the Greens, while
they often did so in the aforementioned cases (Meguid 2005; Spoon 2011;
Spoon, Hobolt, and de Vries 2014).
41 Though the Green party in Canada now holds one seat since the 2012 Canadian federal
election.
69
4.3 rhetorical activities, manifesto positions & policy
In summary, I expect that the nuclear issue reflects, to a certain extent,
the traditional divide between party families, with left parties leaning towards anti-nuclear positions and right parties leaning towards pro-nuclear
positions. Because of this, parties’ rhetorical activities on the nuclear issue
should also mirror a left-right divide after the Fukushima accident. Furthermore, politicians are likely to mirror their party’s manifesto positions
when talking in order to not blur the party message between elections.42
4.3.1
Linking parties’ rhetorical activities & policy outputs
After having outlined my expectations for the descriptive part of the chapter, the following section will derive the hypotheses to link parties’ rhetorical activities to their attempts to influence policy making in their respective
countries. If parties’ rhetorical activities are structured policy talk and do
not only reflect cycles of issue attention, the question arises which binding
meanings rhetorical activities can have for political parties. The question
is whether rhetorical activities are sheer symbols or whether they represent cues for policy-making. Are rhetorical activities congruent to parties’
more substantive policy outputs (e.g. legislative proposals; bills)? And if
so, which parties are likely to keep their rhetoric congruent to their policy
proposals?
The literature shows that parties which dedicate more attention to a certain topic, are subsequently also more likely to provide policies on the same
issue. This has been shown for governmental speeches (Bevan, John, and
Jennings 2011) and for salient topics in party manifestos (Budge and Hofferbert 1990). Subsequently we can assume that parties’ rhetorical activities
should also reflect parties’ intentions to influence policies in a congruent
manner. While parties’ rhetorical activities do not state a credible commitment comparable to speeches or party manifestos, they still raise audience
costs through voters’ attention to party talk (Fearon 1997) – as has been
argued above in the last section of this chapter:
H1 : Parties are likely to deliver political decisions which are congruent to their rhetorical activities.
42 I do not stipulate hypotheses here, since the following analysis will remain descriptive in
most regards.
70
4.3 rhetorical activities, manifesto positions & policy
Yet, whether parties follow through with their talk might be moderated
by parties’ “raison d’être”. Parties might talk about issues in order to attract
voters or accommodate issues (Meguid 2005, 2007) without perceiving the
necessity to push for congruent policies. They could use rhetorical actions
as symbols to persuade voters and they might manipulate politics in a
way which enforces their electoral goals without intending to alter policy
(Riker 1996). Thus, by “talking” they strategically change public discourses
and voters subsequently hold them competent on a topic without parties
ever having aimed to implement policies. In contrast, parties which are
already perceived as competent for a certain issue are likely to push for
congruent policy implementation. In the case of nuclear energy, which has
been largely framed as an environmental issue after Fukushima, Green
parties are traditionally perceived by voters as being competent (Walgrave,
Lefevere, and Nuytemans 2009).43 They own the issue of nuclear energy
and profit from their attempts to push for congruent policies, since this
draws more media attention on an issue from wich Green parties are likely
to benefit electorally:
H2 : In the case of nuclear energy Green parties are more likely
to influence policy implementation congruent to their rhetorical
activities.
As has been argued elsewhere, government and opposition parties find
themselves in completely different situations when it comes to setting issue agendas (Green-Pedersen and Mortensen 2010; Walgrave, Lefevere, and
Nuytemans 2009). Governments need to be responsive to ever changing
environmental conditions, since they are held accountable for a nation’s
well-being. Therefore, they often cannot remain silent on salient issues, otherwise voters apprehend the government as being incapable of delivering
solutions for relevant issues (Green-Pedersen and Mortensen 2010: 262).
43 I am aware of the debate around the dimensions of issue ownership. Thus, the argument
put forward here mainly reflects on the associative dimension of issue ownership, which
states that certain party families are almost intuitively linked to being competent on a certain issue (Walgrave, Lefevere, and Tresch 2012). For instance in a survey in Belgium 87 %
of the Flemish respondents spontaneously linked Green parties to environmental issues,
while only 45 % linked Liberals to taxes (Walgrave, Lefevere, and Tresch 2012: 774-779). I
conclude that it is a fair assumption that Green parties are representing a rather clear cut
case for issue ownership of environmental topics. Furthermore, as Tresch, Lefevere, and
Walgrave (2015) show parties appear to be unable to steal previously associatively owned
issues – such as the environment for Green parties.
71
4.4 modelling specifications
The Fukushima shock not only incrementally raised the salience of nuclear
energy as a topic, but also pressures on several Western governments to address questions on its safety. While government parties enjoy less freedom
than the opposition in selecting their agendas and therefore might have
been forced to talk about nuclear energy after the explosions in Fukushima,
they can still choose how to frame an issue. Thus, government parties are
likely to share rhetorical positions which they subsequently can also follow
suit. In contrast, opposition parties are more susceptible to say one thing
and then do another. They feel more independent from public pressures.
Thus, they are less likely to deliver policy outputs in line with their talk. In
turn, voters are not likely to measure opposition parties on basis of their
policy outputs, because they hold the government accountable to deliver
policy.
H3 : Governmental parties are more likely than opposition parties to attempt to influence policy implementation congruent to
their rhetorical activities.
4.4
modelling specifications
This chapter makes use of the measurement of parties’ rhetorical positions
as proposed in detail in the third chapter of the dissertation. First, I look
into descriptive evidence to outline how parties’ rhetorical positions are
structured within party families and how they are linked to party manifestos.
In order to understand whether this rhetorical measure is in line with
parties’ policy outputs, a measurement of such outputs is needed. Parties’
policy outputs again come from the ResponsiveGov data. I used parties’
policy proposals, motions, administrative decrees, and legislative acts to
measure their policy outputs. As outlined above, all countries included in
the study had nuclear programs. Therefore, the status quo is a pro-nuclear
policy. Thus, it seems counterfactual to analyse whether parties attempted
to extend the pro-nuclear situation of their country by trying to bring policies to parliamentary floors which are even ‘more pro-nuclear’ than the
status quo. Empirically, such examples are rarely found in the data-set. For
this reason I excluded all pro-nuclear legislative attempts from the analysis
72
4.4 modelling specifications
and only focussed on the anti-nuclear ones. Yet, according to the dataset
and in line with the argument put forward here, only a handful of such
attempts were actually made. Instead, the binary dependent variable only
accounts for parties’ attempts to introduce ‘anti-nuclear policies’. Thus, the
variable is ‘1’ for a month in which a party produced an anti-nuclear output
and 0 otherwise.
A logistic regression will be calculated to estimate the congruence between parties’ rhetorical activities and policy outputs. The underlying data
are measured across 67 parties per month (across the whole time period
included in the ResponsiveGov data) and therefore should be treated as
pooled time-series-cross-section (tscs) data. To account for the temporal
correlation in the data (autoregression), a lagged dependent variable (LDV)
is usually used in standard ordinary least squares (OLS) regression models (Beck and Katz 1995). Besides criticism that such LDVs tend to dominate results and might thereby camouflage other relationships in the model
(Achen 2000; Plümper, Troeger, and Manow 2005), they are insufficient for
controlling serial correlation in event count models and bernoulli processes
as is the case for the dependent variable used in the following analysis
(Brandt et al. 2000; Brandt and Sandler 2012; Beck and Epstein 2002). Thus,
I did not include a lagged dependent variable in my models. In addition to
standard logistic regression techniques, I followed the suggestions by Stewart (2006) and also ran logistic random effects models and country fixed
effect models.44
The measures of the independent variables are straightforward. Green
parties are measured by a dummy variable which takes on ‘1’ in case the
party belongs to the Green party family and zero otherwise. The same is
true for government parties, which is ‘1’ once a party is in government and
‘0’ otherwise. The hypotheses section introduced three hypotheses, two of
them being about interaction effects with parties’ rhetorical activities. Thus,
I constructed interaction effects with both the government and the Green
party variable to estimate whether such a moderation effect can be found
in the data. Finally, since my main independent variable and the dependent
44 Dickey Fuller tests reveal that the dependent variable follows a non-stationary structure.
Due to the incidental parameters problem, party fixed effects models are not a suitable
alternative. Using party fixed effects reduces the included sample in the analysis by 90 %
due to correlation with the main explanatory variables. The country fixed effects models
already reduce the included observations by 50 % due to perfect prediction of the failure.
73
4.5 results
variable might be affected by party size, with larger parties having more
access to the media and/or producing more policy outputs, all models
control for party size measured as the seats a party currently holds in the
national parliament. Since drafting policy implementation takes time, all
independent variables have been lagged by one month.
4.5
results
We now turn our attention towards the results derived from these specifications. First, I describe how parties talk between elections. Second, the
hypotheses regarding congruence between rhetorical activities and policy
outputs will be tested using logistic regression models.
Table 2 presents parties’ rhetorical activities across party families after
the Fukushima shock. The upper part of table 2 reports the relative frequencies of pro-, neutral- and anti-nuclear rhetorical activities. Thus, this
table does not rely on the measurement as proposed in chapter 3 of the
thesis, but instead presents the position of all rhetorical activities before applying the log-transformation as proposed in the last chapter.45 The lower
part of table 2 reports the means and standard deviations of the rhetorical
position measurement as described in the last section. The fact that across
all party families rhetorical activities about nuclear energy are either positive (+) or negative (-), but seldom neutral (0), is eye-catching.46 Hence,
parties appear to use rhetorical activities to share cues on their positions.
Thus, it appears that parties do not generally attempt to conceal their position by just talking without holding a clear-cut position on a matter. In
45 The ResponsiveGov data also coded statements which are related to the topic of nuclear
energy, but do not give a clear cut position on it. The table also reports these statements.
However, I decided to omit these statements from the rhetorical position paper. First,
because most of these statements are related to nuclear energy, but do not outline a specific
position on the issue. For example, in Germany a lot of the debate after Fukushima was
about potential legal issues in relation to Merkel’s decision to shut-down a couple of
nuclear reactors immediately. This debate was purely driven by the fact that German law
would force the Chancellor to have a vote on the matter in the Bundestag. While related to
the topic of the juncture coded by the ResponsiveGov, such legal difficulties do not reveal
actors’ true position on the nuclear issue, but only on how the government should have
dealt with the issue from a legal point of view. Comparable situations can be found in
other countries as well.
46 The only party family that did not hold up to this trend is the regional family. Yet, the
number of rhetorical activities for this family is fairly low with 10 rhetorical activities in
total.
74
12.3
14.5
73.2
100
138
-1.24
1.37
-1.65
1.05
Communist
5.4
19.4
75.2
100
423
Ecologist
80.0
10.0
10.0
100
10
0.93
0.93
0.1
1.91
Ethnicregional
30.0
24.1
45.9
100
623
Social
democratic
0.93
1.33
39.6
11.1
49.3
100
371
Liberal
1.14
1.38
26.8
21.7
51.4
100
451
Christian
democratic
Source: Author’s own, party family categorisation taken from comparative manifesto data.
+
0
Total %
N
rhetorical position
mean
standard deviation
rhetorical activities
Table 2: Party families & rhetorical activities
1.31
0.97
69.0
27.6
3.4
100
29
Radical
Right
1.6
0.56
89.8
4.5
5.7
100
88
Agrarian
1.81
0.84
84.0
12.9
3.1
100
549
Conservative
4.5 results
75
4.5 results
the case of nuclear energy, parties tend to reveal their positions through
rhetoric. Furthermore, the most silent party families have comparably narrow issue stances without an interest in environmental politics (Ethnic Regional; Radical Right; Agrarian). This is important to note once we realise
that there are almost as many observations of regional parties in the data as
observations of Green parties. Since party families in table 2 are sorted by
the party family’s mean rhetorical position (mean), it becomes evident, as
expected, that Green parties are the most prominent and homogenous critics of nuclear energy, followed by the Communist and Social Democratic
parties. The latter present the most heterogeneous group (standard deviation). Thus along the lines of the arguments in the last section, there still
appears to be large chasms across Social Democratic parties on the issue
of nuclear energy. Finally, the Christian Democratic and Conservative parties both largely reconfirmed their pro-nuclear ideology after the nuclear
accident in Fukushima. Yet, the Christian Democrats are internally more
divided on the matter, which is especially due to the position shift of the
Unionsparteien (CDU/CSU) in Germany shortly after the Fukushima meltdown. In summary, parties’ rhetorical positions in table 2 seem to reflect
the expectations that left party families are more critical about nuclear energy, while right party families hold pro-nuclear positions. Thus, the face
validity of the measure appears to be strong and the measure seems to be
reliable across party families as well.
However, looking at party families represents a time invariant view into
party politics. It would be more interesting to look at more volatile concepts, such as party ideology or their manifesto positions. Figure 4 scatters
parties’ left-right placement (x-axis) and their rhetorical positions (y-axis).
In this figure I rely on the rhetorical position measure as outlined in the
third chapter of this thesis. As we can see there is a strong relationship
between parties’ general left-right placement and the ways in which they
talk rhetorically about nuclear energy. In line with the theoretical assumptions, the more to the right a party’s ideology is, the more likely it is to
support nuclear energy and be outspoken about its position. This result
also supports the argument that nuclear energy is a confrontational issue,
which was also suggested in the previous chapter. The confidence interval
of the regression line is small and the positive slope is highly significant
(coef.=0.5; p-values=0.000; std. err.= 0.049). To a large extent, the conflict
76
4.5 results
Figure 4: Relationship between parties’ rhetorical positions & parties’ left-right placements
rhetorical position
5
95% CI
0
Fitted values
ecologist
communist
social democratic
liberal
christian democratic
conservative
nationalist
agrarian
ethnic−regional
−5
−40
−20
0
20
right−left ideological index (cmp rile score)
40
Source: Author’s own.
Note: x-axis relies on rile score coming from comparative manifesto project. y-axis uses parties’ rhetorical
positions as outlined in detail in the third chapter of the thesis. Parties without true rhetorical position are
omitted (=not talking about nuclear energy) to ease and assure readability. However, including these would
lead to similar conclusions.
on the usage of nuclear energy re-symbolises the general left-right divide
between political parties. Again we can see that some party families appear
to be clustered more clearly on both their left-right positioning as a family
as well as on their position on nuclear energy.
It would also be interesting to see the extent to which parties’ rhetorical
activities after the Fukushima accident reflect their manifesto positions for
the elections prior to and after the Fukushima shock. We should assume
that there is some variation between parties’ manifesto positions and the
rhetorical statements put forward by their lead figures. Figure 5 plots the
distributions of parties’ rhetorical activities (as proposed in the last chapter
of this thesis) split by parties’ manifesto positions before the Fukushima
accident on the left hand side and their positions after the accident on the
right hand side respectively. Higher values on the y-axis outline a more
pro nuclear manifesto position. Both graphs report a strong relationship
77
4.5 results
Figure 5: Violinplot of rhetorical positions & manifesto positions
manifesto position
initial manifesto position
final manifesto position
+2
+2
+1
+1
0
0
−1
−1
−2
−2
−5
0
rhetorical position
5
−5
0
rhetorical position
5
Source: Author’s own.
Note: Violinplots across parties’ initial and final manifesto positions. ‘+2’ indicating a ‘very pro-nuclear’
manifesto position and ‘-2’ a ‘very anti-nuclear’ manifesto position (detailed codebook descriptions can be
found in the appendix section A.1.2 on page 160). Violinplots plot the estimated kernel density of rhetorical
positions with box plots. Marker is the median, box indicates interquartile range and spikes extend to the
upper- and lower-adjacent values.
between parties’ manifesto positions and their rhetorical positions. Therefore, on the surface, rhetorical activities strongly reflect party manifesto
positions on the nuclear issue. Digging deeper into how parties talk, however, reveals that parties’ rhetorical activities appear to be shifting across
time, with distributions overlapping across party positions. Parties with an
initial pro-nuclear manifesto position partly used anti-nuclear rhetoric after
the Fukushima shock. Furthermore, parties with vague manifesto positions
also used vague rhetoric with medians close to zero and distributions including almost pro- and anti-nuclear talk symmetrically. Finally, due to
some parties changing their manifesto positions in the elections after the
Fukushima shock, both pro-nuclear groups appear to be more cohesive
when their rhetorical activities are compared to their final manifesto positions.
78
4.5 results
Figure 6 provides a final look into the movement of parties’ rhetorical
activities across time for two countries and six parties. The plot on the left
5
5
4
4
3
3
2
2
1
1
rhetorical position
rhetorical position
Figure 6: Movement on the nuclear issue in Spain & France
0
−1
0
−1
−2
−2
−3
−3
IU
−4
EELV
−4
PSOE
PP
−5
Mar11
Jul11
month
Nov11
PS
UMP
−5
Mar11
Jul11
Dec11
May12
month
Source: Author’s own.
Note: The graph on the left hand only pictures line plots for the Partido Popular (PP), Partido Socialista Obrero
Español (PSOE) and Izquierda Unida (IU). The graph on the right hand only pictures line plots for the Union
pour un mouvement populaire (UMP), Parti Socialiste (PS) and Europe Écologie - Les Verts (EELV). The
remaining parties in the French and Spanish systems have been omitted in the graph for the sake of clarity.
Missing values omitted.
shows the Spanish parties and their rhetorical activities across time, while
the one on the right shows the French ones. In the Spanish party system
there is a clear cut left-right divide on the nuclear issue after the Fukushima
incident with the Partido Popular (PP) supporting the use of nuclear energy
and the Partido Socialista Obrero Español (PSOE) and Izquierda Unida (IU)
rejecting it. Yet, even though this clear divide continues, emphasis on the
nuclear issue shrunk during the national election campaign, which started
in June 2011, and this trend continued until the election on 20 November
2011.
A decline in interest in the nuclear issue also occurred, but to a lesser
extent, during the French presidential elections in late April and early May
2012. The Green party remained fairly constant in its position and the level
of attention it gave to the nuclear issue, which again supports issue own-
79
4.5 results
ership theories. Interestingly the rhetorical position of the Parti Socialiste
(PS) appeared to be in permanent motion, with the party even flip-flopping
its position once. Across time, the socialists sent mixed policy messages to
voters and their position on the nuclear issue appeared blurred. During August 2011, Martine Aubry and François Hollande, both seeking nomination
as presidential candidates of the PS, publicly debated their different standpoints on the nuclear future of France, which is perfectly reflected by the
PS’s rhetorical position hitting the x-axis in figure 6. After an explosion at
the Marcoule nuclear site in September, several candidates and members of
the PS expressed their doubts about the safety of nuclear energy. Ségolène
Royal even declared that if elected she would close the nuclear reactor in
Fessenheim. However, in October François Hollande stated that the nuclear
construction site in Flamanville should to be finished. This short-lived unity
of the PS is reflected in an anti-nuclear shift in the rhetorical measurement.
4.5.1
Congruence between rhetorical activities & policy outputs
The last section described how parties’ rhetorical activities on nuclear energy were shaped after the Fukushima shock. It became evident that parties’
rhetorical activities substantially reflected their traditional issue positions
on the nuclear issue. To a large extent the issue of nuclear energy nowadays
mirrors the left-right divide between political parties. Yet, besides showing
that parties largely share policy cues when talking, we still do not know
whether their rhetorical activities are congruent to their policy output. To
test whether this is the case I run logistic regression models with parties’
rhetorical positions as the main independent variable and their policy outputs as the dependent variable.
Figure 7 reports the coefficients and confidence intervals of three logistic
regression models which are also reported in table 3.47 Model “baseline” includes parties’ rhetorical positions, the Green and incumbent party dummy
and party size measured by seats held in parliament. The point estimate of
parties’ rhetorical activities in the baseline model shows a negative coefficient and a small confidence interval, which does not include zero. Thus,
there appears to be a highly significant negative effect of parties’ rhetori47 For the following analysis missing values (parties that did not talk in a given month) are
included as zero values.
80
1098
-4.387∗∗∗
(0.288)
1098
-6.025∗∗∗
(0.824)
-0.620∗
(0.304)
-0.803∗∗∗
(0.172)
1098
-5.379∗∗∗
(0.550)
-0.746∗∗∗
(0.182)
1.450∗
(0.665)
0.836
(0.568)
0.001
(0.003)
(3)
logistic
baseline
1098
-5.057∗∗∗
(0.532)
0.0576
(0.443)
1.520
(0.834)
0.314
(0.594)
0.00
(0.004)
-0.561
(0.586)
-1.303∗
(0.532)
(4)
logistic
interaction
1098
-7.084∗∗∗
(1.857)
0.447
(0.657)
3.324
(2.083)
1.088
(1.216)
0.01
(0.007)
-0.690
(0.853)
-1.560∗
(0.752)
(5)
logistic
interaction re
Standard errors in parentheses.
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
Models (1), (2), (4) report standard pooled logistic regression results.
Models (3) and (5) report pooled logistic regression results with random effects.
Model (6) report pooled logistic regression results with country fixed effects which are not reported in the table.
N
constant
rhetorics X government party
rhetorics X Green party
# seats
government party
Green party
rhetorics
(2)
logistic
re
(1)
logistic
542
-6.066∗∗∗
(1.015)
0.483
(0.455)
2.582∗
(1.046)
1.356
(0.889)
0.00
(0.008)
-0.185
(0.605)
-1.298∗
(0.584)
(6)
logistic
country fe
Table 3: Relationship between rhetorical position & policy outputs: logistic regression models (standard b-coefficients)
4.5 results
81
4.5 results
Figure 7: The relationship between policy outputs & party rhetoric, logistic regression
results across three models (standard b-coefficients)
rhetorical position
Green
gov
# seats
green*position
baseline model
interaction model
gov*position
interaction random effects model
−5
0
5
Pr(anti nuclear policy output)
10
Source: Author’s own.
Note: Markers are coefficients of logistic regression results with horizontal spikes for confidence intervals. The
models are also reported in table 3 models (3), (4) and (5). B-coefficients are reported.
cal activities in the baseline model. This essentially means that the more
anti-nuclear a party’s rhetorical position is, the more likely it is to present
an anti-nuclear policy output, meaning a substantive policy activity which
is in line with its rhetorical position. Thus, ceteris paribus hypothesis one,
that parties are likely to deliver political decisions that are congruent to
their rhetorical activities, is supported with the baseline model. Furthermore, being in government is not a significant predictor of parties’ policy
outputs.
Even though Green parties significantly push for more anti-nuclear policies in the baseline model, the interaction term with parties’ rhetorical activities is insignificant in both interaction models. The confidence intervals
of both models stretch across zero and are insignificant on conventional
82
4.5 results
levels. Thus, there is no evidence that Green parties are more likely to draft
policy proposals in line with their rhetorical position.
Figure 8: Marginal effects plot of interaction between rhetorical activities & incumbency
effects on Pr(policy output)
1.5
1
.5
0
−5
−2.5
0
rhetorical position
2.5
5
Source: Author’s own.
Note: Based on model 2 (logistic interaction) in figure 3 & table 3 on page 81, model (3). Average Marginal
Effects of interaction between incumbency and parties’ rhetorical positions. Party with 95 % confidence
intervals shown as dashed lines.
The effect of governmental parties in the baseline model is also not significant, yet government parties moderate the effect of rhetorical activities
in both interaction models. Parties in government also tend to be more pronuclear on the rhetorical position measure, which suggests that they might
choose their words wisely if it comes to talking about a potential phaseout.48 While the confidence intervals in the random effect models reaches
out to zero, it does not include zero. In both interaction models governments moderate the effect of parties’ rhetorical activities significantly on
the 5 % level. However, we are not only interested in whether the coefficient
of the interaction between incumbency and parties’ rhetorical positions has
a significant effect. Instead, we are interested in the substantive effect of the
interaction and its constitutive terms. For this reason we also need to calculate the standard errors, confidence intervals and coefficient for parties’
48 A Boxplot in the appendix reports the distribution of the rhetorical position measure split
by government status in figure 18 on page 170.
83
4.6 conclusion
rhetorical positions conditional on the incumbency dummy (the marginal
effect).49 In the case of a binary variable, such a marginal effect represents
the changes from 0 to 1, which here is the difference between being in government versus being in opposition. Figure 8 reports this marginal effect.
As we can see, parties in government that produce pro-nuclear rhetoric
are not likely to produce an anti-nuclear policy output. In contrast, the
more anti-nuclear an incumbent party’s positional measure is, the more
likely it is to produce a congruent policy output. This again provides support for hypothesis three, which states that government parties appear to
be more likely to push for policies which are congruent to their rhetorical statements. This effect is also significant once we introduce parties’ pre
Fukushima manifesto positions into the equation.50
4.6
conclusion
This chapter had two purposes. First, I introduced the dataset employed
in the study, thereby explaining the issue selection and party selection of
the thesis. I then unveiled my measure of parties’ rhetorical positions, such
as politicians’ statements to the press, speeches and interviews, and how I
transferred the theoretical concept of protrusion of protest into a statistical
measure. Second, the chapter validated the rhetorical position measures in
a two-step logic. In the first step I placed the issue of nuclear energy into
the context of party competition. Historically the usage of nuclear energy
is a topic that tends to reflect the dividing lines of party families and their
ideology, with party families on the right of the ideological spectrum tending to support nuclear energy and parties to the left rejecting the usage of
nuclear energy. In the second step I constructed a theoretical framework
and derived hypotheses linking parties’ rhetorical positions to their more
substantive policy output in laws, bills and motions. The key theoretical
49 The marginal effect is calculated here by the derivative of the regression function:
∂policy output
∂incumbency = b1 + b3 rhetorical position (see: Brambor, Clark, and Golder 2005: 70). Fur∂policy output
thermore, the standard error of the substantive interaction is given by: σ̂ ∂incumbency =
q
var ( β̂ 1 ) + rhetorical position2 var ( β̂ 3 ) + 2 ∗ rhetorical position cov( β̂ 1 β̂ 3 ).
50 However, as highlighted in the last chapter, introducing both measures, manifesto positions and rhetorical positions, would lead to high multicollinearity in the models reported
above. Therefore, I only run robustness models to confirm that the main results are robust
to such a specification.
84
4.6 conclusion
argument is that voters listen to the policy cues that parties deliver rhetorically. In fact, parties’ media presence should have a sufficient impact on
the public’s voting intensions. Thus, parties should be careful about the
positions they share with the media and should aim to pour their talk into
substantive policy outputs.
This chapter also contributes to the increasingly growing debate on whether
and how parties’ rhetorical activities matter. The descriptive part of the
chapter shows that rhetorical party positions firstly reflect party families’
traditional issue appeals and their manifesto positions. Yet, the descriptive
section of the chapter also reveals that the proposed rhetorical activities
measure deviates from parties’ raison d’être. Political parties’ rhetorical media presence is strongly linked to their general left-right placements and the
issue of nuclear energy strongly mirrors this left-right distinction between
political parties.
Second, the pooled logistic regression analysis was underpinned by the
assumption that parties refrain from saying one thing and doing another.
Rhetorical activities are congruent to parties’ policy outputs (e.g. bills, legislative acts and laws) across several models. Furthermore, incumbent parties appear to be more likely than opposition parties to feed the media with
rhetorical positions that they can subsequently follow through on. This reinforces results of existing literature on the responsiveness of governmental
policy agendas (Green-Pedersen and Mortensen 2010; Walgrave, Lefevere,
and Nuytemans 2009). It is also important to note that the chapter does not
seek to suggest a causal direction running from party rhetoric to policy outputs. It might as well be that parties talk like they walk and not walk like
they talk, meaning that there is a strong congruent correlation between parties’ rhetorical activity and the substantive policy activities they perform.
However, it might also be the case that parties that recently drafted policy are more likely to talk more about the proposed policy. Yet, the causal
direction is not important for the analysis of this chapter. It should rather
be read as a preceding step to more detailed analyses that taking public
opinion and public mobilisation into account, as the following chapters do.
85
5
W H O R E P R E S E N T S P R O T E S T E R S – PA R T I E S A S
C R U C I A L G AT E K E E P E R S T O T H E P O L I T I C A L A R E N A
5.1
introduction
icking up the major question raised in the last chapters, this chapter gives an insight into how protest matters for party competition.
Theoretically, the chapter attempts to bridge the gap between arguments stemming from social movement studies and party politics. This
chapter uses the key theoretical argument discussed in the theoretical chapter (chapter 2) and also explores the major interacting effects of party ideology and public opinion on the influence of protest. I argue that under
certain conditions protests influence rhetorical competition between parties, such as politicians’ statements to the press, speeches and interviews.
First, protest needs to be compelling enough to capture the attention of
political parties. However, if protests succeed in penetrating party agendas,
parties show divergent reactions to them depending on how their ideology relates to the issue relevant for protesters. In the case of the debate
on the civil usage of nuclear energy after the Fukushima meltdown, left
wing parties were more likely to support protesters’ claims, while parties
to the right more often rejected them in an effort to reinforce the status
quo. In contrast to previous studies (Giugni 2004; Agnone 2007), I hypothesise that public opinion forestalls the impact of protest: once protesters’
claims are at odds with the public at large, so will be most of the parties’
rhetorical positions. Unsurprisingly, Green parties are more prone to react
to protest concerning nuclear energy than other party families. However,
being “issue owners” Green parties are more likely to speak out against
nuclear energy, even if there is no protest. Thus, Green parties understand
increasing protest as a window of opportunity to set nuclear energy on top
of the political agenda of the party system.
In contrast to previous research, this chapter not only answers the question as to when parties increase attention to an issue, but also how an
P
86
5.2 protest, public opinion & party ideology
increase of attention is shaped, namely whether a party supports or rejects protesters’ claims. Using the ResponsiveGov data, I ran time-seriescross-section models across 67 parties and 24 months in 12 democracies
on parties’ rhetorical positions on nuclear energy after the accident at the
Fukushima nuclear plant, finding support for my theoretical framework.
The next section gives an overview of the literature on the influence of
protest and party position taking. Section three introduces the theoretical
framework of the study and derives hypotheses that will be tested in section five. Section four describes the analytical strategy of the paper and
section six presents the results of several regression models. The final section concludes and points out avenues for future research.
5.2
protest, public opinion & party ideology
Research shows that parties’ responses to shifts in public opinion are moderated by their ideology (Adams et al. 2004), with left parties being less
responsive to shifts in the public mood (Adams, Haupt, and Stoll 2008).
Niche parties, such as Green, Communist and Radical Right parties, are
also understood to respond to particular fractions of the public while remaining unresponsive to the public at large (Meguid 2005; Adams et al.
2006). Furthermore, whether or not we regard political parties and legislators as a collective potentially changes our conclusions about mechanisms
of responsiveness (Weissberg 1978; Hill and Hurley 1999). In this vein, models more often theorise that neither parties nor parliaments are single entities, but should be disentangled into their constituent parts. Yet, less effort
has been made to disentangle public opinion. Political science has mostly
stressed the importance of the mean voter as the driving vehicle for political as well as social change and has thereby ignored the mobilised masses
(Burstein 2006).
Using the metaphor of a thermostat, the public has been interpreted as
a single and unified unit, screaming “too hot” if it demands less public
spending from the government and screaming “too cold” if it senses not
enough public spending (Soroka and Wlezien 2010; Bølstad 2012; Wlezien
1995).51 Besides the lingering interpretational problem that an increased
51 The “most important problem” survey questions represent the temperature in this
metaphor.
87
5.2 protest, public opinion & party ideology
rank of a problem by the public might not go hand in hand with the public
claiming more spending – but potentially for other policy instruments not
linked to budgetary changes – the signals sent by it appear multifaceted
and are often contradictory in nature.
The public articulates its preferences not only by answering questions
in surveys, but also through a range of other actions such as direct interactions with representatives during their visits to constituencies (Fenno 1977),
via letter exchanges with their representatives (Cain, Ferejohn, and Fiorina
1984), via the internet (Adler, Gent, and Overmeyer 1998) or by means of
collective actions. The sheer amount of possibilities at the public’s disposal
to share its concerns suggests that representatives face at least some extent of variation in the expression of the public mood, contingent on which
sources they consult. Thus, the surveyed public opinion might be at odds
with the mobilised opinion. Transferred to the thermostat metaphor, the
public, measured through surveys, might scream “too hot” while the mobilised masses scream “too cold”.
In the set of informational resources, demonstrations have become an
ever more important indicator of public mood swings. Since the large scale
civil rights and anti-Vietnam social movements in the 1960s and 1970s, the
anti-nuclear and pacifist demonstrations in the 1980s, people have become
increasingly involved in demonstrations, boycotts, strikes and petitions (Inglehart 1977; Barnes et al. 1979). Thus, while protests have become a common tool in the masses’ toolbox for articulating grievances, the impact of
demonstrations have remained largely a concern of sociologist (Soule and
King 2006; King, Bentele, and Soule 2007; Olzak and Soule 2009) and have
so far been neglected by scholarly work on the positions of political parties
(McAdam and Su 2002: 696).
Most earlier examinations on whether protests matter proclaim that protests
directly affect policy decisions. These studies largely disregarded the role
that political parties play in democracies (Gamson 1990; Morris 1993; Barkan
1984), especially their central role in the chain of delegation between voters
and policy-making (Müller 2000).52 In contrast, later studies emphasised
the relevance of various environmental factors (political opportunity structures) mediating or even forestalling the influence of protest (Kriesi et al.
1992; Amenta, Dunleavy, and Bernstein 1994). Thereby the concept of politi52 Admittedly more so in parliamentarism than in presidentialism (Strøm 2000).
88
5.2 protest, public opinion & party ideology
cal opportunity structures has been defined differently across the literature
(Meyer 2004: 125-126), encompassing diverging sets of actors such as political parties (Rucht 1996), political institutions (Kitschelt 1986) and resources
at demonstrators’ disposal (Andrews 2001).53 Until today though, most research has understood political actors as facilitators for protests, but has
not sought to understand the relationship between parties’ policy preferences and protesters’ demands (Rucht 1996: 185-186).
The literature on protests’ influence on policy finds mixed results and
suffers a range of shortcomings (for a detailed overview, see: Amenta et al.
2010; Giugni 1998). First, most studies are single country analyses (Gillion
2013; Walgrave and Vliegenthart 2012), often bound to the US case (Amenta,
Carruthers, and Zylan 1992; McAdam and Su 2002; King, Bentele, and
Soule 2007) and leave party democracies and institutional diversity untouched. Only a few exceptions can be found which take a comparative
perspective across time, countries and diverse issues (Giugni 2004, 2007).
Second, the preferred method of analysis has often been detailed case studies. Thus, methodological diversity is lacking and generalisability becomes
unfeasible. Finally, the most crucial limitation might be the focus on the
impact of protest on policy outcomes, the final stage of the policy cycle.
This influence might be the most unlikely influence of protest and even in
cases where such an influence could exist, conclusively proving it appears
to be challenging.
However, looking into earlier stages of the policy cycle has rarely been
done even though there are several reasons that the agenda setting stage
of politics matters for protesters and scholars alike. First, the link between
protesters’ claims and political parties’ agendas decides whether and how
protest might affect policy. Located at the beginning of the policy cycle,
the agenda setting stage appears to be a crucial waypoint for protesters’
success (for comparable claims see: Soule and King 2006; King, Bentele,
and Soule 2007). Protests might explicitly seek to launch the debate about
an issue demonstrators care about in the first place.54 Parties can then be
53 Thus, while for instance Kitschelt (1986) advocated a narrow understanding of political opportunities by preliminary focusing on the institutional openness for protesters’ claims of
state structures, Kriesi et al. (1995) also included the alliance structure between new social
movements and positions of left parties in their study on social movement mobilisation in
four Western democracies.
54 The influence of protest on parties’ policy agendas can be understood as a minimal goal
of protests, while affecting policy outcomes is the final aim. Effectively, in case the latter
89
5.3 how protest matters – parties between policy & vote
understood as the crucial gatekeepers for absorbing and to implementing
their policy claims. Second, establishing a causal link between protest and
party agendas seems to be more feasible than showing a long-term impact
on policy outcomes which often depends on complex veto-player constellations (Tsebelis 2002), economic conditions (Garrett 1998) and a range of
other factors. Finally, if establishing a plausible influence of protest on party
agendas fails, we have little reason to believe that protests have an impact
on the final stage of policy cycles, namely policy outcomes.
In fact, lately a handful of studies have attempted to link protest with
the agendas of political actors (Gillion 2013; Walgrave and Vliegenthart
2012; Olzak and Soule 2009; King, Bentele, and Soule 2007; Soule and King
2006). Indeed, their results lend support to the aforementioned argument
that the influence of protest decreases along the policy cycle. Thus, protest
tends to have its strongest impact at the agenda-setting stage (Walgrave and
Vliegenthart 2012: 150). Baumgartner and Mahoney (2005) even claim that
much of the variation in the US government’s agenda is due to the impetus
of social movements and their organisations. However, the cited studies
neglect the importance that party ideology might play in mediating the
effects of protest on party agendas.55
5.3
how protest matters – parties between policy & vote
My theoretical argument departs from the idea that protest affects all parties equally. In contrast, I argue that ideology shapes the way that political
actors perceive their environment and subsequently how parties react to
different impetuses. This leads parties to react differently to protest depending on how their own ideology relates to the claims of the demonstrating
masses (Adams et al. 2006; Adams, Haupt, and Stoll 2008; Ezrow 2010).
Today parties are overwhelmed with information about their voters’ preferences. Not only are they confronted with the classical instrument of pubgoal is reached, a protest’s raison d’être no longer exists. This means there is no longer a
reason to protest, when we define protesters’ goals as being policy influence.
55 Yet, again only one of the cited studies looks into a party democracy – Belgium – (Walgrave and Vliegenthart 2012), and not a single one seeks a comparative research design
across countries. Finally, while the studies on the US case differentiate between the policy agendas of Democrats and Republicans, party ideology and typologies are treated as
controls only.
90
5.3 how protest matters – parties between policy & vote
lic opinion surveys, but mass mobilisation against governmental decisions
has also increased. The media regularly shares public opinion surveys
on salient issues and politicians might be bombarded with information
through social media; Twitter, Facebook etc. Given the richness of informational resources, political actors (a) are rarely capable of seeking all potential information and thus react only to those signals they perceive as
legitimate and penetrating enough, (b) might face the task of responding
to contradictory information, (c) rely on their party’s ideology as a filter to
deal with their complex environment. Parties have reasons to follow and
adapt to what the public wants, otherwise they might compromise their
electoral survival (Stimson, Mackuen, and Erikson 1995). Consequently the
question arises as to how how different impulses from the public affect ideologically diverse party agendas?
As has been debated in detail in the theoretical chapter of this thesis,
protests are often tailored to a certain topic and can present a clear demand for policy change on a given issue (Schumaker 1975: 490). However
they might be small in size (DeNardo 1985), a single one day event, or not
be disruptive enough to gain the attention of the media (Smith et al. 2001).
They are prevalently characterised by low protrusion and remain hidden
for most parties. Yet, with increasing mobilisation, support and disruption,
the visibility of protests increases. As a result, parties are more likely to
be forced to adapt their agendas in order to sufficiently address demonstrators’ claims. Parties tend to pick up issues from public claims that reassure their ideology and therefore their ideology restricts their attention
(Walgrave and Nuytemans 2009: 192). Under certain circumstances parties
feel it necessary to react to their environment in order to secure their electoral survival irrespective of their ideology. Ignoring protesters in such a
scenario could lead to more mobilisation on the matter, increase salience
on an unfavourable issue, or even lead to shifts in public opinion surveys
(Burstein 1979). Therefore, in the long run remaining silent on the issue
might be more costly than addressing it.
Still, while parties may adapt their attention when facing increased protrusion of protests, their openness to them depends on the topic and demonstrators’ claims. Parties might understand protests as a challenge to their
ideological position or a window of opportunity to use demonstrations to
legitimise their policy positions or to harm their competitors. Thus, when
91
5.3 how protest matters – parties between policy & vote
increasing the attention given to protesters’ grievances, they deliberately
choose whether or not to confront protesters or to support their claims
depending on how their own ideology relates to the issue at stake.
Traditionally parties to the left, e.g. Social Democrats, Green and Communist parties, have enjoyed strong ties to trade unions and/or “new social
movements” (Tarrow 1994; Kitschelt 1994; Kriesi et al. 1995; Hamann, Johnston, and Kelly 2012: 1038-1039; Hamann, Johnston, and Kelly 2013). Green
parties emerged from social movement organisations calling for nuclear
phase-out, pacifism and environmental protection in the early 1980s. Furthermore, since many protest activities, especially strikes, are organised or
supported by unions or new social movements, parties with a left ideology
might perceive such protests as representative of their core voters’ preferences.56 In contrast, parties on the ideological right are associated with
conservatism and a tendency to support and underpin the status quo instead of pushing for reforms. This is especially the case for issues they are
unsure how to relate to or how their core voters relate to (Rokkan 1970),
such as nuclear energy. Christian Democratic and ideological right mainstream parties also largely neglected new social movements issues at the
beginning of their rise (Meguid 2005, 2007).
Thus, in the case of protest on issues related to “new social movements”
such as nuclear energy, parties on the left of the ideological spectrum are
likely to support protesters’ claims and understand increasing protest mobilisation as a window of opportunity for favourably shaping the public
debate in their interests. In contrast, parties with right wing ideologies perceive protests on new issues as a challenge to their core ideology. They
will also increase attention to the issue but try to argue against protesters’
claims. I argue that with increasing protrusion of protest debate between
left and right parties will become more polarised, while having only small
effects on parties located in the middle of the ideological spectrum.
H1( party polarisation hypothesis) : The more left (right) a party’s ideology the more likely it is to support (reject) protesters’ claims on
new issues.
56 This already suggests a bias in the issues protesters pay attention to. Empirically protest
appears to be more often bound to political interests of the left than of the right. Thus, in
a model containing “all issues” in the world, left parties might be generally more likely to
support protesters and their claims.
92
5.3 how protest matters – parties between policy & vote
Several scholars have emphasised that the impact of protest on policy
and politics depends on favourable environmental circumstances, and public opinion has become the major external factor more recent studies have
drawn upon (Giugni 2007; Agnone 2007; McAdam and Su 2002; Burstein
1999).57 Yet, previous research stressed the reinforcing capabilities that public opinion has on the meaningfulness of protest, naming such effects “the
amplification mechanism” (Agnone 2007) or “joint effect of public opinion
and protest” (Giugni 2007, 2004). Thus, these scholars focussed mainly on
a positive interaction between public opinion and protest. In fact, disagreement to protesters’ claims by the public at large might be the strongest
obstacle for the successful impact of demonstrations.
Yet, parties that are motivated primarily by re-election (Mayhew 1974) or
vote maximisation (Downs 1957; Strøm 1990) are likely to support the signal shared by the public at large while rejecting protesters’ claims. While
protests with high protrusion are more likely to affect parties’ talk, they
might still be read as only representing the interests of a noisy minority while the silent masses support other policy reforms or the status quo.
Protest then increases the salience parties designate to the issue at stake
during protest, but does not lead to increased support by party elites, irrespective of whether the silent masses are well informed or not. In order to
keep or extend their public support, parties are prone to react to the public
at large and neglect the interests of the demonstrating minority. Thus, I
argue once the signal sent by the mobilised protesters and public opinion
contradict each other, parties are more likely to represent the interests of
the public at large and ignore protesters’ claims. I expect the forestalling
mechanism of public opinion to be stronger than its amplification mechanism.
H2( f orestalling hypothesis) : If the percentage of respondents in public opinion surveys disagreeing with protesters’ claims is large
enough, an increase in the protrusion of protest will lead parties
to refuse protesters’ claims more.
As outlined above, Green parties emerged from social movements in the
1980s calling for a nuclear phase-out. Green parties, thus, can be under57 Earlier studies on protest regularly omitted public opinion measurements from their analyses (Gamson 1990).
93
5.4 modelling specifications
stood as owning the issue of nuclear energy. They are largely associated
with it and perceived as competent by voters on the issue (Walgrave, Lefevere, and Nuytemans 2009; Tresch, Lefevere, and Walgrave 2015). Issue
owners attempt to manipulate politics with their rhetoric in a way that
is advantageous for their vote-seeking efforts (Riker 1996). To remind voters of their competence on the matter and in order to establish a strong tie
between the party brand and the issue of nuclear energy, Green parties often talk about the issue in order to keep their ownership or to influence the
public debate beneficially for their vote-seeking goals. Thus, Green parties
are less likely to react to protest since they are already rhetorically pushing
for nuclear phase-out policies, irrespective of whether protesters take to the
streets.
H3(Green party hypothesis) : Green parties do not adapt their attention
to the nuclear issue when protest frequency and size change.
5.4
modelling specifications
To test my theoretical arguments, we need to examine if parties rhetorically respond to protests. Thus, I estimate a model in which the rhetorical
position measurement proposed in the methods chapter is the dependent
variable and party characteristics, the protest-index, and several controls
the independent variables:
RPit = b0 + b1 Party lrit + b2 Protestit + b3 Party lrit ∗ Protestit
+b6 Public Opinionit + b7 Public Opinionit ∗ Protestit
+b8 Green Partyit + b9 Green Partyit ∗ Protestit
(2)
+b10 ζ it + eit
with ζ being a vector of controls outlined below. The hypotheses derived
in the theoretical section of this chapter are translated into interaction
terms, each indicating the respective moderation effect of party ideology
on protest, public opinion and Green parties. I first test each hypothesis in
a separate model before estimating a model including all interaction terms
as outlined in the equation above.
94
5.4 modelling specifications
Since the data are time-series-cross-sectional, with each party being observed over an average of 17.6 months, the Markov assumptions of standard OLS regression analysis are violated. Indeed, autocorrelation tests reveal that the null hypothesis of no serial correlation needs to be rejected.58
Further test-statistics show that the data are heteroscedastic and stationary.59 I ran models with panel-corrected standard errors (PCSE) combined
with a Prais-Winsten transformation to address the issues of heteroscedasticity, serial correlation within partyi and contemporaneous correlation (correlation of the errors of partyi and party j at time t). I opted for these model
specifications instead of a lagged dependent variable specification, since
the latter tends to absorb a huge amount of the variation of the dependent
variable (LDV) (Achen 2000; Plümper, Troeger, and Manow 2005). However,
the results of the models employed are robust to different sub-samples and
estimation strategies, e.g. including a LDV (Beck and Katz 1995); or using
Driscoll-Kraay standard errors which tend to be more consistent for time invariant covariates (Driscoll and Kraay 1998).60 Furthermore, as unobserved
heterogeneity potentially infringes on the results, I include country fixed
effects in my models. This also helps to address media bias towards reports on protest across countries. Since the study is interested in the effect
of protest on party rhetoric, I opted not to lag the independent variables.
In line with previous research I assume a temporally close translation from
protest into party talk (Walgrave and Vliegenthart 2012). However, the results are robust to lagging the protest-index by one month.
Public opinion is measured by subtracting the percentage of respondents
who disagree with the usage of nuclear energy from the percentage of
respondents who favour it. Thus, a higher value on the public opinion
scale indicates a more pro nuclear public mood. Public opinion polls have
58 To be precise a pooled Wooldridge test (Wooldridge 2013) is not significant and thus
the null hypothesis of no autocorrelation cannot be rejected. Since H0 does not test for
autocorrelation, not rejecting H0 does not entitle me to accept H A . Using Wooldridge tests
on a country by country basis discloses that H0 needs to be rejected for several parties
in several countries. Thus, I decided to proceed with caution and to control for an AR-1
autocorrelation structure. The standard errors of these model are also more conservative
when compared to models not controlling for autocorrelation. Even though the p-values
do not extensively differ from each other.
59 Breusch-Pagan and Cook-Weisberg’s tests for heteroscedasticity were employed and reject
the null hypothesis of constant variance. Significant Unit-roots tests (Fisher-type test based
on ADF test) reveal that the data are stationary.
60 Robustness tests are reported on page 14 in table 14 in the appendix.
95
5.4 modelling specifications
been selected to either reflect a respondent’s position towards the status quo
policy of a given country or her/his general opinion on the usage of nuclear
energy and also come from the ResponsiveGov data. Comparable to other
studies, the number of respondents naming environmental issues as the
most important problem/issue in each country has been used to measure
the salience of the issue of nuclear energy (Jennings and Wlezien 2011).61
The results reported in the analysis section are also robust to a different
specification of the salience variable illustrated by Lax and Phillips (2009,
2012). Using the overall amount of news articles on the nuclear issue as a
measure of salience, thus, supports the results described in the next section.
As suggested in the second set of hypotheses, parties might only respond to
issues that are salient, ignoring protest and public opinion on non-salient
issues. The debate on the usage of nuclear energy after the accident in
Fukushima has mainly been framed as an environmental issue by political
parties. Thus, using the number of people naming the environment as the
most important problem facing the nation should depict the salience of the
nuclear issue for the public well.
Parties’ left-right position come from the comparative manifesto project
(MARPOR) (Budge et al. 2001; Klingemann et al. 2006; Volkens et al. 2012).
While most studies used the rile score coming with the MARPOR data to
estimate parties’ left-right positions, here I rely on Lowe et al.’s (2011) suggestions and use a logit scale to measure parties’ ideological left-right placement. Parties’ left-right placement runs from -2 to 1.4 with higher values
indicating parties with a more right wing ideological position (µ=-.4; σ=.7).
Given that the MARPOR has been criticised for showing erroneous leftright placement of parties in several instances (Laver 2001: 66-75; Pelizzo
2003), I cross validated the measurement and results with the Chapel Hill
expert survey, the most extensive expert survey on parties’ left-right placement (Bakker et al. 2015). Both measurements are highly correlated (r=.7)
and using the Chapel Hill expert judgements instead of the MARPOR leftright placement does not change the conclusions to be drawn from the
following analysis.62 I chose the MARPOR data above the Chapel Hill ex61 Data stem from the Eurobarometer for European countries; Gallup for the USA; Focus
Canada Report for Canada; Sorgenbarometer for Switzerland. A detailed overview on the
sources and date of fieldwork can be found in the appendix on 168 in table 12.
62 For the Swiss Bürgerlich-Demokratische Partei (BDP) and the French Nouveau Centre
(NC) the MARPOR does not provide manifesto codes, since both parties had not yet
96
5.5 results
97
pert surveys as the latter only includes European countries. Green parties
have been measured by using a dummy variable, which is ‘1’ in the case a
party belongs to the Green party family and ‘0’ otherwise.
Several controls are included in the subsequent analysis. Parties might
be prone to be influenced by the status quo of the energy reliability of the
country they are acting in. Thus, instead of calling for reforms they might
favour the more convenient current situation, which does not force them to
replace existing policies. I used the amount of energy produced by nuclear
reactors in each country to measure the status quo. The data come from
the World Nuclear Association and are available on an annual basis.63 Finally, I decided to standardise the unlogged variables by subtracting the
mean from a given variable and dividing by its standard deviation. While
standardisation does not change the substantial results of the regression
analysis it eases comparisons between coefficients. Above all it mitigates
multicollinearity issues, which potentially linger due to several interaction
terms and their constituting base terms being included in the regression
analysis (Gelman 2008).
5.5
results
Table 4 reports the results of the analysis. Column 1 is a baseline model containing all variables without any interaction. Column 2 adds an interaction
term between parties? left-right placement and the protest-index, model
3 includes the interactions between public opinion and the protest-index.
Model 4 reports the results for both interactions within a single model.
Columns 6 and 7 report models for testing the Green party hypothesis.
Column 1 reports the baseline model. The positive and statistically significant coefficient of the protest measure suggests that parties react with
pro-nuclear statements when they face higher protrusion of protest. Thus,
in the baseline model increased protest on the nuclear issue leads parties
competed in elections previous to the Fukushima disaster. Therefore, I used the Chapel
Hill expert judgements to linearly interpolate their MARPOR left-right placement. Due to
the high correlation of both measurements, the Chapel Hill expert survey appears to be a
reliable source for linear interpolation of the two values. Excluding the two parties from
the analysis does not change the findings.
63 http://www.world-nuclear.org/info/Facts-and-Figures/Nuclear-generation-by-country/,
last checked: 07.01.2015.
0.578
(0.457)
1165
0.245
0.573
(0.450)
1165
0.260
(2)
polarisation
0.0204
(0.061)
-0.898∗∗∗
(0.106)
0.275∗∗∗
(0.064)
0.0735
(0.120)
0.0879
(0.165)
-1.247
(1.033)
0.257∗∗∗
(0.063)
Panel corrected standard errors in parentheses;
all models use ar(1).
CFE omitted from table.
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001.
N
R2
constant
Green X survey
Green X protest
survey X protest
ideology X protest
nuclear share
mip
survey
protest
Green
ideology
(1)
baseline
0.138
(0.072)
-0.894∗∗∗
(0.108)
0.142∗∗
(0.045)
0.0933
(0.122)
0.130
(0.164)
-1.369
(1.047)
0.705
(0.428)
1165
0.257
0.310∗∗
(0.113)
(3)
deflation
0.142∗
(0.071)
-0.890∗∗∗
(0.106)
0.129∗∗
(0.047)
-0.0496
(0.125)
0.0848
(0.161)
-0.954
(0.968)
0.679
(0.422)
1165
0.272
(4)
mutual
0.0284
(0.059)
-0.896∗∗∗
(0.104)
0.245∗∗∗
(0.055)
-0.0551
(0.121)
0.0416
(0.170)
-0.950
(0.952)
0.245∗∗∗
(0.056)
0.294∗∗
(0.105)
Table 4: The effect of protest on party agendas, regression results
-0.656∗∗∗
(0.117)
0.290
(0.197)
0.545
(0.450)
1165
0.273
(5)
Green party
0.136
(0.072)
-0.589∗∗∗
(0.113)
0.200∗∗∗
(0.046)
0.0710
(0.120)
0.133
(0.161)
-1.418
(1.035)
0.545
(0.450)
1165
0.271
-0.673∗∗∗
(0.120)
(6)
Green party
0.139
(0.072)
-0.591∗∗∗
(0.115)
0.202∗∗∗
(0.047)
0.0974
(0.121)
0.133
(0.162)
-1.396
(1.037)
5.5 results
98
5.5 results
to speak out against protesters’ claims. This is an interesting finding and
adds nuance to previous research’s findings, which have only focused on
measuring the amount of reactions by political parties. Previous research
interpreted the increased attention by parties to protesters’ claims as a signal of responsiveness of political actors (Walgrave and Vliegenthart 2012).
In fact, the results reported in model 1 here suggest that parties seem to
reject protesters’ claims. Thus, parties are following the exact opposite strategy by rejecting protesters’ claims.64 Furthermore, since parties’ left-right
position is not significant, their rhetorical positions are not purely driven
by their ideology but by other external factors as well. In contrast to the
univariate regressions in the last chapter (chapter 4), the difference of party
ideology is washed away once we control for other factors that influence
parties’ rhetorical positions. Public Opinion also shows no significant effect on party rhetoric. This result adds to the mixed findings of previous
research on parties’ responsiveness to public opinion (Manza and Cook
2001). However, as further analyses reveal, this effect is mainly driven by
the low salience of the issue of nuclear energy within most countries included in the sample. Once we take salience into account, parties are more
likely to talk pro-nuclear the more this position is supported by the public
at large (Lax and Phillips 2009, 2012). This again contributes to the results
of previous research, suggesting that parties and governments are mainly
responsive to salient issues, while refusing to react to public opinion on
low salient topics (Uba 2009; Burstein 2014). Furthermore, unsurprisingly,
Green parties are more likely to talk anti-nuclear than all other party families.
Model 2 adds an interaction term between the protest-index and parties’
left-right placement to test the party polarisation hypothesis. As suggested
in hypothesis two, parties react differently to protest contingent on how
their general left-right ideology relates to the issue of nuclear energy. The
more left a party’s ideology, the more likely it is to support protesters’
anti-nuclear claims. In contrast, the more right the party’s ideology is, the
more likely it is to reject the demands put forward in protests. Figure 9
visualises this result, plotting the marginal effect of the protest-index con64 Note that a negative binomial model with the amount of rhetorical actions by political
parties supports the findings of previous research. The higher the protrusion of the issue
at stake is, the likelier political parties are to focus their attention on it during protest.
99
5.5 results
ditional on parties’ left-right placements. The density function of parties’
left-right placement is in the background. In the line of hypothesis two,
party competition on the nuclear issue becomes more polarised with increasing protrusion of protest. Instead of all parties reacting in favour of
protesters, parties to the right tend to underpin the status quo when facing
increased protest on nuclear energy. In contrast, left parties are more likely
to support the protests in an effort to benefit from the popular support of
their ideology. This latter effect appears for parties with a left ideology of
‘-1’ or more on the standardised left-right placement (e.g. the Scandinavian
Social Democratic and Green parties or the Spanish PSOE). Parties tend
to perceive protest against nuclear energy either as a window of opportunity (left parties) or a threat (right parties) and react accordingly. Admittedly, issues of endogeneity might linger, with protesters being potentially
motivated and inflamed by parties’ rhetorical positions. Yet, in this vein
protesters’ activities would depend on right parties’ pro-nuclear talk, while
the support of left parties would belittle protesters’ motivations to take to
the streets. Furthermore, running a model with the protest-index as the
dependent variable does not find support for such a reading of the results.
Column 3 tests the deflating effect of public opinion. The theoretical section argued that parties are more likely to support the public at large if
protest and public opinion disagree with each other. Recall that public
opinion is measured here as the difference between the people favouring
nuclear energy and those rejecting it. Figure 9b shows that with an increasing gap between those who favour the usage of nuclear energy and those
who reject it, higher protrusion of protest results in parties supporting the
public at large while rejecting protesters’ claims. This effect is significant
across the whole range of pro-nuclear public opinion. Intriguingly parties
also reject protesters’ claims if the public supports protesters’ anti-nuclear
stance. It takes a gap between the people supporting nuclear energy and
those rejecting it of more than 16 per cent for the deflating effect to become
insignificant (the horizontal line in 9b running through ‘-16’ on the x-axis
marks this threshold). Thus, parties appear to be mainly vote driven. They
are not interested in taking chances by supporting a strong, mobilised minority, but rather stick to the position of the status quo (the usage and support of nuclear energy). Simultaneously testing both the party polarisation
and deflating hypotheses (column 4) shows that even though the polarisa-
100
5.5 results
1
.6
.5
.4
0
.2
−.5
density party ideology
Effects on Fitted Values
Figure 9: Marginal effect of polarisation & deflating hypothesis
0
−2
−1
0
1
party LR position
(a) Party polarisation hypothesis
1
.02
.5
.01
0
density public opinion
Effects on Fitted Values
.015
.005
−.5
0
−55
−40
−20−16
0
PO: % favoring nuclear energy
20
40
(b) Deflating hypothesis
Source: Author’s own.
Note: Solid lines report marginal effects surrounded by the 95 % confidence interval as
dotted lines. Dashed lines report density function of the variable plotted on the x-axis.
Independent variables kept as observed (at their means).
101
5.5 results
tion effect of protest takes away power from the coefficient of the deflating
effect of public opinion, both effects remain significant.
Models 5 and 6 reject the hypothesis that Green parties do not increase
their attention to their anti-nuclear position when facing protest. Green parties read protest on nuclear energy as a window opportunity. Issues related
to the usage of nuclear energy remain at the core of Green parties’ ideology.
The Green party variable is highly significant across all models and has a
strong negative coefficient. In order to be perceived competent on the nuclear issue Green parties continuously remind voters of their anti-nuclear
position and extend their anti-nuclear rhetoric with increasing protrusion
of protest. This result speaks to previous studies on niche party behaviour,
showing that niche parties are less prone to be responsive to public opinion shifts, but might in fact be more responsive to the interests of their
core-voters (Adams et al. 2006; Ezrow 2010). As model 7 shows, an interaction term between Green parties and public opinion is not significant,
while the interaction between Green parties and the protest-index is. Thus,
the results enhance our knowledge on niche party behaviour.
Figure 10 compares the coefficients of model four in table 4. Recall that
all independent variables have been standardised to ease comparisons. Being a Green party has the strongest effect on parties’ decision to express an
anti-nuclear point of view, while the effect of the three way interaction between the protest-index, public opinion and party ideology is considerably
smaller. Finally, the status quo of a country’s reliability on nuclear energy
does not have a significant effect on party rhetoric.
The previous results gave insight into how parties adapt their rhetorical
position when facing protest. However, the measurement of parties’ rhetorical positions intentionally conflates party positions with the salience they
attach to protest. Thus, the results do not address the question of whether
parties might also become more congruent in their position when facing
protest, as the dependent variable takes into account party positions and
the salience they attach to an issue.
To disentangle salience and position, the figures (11 a) & b)) below employed the same set of independent variables as specified above, but a
different specification of the dependent variable. In line with the theoretical reasoning on congruence, I conceptualise congruence as the percent-
102
5.5 results
Figure 10: Polarisation & deflating hypotheses, coefficient plot of fully specified model
green party
public opinion
protestindex
public opinion * protestindex
party ideology
party ideology * protestindex
−1
−.5
0
.5
Source: Author’s own.
Note: figure reports the coefficients of model (4) on page 98 in table 4. Spikes mark 95 %
confidence interval.
103
5.5 results
age of statements in favour of the protesters’ anti-nuclear claims.65 Therefore,
the higher this congruence measure is, the higher the percentage of antinuclear statements made by a political party. Variables measured in percentage are usually not approximately normally distributed but tend to
be skewed towards the extremes. This is also the case for the congruence
measure employed here, as the density function of the variable in the appendix shows (see: figure 20 on page 172). Using standard OLS regression
techniques might result in highly biased estimates since the errors of the
model tend not to be distributed approximately normally, which violates
the Gauss Markov assumptions of standard OLS regressions.
To address this issue I use generalised estimating equation models (GEE)
(Liang and Zeger 1986; Zeger, Liang, and Albert 1988). The advantage of
GEE models have been extensively discussed elsewhere (Zorn 2001). For
the calculations here two major advantages outweigh other calculation techniques. First, violations of the normality distributions of the errors are not
an issue for GEE models. Second, users of such models can specify their
own link function and correlation structure within cases. Due to the approximately bimodal distribution of the dependent variable, I use a logit
link function and a bernoulli (binomial) distribution to model the dependent variable. Furthermore, I include robust standard errors for each party
to control for clustering of the errors within parties.66
Figure 11 summarises the major findings from these models. The full
models can be found in the appendix on page 176 in table 15. Figure 11a
tests the contingent effect of ideology. The results of the congruence model
correspond to the results outlined in the rhetorical position measure. The
more right wing a party’s ideology is the smaller the probability of a higher
percentage of anti-nuclear statements. The interaction is significant, but the
effect shades away for parties located on the extremes of the right. The confidence interval overlaps with zero starting at a value of 0.8 of party ideol65 We can only conceptualise congruence if we define protesters as having a single ideological position. Otherwise, parties cannot be congruent to all protests, but only to specific subsamples. Conveniently in the case of nuclear energy pro-nuclear protest never outscores
anti-nuclear protest. Thus, the protrusion of protest only takes on anti-nuclear positions.
66 I did not employ country fixed effects as these result in perfect prediction of the failure
of the dependent variable and subsequently to the exclusion of various cases. I use an
exchangeable correlation structure within parties. However, I follow Zorn (2001) and calculated other correlation structures as well. The results reported here a roust to other
correlation structures (unstructured; stationary; ar1).
104
5.5 results
Figure 11: Marginal effect of polarisation & deflating hypotheses, results of GEE models
.25
Effects on Pr(Congruence != 0)
.2
.15
.1
.05
0
−2
−1
0
party LR position
1
2
(a) marginal effect of interaction between protest and party ideology
Effects on Pr(Congruence != 0)
.1
.05
0
−.05
−.1
−.15
−60
−40
−20
0
PO: % favoring nuclear energy
20
40
(b) marginal effect of interaction between protest and public
opinion
Source: Author’s own.
Note: Figures report marginal effect of the interactions. Dotted lines mark 95 % interval.
Independent variables kept on their means. Table 15 and figure 21 in the appendix on
page 174-176 report the full results.
105
5.6 conclusion
ogy (e.g. Vlaams Belang in Belgium). Thus, these results actually add nuance to my previous findings since they suggest that polarisation only happens on the extremes and in complex ways for the right-wing, but not for
the centre-left or for the centre-right parties. Figure 11b also supports the
assumption that parties’ reactions to protest are contingent on the support
of protesters’ claims amongst the general public. In summary, parties not
only adapt the salience they attach to the issue of nuclear energy, but also
adapt their general position to protesters’ claims. Again, this result adds
some nuance to the previous findings. Parties will not clearly express an
anti-nuclear position until public opinion is overwhelmingly anti-nuclear.
However, some politicians already outline anti-nuclear rhetoric before the
anti-nuclear mood is dominant amongst the wider public.
5.6
conclusion
This chapter aimed to bridge the gap between studies focusing on the impact of protest on policies and research linking public opinion and policies.
In contrast to previous research, I argued that protest does not have the
same effect for all parties. Parties are understood to perceive changes in
their environment differently depending on their ideology, with left parties being more responsive to protest, while right parties attempt to debate
and underpin the status quo in the case of nuclear energy. Furthermore, I
hypothesised that public opinion has a deflating effect on the influence of
political protest on party rhetoric.
The analysis of party rhetoric across twelve countries and two years underpins both theoretical arguments. The analysis relied on the dataset presented in chapter 3. Using the protest-index and parties’ rhetorical positions conceptualised in chapter 3, I show that protest has an influence on
rhetorical party competition. I find that the influence of protest on parties’
rhetorical position on nuclear energy is contingent on two major factors.
First, parties react differently to protest depending on how their ideology
relates to the nuclear issue. Parties to the left support protesters’ claims,
while parties to the right tend to reject them. Thus, the latter parties tend
to resist public pressure through political protest by sticking to the status
quo. Furthermore, public opinion seems to be an important obstacle for
106
5.6 conclusion
protesters. If the public rejects the mobilised protesters the odds are high
that political parties will also reject protesters’ claims in an effort to secure
votes and policy.
However, more research is needed in order to be able to strengthen these
arguments. The moderating effect of protest on public opinion especially
needs to be studied in more detail, as it appears that protest mainly interacts with the opinion of the public at large if it is at odds with public
opinion. The study of protest also needs to be expanded to a wider range of
issues than just nuclear energy, for example, foreign policy and economic
issues. Chapter 7 address this gap partly by looking into three different issues. In this regard the strong effects found in the analysis on party rhetoric
are telling and might be promising for future studies to look into the impact of protest on economic and foreign policy issues.
107
6
D O E S T I M I N G M AT T E R ? T H E E L E C T O R A L C Y C L E &
PA R T I E S ’ R H E T O R I C A L R E A C T I O N S T O P U B L I C
CLAIMS
6.1
introduction
lections are the most obvious and often theorised mechanism for
holding representatives accountable in modern democracies (Ferejohn 1986; Powell 2000). On election day the electorate decides
whether to keep its previous government in office or swap. In turn, elections present an opportunity for politicians and their organisations to respond to public grievances and desires in an effort to maximise votes and
secure office benefits (Downs 1957; Strøm 1990). In this respect the timing
of elections should affect and potentially moderate politicians’ political behaviour. Elections bring politicians and their statements into the limelight
of the media and the public. Thus, politicians should carefully choose when
and how to respond to public claims when elections are near.
The mechanism of an electoral cycle has been studied for American Presidents and the Congress in several studies (Kuklinski 1978; Tufte 1978; Cohen 1999; Canes-Wrone and Shotts 2004; Rottinghaus 2008).67 Yet, to my
knowledge only a handful of studies have sought to investigate the effects
of the temporal proximity of elections in comparative perspectives (Jennings and Wlezien 2015; Bernardi 2014). This paper aims to address this
research gap by arguing that political parties’ rhetorical reactions, such as
politicians’ statements to the press, speeches and interviews, are moderated
by the temporal proximity of elections.
I argue that the electoral cycle has a moderating effect on how parties
react to divergent public claims. Given that parties are understood to maximise their electoral payoffs, they should be more likely to respond to the
public at large when elections are near. Due to their frequently low turnouts
and form, protests are often assumed to reflect minorities’ interests on is-
E
67 Often also called “electoral business cycle” (Canes-Wrone and Park 2012).
108
6.2 the electoral cycle and responsiveness
sues which are less salient than other for the public in general (Gillion
2013). Therefore, parties should be less responsive to protesters’ interests,
which often reflect minority interests, the smaller the temporal proximity
to an election is. The theoretical argument is tested on the dataset of parties’ rhetorical reactions to public claims after the Fukushima meltdown. I
find partial support for my theoretical argument. While the data support
the argument that parties become more responsive to the public at large
the closer the next election is, the hypothesis of a decreasing effect of parties’ responsiveness to protest finds no support. Only when the salience of
the nuclear energy issue amongst the public is included in the moderating
effect of the electoral cycle does the effect of the electoral cycle on parties’
rhetorical responsiveness to protesters’ claims become visible. However the
effect of this three-way interaction runs against the hypothesis derived in
the theoretical section, which is likely to be the case due to the low mean
salience of the nuclear issue across the sample.
The following section reviews the literature on the mechanism of the
electoral cycle. The third section outlines the theoretical argument of how
the electoral cycle affects parties’ rhetorical reactions to public claims. The
fourth section describes the data used in the analysis, while the fifth section
shows the results of the pooled regression analysis. Section six concludes
the chapter.
6.2
the electoral cycle and responsiveness
Understanding the interdependence between public demands and policy is
central to the study of democracy (Dahl 1971). As we have seen in chapters
2 and 5 of this thesis, a fairly large number of studies show that governments adapt their policies towards public claims (Miller and Stokes 1963;
Page and Shapiro 1983; Stimson, Mackuen, and Erikson 1995; Wlezien 1995;
Soroka and Wlezien 2012). Thus, we seem to know that public opinion
has an influence on what policies governments implement (Burstein 1998).
Even though these studies enhance our theoretical and empirical comprehension of governments’ accountability to the public, they largely rely on
annual spending or comparable budgetary measurements to judge how responsive governments are to the public. Hence, far less research scrutinises
109
6.2 the electoral cycle and responsiveness
the daily responses politicians give to the people in their press releases, interviews, speeches, and press conferences. Furthermore, the time intervals
of the measures employed in these studies make it difficult, if not impossible, to determine whether and how parties adapt their position along the
electoral cycle.
However, the effects of an “electoral cycle” are well documented for
American Presidents and in work on legislators’ roll call behaviour in the
USA. Legislators are shown to be more responsive to public opinion in
the years they face re-election, likely due to an effort to secure the latter (Kuklinski 1978; Bernstein, Wright, and Berkman 1988; Thomas 1985;
Ahuja 1994; Shepsle et al. 2009). Similar theoretical claims have been made
for American Presidents (Canes-Wrone, Herron, and Shotts 2001), but studies have found mixed empirical results for the US case (Tufte 1978; Cohen
1999; Canes-Wrone and Shotts 2004; Rottinghaus 2008). As a result, political science is stuck for an answer on how the electoral cycle affects the
responsiveness of parties in comparative perspective.
On top of that, the current research agenda dealing with representatives’
responsiveness to public claims is primarily concerned with how public
opinion surveys affect the daily routine of legislators. Certainly, public
opinion is not the only factor seeking to influence politicians’ standpoints,
even though it is usually understood as the most distinct one. Just like the
agendas of legislative chambers can be subdivided into several constituting parts (e.g. parties, single MPs, see: Achen 1977, 1978; Weissberg 1978),
the public agenda appears to be more complex than single point estimates
measured through public opinion polls suggest. People attempt to affect
politics through various channels and protest has become a profound strategy for seeking such influence, especially on issues linked to postmaterialism, such as the environment, pacifism and various cultural/social issues
(Kriesi et al. 1995; Inglehart 1977). Yet again, sociological research dealing
with the influence of social movements neglected the moderating effect that
the electoral cycle might have on protests’ success (Andrews 2001; Giugni
2007; Agnone 2007).
110
6.3 how timing matters
6.3
how timing matters
The successful influence of public claims on party agendas is likely to depend on timing. Most research on the electoral cycle points out that responsiveness becomes more likely the closer the date of the next election is
(Kuklinski 1978: 173-174; Thomas 1985: 97; Canes-Wrone and Shotts 2004:
693). Several mechanisms might stand behind such an electoral cycle effect.
First, citizens’ voting preferences are preliminary driven by recent policy
decisions and positions as they are most likely to remember these (Thomas
1985: 97). This tends to lead politicians to act particularly cautiously and
responsively to public claims when facing elections in an effort to secure
(re-)election. In contrast, representatives have more leeway early in the legislative term (Elling 1982: 76). Thus, they can afford to deviate from the
interest of the public during the first part of their term, but rarely during
the second.
Second, periods of electoral campaigning result in the media focusing
their attention on the emphasised issues during campaigning and especially on the ideological position of political parties (Harris, Fury, and Lock
2006). Thus, voters are more frequently exposed to messages by parties
during campaign periods. In turn, politicians will try to enhance their popularity with their statements.
Third, politicians might make promises in electoral campaigns which
they cannot implement before the election date. As such, electoral campaigns can deliberately be used by politicians to give promises for the midand long-term future without the pressure to deliver policy in the short
term. This last argument appears to be an especially important theoretical
mechanism when trying to explain rhetorical reactions by parties, which
this chapter seeks to do.
Yet, politicians are exposed to a divergent set of actors seeking to influence their positions and decisions. Thus, even though they face a single
electorate, voters express their opinions through multi-facetted channels. I
argue that parties are not equally responsive to all of these opinions across
the electoral cycle. Parties’ responsiveness depends on the temporal proximity and channel through which the public seeks influence.
Assuming that parties aim to maximise their vote-share come election
day (Mayhew 1974; Stimson, Mackuen, and Erikson 1995), parties should
111
6.3 how timing matters
be responsive to those fractions of the public indicating the highest possible electoral payoff at the ballot box. Thus, the closer the next election the
more likely parties are to represent the public at large. Consequently, the
closer the election day the less likely parties are to position themselves in
line with minorities in the electorate. Representing interests of minorities
in this scenario comes with the risk of disagreeing with the majority opinion. Even if this is not the case, representing minority opinions taking to
the street might result in counter-mobilisation by the silent masses. Subsequently parties might be punished by voters for being unresponsive to
their preferences.
H1a : The closer the next election, the more likely parties are to
represent the public at large – as measured through public opinion polls.
H1b : The closer the next election, the more likely parties are to
ignore protesters.
Clearly, protest does not always reflect the opinion of a minority. The
grievances shared during protest events might also be rooted in the general
public. This can manifest in large numbers of participants during protest
events. However, the number of participants during protests does not necessarily mirror the support of protesters’ preferences by the public. Turnout
during protest is also a function of a wide set of other factors which can be
broadly summarised under the term of costs and collective action problems
(Tucker 2007; Ostrom 1998). Thus, the first hypothesis might be moderated
by the amount of people regarding the issue at stake during protest as
important. Parties might not perceive protest as representing minorities in
case a larger fraction of the public perceives the issue at stake as a relevant issue to be solved by the political elite. Therefore, parties might meet
protesters with open ears even though the upcoming elections are close if
they sense that the public supports the protesters’ grievances. Parties might
perceive such an occasion as an opportunity to gain new voters, without
ignoring the interests of the public at large or as a means to satisfy public
grievances before facing them during elections. On top of that, a comparable argument can be made for parties’ responsiveness to the majority public opinion. The literature suggests that governments’ responsiveness to the
public at large depends on the salience of the issue (Wlezien 1995; Soroka
112
6.4 model specifications & measuring the electoral cycle
and Wlezien 2010). Thus, the effect of the temporal proximity of the next
election on parties’ responsiveness to protest and public opinion is likely to
be moderated by the salience of the issue at stake. Only if the percentage of
the public that regards the issue as important is large enough will parties
be responsive to public opinion polls when facing an election. In contrast, if
the public regards an issue as unimportant, parties will become less likely
to be responsive to public opinion along the electoral cycle.
H2a : Only if the public at large regards the issue as salient will
parties be more responsive to the public at large the closer the
next election.
H2b : Parties are more likely to be responsive to protest the closer
the next election, if a sufficient amount of the public perceives
the issue at stake during protest as salient.
6.4
model specifications & measuring the electoral cycle
Previous research has employed various measurements of the electoral cycle. For instance, studies on the American presidency have split data-sets
into first and second term periods (Canes-Wrone and Shotts 2004; Rottinghaus 2008). Using dummy variables to mark the pre-electoral quarter has
been the most common approach for measuring the electoral cycle (CanesWrone and Park 2012). Yet, such a dummy approach does not translate
the theoretical argument of a complete business cycle of elections empirically. Rather it captures a spasmodic adaption of parties’ behaviour during
campaign periods.
Thus, instead of following this approach, I include a time variable to
measure the electoral cycle. I measure the electoral cycle by counting the
months until the next election for each country starting with the Fukushima
accident. Thus, the cycle variable reaches its maximum (52) directly after
a national election and its minimum (0) on the month of the next national
election. The lower the value of the electoral cycle variable, the temporally
closer the next election is. Also the countries included in the study were
at different and distinct periods of the electoral cycle when the Fukushima
meltdown happened.
113
6.4 model specifications & measuring the electoral cycle
114
Several controls are included in the following analysis. This chapter uses
the data outlined in chapter 3 and used in the previous empirical chapter
(chapter 5). Again I employ two measures for public opinion, one measuring the ideological position of the public and the other the salience of environmental issues. I include binary variables for Green parties and government parties. While Green parties are expected to support the anti-nuclear
stance of protesters, incumbent parties are suspected to speak against demonstrators’ interests. I control for party ideology by using the logged rile score
stemming from the manifesto data project (Lowe et al. 2011). Finally, parties might be influenced by the status quo of a country’s reliability on nuclear energy. Thus, instead of calling for reforms they might favour the
more convenient current situation, which does not force them to replace
existing policies. I used the amount of energy produced by nuclear reactors in a given country to measure the status quo. The data come from the
World Nuclear Association and are available annually.68 Finally, I decided
to standardise the unlogged variables by subtracting the mean from a given
variable and dividing the result by two standard deviations (Gelman 2008).
To test my theoretical arguments I examined whether parties’ responsiveness to public claims vary along the electoral cycle. Thus, I estimate
a model in which the rhetorical position measurement is the dependent
variable, the electoral cycle, the protest-index, public opinion and several
controls are the independent variables:
RPit = b0 + b1 Electoral Cycleit
+b2 Protest − indexit + b3 Protest − indexit ∗ Electoral Cycleit
+b4 Public Opinionit + b5 Public Opinionit ∗ Electoral Cycleit
(3)
+b6 ζ it + eit
With ζ being a vector of controls outlined above. The hypotheses derived in
the theoretical section are translated into interaction terms, each indicating
the respective moderation effect on the influence of public claims through
the electoral cycle.
I then rely on the same modelling strategy used in the previous empirical chapter (chapter 5), since I use the same data in both chapters. I ran
68 http://www.world-nuclear.org/info/Facts-and-Figures/Nuclear-generation-by-country/,
last checked: 07.01.2015.
6.5 results
an OLS regression with panel-corrected standard errors, an first-order autoregressive (ar(1)) correction and country fixed effects. Again, the results
of the models employed are robust to different sub-samples and estimation strategies, e.g. including a lagged dependent variable (Beck and Katz
1995); or using Driscoll-Kraay standard errors which tend to be more consistent for temporal invariant covariates (Driscoll and Kraay 1998). Since
the study is interested in the effect of protest on party rhetoric, I chose to
not lag the independent variables. In line with previous research I assume
a temporally close translation from protest into party agendas (Walgrave
and Vliegenthart 2012).
6.5
results
Table 5 reports the results of the analysis. Column 1 is a baseline model
containing all variables without any interaction. Column 2 and 3 add interactions between the protest-index, public opinion and the electoral cycle.
Model 4 and 5 subsequently add the most important issue measure to the
interaction.
Recall that the cycle variable decreases the closer the next election is.
Thus, parties tend to have a more pro-nuclear rhetorical position the closer
the next election is as the significant and negative effect of the cycle variable in column 1 shows. Yet, this effect is not stable across all models and
is only significant at the five per cent level. The effect of most controls corresponds to the assumptions made earlier, as parties in government tend
to talk more in favour of nuclear energy and Green parties talk against it.
Both results are in line with the assumptions outlined in the methodological section of this chapter and support the results of chapter 5. However, an
increase in the protest-index by a single unit leads to a significant increase
of pro-nuclear rhetoric by a factor of 0.153 (maximum of rhetorics=4.5). Yet,
this effect is comparably small in relation to other coefficients in the model
and its significance varies across the models. Furthermore, the result of
chapter 5 showed that parties adapt their rhetorical agenda to protest depending on their general left-right ideology. Once the model includes an
interaction between the protest-index and party ideology, it becomes clear
that left parties support protest in their rhetoric, while right parties reject
115
6.5 results
Table 5: The moderation effect of the electoral cycle, regression results
cycle
protest
survey
mip
government
rile
Green
nuclear share
(1)
baseline
(2)
cycle
X protest
(3)
cycle
X survey
-0.194∗
(0.093)
0.153∗∗
(0.049)
0.120
(0.128)
0.0707
(0.177)
0.549∗∗∗
(0.079)
0.109
(0.069)
-0.834∗∗∗
(0.098)
-1.244
(1.027)
-0.211
(0.108)
0.163∗∗
(0.052)
0.121
(0.127)
0.0668
(0.177)
0.548∗∗∗
(0.079)
0.108
(0.069)
-0.834∗∗∗
(0.098)
-1.261
(1.030)
0.0421
(0.111)
-0.125
(0.091)
0.122∗
(0.051)
0.224
(0.116)
-0.0401
(0.164)
0.576∗∗∗
(0.083)
0.114
(0.069)
-0.825∗∗∗
(0.099)
-1.808
(1.012)
cycle X protest
(4)
cycle
X protest
X MIP
-0.141
(0.103)
0.191∗∗∗
(0.053)
0.0978
(0.127)
0.0414
(0.176)
0.559∗∗∗
(0.079)
0.110
(0.068)
-0.831∗∗∗
(0.097)
-1.201
(1.153)
-0.137
(0.119)
-0.745∗∗∗
(0.202)
cycle X survey
cycle X mip
0.0739
(0.197)
0.0172
(0.106)
-0.760∗
(0.312)
protest X mip
cycle X protest X mip
survey X mip
cycle X survey X mip
N
R2
1165
0.283
1165
0.283
1165
0.292
Panel corrected standard errors in parentheses.
Country fixed effects omitted from table; USA used as reference.
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
1165
0.290
(5)
cycle
X survey
X MIP
-0.117
(0.087)
0.124∗
(0.050)
0.184
(0.117)
-0.00194
(0.170)
0.577∗∗∗
(0.083)
0.114
(0.069)
-0.825∗∗∗
(0.098)
-2.042∗
(1.028)
-0.754∗∗∗
(0.187)
-0.0314
(0.218)
-0.245
(0.213)
-0.0596
(0.483)
1165
0.294
116
6.5 results
protesters’ claims. Finally, the status quo measured through the amount of
nuclear energy used by a given nation does not have a statistically significant effect on parties’ rhetorical positions.
Column 2 reveals that the hypothesis of a decreasing effect of protest on
parties’ rhetorical positions needs to be rejected. The interaction between
the electoral cycle and the protest-index is not statistically significant. However, the coefficient of the interaction holds the expected sign; the closer the
next election, the less likely parties are to support protesters’ claims. Yet,
even looking into a marginal effect taking into account the entire covariance of the fully specified interaction term does not reveal a significant
interaction effect. Thus, the data find no support for the first hypothesis for
the protest-index. Interestingly, this is not true for the interaction between
the support of the public for the usage of nuclear energy and the electoral
cycle. The interaction is significant. The closer the election, the more likely
parties are to adapt their rhetoric to the position of the public.69 This effect
is significant and robust across different model specifications. Thus, the
results underpin the theoretical assumption of an electoral cycle effect for
parties’ reaction to the public at large, but reject such an effect in relation
to protest.
Yet, this effect of the electoral cycle becomes visible once the moderating
effect of the salience of the nuclear energy issue is included in the interaction. Column 4 and 5 in table 5 reveal that the salience of the nuclear
issue has an important moderating effect on parties’ rhetorical reactions to
protest.
Figure 12 illustrates the significant effect between protest, the salience
of the issue and the electoral cycle. Figure 12 shows that parties are more
likely to reject protesters’ claims the nearer the next election and the higher
the salience of environmental issues. In sharp contrast to the theoretical argument, political parties actually become more outspoken in their support
of nuclear energy the closer the next election and the higher the protrusion of protest is (cycle=11). In contrast, 24 months before the elections the
slope of the line is flat and approximately zero, suggesting that political
parties ignore protesters even when the public holds the environment to be
a highly salient issue. The slope becomes negative roughly two years (26
69 To assure readability and not overload this chapter, the graphical interpretation of this
interaction is reported in the appendix on page 177 in figure 22.
117
6.5 results
Figure 12: Marginal effect of interaction between protest, cycle & mip
1.5
Effects on Fitted Values
1
.5
0
−.5
cycle=11
cycle=24
cycle=38
−1
.01
.05
.1
.15
.2
MIP in %
Source: Author’s own.
Note: Solid lines report marginal effects surrounded by the 95 % confidence interval as
dotted lines. All other covariates kept at their means. Figure uses the mean and +/- one
standard deviation of the electoral cycle.
months) before the next election. Thus, political parties react in line with
protesters’ claims if an election is two or more years ahead and the higher
the salience of the nuclear issue. Therefore, the data reject the second hypothesis. The reactions of political parties to protesters’ claims along the
electoral cycle are contingent on how important the public ranks environmental issues, but in the opposite direction than hypothesis two suggested.
Parties might curry favour with protesters far away from elections, but as
the day of facing the electorate nears, the more likely they are to speak out
against protesters’ claims, especially if environmental issues become more
salient. At first sight this results sounds counterintuitive. However, even
though there is considerable variation in the salience of environmental issues across the sample compared to other issues, environmental issues are
118
6.5 results
ranked low salience throughout the sample with Sweden and Finland being
the only two countries in which environmental issues are ranked as important by ten per cent or more of the public. Thus, protests on issues related
to the environment largely still reflect the interests of a mobilised minority
during the time period analysed in this chapter. This might explain why political parties are especially interested in sticking to the pro-nuclear status
quo instead of experimenting with claims to reform it.
This effect remains significant once a four-way interaction is introduced
with the expired time since the Fukushima shock. Furthermore, the twelve
countries included in the sample were in distinctively different periods of
the electoral cycle when the accident at Fukushima took place, ranging
from one month until the next election in Finland all the way to 52 months
in the UK, as can be seen in figure 13. While the period from 28 to 38
Figure 13: Histogram of cycle variable
.03
Density
.02
.01
0
0
10
20
30
40
50
cycle
Source: Author’s own.
months is very densely filled, so are periods closer to the next election
(especially one month before the election). Column 5 does not find such
an effect for the position of the public. In this way, there is no contingent
effect of the salience of an issue on parties’ rhetorical reactions along the
electoral cycle. Thus, while parties adapt their reactions to the noise of
protesters along the electoral cycle contingent on the salience of the issue
119
6.5 results
at stake during protest, parties become more likely to react in favour to
public opinion the closer the next election.
These effects also find support in the congruence model. Like in the last
chapter, I again estimated GEE models using a measure of rhetorical congruence as the dependent variable and the same set of independent variables as in the models reported in table 5. The detailed results and estimates
derived from these models can be found in the appendix on page 178 in
table 16. Due to the similarity of the results, I limited the discussion here
to the robustness of the most important results discussed above. Both significant interactions described in the paragraphs above are also significant
in the congruence models. Thus, political parties are more congruent to
the positions of the public the closer the next election is. Furthermore, figure 14 shows that the rhetorical congruence of political parties to protesters’
claims again depends on the salience of environmental issues and the closeness of the next election.
Political parties are most likely to utilise a higher percentage of antinuclear rhetorical activities the further away the next election is and the
lower the salience of environmental issues. This effect decreases with the
approach of election day. Interestingly the slope of the interaction becomes
flat the further away elections are. This suggests that the further away elections are, the more willing political parties are to be congruent to protesters
claims, irrespective of the salience of the issue protested upon. Again the
congruence models add some nuances to the results of the first set of models.
120
6.6 conclusion
Figure 14: Marginal effects of moderation through the electoral cycle
Effects on Pr(Congruence != 0)
.3
.2
.1
0
cycle=11
cycle=24
cycle=38
−.1
.01
.05
.1
.15
.2
MIP in %
Source: Author’s own.
Note: Solid lines report marginal effects surrounded by the 95 % confidence interval as
dotted lines. All other covariates kept at their means.
6.6
conclusion
This chapter enhanced our knowledge of the moderating effect of the electoral cycle on parties’ reaction to public claims measured through public
opinion data and popular protest. In doing so it depicts a first step towards
a larger understanding of the effect of the electoral cycle in comparative perspective. Previous research has largely focused on the American case and
rarely sought to understand the effect of an electoral cycle on the responsiveness of political parties to public claims in a comparative perspective.
I theorised that parties’ rhetorical reactions are contingent on the temporal proximity of elections and differ for public opinion and protest. While
parties are more likely to be responsive to the public at large the closer the
next election is, they become less likely to react in favour of protesters the
121
6.6 conclusion
nearer the next election becomes. The results partly support this claim in
as much as parties become more likely to react to public opinion the closer
the next election. Yet, there is no significant effect of the electoral cycle on
parties’ reactions to protest. Even though the direction of the coefficient
of the interaction between protest and the electoral cycle corresponds to
the theoretical assumptions, this effect is not statistically significant in the
models calculated for this chapter.
Yet, controlling for the salience of the issue at stake reveals that parties
become less likely to support protesters the nearer the next election in case
the public at large ranks the salience of environmental issues highly. In
this regard the effect of the electoral cycle is strong and substantially interacts with the effect of protest; instead of achieving support for their cause,
protesters’ interests are disregarded in an effort to secure votes at the ballot
boxes. I suspect that this effect is due to the low overall salience of environmental issues. Due to this political parties might shy away from supporting
anti-nuclear protest as they suspect that the public at large believes that it
is more important to discuss and address other issues.
Yet, further research is needed to extend our knowledge of the moderating effect of an electoral cycle. This chapter relies only on data about
parties’ rhetorical position on nuclear energy. Even though the salience of
the issue varies considerably across countries, different dynamics of the
electoral cycle might be found for different issues (Burstein 2014).
122
7
D O E S P R O T E S T M AT T E R I N T H E L O N G - R U N ? F R O M
M AT E R I A L I S M T O P O S T M AT E R I A L I S M , 1 9 8 1 - 1 9 9 6
7.1
introduction
his chapter scrutinises the long-term relationship between party agendas and political protest. It observes the salience parties attach to
issues raised at protests across a longer period than the previous
two empirical chapters. It aims to understand the link between protests on
environmental, military and labour issues and the emphasis parties have
attached in their manifestos to these issues from 1981 to 1996 across 17
West European countries.
Chapters 5 and 6 showed that protests have a strong impact on rhetorical party agendas. I demonstrated that protests do not only lead parties to
talk about the issue at stake during protest, but might also move parties to
support or reject protesters’ claims contingent on how a parties’ ideology
relates to the issue brought up by protesters. Furthermore, chapter 6 suggested that parties’ reaction to protest depends on the temporal proximity
of elections. Both chapters looked at parties’ rhetorical reactions to protests.
Reflecting on arguments discussed in chapter 4, rhetorical reactions are
not always followed by substantive activities, such as drafting legislation
or policy change. Even though the chapter showed that parties’ rhetorical
positions are a solid predictor of parties’ more substantive legislative activities, most rhetorical reactions are made by a single individual and are,
thus, not commonly subject to discussion and formal decision making processes within the party. I then understood the amalgam of these activities
as reflecting the rhetorical position of a political party.
This chapter takes one step forward and looks at the emphasis political
parties attach to different issues in their manifestos. The most often used
approach to interpret the role of manifestos in modern democracies, the
mandate theory, understands party manifestos as “coherent statements of
policy commitment” (Harmel 2011: 1). Manifestos not only link parties to
T
123
7.1 introduction
their voters during electoral campaigns (Adams et al. 2004), they also often
bind representatives of political parties to the party line (Kavanagh 1981)
and are frequently followed by policies coherent with party manifesto positions (Hofferbert and Budge 1992). Thus, contrary to rhetorical reactions,
manifestos might have a binding force for party members, are subject to
more veto points within parties, are drafted specifically during the campaign period, and consequently are also more stable over time. Thus, this
chapter analyses whether political protest influences the manifestos that
parties use to compete against each other in elections with.
To answer this question, this chapter investigates a period of recession,
high unemployment, radical political reforms and social change in advanced Western democracies (1981 - 1996). Thereby, I will again start from
the key theoretical argument that protest is not usually protrusive, but
rather mostly characterised by low numbers of participants, short duration, and a lack of newsworthiness. Only if social movements are able to
overcome these issues are they likely to penetrate into party agendas. However, the main argument presented in this chapter aims to understand the
joint effect of protests on different issues. Instead of investigating each issue separately, I argue that which issue parties emphasise depends not only
on the protrusion of protest on this respective issue, but also on how protrusive protests on other issues are. Merging the comparative manifesto
dataset (MARPOR) (Klingemann et al. 2006) and the European Protest and
Coercion dataset (EPCD) (Francisco 1996), I test my theoretical framework
across 150 parties competing in 80 elections in 17 democracies. The results
reveal that during the 1980s political parties adapted their agendas to the
issues at stake during protests. Thus, the attention seeking assumption
also finds support in this chapter. Interestingly, neither parties’ ideology
nor public opinion have a significant moderation effect on the influence of
protest. This chapter also addresses the causal relationship between political protest and party positions. Using an instrumental variable model, I
show that political parties adapt to political protest and not the other way
around.
The next section summarises the main contributions in the literature to
the question of how parties adapt their manifesto positions to public opinion and protest. Section three discusses the theoretical framework of the
study, which deviates slightly from the framework used in the previous
124
7.2 parties, postmaterialism & issue emphasis
empirical chapters. Section four describes the data and modelling strategy.
The remainder of the chapter tests the theoretical argument and concludes
7.2
parties, postmaterialism & issue emphasis
Traditionally scholarly work has been interested in the adaption of parties’
ideological positions to public opinion shifts on the general left-right scale
(Adams et al. 2004; McDonald, Mendes, and Budge 2003). Also, besides
ideological shifts alteration in the salience parties attach to different issues
in response to public opinion has received increased attention in the last
decade (Spoon 2012; Spoon and Klüver 2015). In both instances scholars
tend to agree that parties adapt to public preferences. Yet, parties’ adaptation is often contingent on their ideology or the type of party. For instance,
Adams, Haupt, and Stoll (2008) find that left parties are more prone to be
stable in their policy positions across time. Ezrow’s (2010) results show that
niche parties are more likely to be responsive to their core voters instead of
being responsive to the public at large.
However, less research attempts to scrutinise the effect of protest on
party agendas (Gillion 2013) and even fewer studies attempted to examine the joint effect of public opinion and protest on party agendas (but
see: Walgrave and Vliegenthart 2012; King, Bentele, and Soule 2007). Furthermore, as discussed elsewhere, the selection of issues analysed in these
studies is biased (Uba 2009; Burstein and Linton 2002; Burstein 2014). First,
like the previous chapters of this thesis, which focus on the issue of nuclear energy, single issue studies are the rule and not an exception in the
field (Gillion 2012; Olzak and Soule 2009; Johnson, Agnone, and Mccarthy
2003; McAdam and Su 2002; Agnone 2007; Amenta, Carruthers, and Zylan
1992; Andrews 2001; Burstein 1979). Thus, methods and operationalisation
vary across studies and issues. Consequently, readers are left behind with
mixed results and their understanding of the influence of protest are largely
bound to specific issues such as welfare policies, civil rights issues or, as
here, nuclear energy.
Second, the variety of issues are neither selected randomly (but see:
Burstein 2014), nor subject to specific selection criteria. Hence, studies tend
to focus on highly salient issues and rarely use the cross sectional varia-
125
7.2 parties, postmaterialism & issue emphasis
tion of issues as has been done by this thesis. However, the probability of a
strong impact of public opinion on these issues is higher, since political actors are likely to satisfy public opinion on these electorally decisive issues
as has been shown elsewhere and in the previous chapter of this thesis as
well (Lax and Phillips 2009, 2012). In fact, research indicates that in case
public opinion disagrees with protesters’ claims on salient issues, political
parties tend to ignore protesters’ claims and instead support the claims of
the public at large to maximise their vote share on election day (Giugni
2007: 65-67). Furthermore, research investigating more than a single issue
has tended to understand and analyse each issue separately. Giugni (2004),
for instance, analyses the ecology, antinuclear and peace movements in separate models.
Yet, this seems to be particularly misleading as we should not think of
the mobilisation and salience of a particular issue to be independent of
the mobilisation and salience of other issues. Both time and resources are
scarce, leading parties and the public to care about certain issues, while
ignoring others. Thus, if the masses take to the streets against nuclear energy, they are unlikely to simultaneously mobilise on a large scale against
unemployment.
In fact, since the early 1970s some scholars have pointed out that public
priorities are changing considerably, which has resulted in the public caring more about some issues and less about others. Inglehart (1971, 1977)
argues that with more and more people enjoying prosperity and especially
secured prosperity, post-war generations began to redefine their values and
political interests. This led the public to care less about traditional values,
those linked to economic, material issues, and give priority to postmaterialist issues, such as the environment, peace and rights of minorities (Inglehart 1971, 1977). Inglehart further suggests that even though society as a
whole adapted its values, individuals’ value orientations tend to be very
stable across time. Thus, his argument departs from the idea of a generational change and not from a short-term fluctuation of individual values.
This development was accompanied by an adaption of the channels used
to share public grievances. What was once known as a means of unconventional participation became ever more conventional (Barnes et al. 1979).
Thus, people more frequently joined protests, strikes and signed petitions.
126
7.3 from materialism to postmaterialism
Yet, while there is a debate about if and how there is a rise of public
concern for postmaterialist issues, we know little about how political parties responded to postmaterialist values by adapting their issue profiles.
Research looking into this period of transition has tended to explore theoretical arguments of (re)-alignment or the transformation of the political
space (Carmines 1991; Stoll 2010; Kitschelt 1994). Some studies have assumed that the increase of postmaterialist values led new parties to emerge.
Yet, in their study on the emergence of new parties, Harmel and Robertson
(1985) do not find a significant relationship between the number of new parties and the percentage of postmaterialists within a society. Other scholars
have found stronger connections between voters supporting postmaterialist values and the vote for new, small, left parties (Müller-Rommel 1985).
Finally, in their single issue study on parties’ emphases of environmental
issues, Spoon, Hobolt, and de Vries (2014) found that factors of party competition such as the size of Green parties especially induced parties to stress
environmental topics.
7.3
from materialism to postmaterialism
Thus, in summary the question remains as to whether or not only citizens
changed their priorities, but also whether parties responded accordingly
by congruently shifting their issue priorities.
Starting from the theoretical framework presented in the last chapters, I
argue that protests will only have an impact on parties’ emphasis of materialist and postmaterialist values if their protrusion is high enough. Even
though protests often focus on a certain topic and can present a clear demand for policy change on a given issue (Schumaker 1975: 490), they might
be small in size (DeNardo 1985), a single one day event or not disruptive
enough to gain the attention of the media (Smith et al. 2001). Thus, they are
often not protrusive and remain unregarded by most external actors. Parties should be willing to ignore protesters’ claims if they sense that protest
reflects the interests of a minority and mobilisation is subject to high fluctuation across time. While political parties have an interest in listening to
the masses to try to secure votes and office, representing specific fractions
of society or a minority could harm their vote- and office-seeking goals.
127
7.3 from materialism to postmaterialism
Yet, with increasing mobilisation, support and disruption, the visibility of
protests in the media grows. As a result, parties are more likely to be forced
to adapt their agendas to secure the named goals.
The argument put forward in this chapter, however, slightly deviates
from the arguments made in previous chapters. Here I argue that the protrusion of protest on postmaterialist values first needs to outweigh the protrusion of materialist protests. Thus, parties are only likely to adapt their
manifesto positions to postmaterialist values if the protesting public sends
a clear signal of the necessity of addressing postmaterialist values. During
the period under observation in this chapter materialist protest on labour
issues occurred frequently. Thus, parties might still have perceived the need
to adapt to materialist priorities instead of shifting their attention to postmaterialist ones.
H1 : The higher the protrusion of protest on (post)-materialist issues, the more likely parties are to emphasise (post)-materialist
values in their manifestos.
The theoretical framework provided in the chapters 2, 5 and 6 argued
that this effect might be contingent on two major factors. First, parties’ general left-right ideology might impact the way in which they react to protest.
However, as Inglehart himself famously noted, traditional cleavages are expected to erode and be replaced by new postmaterialist ones (de-alignment
and re-alignment) (Stoll 2010; Inglehart 1971). Furthermore, recall that this
chapter is not interested in parties’ ideological positioning, but whether
parties prioritise postmaterialist or materialist issues. Clearly one would
expect that party ideology affects parties’ positional reaction to protest. Yet,
in the case of salience a priori there is no reason to assume that parties on
the left will react differently to postmaterialist protest than parties located
on the right of the ideological spectrum. For instance, both parties of the
left and right might attach more salience to military issues in response to
popular protest against missile deployment, even though their positions on
the matter might be entirely different.
H2 : The extent to which parties emphasise postmaterialist values in response to postmaterialist protests is not contingent on
their general left-right placement.
128
7.4 data & methods
Second, instead of reacting to protesters alone, parties might only support protesters’ claims if the public at large supports the protesters’ claims
in the first place. Thus, the percentage of postmaterialists within a society
might be an important obstacle for the success of protest. Only if enough
people within a society support postmaterialist claims, might parties be
willing to support protesters’ postmaterialist claims.
H3 : The protrusion of protest will lead to an increase in parties’
emphases on postmaterialist issues if the percentage of postmaterialists in a society is sufficiently high.
7.4
data & methods
To analyse the effect of protest on party agendas longitudinal data of both
party agendas and protest are needed. I measure party agendas by relying on manifesto data stemming from the Comparative Manifesto Project
(MARPOR) and protest by using the European Protest and Coercion dataset
(EPCD). These data allow comparisons of the impact of protest across 17
democracies and 15 years (1980-1995).70 The following analysis is based on
three major issues aiming to reflect the silent revolution from materialism
to postmaterialism. I chose three issues which can be measured reliably
on the protest- and party-side: the environment, military and labour rights.
Other issues, such as gay rights and feminism, might only be found on
the protest-side, but not specifically coded in the party manifesto dataset
employed here. Thus, I decided to stick to those issues which reliably represent the trajectory between post-materialism and materialism for both data
sources, EPCD and MARPOR.
As suggested in the introduction of this chapter, party manifestos are central to the studies of parties, and more specifically party positions. Manifesto-based research found large appeal since the 1990s and has been applied
in diverse institutional settings across most advanced Western democracies
(Budge and Hofferbert 1990; Adams et al. 2006; Meyer 2013). Thus, party
manifestos are well suited to measure the salience parties attach to different issues. I use the MARPOR data in this chapter. The MARPOR data is
70 Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg,
Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom.
129
7.4 data & methods
based on a content analysis, which codes ‘quasi-sentences’ of party manifestos into 56 broader issue categories (Budge et al. 2001; Klingemann et al.
2006). MARPOR aims to measure the salience of different issues across parties, countries and time. It is the most proficient data source containing
longitudinal and cross-national data of party manifestos. It can be used to
analyse party positions across time and nations from 1943 until 2013.71 As
Laver (2001: 66-75) has shown, using MARPOR to derive left-right positions
of parties may result in flawed estimates. However, in contrast to spatial
modelling of party positions, this chapter is interested in issue emphasis
(Budge and Farlie 1983). While other issues still linger with the MARPOR
data, such as that nearly 30 % of the content of manifestos in Denmark
are deemed impossible to be coded (Hansen 2008: 201), using MARPOR
in line with its original intents seems less problematic than deriving ideological left-right placements of parties from the data. Furthermore, until
today the MARPOR data remain the most inclusive and broad dataset for
estimating which issues political parties emphasise in their manifestos.
Table 6 reports all issue categories utilised to measure the three issues
included in the following analysis. The first column reports the name of
the issue, the second column the MARPOR codes used to describe the issue, the third reports the three largest issue codes employed to categorise
protests into the same issue, and the final column reports the number of
protests found belonging to each issue.72 I grouped the MARPOR codes
to perfectly match the issue categories of the EPCD, which results in three
broader dimensions being used in the analysis. The environmental dimension is covered by “environmental protection” and “anti-growth economy”
on the party-level. Most protests belonging to the environmental issue are
directed against nuclear energy, ecology in general and animal rights. The
general goal of “peace” and positive and negative references to the “military” account for the military dimension. Recall that this chapter is interested in issue emphasis and, thus, collapses positional categories such
as supporting or rejecting the use of the military into a single dimension.
71 The timeframe depends on countries and election years.
72 Retrieving the issue categories for each protest event is cumbersome. For some of the
categories roughly 50 different codes have been used. I also checked for misspellings
such as “worker’s rights” instead of “workers’ rights”. Listing all categories used goes
beyond the limits of the table and the chapter. An all-encompassing list is available in the
replication material.
130
Source: Author’s own.
per501(environmental protection)
per416(Anti-Growth Economy: Positive)
per106(peace)
per105(military: negative)
per104(military: positive)
per701(labour groups: positive)
per702(labour groups: negative)
per410(economic growth: positive)
environment
military
employment
MARPOR issue categories
Definition of issue
Table 6: MARPOR and EPCD codes used
anti-nuclear (5%)
ecology (2.8%)
animal rights (0.8%)
nuclear weapons (5%)
missile deployment (1.4%)
nuclear arms (0.8%)
unemployment (5.5%)
wages (3.7%)
labor policy (0.9%)
EPCD largest issues per category
49713 (56%)
33777 (38%)
6667(7.5%)
8123(9.2%)
N(% of total)
7.4 data & methods
131
7.4 data & methods
Even though there are also protests that reject the use of force and financial
support to the military in general, the largest share of protests about the
military are against nuclear weapons and missile deployments. Finally, I
capture the issue of employment (labour rights) on the party level with the
manifesto codes for labour groups and economic growth. The lion’s share
of protests coded in the EPCD data speaks to the dimension of labour rights
and unemployment. In total, these three dimensions account for 56 % of all
protests recorded in the EPCD data and, thus, cover a large part of the
actual dataset.
Since this chapter tries to disentangle the salience parties attach to materialist vs. postmaterialist issues, I subtract the postmaterialist issue emphasis
from the materialist issue emphasis:
post- vs. materialism = (environment + military) − employment
(4)
I also apply the same logic to the information on protest stemming from
the EPCD to ensure the comparability of the scaling of both variables. Data
on the protrusion of protest stem from the EPCD (Francisco 1995, 1996,
2009).73 EPCD data are manually coded protest-event data.74 Coded are
all reported protests and repressive events in a range of newspapers, irrespective of their size and shape. The Collection is based on 500 newspapers, trade publications, and wire services electronically available using
Lexis/Nexis (Nam 2006: 283; Reising 1999: 325). The data also capture local news outlets as well as major national outlets. The data report for each
day for each event. They have also been compared to other protest event
data-sets and tend to outperform those in terms of validity and reliability
(Nam 2006; Braun and Koopmans 2010: 114). Furthermore, the EPCD is
also the longest available time series to date on protest events covering a
larger number of countries.
73 The data is available for download on Ronald Francisco’s webpage: http://web.ku.edu/
~ronfrand/data/.
74 The manual coding was assisted by machine-coding: “Most coders use the KEDS software
to code the data. We then transfer the machine-assisted codes to an Excel spreadsheet,
which is the primary coding document. We also code using Word and Excel simultaneously, using Excel’s artificial intelligence to recognise each word in the column and to
place suggestions while writing. Both of these methods increased the efficiency of coding”
(EPCD Research Group N.d.: 1-2).
132
7.4 data & methods
As discussed in the methods chapter, I decided to capture the protrusion
of protest with an index based on eight dummy variables. Here I use the
same eight characteristics.75 I then lagged the protest-index by one year to
ensure that the protest events captured by the index occurred before parties
drafted their manifestos.76 Figure 15 pictures the distribution of the protestindex across time for the sample of countries included in the dataset. Before 1992 protest on materialist issues mostly outweighed postmaterialist
protests. Interestingly the variation is especially high in 1986, the year of
the Chernobyl accident. Looking into the data reveals that a range of countries experienced a major increase of protrusive protest on environmental
issues in 1986, while in other countries protesters stuck to pushing for the
rights of the working classes.
Like most protest event data, the EPCD relies on codes originating from
the media. Even though media data still represents the gold-standard for
protest event coding, cross validation of different data-sets varies considerably depending on which media source was employed (Nam 2006). Information on participants shows particularly considerable variation. Thus,
using binary variables to measure the presence of certain characteristics
mitigates potential biases, e.g. protest duration usually increases with its
endurance, but rarely contradicts the fact that a protest lasted longer than
a single day. This also applies to the EPCD data, which often reports the
number of participants for an event as ending with a ‘1’ e.g. 1001. According to the codebook this means that the respective protest event is reported
in the media to have had “1000 or more protesters”. Thus, the EPCD’s coding
rules directly translate into my logic of coding certain protest characteristics as binary.
75 Due to the codebook of the EPCD dataset the duration of protest cannot be measured the
same way as in the previous chapters. As I have explained the EPCD data reports each
protest on a daily basis. Thus, the unit of observation is not an event, but a day. Instead,
here I used an approximation to measure the duration of protest. In case a protest on a
specific day is on the same topic as the last protest reported in the EPCD data, I coded the
duration as ‘1’, and ‘0’ otherwise.
76 Rearranging the index into legislative periods reveals similar results, but restricts the sample size to fewer observations since only elections after 1984 can be included in the analysis.
Furthermore, such a set-up would assume that the behaviour of parties at electiont is affected by protests that happened four years prior. This assumption seems very strong and
it sounds more convincing that parties are affected by protests that occur a year before an
election. Thus, I decided to use a yearly protest-index.
133
7.4 data & methods
Figure 15: Protest-index & postmaterialism across time
10
UK
5
std. protestindex (post−mat)
UK
Netherlands
UK
UK
UK
UK
Switzerland
Germany
Germany
Germany
UK
Germany
0
Finland
−5
Ireland
France
Ireland
France
−10
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
(a) Distribution of lagged protest-index, 19811996
40
postmaterialism vs. materialism
20
0
−20
−40
−60
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
(b) Distribution of lagged postmaterialist index (imputed values), 1981-1996
Source: Author’s own.
Note: Figure (a) shows the distribution of characteristics for each protest coded within
the ResponsiveGov project. Figure (b) reports average protest-index per country and
month. Figure b) omits USA (=1.0) and Canada (=2.2) to ensure readability. Protest-index
is split into quantiles.
The public’s position is measured using Inglehart’s standard two item
postmaterialism index (Inglehart 1971). I draw on three public opinion surveys to gather the information on each country in my sample (World Value
Survey, Eurobarometer, European Social Survey).77 I then subtracted the
percentage of people supporting materialist values from the number of
people supporting postmaterialist values. However, the data is not avail77 An overview about the sources employed and the time periods covered can be found in
the appendix on page 184 in table 19.
134
7.4 data & methods
able for a range of countries across the whole time span included in the
study. I used a multiple imputation model to draw ten observations for
each missing value in the dataset (a very detailed description of the pattern
of missings, the theory behind multiple imputation and my imputational
model can be found in the appendix 7 on page 123). Research shows that
using multiple imputation models is the more reliable approach for achieving unbiased estimates than listwise deletion in the case of missing data
(King et al. 2001; Gelman, King, and Liu 1998; Honaker and King 2010). I
again lagged the postmaterialist index by one year for the same reasons
described above. Figure 15 pictures the distribution of the postmaterialist
index of the sample across time. Comparable to previous research, the ten
imputed datasets reveal an increase of postmaterialist values, with a sharp
decline in the early 1990s after the fall of the USSR (Fuchs and Rohrschneider 1998).
Several controls are included in the following analysis. First, the reaction
of incumbent parties to protesters might differ in comparison to the reaction of opposition parties. Parties in government face different trajectories
and could be punished for not delivering on their electoral agenda. Thus,
I include a dummy variable which is ‘1’ if a party is in government at the
time of the election. Second, to test the assumption that there is no significant effect of party ideology on parties’ reactions to protest, I include the
lagged logged rile scale as proposed in Lowe et al. (2011). Unfortunately
expert surveys on party positions only cover a fraction of the whole sample and, thus, are not a valid alternative for this chapter. Third, I control for
party size measured with vote share, again lagged by one legislative period.
Smaller parties might interpret the postmaterialist revolution as a window
of opportunity. Therefore, smaller parties might have already adapted their
agenda to focus more strongly on postmaterialist values, while larger parties stuck to a catch-all approach by focusing on the more salient materialist
issues. Fourth, parties might adapt their issue attention to real world events.
For instance, parties might talk more about labour rights if the rate of unemployment is higher. Equally, they might increase the salience they attach
to environmental issues if focusing events such as the Chernobyl accident
or the first Iraq war occur. I included a binary variable which is ‘1’ in the
135
7.4 data & methods
case such a focusing event occurred before an election (Birkland 1998).78
Lastly, the unemployment rate is included in all analyses.
Like in the previous empirical chapters, I standardised all variables either by logging (party emphasis, protest-index) or by following Gelman
(2008) and subtracting by two standard deviations (vote share, percentage
of postmaterialists).
7.4.1
Modelling strategy
Due to the hierarchical structure of the data I decided to use a multilevel linear regression model. In contrast to the previous empirical chapters (5 & 6),
the data are not monthly time-series, but rather legislative terms. Furthermore, the data on postmaterialism used in the following analysis shows
a high number of missing values. Since panel-corrected standard errors
with an ar(1) correction are not computable with imputed data, I relied on
a multilevel linear regression model. This model also comes with the advantage of being able to calculate random intercepts on the country and
party level (Steenbergen and Jones 2002). Thus, random intercepts portray
the different baseline propensities for each party and country moderating
the traditional intercept we know from a standard linear regression context. I model the salience parties attach to postmaterialist values using the
following equation:
salience pt = b0 + b1 protest − indexct + b2 postmaterialismct
+(b3 protest − indexct ∗ postmaterialismct ) + b4 vote share pt
+b5 incumbency pt + b6 ideology pt + b7 f ocusing eventct
(5)
+b8 unemploymentct + b9 ζ cpt + γc + γ p + ecpt
where γc and γ p are the random intercepts, ζ cpt the vector of controls
outlined above. Equivalent models have recently been used elsewhere to
deal with parties’ time-series-cross-sectional data (Tavits and Potter 2015;
Greene 2015). The hypotheses are tested with interactions of the independent variable of interest with the protest-index, represented here by the
78 The following wars and focusing events were coded: Tchernobyl (1986); Piper Alpha Accident (1988); Exxon Valdez accident (1989); Bosnia war (1993-1996); Gulf war or Operation
Desert Shield (1991-1992).
136
7.5 results
interaction between the protest-index and postmaterialism. In contrast to
the previous empirical chapters, the data used in this chapter are strongly
unbalanced since parties do not participate in every election and may even
drop out of existence. Thus, using a maximum likelihood approach is more
suitable for producing unbiased estimates than panel corrected standard errors. Furthermore, test statistics reveal that the data are not heteroskedastic,
meaning correcting for this issue is unnecessary here.79 The model set-up
chosen does not include a lagged dependent variable, since the lagged dependent variable would be correlated with the random intercept on the
party level. Therefore, the random intercept already takes the autocorrelation on the party level into account.80 Finally, due to imputing the missing
values for the postmaterialism variable, all results are “pooled” across the
ten datasets (King et al. 2001).
7.5
results
Table 7 reports the regression results for testing the hypotheses derived in
the theoretical section. Models one to four are based on the ten multiple
imputation datasets, while model five uses the sample without missings
only.
The models support hypothesis one. The higher the protrusion of protest,
the more likely parties are to emphasise postmaterialist values instead of
materialist ones. This effect is statistically significant on the five per cent
level in all models. Note also that this effect does not rest on the inclusion
of the postmaterialist index as model one does not include the index. Once
the postmaterialist index is included in model 2 the results remain stable to
different model specifications. The effect is also similar in size and significance in the non-imputed dataset. Given that all variables are standardised,
it becomes apparent that the effect of the protest-index is significant but
comparably small in size. Nevertheless, increasing protrusion of protest
leads parties to emphasise postmaterialist values more and de-emphasise
the interests of the working class.
79 A Breusch-Pagan test is not statistically significant. Thus, the H0 of constant variance
cannot be rejected.
80 A Wooldridge type test is statistically significant on the five per cent level. Therefore,
the H0 of no serial autocorrelation needs to be rejected. Furthermore, a fisher test is also
statistically significant, rejecting the hypothesis of non-stationarity.
137
1.141
(0.138)
0.360
(0.119)
0.610
(0.0910)
1.061
(0.0413)
0.0468
500
10
1.092
(0.158)
0.492
(0.133)
0.620
(0.0917)
1.071
(0.0413)
0.000781
500
10
0.0659∗
(0.0256)
-0.437∗∗
(0.164)
0.241
(0.141)
-0.327∗∗∗
(0.0682)
0.329∗
(0.147)
-0.0533
(0.148)
-0.424∗
(0.184)
0.597∗∗∗
(0.158)
0.0540∗
(0.0258)
-0.460∗∗
(0.167)
0.272
(0.142)
-0.320∗∗∗
(0.0693)
0.313∗
(0.144)
0.163
(0.137)
-0.527∗∗
(0.189)
500
10
1.087
(0.159)
0.496
(0.134)
0.616
(0.0923)
1.072
(0.0414)
0.000732
protest X
ideology
0.0628∗
(0.0280)
-0.437∗∗
(0.163)
0.242
(0.141)
-0.333∗∗∗
(0.0729)
0.330∗
(0.147)
-0.0495
(0.149)
-0.426∗
(0.185)
0.597∗∗∗
(0.158)
-0.00836
(0.0272)
Standard errors in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
All models are multi-level models with random intercepts for party and country;
constants omitted from table.
Model (5) restricted to dataset with missings.
RVI
log. lik.
χ2
N
Imputations
σparty
σcountry
lnα
constant
protest-index X postmaterialism
protest-index X ideology
postmaterialism index
unemployment
war
accident
ideology
government
vote share
protest-index
postmaterialism
baseline
Table 7: Postmaterialism vs. materialism, regression results
500
10
0.00821
(0.0537)
1.136
(0.141)
0.358
(0.120)
0.611
(0.0911)
1.061
(0.0413)
0.0508
protest X
postmaterialism
0.0651∗
(0.0259)
-0.438∗∗
(0.164)
0.242
(0.141)
-0.328∗∗∗
(0.0690)
0.325∗
(0.149)
-0.0494
(0.150)
-0.425∗
(0.185)
0.606∗∗∗
(0.170)
-548.4
53.01
356
0
1.291
(0.166)
0.468
(0.131)
0.437
(0.105)
1.025
(0.0474)
0.984∗∗∗
(0.175)
excluding
missings
0.0562∗
(0.0284)
-0.384∗
(0.165)
0.183
(0.153)
-0.242∗∗∗
(0.0733)
0.127
(0.158)
-0.213
(0.147)
7.5 results
138
7.5 results
In contrast to protest, the postmaterialist values of the public at large
have a very strong effect on the extent to which parties emphasise postmaterialist values. Model 2 reports a strong and highly significant effect of the
imputed postmaterialist index on the salience parties attach to postmaterialism. This effect also remains robust across all model specifications. Interestingly, the size of the effect substantially increases in the non-imputed
dataset. Thus, parties tend to emphasise environmental concerns and issues
related to the military more if the public at large supports postmaterialist
values. However, the amount of support for postmaterialist values by the
public at large provides no contingent effect on the influence of protest.
The interaction effect between the postmaterialist index and the protestindex is not statistically significant in model four. According to Brambor,
Clark, and Golder (2005: 70-71), establishing conclusions based only on the
significance of the interaction term can be misleading. Here we are not interested in whether the interaction term itself and its constitutive terms are
statistically significant, but rather in the marginal effect of the postmaterialism index conditional on the protest-index. As Brambor, Clark, and Golder
(2005: 74) notes:
“The analyst cannot even infer whether x (independent variable)
has a meaningful conditional effect on y (dependent variable)
from the magnitude and significance of the coefficient on the
interaction term either. As we showed earlier, it is perfectly possible for the marginal effect of x on y to be significant for substantively relevant values of the modifying variable z (moderator) even if the coefficient on the interaction term is insignificant. Note what this means. It means that one cannot determine
whether a model should include an interaction term simply by
looking at the significance of the coefficient on the interaction
term. Numerous articles ignore this point and drop interaction
terms.”
Thus, they suggest that we should not interpret the significance of constitutive and interaction terms separately, but rather the joint effect of the
∂issue emphasis
standard error of ∂postmaterialism = β 1 + β 3 protest-index (Brambor, Clark,
and Golder 2005: 70; Braumoeller 2004). Figure 16b reports exactly this
marginal effect of the interaction. The figure shows some support for hy-
139
7.5 results
Figure 16: Marginal effects of interaction terms with protest-index
prediction of postmaterialist emphasis (fixed portion)
0
−.2
−.4
−.6
−.8
−6
−4
−2
0
protestindex
2
4
6
(a) Marginal effect interaction ideology X protest-index
prediction of postmaterialist emphasis (fixed portion)
1.5
1
.5
0
−.5
−6
−4
−2
0
protestindex
2
4
6
(b) Marginal effect interaction postmaterialism X protest-index
Source: Author’s own.
Note: Figures based on table 7, models 3 and 4 respectively. Dotted lines report 95 %
confidence interval. Marginal effects are built on the fixed parts of the multilevel models
only. Fixed part of the models only.
pothesis three. There appears to be a significant contingent effect on the
influence of protest contingent on the percentage of postmaterialists in a society. This supports the results found in previous chapters. Yet, the slope of
140
7.5 results
this contingent effect is very flat, meaning that this contingent effect might
be significant, but its effect is almost negligible (Rainey 2014). Hence, even
though the figure reports a statistically significant result for most values of
the protest-index, it is only of limited practical relevance. The effect of the
protest-index on parties’ emphases on postmaterialist issues is not contingent on a one-unit increase of the postmaterialism index (indicated by the
flat slope of the line of the marginal effect in figure 16b and the fact that
there is a positive and significant effect of the baseline term of the protest
index in model 1).
Party ideology is a strong predictor of how much parties emphasise postmaterialist values. The logged rile scale (ideology) is significant in all models and the prefix of the coefficient also supports the assumption that the
more right wing a party’s ideology is, the less likely it is to support postmaterialist values. I assumed that party ideology does not change how parties
react to increased and radicalised protest. This assumption stands in stark
contrast to the assumptions and results presented in previous chapters. Yet,
since this chapter is interested in issue emphasis and not in pro- or antipositions as the chapters before are, a contingent effect of party ideology
became less apparent. Looking into model 3 supports this claim, since the
interaction effect is not significant on conventional statistical levels. Having
said that, the same logic as outlined for the interaction with the postmaterialism index applies here as well. Figure 16a reports the marginal effect
of parties’ ideology on the protest-index. However, the same conclusions
apply for the interaction between parties’ ideology and the protest-index
as outlined above. Even though there is a significant effect for a large share
of parties’ ideology on the protest-index, the slope is fairly flat. Nevertheless, it partly rejects the hypothesis that there is no contingent effect of party
ideology on how parties react to protest. The more right wing a parties’ ideology is, the more likely it is to emphasise materialist values. Thus, there
is a contingent effect of party ideology, but again this effect is negligible.
Thus, the results do not lend support to hypothesis two.
Most control variables behave as expected. In the wake of large environmental accidents, parties emphasise postmaterialist values – specifically
environmental issues. This effect is significant on the five per cent level in
most models. In contrast, being at war does not have a statistically significant effect on how much parties emphasise postmaterialist issues. In line
141
7.5 results
with other research, I find that unemployment forces parties to emphasise
materialist issues. Ceteris paribus parties follow the economic conditions in
their country and adapt their issue appeals accordingly. Finally, the larger
a party is in vote size, the less likely it is to emphasise postmaterialist issues. Thus, large, catch-all parties still tend to focus on materialist issues
to attract the median voter and the largest possible share of voters (Downs
1957; Kirchheimer 1990). Finally, government parties are not more likely to
emphasise materialist values.
I also ran a range of robustness tests. First, I limited the sample to parties
that gained more than five per cent of the vote and five seats. The major interpreations of the results do not change. Second, I conducted interactions
of the protest-index with parties’ incumbency and their vote share. Neither
of these interaction terms add to the story outlined above. Third, restricting the sample to those countries with a sufficient amount of non-missings
in the postmaterialism index does not change our conclusions either, as
column five in table 7 already suggests. Fourth, running a regression with
robust standard errors for elections and a lagged dependent variable also
supports the results shown above.
7.5.1
The causal relationship between protest & party manifestos
The overarching theoretical argument of this thesis assumed that parties
react to public claims. Thus, I argued that the causal direction runs from
the people to the political parties and not the other way around. However,
the causal relationship between public opinion, protest and political parties is extensively debated in the literature (Hakhverdian 2012; Hobolt and
Klemmensen 2005; Gabel and Scheve 2007; Page and Shapiro 1983). Yet,
besides Gabel and Scheve (2007), the named studies tend to agree that the
causal link runs from public opinion to parties and more specifically governments.81
Apart from such general claims on the causal link between public opinion and political elites, postmaterialism has faced extensive criticisms suggesting that the concept is at least highly correlated with, if not endogenous
81 Note, however, that Hakhverdian’s (2012: 1400-1402) results suggest that the causal effect
conditionally depends on the popularity of governments, with popular governments being
more likely to lead public opinion.
142
7.5 results
to, others factors, specifically to changes in unemployment rates, inflation
and income (Ward et al. 2015: 1237-1238; Clarke and Dutt 1991; Duch and
Rusk 1993).82
This chapter proposes two answers to these claims. First, it gives a purely
theoretical answer. As this thesis is not interested in policies or actual
policy-making, but rather rhetorical reactions by political parties, a causal
link from party rhetoric to increased protests becomes less theoretically
convincing. Certainly, protesters might react to rhetorical activities by politicians, but a reaction to politicians by the public becomes more likely the
further down the policy cycle politicians’ activities are. Drafting legislation not only raises more attention by the media, but also by the public as
severe and tangible consequences follow from legislation but not by necessarily from party talk. Recall that most rhetorical activities by politicians
scrutinised in this chapter are responses to the media and as such are
already a reaction by political elites and not a self-initiated action. Thus,
theoretically it sounds more plausible that parties’ rhetoric is largely a reaction to external influences. This argument seems to be especially true for
party manifestos, which are drafted during campaign periods and largely
ignored by the public (Bara 2005). In summary, it seems less theoretically
plausible that the public reacts to statements made within manifestos, than
that parties adapt their positions to public claims.
Second, an empirical answer is given. I run instrumental variable models
to estimate the causal relationship between the protrusion of protest and
party manifesto positions. Instrumental variables are based on the idea
that the researcher finds a variable (or more than one variable) that is sufficiently correlated with the independent variable of interest, but not correlated (or only weakly correlated) with the dependent variable. Thus, the effect of the instrument on the dependent variable should only be indirect (exclusion restriction, see: Menaldo 2012: 716). Whether or not this assumption
is met can be tested in the first stage of the instrumental variable models.
Furthermore, instrumental variables should not be correlated with the error term of the second stage of the instrumental variable model. Notice that
this condition cannot be tested. Thereby, instrumental variables expurgate
the endogenous variable, here the protest-index, of variation which does
82 Furthermore, some authors cast doubts on the validity of the measurement of postmaterialism (Davis and Davenport 1999).
143
7.5 results
not originate from the exogenous variables (Gabel and Scheve 2007: 1016).
Below, I use a two-stage least squares instrumental variable approach with
random effects. The first stage thereby tests the exclusion restriction, while
the second stage reports the final effect of the instrument on the dependent
variable.
I used two instruments for protest here. The first instrument is based
on the idea that certain institutional settings of political systems might foster protest, but not influence the positions parties take in their manifestos.
Nam (2007) argues and shows that the power of the legislature is an important institutional factor to channel protest. His results suggest that in
systems with weak legislatures, protests are more likely to occur since the
parliament is not able to channel protesters’ requests through the government. Thus, protest is more likely to occur in political systems lacking a
powerful legislature that can force the government to react to protest. To
measure the power of a legislature Nam (2007: 104-106) retrieves a rich
amount of data based on the constitutional setting of the countries included
in his sample.83 He uses an ordinal scale, which depends on the characteristics of the legislature’s power (initiative or restrictive). I use his coding
for legislative power as an instrument in my models and refer readers to
extract any further information on his coding from the original text. While
I assume a comparably strong effect of the legislative power variable on
the size of the protest-index, I do not find a correlation with the extent to
which parties emphasise postmaterialist values. Also, theoretically, there is
no reason to assume that parties working in a legislature with more power
are less likely to emphasise postmaterialist values. The second instrument
is the amount of repression a state uses when reacting to the challenges
put forward by protestors. I used the political terror scale to measure the
amount of repression used in each country included in the study here.
Unfortunately the data do not cover the whole sample of time periods included in the study leading to the omission of a range of observations. On
first sight one should assume that repression does not vary largely across
the sample of countries used in this study. Indeed of the five point scale
used in the political terror scale only three levels have been used to classify
the countries included here (29.9 % “countries under a secure rule of law”;
83 His sample does not include Finland and the UK. Finland is not included in his study,
while the UK lacks the constitution to be used to code the de jure power of the legislature.
144
7.5 results
29.3 % “limited amount of imprisonment for nonviolent political activity”;
1.7 % “extensive political imprisonment”). This shows that even though the
two most radical methods of repression are not found in Western democracies, a sufficient amount of variation is found in the data. I assume that
increased repression either happens because the state increases its repressive reactions to the protesters on the streets or because it leads people in
democracies to increase their resistance against the state (for a fully fledged
theoretical argument, see for instance: Bischof and Fink 2015). Again, there
is no reason to assume that the amount of repression in a country affects
how parties emphasise postmaterialist values.
Table 8 reports the results of a two-stage least squares instrumental variable approach. I run three models using random effects on the party-level.
Model 1 is fully reported. However, since the results of the first stage regression are representative of the first stage of the remaining two models, I omit
it from the remaining models. Model two also includes robust standard errors per party. The first two models use the first imputed dataset, while
model three reports the results for the datasets without missings only.84
Looking at the first stage of the first model we can see that repression
appears to be an excellent instrument for the protest-index. The level of
repression is not only highly correlated with the protest-index, but also
holds the second strongest coefficient in the model. Legislative power is
also a significant predictor of the amount of protest, but only significant
on the five per cent level. Thus, both instruments appear to be suitable
instruments for protest.
The results of all three models suggest that the causal link runs from
the protesters to parties’ emphases in manifestos and not the other way
around. Thus, I find support here for this thesis’ overarching argument
that protest causally relates to party positions. I also find support for the
direction suggested in this thesis, namely moving from the people taking
to the streets to the parties, and not the other way around. This conclusion
is also supported by the non-imputed dataset reported in model three in
table 8.
84 It is not possible to run instrumental variables on several imputed datasets. Thus, the
standard errors of the first two models are biased. However, the third model underpins
the major conclusions drawn from the first two models.
145
7.5 results
Table 8: Testing causality between protest & party agendas, instrumental variable models
Model 1
DV=
protest-index
repression
legislature
vote share
government
ideology
accident
war
unemployment
postmaterialism
constant
R2
N
Imputations
1st
protest-index
1.334***
(0.308)
0.287*
(0.121)
0.350
(0.376)
-0.494
(0.349)
-0.024
(0.193)
0.485
(0.415)
-0.276
(0.307)
-1.765***
(0.390)
1.120**
(0.424)
-4.845
(0.861)
235
1
Model 2
Model 3
2nd
issue emphasis
0.214*
(0.108)
2nd
issue emphasis
0.214*
(0.091)
2nd
issue emphasis
0.296***
(0.078)
-0.390
(0.202)
0.390*
(0.195)
-0.267**
(0.102)
0.198
(0.222)
-0.119
(0.165)
-0.0604
(0.251)
0.549**
(0.208)
1.193***
(0.187)
0.21
235
1
-0.390*
(0.173)
0.390
(0.200)
-0.267*
(0.120)
0.198
(0.183)
-0.119
(0.165)
-0.0604
(0.214)
0.549**
(0.203)
1.193***
(0.140)
0.21
235
1
-0.302
(0.179)
0.397
(0.225)
-0.266**
(0.102)
0.0275
(0.262)
0.252
(0.203)
0.217
(0.227)
0.599**
(0.219)
1.114***
(0.183)
0.24
187
0
Standard errors in parentheses.
All models use random effects for parties;
Model 2 same as model 1 but with robust standard errors for parties;
Model 2 same as model 1 but with missings instead of imputed values.
146
7.6 conclusion
7.6
conclusion
This final empirical chapter had three major purposes. First, it tested whether
the effect of the prostestindex also appears in other data sources and issues.
Second, it tried to understand how one of the major shifts of public priorities, the silent revolution, affected party emphasis on postmaterialist values.
Third, it developed a first test for the causal relation between political parties and protest.
The chapter relied on the same core theoretical arguments put forward in
the theoretical chapter of the thesis. Again, I assumed that parties become
more likely to respond to protest the higher the protrusion of protest is.
Yet, this chapter deviated from the previous chapters by not looking at a
single issue, but rather at the struggle for attention between three different
issues. Thus, I understood postmaterialist values to have an influence on
party agendas only if it managed to outscore materialist issues.
Empirically the chapter used three major data sources. I merged the
MARPOR data, EPCD data and several public opinion polls. I ran a multilevel model to test my assumptions on 150 parties competing in 80 elections
in 17 democracies. I found strong evidence that the effect of the protestindex is consistent across the data sources employed. The results of this
chapter showed that parties adapted their agendas to popular protest on
postmaterialist values. Interestingly, this effect is also robust over a range
of modelling specifications and neither interacts with the percentage of
the public supporting postmaterialist values nor with party ideology. Thus,
this chapter supports the previous empirical chapter’s major finding that
protest matters for political parties. Furthermore, protest not only influences parties’ rhetoric, but also their more binding policy commitments
put forward in their manifestos. Parties tend to adapt their positions to
major popular protest on materialist and postmaterialist values. Yet, the coefficient of this effect is smaller than the results of the previous chapters. In
the case of postmaterialism the multiply imputed percentage of the public
supporting postmaterialist values had a far stronger effect on party agendas than political protest. This effect is also consistent once the missing
values for the postmaterialism index are excluded from the analysis.
Finally, the instrumental variables used in the empirical section of the
chapter supported the thesis’ major claim that parties adapt their issue em-
147
7.6 conclusion
phasis to protest and not the other way around. While the previous chapter
assumed that this is the causal direction of the link between party agendas
and protest, this chapter added flesh to the theoretical bones outlined in
chapter 2 of the thesis.
Future research should build on the results of this chapter and look more
carefully at the relationship of protest mobilisation on different issues and
the ways in which these different issues relate to each other. While there
is a rich body of literature on the relationship between issues salience in
the media and party emphases (Soroka 2012, 2002b,a; Walgrave 2008), we
know very little about how protests on different issues win the battle for
attention against each other.
148
Part IV
Parties and Protest: What have we
learned?
149
8
CONCLUSION
his thesis addresses two main research interests: Does protest influences the agendas of political parties? And if so, how does political protest
affect party rhetoric? This chapter outlines the conclusions we can
draw from this thesis. What have we learnt from the previous chapters and
from extensively looking into several data sources?
First, we learnt that protest has implications for party competition in
western democracies: protest does matter. The thesis showed that demonstrators can influence the agenda-setting stage of political systems. However,
protest needs to be protrusive to influence the political agenda. Consequently, protests can only achieve influence when they are large enough,
long enough, have the right organisation and level of disruptiveness. In
contrast to research linking political parties with public opinion data, this
thesis shows that the influence of protest is a mechanism in flux. A lot of
protest events do not accomplish any of the aforementioned requirements.
In this sense protest needs to solve common collective action problems before actually exerting an impact on politics.
The study’s second major finding is that the influence of protest is contingent on other factors. While protest with increased protrusion influences
the political agenda, it does so in different ways depending on the ideologies of political parties, the position of public opinion on the issue and
partly on the stage of the electoral cycle it occurs in.
T
8.1
how protest matters
Each of the empirical chapters addressed at least one of the following four
key research questions:
1. Does protest affect party rhetoric?
2. On a smaller note, are parties’ rhetorical activities cues for
substantive policy-making?
150
8.1 how protest matters
3. Are there contingent factors that moderate the relationship
between political protest and party rhetoric (e.g. party ideology; public opinion)?
4. Does protest have an effect on more binding party commitments such as issue emphasis in party manifestos?
Chapters 5 and 7 addressed the question of whether protest affects the
agenda of political parties. Both chapters find support for the theoretical
argument that protest influences the political agenda if its protrusion is
significant enough. Protrusion is conceptualised as the achievement of four
characteristics (size, duration, organisation & disruptiveness). Interestingly
both chapters find support for the protest attention hypothesis even though
they used different data sources. While chapter 5 used the ResponsiveGov data and estimated the effect of protest on parties’ rhetorical activities (press releases, press conferences, interviews & speeches) chapter 7
merged MARPOR data with the EPCD dataset. The latter chapter showed
that protest is also an important source of information for parties’ more
binding party manifestos. The results find that protest has a robust impact
on the extent to which parties emphasise postmaterialist values in their
manifestos.
Chapter 4 aimed to answer the second question listed above. Thus it can
also be understood as a validation for the measure of parties’ rhetorical
positions employed throughout most of the thesis. Previous research has
understood parties’ rhetorical activities, such as statements to the press,
speeches and interviews, as sheer symbolism signalling the cognition of
public interests without pursuing more substantial policy steps. As a result few studies seek to embrace party rhetoric as a signal of substantial
policy efforts between elections. The chapter suggests that by understanding rhetoric as a policy cue party scholars can gain important information
about parties’ policy preferences between elections. I show that rhetorical
party positions appear to be strongly structured around a) party manifesto
positions and b) party families’ traditional stance on the usage of nuclear
energy. Yet, even though party rhetoric mirrors general ideological divides
between political parties, there is also considerable variation between parties’ general ideology and how they talk about nuclear energy between
elections. Second, by linking parties’ rhetorical positions to their policy
151
8.1 how protest matters
outputs (e.g. bills, legislative acts and laws), the analysis shows that parties walk like they talk, with their policy efforts being mostly congruent
with their rhetorical positions. In summary, parties’ rhetorical activities appear to be policy cues for the issue of nuclear energy, which is studied
throughout most of this thesis.
I then used parties’ rhetorical positions to understand how political protest
affects rhetoric and which factors moderate this relationship in order to
address the third question listed above. In chapter 5 I hypothesise that
increased protest leads to a polarisation of party systems. While parties
increase their attention to the issue at stake during protest, they respond
differently to protest depending on how their ideology relates to protesters’
demands. Traditionally parties of the left, e.g. Social Democratic, Green and
Communist parties, enjoy strong ties to trade unions and “new social movements” (Hamann, Johnston, and Kelly 2012: 1038-1039; Hamann, Johnston,
and Kelly 2013; Kriesi et al. 1995). Furthermore, since many protests are
organised or supported by unions or new social movements, parties with
a left ideology might perceive such protests as representative of their core
voters’ preferences. In contrast, right wing parties are associated with conservatism and thus a tendency to support the status quo instead of pushing
for reform. Apart from party ideology I suggest that the public’s position
is an important contingent factor for political protest. Thus, if protest is
not supported by the public, parties should ignore or reject the position of
the protesting minority in an effort to secure votes and office. The analysis
finds support for both of these theoretical claims. Traditionally responsive
to protesters’ claims, left wing parties understand increased protest as a
window of opportunity to shape the policy debate that is beneficial for
their goals, while right wing parties perceive protest as a threat to their
ideological position on the usage of nuclear energy.
Chapter 6 looked at the most obvious and often theorised mechanism
for keeping representatives responsible: elections (Ferejohn 1986; Powell
2000). On election day the electorate makes a decision about whether to
keep its previous representatives in office or to reshuffle its parliament and
incumbent cabinet. Politicians and their organisations are more likely to
respond to public grievances the closer the next election. While the mechanism of an electoral cycle has been studied for American Presidents and
the Congress in several studies (Kuklinski 1978; Tufte 1978; Cohen 1999;
152
8.2 contribution to the discipline
Canes-Wrone and Shotts 2004; Rottinghaus 2008), only a few studies exist
scrutinising the moderating effect of the electoral cycle in comparative perspective. This chapter attempted to address this specific gap. I argued that
the electoral cycle has a moderating effect on how parties react to divergent
public claims. Given that parties are assumed to maximise their electoral
payoffs, they are more likely to respond to the public at large when elections are near, and have more leverage in their responses between elections.
This leverage often leads parties to be more open to minority interests inbetween elections. While the data support the argument that parties are
more likely to be responsive to the public at large the closer the next election is, the hypothesis of a decreasing effect of parties’ responsiveness to
protest finds no support. The effect of the electoral cycle on parties’ rhetorical reactions to protesters’ claims only becomes visible if the salience of the
issue for the general public is included. However, the results of this chapter
remained mixed and did not find a clear effect of the electoral cycle.
In the last empirical chapter (chapter 7) I further underpinned my results
by using different datasets to measure the protrusion of protest and parties’
reactions to it. The chapter looked into 150 parties competing in 80 elections
across 15 years. While the chapter supports the thesis’ general claim that
parties adapt their positions to popular protest if protest is encompassing
enough, neither public opinion nor party ideology had a substantial contingent effect on parties’ issue emphases in election manifestos. The data used
in the chapter also allowed for the testing of an implicit question underlying the whole thesis, which is the question of causality. Does protest influence party positions? Or, do the agendas of political parties lead to more
protest? Using an instrumental variable model that measures the amount of
repression and the legislative power of the countries included in the study
as an instrument for protest, I show that parties are causally affected by
protest and not the other way around.
8.2
contribution to the discipline
The findings of this thesis have implications for theories linking politics
and policy with public opinion, as protest might be an important resource
for parties and their position taking. Previous research in political science
153
8.3 suggestions for further research
has largely ignored the potential influence of protest on political parties.
Thus, omitted variable bias might drive at least some of the theoretical
ideas and findings in the discipline. Protest might have supported some of
the larger public opinion shifts in the last decades and empirical research
rarely engaged with the influence of protest on party competition.
This is important since political protest is a conventional means of participation in Western democracies. Furthermore, taking to the streets is an active form of participation. Costs are involved if people organise themselves
in marches, vigils, blockades and other forms of protest. People actively
seek to communicate their grievances to the public, media and political
actors. In contrast, people can answer questions in surveys via the phone
without actively seeking to influence anything. Public opinion polls might
capture what Stimson (1991) calls the policy mood, but we might want
to go beyond a general mood and try to disentangle the concept of a single entity of public opinion into multiple dimensions interacting with each
other.
Theoretically the thesis linked two major research strands in contemporary social science, namely literature on political parties with the literature
on social movements and protest. It thereby brought together a rich body of
literature and sought to enhance theoretical reasoning by combining both
strands.
8.3
suggestions for further research
Chapter 2 reviewed the existing literature in detail and implicitly stated
that political science and sociology have, to a certain extent, a shared research agenda, without sharing each others’ insights. Political science is
traditionally interested in how political actors react to external stimuli
(Miller and Stokes 1963; Erikson 1978; Stimson, Mackuen, and Erikson 1995;
Soroka and Wlezien 2010; Lax and Phillips 2012), while sociology is often interested in how these external stimuli are shaped and more recently
also in how these stimuli seek to influence political systems (Tarrow 1994;
Giugni, McAdam, and Tilly 1999; Agnone 2007; Soule and Olzak 2004).
This thesis was a first attempt to merge these two traditions to understand the wider set of channels through which the public may actively seek
154
8.3 suggestions for further research
to influence politics and how these channels might interact with each other.
It is not enough to only question how public opinion as a single entity
influences political parties. During the last decades people have increasingly used other channels to articulate their grievances. Using the famous
metaphor of the thermostat used by Soroka and Wlezien (2010): it is not
enough to ask when the public states that the temperature is too hot or too
cold in a survey. We need to know how different forms of screaming too
hot and too cold relate to each other and subsequently which channels are
more successful in communicating the temperature to political parties and
governments.
More research needs to be done to fully understand the relationship between political parties and protest. Like most research in social science, this
thesis comes with a range of limitations that should be addressed in future
research.
First, this thesis relies solely on observational data. However, if we truly
care about the causal relationship between political actors and protest we
need to use different methods and datasets to fully understand the link
between protest and political parties. Researchers might use experimental designs to judge how politicians react to different forms of protest.
For instance one might present pictures and videos of different protests
to politicians and record their reactions to them. Quasi-experimental designs might also help to further address the causal relationship between
popular protest and political parties. Also regression discontinuity designs,
instrumental variables as used in the last chapter of the thesis, and granger
causality tests could enlarge our knowledge on the effects of protest on
parties behaviour.
Second, most of the empirical chapters are single issue studies. Even
though the cross country comparison ensures a larger variation of the
salience of the nuclear issue is taken into account, other factors such as
issue characteristics might influence the results (Soroka 2002b). As the last
empirical chapter showed, looking into a larger sample of issues and how
these issues relate to each other might result in slightly different results and
theoretical contributions. Thus, future research should follow this road to
enlarge our knowledge on this particular question.
Third, protest might not only influence the agenda setting of political
parties but also that of other external actors such as the media. On top of
155
8.3 suggestions for further research
that political protest could also impact the landscape of political systems.
We know that some political parties, e.g. the Greens, grew out of popular
movements. However, political science does not know enough about the
circumstances under which such an organisational transition from a movement organisation to a political party takes place and which factors hinder
or help this process. Protest might also struggle to seek influence on policymaking due to counter lobbying efforts. In summary, protest might have
a range of other implications on political systems, which we have not yet
recognised.
156
Part V
Appendix
157
A
APPENDIX
a.1
a.1.1
appendix chapter 3:
Re-coding actors’ positions
All actor positions within the ResponsiveGov project are coded in relation to governments’ initial policy positions. Government positions are
coded on a pre-defined five point scale – ranging from ‘very progressive’
to ‘very conservative’. Based on governments’ manifestos, coalition agreements or policy documents prior to the Fukushima accident, whether they
are favouring nuclear energy or whether they intend to phase-out nuclear
energy at a certain point in time, they are then matched into the five
point scale to assure comparisons between countries. Progressive positions
thereby outline positions against nuclear energy and conservative positions
outline positions in favour of nuclear energy.
Coders are then asked to place all actor positions during an event again
on a five point scale, ranging from ‘(-2) an actor’s position is radically more
progressive’ to ‘(2) an actor’s position is radically more conservative’ than the
government’s initial policy position – with ‘(0) being the same position as the
government’s initial position’. In the case of Fukushima, statements coded
as progressive on the scale thereby are more against nuclear energy than
the initial government position, while conservative statements picture more
support for nuclear energy than the initial government position. Since this
paper is interested in party positions, the original coding scale does not
allow a party based comparison across countries due to varying initial government positions. Events outlining a more progressive position than the
government might still be in favour of nuclear energy depending on the
government’s initial position in the country. Therefore, I recoded the scale
according to the rules outlined in table 9 (see also: Lühiste et al. 2014: 1216). For example in the case of Italy, the Berlusconi government’s position
was to re-enter nuclear energy and construct new nuclear plants – which
158
A.1 appendix chapter 3:
159
Table 9: Detailed rules for recodes of rhetorical positions
Initial government position
-2
(Radically more progressive)
-1
(slightly more progressive)
1
(Slightly more conservative)
2
(Radically more conservative)
statements’ original classification
-2
-1
0
1
2
-2
-1
0
1
2
-2
-1
0
1
2
-2
-1
0
1
2
position after recode
Anti-nuclear
Anti-nuclear
Anti-nuclear
Anti-nuclear
Pro-nuclear
Anti-nuclear
Anti-nuclear
Anti-nuclear
Pro-nuclear
Pro-nuclear
Anti-nuclear
Anti-nuclear
Pro-nuclear
Pro-nuclear
Pro-nuclear
Anti-nuclear
Pro-nuclear
Pro-nuclear
Pro-nuclear
Pro-nuclear
Source: Author’s own, based on Lühiste et al. (2014: 12-16).
was firstly coded as ‘(2) very conservative’. I then recoded any action by a
politician which was classed between ‘-1’ and ‘2’ an activity supporting nuclear energy, while any action classified as ‘-2’ into an activity rejecting the
use of nuclear energy. By this procedure all policy positions are then coded
as either supporting or rejecting nuclear energy.
A.1 appendix chapter 3:
a.1.2
Codebook for party positions:
In the case of coding parties (pre-Fukushima) nuclear energy policy positions, the following categories applied:
-2 = Very anti-nuclear: The party is completely against the use of nuclear
energy. If there are no nuclear power plants but some political or
public stakeholders want the party to build (or allow the building) of
nuclear power station, then code ‘-2’only if the party is fundamentally
against such plans. If the country has nuclear power station(s), code
‘-2’only if the party is committed to close ALL the nuclear power
stations in the near future (within the next 5-10 years).
-1 = anti-nuclear: The party is against the use of nuclear energy but is
not very radical in its views. If there are no nuclear power plants
but some political or public stakeholders want the party to build (or
allow the building) of nuclear power station, then code ‘-1’if the party
is in principle against such plans but may be open for some talks and
negotiations. If the country has nuclear power station(s), code ‘-1’only
if the party indicates that it is committed to close some or all nuclear
power stations (the plans do not have to be very clear and the planned
closure can be in more than 10-years time).
0 = No position / neutral / vague: Try to avoid using this category. Use it
only if the document used for coding the party position EXPLICITLY
mentions that the party does NOT have a clear position on nuclear energy. See the notes below for coding partys policy position when NO
references are made in ANY document to partys position on nuclear
energy.
1 = pro-nuclear: The party is in favour of the use of nuclear energy but is
not very radical in its views. If there are no nuclear power plants but
some political or public stakeholders want the party to build (or allow
the building) of nuclear power station, then code ‘1’if the party is in
principle in favour of such plans but may be open for some counterarguments from parties and interest groups against such plans. If the
country has nuclear power station(s), code ‘1’only if the party does
160
A.1 appendix chapter 3:
NOT indicate that it wants to close some or all nuclear power stations
BUT also does NOT plan to build any new nuclear power plants.
2 = Very pro-nuclear: The party is strongly in favour of the use of nuclear
energy. If there are no nuclear power plants but some political or
public stakeholders want the party to build (or allow the building)
of nuclear power station, then code ‘2’if the party is the principal
stakeholder advocating for the building of nuclear power plants and
is not open for any discussions to halt these plans. If the country has
nuclear power station(s), code ‘2’only if the party does NOT indicate
that it wants to close ANY of the nuclear power stations within the
next 20 years, sees no alternative to nuclear energy, and plans to build
or is building new nuclear power stations.
161
Écologistes Confédérés pour
Ecolo
Finland
Canada
Centre démocrate humaniste
cdH
Kristillisdemokraatit
Suomen Keskusta
Kansallinen Kokoomus
Perussuomalaiset
KESK
KOK
Ps
New Democratic Party
NDP
KD
Liberal Party
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
5&5
Continued on next page
Vlaamse Liberalen en Democraten
VLD Open
LP
Vlaams Belang
VB
Conservative Party
Socialistische Partij Anders
SP.A
CP
Parti Socialiste
PS
Bloc Québécois
Nieuw-Vlaamse Alliantie
N-VA
BQ
Mouvement Réformateur
MR
l’Organisation de Luttes Originales
Christen-Democratisch en Vlaams
CD&V
Belgium
party name
abbreviation
country
Table 10: Parties included in the Fukushima juncture
x
gov
gov supporter
*
right censored
Inclusion criteria
A.1 appendix chapter 3:
162
Italy
Germany
France
country
Table 10 – Continued
Nouveau centre
Parti communiste française/
NC
PCF
Sozialdemokratische Partei Deutschlands
SDP
x
x
x
x
x
x
x
x
x
x
x
x
x
x
5&5
Continued on next page
Die Linke
Linke
Lega Nord
Die Grünen
Grüne
LN
Freie Demokratische Partei
FDP
Christlich-Soziale Union
Christlich Demokratische Union/
Union pour un mouvement populaire
UMP
CDU/CSU
Parti Socialiste
PS
Front de gauche
Mouvement démocrate
MoDem
Vihreä Liitto
VIHR
Front National
Vasemmistoliitto
VAS
FN
Suomen Ruotsalainen Kansanpuolue
SFP
Europe Écologie-Les Verts
Suomen Sosialidemokraattinen Puolue
SDP
EELV
party name
abbreviation
x
x
gov
gov supporter
**
right censored
Inclusion criteria
A.1 appendix chapter 3:
163
Sweden
Spain
Netherlands
country
Table 10 – Continued
Centerpartiet
Folkpartiet liberalerna
Kristdemokraterna
Miljöpartiet de Gröna
Moderata samlingspartiet
FP
Kd
MP
MSP
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
5&5
Continued on next page
Partido Socialista Obrero Español
PSOE
C
Partido Popular
Volkspartij voor Vrijheid en Democratie
VVD
PP
Socialistische Partij
SP
Izquierda Unida
Partij voor de Vrijheid
PVV
IU
Partij van de Arbeid
PvDA
Convergència i Unió
GroenLinks
GL
CiU
Democraten 66
Unione di Centro
UdC
D’66
Il Popolo della Libertà
PDL
Christen Democratisch Appèl
Partito Democratico
PD
CDA
party name
abbreviation
gov
gov supporter
right censored
Inclusion criteria
A.1 appendix chapter 3:
164
Democratic Party
Republican Party
DEM
REP
Liberal Democrats
LIB
Sozialdemokratische Partei der Schweiz
SPS
Labour Party
Schweizerische Volkspartei
SVP
LAB
Grüne Partei der Schweiz
GPS
Conservative and Unionist Party
Freisinnig-Demokratische Partei der Schweiz
FDP
CON
Christlichdemokratische Volkspartei
Vänsterpartiet
V
CVP
Sverigedemokraterna
SD
Bürgerlich-Demokratische Partei Schweiz
Socialdemokraterna
SAP
BDP
party name
abbreviation
Legend: *=19.1% & 39 seats in 2011; **=system relevance; ***=5.7% & 20 seats in 2011.
Source: Author’s own.
United States
United Kingdom
Switzerland
country
Table 10 – Continued
x
x
x
x
x
x
x
x
x
x
x
x
5&5
x
gov
gov supporter
***
right censored
Inclusion criteria
A.1 appendix chapter 3:
165
A.1 appendix chapter 3:
Table 11: Factors favouring party rhetoric on nuclear energy
seat share
vote share
government party
constant
log likelihood
N
(1)
1=Did party talk about
nuclear energy in a given month
1.011∗∗∗
(0.002)
0.986
(0.010)
1.191
(0.167)
0.322∗∗∗
(0.039)
-681.849
1162
Exponentiated coefficients; Standard errors in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
166
A.1 appendix chapter 3:
a.1.3
MIP/MII questions used in the study
Question wording:
• Eurobarometer Q: What do you think are the two most important issues facing (OUR
COUNTRY) at the moment?
A: The environment
• Focus Canada Q: What is the most important problem facing Canada today?
A: Environment/pollution/climate change/Kyoto
• GALLUP Q: What do you think is the most important problem facing this country
today?
A: Environment/pollution
• Sorgenbarometer Q: Auf diesen Kärtchen sehen Sie einige Themen, über die in der
letzten Zeit viel diskutiert und geschrieben worden ist: Sehen Sie sich bitte alle
Kärtchen an, und legen Sie mir dann von allen Kärtchen jene fünf heraus, die Sie
persönlich als die fünf wichtigsten Probleme der Schweiz ansehen.
A: Umweltschutz
For more information on the sources and timing of fieldwork, please consult the next
page.
167
1. November 2010
1. Mai 2011
1. November 2011
1. Mai 2012
1. November 2012
1. Mai 2013
1. Oktober 2010
1. Dezember 2011
1. November 2012
1. Januar 2010
1. Januar 2011
1. Januar 2012
18. September 2010
Eurobarometer
BE, DK, FN, FR, DE, IT, NL, ES, SW, UK
Sorgenbarometer
Switzerland
GALLUP
USA
Focus Canada Report
CA
date fieldwork ended
source + countries
Table 12: Overview of sources for MIP/MII questions
URL link: Security-Finder
URL link: environicsinstitute
URL link: environicsinstitute
URL link: environicsinstitute
URL link: GALLUP
URL link: GESIS
source
A.1 appendix chapter 3:
168
A.2 appendix chapter 4:
a.2
appendix chapter 4:
Figure 17: Diagnostics and descriptives of rhetorical position measurement
quantile−normal plot
5
150
histogram
Frequency
50
100
114
0
73
52
1414
3 6
24
4 1 2
−4
0
4
11
−5
31
31
3 3 4 5
−5
0
5
boxplot
−5
Source: Author’s own.
0
5
−2
0
2
Inverse Normal
4
6
169
A.2 appendix chapter 4:
Figure 18: Rhetorical position split by government status
rhetorical position
5
0
−5
opp
Source: Author’s own.
gov
170
A.2 appendix chapter 4:
Figure 19: Marginal effects plot of interaction between rhetorical activities & incumbency
(controlling for manifesto positions)
Effects on Pr(policy output)
1.5
1
.5
0
−5
0
rhetorical position
5
Source: Author’s own.
Note: Based on model 2 (logistic interaction) in figure 3 & table 3 on page 81, model (3) but with controlling for
parties’ manifesto positions before the Fukushima accident. Average Marginal Effects of interaction between
incumbency and parties’ rhetorical positions party with 95 % confidence intervals as dashed lines.
171
A.3 appendix chapter 5:
a.3
appendix chapter 5:
Figure 20: Densityplot of congruence
4
Density
3
2
1
0
0
.2
.4
.6
congruence
kernel = epanechnikov, bandwidth = 0.0669
Source: Author’s own.
.8
1
172
A.3 appendix chapter 5:
Table 13: Effect of protest on amount of party rhetoric, negative binomial regression
party ideology (rile)
Green
protest
survey
mip
nuclear share
CA
FI
DE
NL
UK
constant
lnα
N
(1)
-0.433∗∗∗
(-3.84)
0.310
(71)
0.732∗∗∗
(8.09)
0.737∗∗∗
(4.16)
0.325
(0.79)
-2.151
(-0.83)
-2.953∗∗∗
(-6.43)
-2.786∗∗∗
(-3.37)
117∗
(2.15)
-2.542∗
(-2.41)
-0.694∗
(-2.41)
0.133
(0.14)
0.364∗∗∗
(4.02)
1165
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
All models negative binomial regression with country fixed effects.
173
A.3 appendix chapter 5:
Figure 21: Congruence & the polarisation hypothesis, results of GEE model
Hypotheses: + positive effect, − negative effect
−
ideology
+
green
+
protestindex
+
public opinion
mip
nuclear share
Interaction Effects
ideology*protest
−
survey*protest
base
polarization
green*protest
deflating
green
green*survey
−2
Source: Author’s own.
Note:
−1
0
1
2
3
174
A.3 appendix chapter 5:
Table 14: Robustness tests for polarisation hypothesis, regression results
ideology
protest-index
ideology X protest-index
(1)
-0.0153
(0.161)
0.248∗∗∗
(0.054)
0.190∗∗
(0.065)
(2)
-0.00155
(0.046)
0.112
(0.070)
0.100∗∗
(0.038)
lagged protest-index
green
survey
mip
survey X mip
-1.201∗∗
(0.371)
0.00934
(0.082)
0.0757
(0.156)
-0.263
(0.154)
-0.460∗∗∗
(0.091)
-0.00586
(0.113)
0.0431
(0.117)
-0.135
(0.158)
0.0971
(0.068)
-0.900∗∗∗
(0.119)
-0.154
(0.499)
lagged rhetoric
-0.544
(0.972)
0.463∗∗∗
(0.074)
lagged ideology
lagged survey
lagged survey X lagged mip
lagged nuclear share
-1.662
(1.127)
0.263
(0.410)
0.551
(0.474)
0.180∗∗
(0.070)
0.274∗∗∗
(0.083)
0.356
(0.501)
1098
1098
1165
protest present
ideology X protest present
σ̂cpe
σ̂p
σ̂c
log lik.
N
-0.902∗∗∗
(0.110)
0.0415
(0.121)
0.112
(0.180)
-0.274
(0.144)
0.0248
(0.058)
0.148∗
(0.060)
0.0709
(0.092)
-0.364∗
(0.148)
-1.508
(1.081)
lagged ideology X lagged protest-index
constant
(4)
0.0316
(0.062)
0.352∗
(0.158)
lagged mip
nuclear share
(3)
0.0220
(0.309)
0.464
0.654
0.876
-1292.025
1165
Model 1 multi-level model with country and party random intercepts and ar(1) correction.
Model 2 PCSE model with lagged dependent variable.
Model 3 PCSE model with ar(1) correction and all IVs lagged.
Model 4 PCSE model with ar(1) correction and protest dummy;
(1=protest Yes) included instead of protest-index.
Standard errors in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
175
A.3 appendix chapter 5:
Table 15: Congruence & the polarisation hypothesis, results of GEE model
ideology
green
protest-index
survey
mip
nuclear share
(1)
congruence
-0.887∗
(-2.56)
1.870∗∗
(3.11)
0.699∗∗∗
(6.60)
0.285
(0.75)
-0.0579
(-0.19)
-0.621
(-1.56)
ideology X protest
(2)
congruence
-1.016∗
(-2.56)
1.831∗∗
(2.99)
0.779∗∗∗
(5.13)
0.284
(0.74)
-0.0693
(-0.22)
-0.612
(-1.55)
0.113
(0.64)
survey X protest
(3)
congruence
-0.873∗
(-2.29)
1.797∗∗
(2.68)
0.547∗∗∗
(6.11)
0.884∗
(2.34)
-0.0277
(-0.09)
-0.355
(-0.85)
-0.889∗∗∗
(-5.08)
green X protest
green X survey
constant
N
Groups
-3.027∗∗∗
(-11.23)
1165
67
t statistics in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
-3.124∗∗∗
(-10.72)
1165
67
(4)
congruence
-0.879∗
(-2.51)
0
(.)
0.721∗∗∗
(6.55)
0.352
(0.79)
-0.0445
(-0.14)
-0.618
(-1.59)
-2.992∗∗∗
(-10.41)
1165
67
-0.0753
(-0.24)
-0.365
(-0.49)
-3.047∗∗∗
(-12.26)
1165
67
176
A.4 appendix chapter 6:
a.4
appendix chapter 6:
Figure 22: Margins at means of interaction between electoral cycle & public opinion
Adjusted Predictions with 95% CIs
2
Fitted Values
1
0
cycle=election (0)
cycle=midterm (20)
−1
cycle=after election (42)
−55
0
PO: % favoring nuclear energy
40
Source: Author’s own.
Note: Solid lines report marginal effects surrounded by the 95 % confidence interval as
dotted lines. All other covariates kept at their means.
177
A.4 appendix chapter 6:
Table 16: Effect of electoral cycle on congruence, results of GEE model
cycle
protest-index
survey
mip
gov
rile
green
nuclear share
(1)
congruence
0.660∗
(0.261)
0.715∗∗∗
(0.127)
0.183
(0.317)
-0.0910
(0.314)
-0.584
(0.504)
-0.874∗
(0.360)
0.743∗∗
(0.612)
-0.524
(0.396)
cycle X protest-index
(2)
congruence
0.437
(0.298)
0.782∗∗∗
(0.126)
0.224
(0.313)
-0.154
(0.331)
-0.593
(0.525)
-0.946∗
(0.373)
0.703∗∗
(0.606)
-0.453
(0.405)
0.337
(0.201)
cycle X survey
(3)
congruence
0.602∗
(0.272)
0.746∗∗∗
(0.133)
0.133
(0.315)
-0.0105
(0.312)
-0.683
(0.537)
-0.854∗
(0.360)
0.761∗∗
(0.606)
-0.603
(0.394)
(4)
congruence
0.758∗∗
(0.264)
0.882∗∗∗
(0.135)
-0.154
(0.437)
-0.235
(0.356)
-0.730
(0.548)
-0.996∗∗
(0.378)
0.859∗∗∗
(0.549)
-0.820
(0.433)
0.541
(0.400)
076
(0.567)
2.118∗∗
(0.733)
-0.458
(0.696)
2.236
(0.589)
113.0
1165
cycle X mip
survey X mip
cycle X survey X mip
χ2
N
82.29
1165
Standard errors in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
98.50
1165
78.58
1165
178
A.5 appendix chapter 7:
a.5
appendix chapter 7:
Table 17: Summary statistics (imputed dataset)
Variable
post vs. mat salience
post vs. mat position
protestgeneral
std. lagged postmaterialism index
government party
lagged log rile position
war
accident
std. election counter
std. lagged vote-share
std. lagged unemployment rate
Mean
1.302
0.252
-0.384
0
0.288
-0.399
0.042
0.123
0
0
0
Std. Dev.
1.521
1.209
2.346
0.5
0.453
0.96
0.201
0.328
0.5
0.5
0.5
Min.
-2.896
-2.916
-4.331
-1.297
0
-6.102
0
0
-1.145
-0.548
-0.788
Source: Author’s own.
Figure 23: Overview of lagged environmental protest-index, 1981-1996
10
UK
std. protestindex (environment)
8
UK
UK
UK
UK
UK
UK
UK
UK
UK
UK
UK
6
Germany
Germany
UK
UK
Germany
Germany
Germany
Germany
Germany
Germany
Switzerland
4
Germany
2
0
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
Source: Author’s own.
Max.
6.208
4.047
7.168
1.585
1
2.532
1
1
1.139
1.332
1.385
N
6006
6006
6006
5848
6006
5500
6006
6006
6006
5500
6001
179
A.5 appendix chapter 7:
Figure 24: Overview of lagged pacifist protest-index, 1981-1996
8
UK
UK
UK
UK
UK
UK
UK
UK
Germany
std. protestindex (pacificsm)
6
UK
UK
Germany
Netherlands
Germany
Germany
Germany
Germany
Netherlands
4
Germany
Spain
Germany
Germany
Germany
Germany
Germany
Germany
France
2
France
Switzerland
UK
Belgium
Italy Greece
France
Spain
0
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
Source: Author’s own.
Figure 25: Overview of lagged labor protest-index, 1981-1996
UK
8
UK
UK
UK
UK
France
Germany
6
std. protestindex (labor)
UK
UK
4
2
0
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
Source: Author’s own.
180
A.5 appendix chapter 7:
Figure 26: Overview of lagged postmaterialism index (non-imputed values), 1981-1996
40
postmaterialism vs. materialism
20
Finland
Denmark Denmark
Denmark
Denmark
Netherlands
0
−20
Portugal Portugal
Portugal
−40
Italy
−60
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
Source: Author’s own.
a.5.1
Imputation of postmaterialist index
The core idea of multiple imputation models is that any case in a sample can be replaced
by a new randomly chosen case from the same source population (Donders et al. 2006).
Thus, in the case of a missing value in a variable this missing is replaced by a value drawn
from an estimate of the distribution of this variable. This process is then called imputation.
In the case of multiple imputation, not only a single estimate is used to replace the missing,
but various estimates are used.
While single imputation results in optimistic estimates – with overconfident standard
errors – multiple imputation yields correctly specified standard errors and confidence intervals. Thanks to Markov Chain Monte Carlo (MCMC) methods obtaining such computationally demanding estimates is nowadays possible in most statistical software packages
(I use Amelia II in R). Given the number of missings in my data (31.99 % of country-years
are missing for the postmaterialism index), I decided to impute ten imputed values for
each missing value. Following Rubin (1987: 114), the efficiency of an estimate based on
m imputations is approximately: (1 + y/m)−1 , with y being the rate of missings of the
imputed variable and m the quantity of imputed datasets. In my case, this results in an
efficiency of 0.97, which is sufficient according to Rubin.
In the case of missing at random (MAR) as assumed in my model, imputation models provide unbiased estimates of the results. MAR assumes that the probability that an
observation is missing depends on information for that observation that is available in
the dataset. I gathered information on variables used elsewhere to explain the percentage
of post-materialists in a society (Inglehart 2008; Inglehart and Abramson 1999; Duch and
181
A.5 appendix chapter 7:
Rusk 1993; Inglehart 1977). Especially Duch and Rusk (1993) show that the number of
post-materialists within a society is best explained with macroeconomic indicators – such
as unemployment, GDP per capita and inflation. Most of these variables are fully specified
in my imputational dataset (four country-years are missing in the case of Luxembourg’s
unemployment rate).
The dataset used to impute the postmaterialism index is a time-series-cross-section
(tscs) dataset with country-years as observations. Usually such data comes with a range of
econometric issues to be taken into account if one runs regressions on tscs data. However,
exactly these issues are an advantage to draw imputed values for the data. Thereby, it is
important to annotate that the prediction of imputations is not causal King et al. (2001:
51). Given the tscs structure of my data my imputation model makes use of lags and leads
of the postmaterialism-index, unemployment rate, inflation and GDP per capita (Honaker
and King 2010). Furthermore, since all variables in the dataset should have smooth trends
across time, I introduce a polynomial time trend into the model to smooth out the trends
of the variables across time (Honaker and King 2010: 566).
Table 18: Chain lengths of imputations
Imputation 1:
Imputation 2:
Imputation 3:
Imputation 4:
Imputation 5:
Imputation 6:
Imputation 7:
Imputation 8:
Imputation 9:
Imputation 10:
155
88
134
136
85
98
95
148
132
133
Source: Author’s own.
Figure A.28(a) reports overimputed values of the post-materialist index. Overimputing
is a technique developed by Honaker and King (2010) to judge the fit of the imputation
model. It treats each observed value as if it had actually been missing. For each observed
value it then generates several hundred imputed values – as if it were missing. Figure
A.28(a) shows the estimates of each observed value against its imputed value. The y = x
line indicates the line of perfect agreement. I also report the 90 % confidence interval
for my imputations. If confidence intervals include the y = x line the values have been
confidently predicted. Figure A.28(a) shows that the imputation model works well for
the postmaterialism index. Note, however, that the model is less efficient in generating
values for societies with very high percentages of post-materialists. This is due to only few
countries having high rates of post-materialists. Yet, this issue only pertains to very few
cases and is not troublesome according to the literature (Honaker and King 2010). Figure
A.28(b) reports the imputations across time for Norway. As can be seen the imputations
182
A.5 appendix chapter 7:
−20 0
−60
Imputed Values
20
Figure 27: Results of multiple imputation
0−.2
.2−.4
−40
.4−.6
−20
0
.6−.8
.8−1
20
Observed Values
−10
−30
−50
postvsmat
10
(a) Overimputed values of postmaterialism index
1985
1990
1995
time
(b) Movement of postmaterialism index for Norway, 1981-1996
Source: Author’s own.
Note: Spikes are 95 % confidence interval in both figures. Black points represent
non-missing values in figure A.28(b), red points imputed values.
follow a smoothed trend across time. Judging by face validity the values seem to fit our ex
ante expectations in most instances.
Sources for postmaterialism index come from Eurobarometer (EBS); European Values
Study (EVS); World Values Survey (WVS), figure 19 gives an overview of the employed
sources: The question and categorizations used to generate the postmaterialism index are
well known and outlined in detail in various publications by Inglehart and others. Inglehart and Abramson (1999) explains the methodology standing behind the postmaterialism
index in extensive detail.
183
A.5 appendix chapter 7:
Table 19: Overview of sources for postmaterialism index
country
EBS
EVS
WVS
Austria
Belgium
Denmark
Finland
France
Germany
Greece
Ireland
Italy
Luxembourg
Netherlands
Norway
Portugal
Spain
Sweden
Switzerland
UK
1980-1994
1980-1994
1980-1994
1980-1994
1980-1994
1980-1994
1980-1994
1980-1994
1980-1994
1990-1994
1985-1994
1985-1994
1980-1994
1990
1990
1981
1982; 1990
-
1981; 1996
1995
1981
1989; 1996
-
Source: Author’s own.
184
0.239
(0.0903)
0.495
(0.0631)
0.776
(0.0301)
0.0660
500
10
10
(2)
0.0689∗∗∗
(0.0207)
-0.212
(0.131)
-0.00406
(0.104)
-0.280∗∗∗
(0.0516)
0.197
(0.123)
-0.651∗
(0.256)
-0.260
(0.136)
0.526∗∗∗
(0.107)
0.667∗∗∗
(0.158)
0.257
(0.0817)
0.370
(0.0698)
0.770
(0.0326)
0.0558
419
(1)
0.0841∗∗∗
(0.0218)
-0.289∗
(0.119)
0.00798
(0.101)
-0.341∗∗∗
(0.0534)
0.0937
(0.132)
-0.560∗
(0.254)
-0.337∗
(0.143)
0.551∗∗∗
(0.112)
10
0.253
(0.0892)
0.524
(0.0679)
0.782
(0.0309)
0.0613
500
0.0303
(0.0517)
(3)
0.0682∗∗
(0.0211)
-0.356∗∗
(0.132)
-0.0415
(0.106)
-0.293∗∗∗
(0.0528)
0.205
(0.124)
-0.670∗
(0.274)
-0.208
(0.137)
0.532∗∗∗
(0.109)
10
0.0476
500
0.495∗∗∗
(0.0366)
(4)
0.0584∗∗∗
(0.0173)
-0.194∗
(0.0851)
0.00766
(0.0898)
-0.173∗∗∗
(0.0424)
0.179
(0.116)
-0.437∗
(0.204)
-0.0877
(0.0850)
0.316∗∗∗
(0.0927)
10
0.0512
500
0.452∗∗∗
(0.0411)
(5)
0.0702∗∗
(0.0232)
-0.197∗
(0.0943)
0.0224
(0.0938)
-0.169∗∗∗
(0.0450)
0.164
(0.125)
-0.567∗
(0.264)
-0.191
(0.171)
0.373∗∗
(0.134)
0.603
500
19
10
0.452∗∗∗
(0.0563)
(6)
0.0702∗∗
(0.0202)
-0.197∗
(0.0807)
0.0224
(0.0834)
-0.169∗∗
(0.0471)
0.164
(0.142)
-0.567∗
(0.210)
-0.191
(0.298)
0.373∗
(0.145)
Model (1) restricted to parties achieving 5 seats | 5 % votes.
Model (4) - (6) included lagged dependent variable.
Model (6) uses robust standard errors per election.
Standard errors in parentheses
∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001
All models omit constant; models (5) and (6) include country fixed effects (none significant), here omitted from the table.
hline RVI
N
clusters
Im
σparty
σcountry
lnα
lagged DV
enp
niche
postmaterialism index
unemployment
war
accident
logged rile
government
vote share
protest-index
Table 20: Robustness models postmaterialism vs. materialism, regression results
A.5 appendix chapter 7:
185
BIBLIOGRAPHY
Aarts, Kees, and Maarten Arentsen. N.d. “Nuclear Power and Politics in the Netherlands.”
In Phasing-out and Phasing-in: The Comparative Politics and Policies of Nuclear Energy in
Europe, ed. Wolfgang C. Müller, and Paul W. Thurner. Oxford, UK: Oxford University
Press p. (unpublished manuscript).
Achen, Christopher H. 1977. “Measuring Representation: Coefficient Perils of the Correlation.” American Journal of Political Science 21 (4): 805–815.
Achen, Christopher H. 1978. “Measuring Representation.” American Journal of Political
Science 22 (3): 475–510.
Achen, Christopher H. 2000. “Why Lagged Dependent Variables Can Suppress the Explanatory Power of Other Independent Variables.” Paper prepared for the Annual Meeting of the Political Methodology Section of the American Political Science Association,
July 20-22.
Adams, Greg D. 1997. “Abortion: Evidence of an Issue Evolution.” American Journal of
Political Science 41 (3): 718–737.
Adams, James. 2012. “Causes and Electoral Consequences of Party Policy Shifts in Multiparty Elections: Theoretical Results and Empirical Evidence.” Annual Review of Political
Science 15 (1): 401–419.
Adams, James, Andrea B. Haupt, and Heather Stoll. 2008. “What Moves Parties?: The Role
of Public Opinion and Global Economic Conditions in Western Europe.” Comparative
Political Studies 42 (5): 611–639.
Adams, James, Michael Clark, Lawrence Ezrow, and Garrett Glasgow. 2004. “Understanding Change and Stability in Party Ideologies: Do Parties Respond to Public Opinion or
to Past Election Results?” British Journal of Political Science 34 (4): 589–610.
Adams, James, Michael Clark, Lawrence Ezrow, and Garrett Glasgow. 2006. “Are Niche
Parties Fundamentally Different from Mainstream Parties? The Causes and the Electoral Consequences of Western European Parties’ Policy Shifts, 1976-1998.” American
Journal of Political Science 50 (3): 513–529.
Adler, E. Scott, Chariti E. Gent, and Cary B. Overmeyer. 1998. “The Home Style Homepage:
Legislator Use of the World Wide Web for Constituency Contact.” Legislative Studies
Quarterly 23 (4): 585–595.
Agnone, Jon. 2007. “Amplifying Public Opinion: The Policy Impact of the U.S. Environmental Movement.” Social Forces 85 (4): 1593–1620.
186
bibliography
Ahuja, Sunil. 1994. “Electoral Status and Representation in the United States Senate: Does
Temporal Proximity to Election Matter?” American Politics Research 22 (1): 104–118.
Alcindor, Yamiche, and Marisol Bello. 2014.
“Police in Ferguson Ignite Debate
about Military Tactics.” http://www.usatoday.com/story/news/nation/2014/08/14/
ferguson-militarized-police/14064675/, date accessed: 05/01/2015.
Amenta, Edwin, Bruce G. Carruthers, and Yvonne Zylan. 1992. “A Hero for the Aged? The
Townsend Movement, the Political Mediation Model, and U.S. Old-Age Policy, 19341950.” American Journal of Sociology 98 (2): 1934–1950.
Amenta, Edwin, Kathleen Dunleavy, and Mary Bernstein. 1994. “Stolen Thunder? Huey
Long’s" Share Our Wealth," Political Mediation, and the Second New Deal.” American
Sociological Review 59 (5): 678–702.
Amenta, Edwin, Neal Caren, Elizabeth Chiarello, and Yang Su. 2010. “The Political Consequences of Social Movements.” Annual Review of Sociology 36 (1): 287–307.
Amenta, Edwin, and Yvonne Zylan. 1991. “It Happened Here: Political Opportunity, the
New Institutionalism, and the Townsend Movement.” American Sociological Review 56
(2): 250–265.
Andrews, Kenneth T. 2001. “Social Movements and Policy Implementation: The Mississippi Civil Rights Movement and the War on Poverty, 1965 to 1971.” American Sociological Review 66 (1): 71–95.
Artés, Joaquín. 2013. “Do Spanish Politicians Keep their Promises?” Party Politics 19 (1):
143–158.
Bäck, Hanna, Marc Debus, and Patrick Dumont. 2011. “Who Gets What in Coalition Governments? Predictors of Portfolio Allocation in Parliamentary Democracies.” European
Journal of Political Research 50 (4): 441–478.
Bakker, Ryan, Catherine D. Vries, Erica Edwards, Liesbet Hooghe, Seth Jolly, Gary Marks,
Jonathan Polk, Jan Rovny, Marco R. Steenbergen, and Milada A. Vachudova. 2015. “Measuring Party Positions in Europe: The Chapel Hill Expert Survey Trend File, 1999-2010.”
Party Politics 21 (1): 143–152.
Bara, Judith. 2005. “A Question of Trust: Implementing Party Manifestos.” Parliamentary
Affairs 58 (3): 585–599.
Barkan, Steven E. 1984. “Legal Control of the Southern Civil Rights Movement.” American
Sociological Review 49 (4): 552–565.
Barnes, Samuel H. 1977. Representation in Italy: Institutionalized Tradition and Electoral Choice.
Chicago: University of Chicago Press.
187
bibliography
Barnes, Samuel H., Max Kaase, Klaus R. Allerback, Barbara Farah, Felix Heunks, Ronald
Inglehart, M. Kent Jennings, Hans-Dieter Klingemann, Allan Marsh, and Leopold
Rosenmayr. 1979. Political Action - Mass Participation in Five Western Democracies. Beverly
Hills: Sage Publications.
Baum, Matthew A. 2013. “The Iraq Coalition of the Willing and (Politically) Able: Party
Systems, the Press, and Public Influence on Foreign Policy.” American Journal of Political
Science 57 (2): 442–458.
Baumgartner, Frank R., Christian Breunig, Christoffer Green-Pedersen, Bryan D. Jones,
Peter B. Mortensen, Michiel Nuytemans, and Stefaan Walgrave. 2009. “Punctuated
Equilibrium in Comparative Perspective.” American Journal of Political Science 53 (3):
603–620.
Baumgartner, Frank R., and Christine Mahoney. 2005. “Social Movements, the Rise of New
Issues, and the Public Agenda.” In Routing the Opposition: Social Movements, Public Policy,
and Democracy, ed. David S. Meyer, Valerie Jenness, and Helen Ingram. Minneapolis:
University of Minnesota Press pp. 65–86.
Beck, Nathaniel, and David Epstein. 2002. “Alternative Models of Dynamics in Binary
Time-Series – Cross-Section Models: The Example of State Failure 1.” Prepared for delivery at the 2001 Annual Meeting of the Society for Political Methodology, Emory
University.
Beck, Nathaniel, and Jonathan N. Katz. 1995. “What to Do (and Not to Do) with TimeSeries Cross-Section Data.” American Political Science Review 89 (3): 634–647.
Belchior, Ana Maria. 2013. “Explaining Left-Right Party Congruence Across European
Party Systems: A Test of Micro-, Meso-, and Macro-Level Models.” Comparative Political
Studies 46 (3): 352–386.
Benoit, Kenneth, and Michael Laver. 2007. “Estimating Party Policy Positions: Comparing
Expert Surveys and Hand-Coded Content Analysis.” Electoral Studies 26 (1): 90–107.
Bernardi, Luca. 2014. “Word is silver, but silence is golden? Policy agenda, responsiveness,
and competitive incentives.” Paper prepared for presentation at the ECPR General Conference Glasgow, 3-6 September 2014.
Bernstein, Robert A., Gerald C. Wright, and Michael B. Berkman. 1988. “Do U.S. Senators
Moderate Strategically?” American Political Science Review 82 (1): 237–245.
Bevan, Shaun, Peter John, and Will Jennings. 2011. “Keeping Party Programmes on Track:
The Transmission of the Policy Agendas of Executive Speeches to Legislative Outputs
in the United Kingdom.” European Political Science Review 3 (03): 395–417.
Bevan, Shaun, and Will Jennings. 2014. “Representation, Agendas and Institutions.” European Journal of Political Research 53 (1): 37–56.
188
bibliography
Bianco, William T. 1994. Trust - Representatives and Constituents. Ann Arbor: University of
Michigan Press.
Birkland, Thomas A. 1998. “Focusing Events, Mobilization, and Agenda Setting.” Journal
of Public Policy 18 (1): 53–74.
Bischof, Daniel. 2015. “Towards a Renewal of the Niche Party Concept: Parties, Market
Shares and Condensed Offers.” Party Politics p. forthcoming.
Bischof, Daniel, Maarja Lühiste, and Laura Morales. 2013. “ResponsiveGov Policy-Related
Events Codebook for the Pilot Study: Party Policy Positions.” ResponsiveGov Project
Working Papers http://www.responsivegov.eu/index.php/data.
Bischof, Daniel, and Simon Fink. 2015. “Repression as a Double-Edged Sword: Resilient
Monarchs, Repression and Revolution in the Arab World.” Swiss Political Science Review
21 (3): 377–395.
Bishin, Benjamin G. 2000. “Constituency Influence in Congress: Does Subconstituency
Matter?” Legislative Studies Quarterly 25 (3): 389–415.
Bølstad, Jørgen. 2012. “Thermostatic Voting: Presidential Elections in Light of New Policy
Data.” PS: Political Science & Politics 45 (01): 44–50.
Brace, Paul, and Barbara Hinckley. 1993. “Presidential Activities from Truman through
Reagan: Timing and Impact.” Journal of Politics 55 (02): 382–398.
Brambor, Thomas, William R. Clark, and Matt Golder. 2005. “Understanding Interaction
Models: Improving Empirical Analyses.” Political Analysis 14 (1): 63–82.
Brandenburg, H. 2002. “Who Follows Whom?: The Impact of Parties on Media Agenda
Formation in the 1997 British General Election Campaign.” The Harvard International
Journal of Press/Politics 7 (3): 34–54.
Brandt, Patrick T., John T. Williams, Benjamin O. Fordham, and Brain Pollins. 2000. “Dynamic Modeling for Persistent Event-Count Time Series.” American Journal of Political
Science 44 (4): 823.
Brandt, Patrick T., and Todd Sandler. 2012. “A Bayesian Poisson Vector Autoregression
Model.” Political Analysis 20 (3): 292–315.
Braumoeller, Bear. 2004. “Hypothesis Testing and Multiplicative Interaction Terms.” International Organization 58 (4): 807–820.
Braun, Robert, and Ruud Koopmans. 2010. “The Diffusion of Ethnic Violence in Germany:
The Role of Social Similarity.” European Sociological Review 26 (1): 111–123.
Brouard, Sylvain, and Isabelle Guinaudeau. 2013. “Policy beyond Politics? Public Opinion, Party Politics and the French pro-nuclear Energy Policy.” Paper Prepared for the
General ECPR Conference, Bordeaux.
189
bibliography
Budge, Ian, and Dennis J. Farlie. 1983. Explaining and Predicting Elections - Issue Effects and
Party Strategies in Twenty-Three Democracies. London: George Allen & Unwin.
Budge, Ian, Hans-Dieter Klingemann, Andrea Volkens, Judith Bara, and Eric Tanenbaum.
2001. Mapping Policy Preferences: Estimates for Parties, Electors, and Governments, 19451998. Oxford: Oxford University Press.
Budge, Ian, and Richard I. Hofferbert. 1990. “Mandates and Policy Outputs: U.S. Party
Platforms and Federal Expenditures.” American Political Science Review 84 (1): 111–131.
Burke, Edmund. 1774. “Speech to the Electors of Bristol.” In Works 1. pp. 446–448.
Burstein, Paul. 1979. “Public Opinion, Demonstrations, and the Passage of Antidiscrimination Legislation.” Public Opinion Quarterly 43 (2): 157–172.
Burstein, Paul. 1985. Discrimination, Jobs, and Politics. Chicago: University of Chicago Press.
Burstein, Paul. 1998. “Bringing the Public Back In: Should Sociologists Consider the Impact of Public Opinion on Public Policy?” Social Forces 77 (1): 27–62.
Burstein, Paul. 1999. “Social Movements and Public Policy.” In How Social Movements
Matter, ed. Marco G. Giugni, Doug Mcadam, and Charles Tilly. Minneapolis & London:
University of Minnesota Press pp. 3–21.
Burstein, Paul. 2006. “Why Estimates of the Impact of Public Opinion on Public Policy are
Too High: Empirical and Theoretical Implications.” Social Forces 84 (4): 2273–2289.
Burstein, Paul. 2014. American Public Opinion, Advocacy, and Policy in Congress. Cambridge,
Massachusetts: Cambridge University Press.
Burstein, Paul, and April Linton. 2002. “The Impact of Political Parties, Interest Groups,
and Social Movement Organizations on Public Policy: Some Recent Evidence and Theoretical Concerns.” Social Forces 81 (2): 380–408.
Butler, Daniel M., Christopher F. Karpowitz, and Jeremy C. Pope. 2012. “A Field Experiment on Legislators’ Home Styles: Service versus Policy.” Journal of Politics 74 (02):
474–486.
Cain, Bruce E., John A. Ferejohn, and Morris P. Fiorina. 1984. “The Constituency Service
Basis of the Personal Vote for U.S. Representatives and British.” American Political Science
Review 78 (1): 110–125.
Canes-Wrone, Brandice. 2001a. “A Theory of Presidents’ Public Agenda Setting.” Journal
of Theoretical Politics 13 (2): 183–208.
Canes-Wrone, Brandice. 2001b. “The President’s Legislative Influence from Public Appeals.” American Journal of Political Science 45 (2): 313–329.
190
bibliography
Canes-Wrone, Brandice, and Jee-Kwang Park. 2012. “Electoral Business Cycles in OECD
Countries.” American Political Science Review 106 (01): 103–122.
Canes-Wrone, Brandice, and Kenneth W. Shotts. 2004. “The Conditional Nature of Presidential Responsiveness to Public Opinion.” American Journal of Political Science 48 (4):
690–706.
Canes-Wrone, Brandice, Michael C. Herron, and Kenneth W. Shotts. 2001. “Leadership and
Pandering: A Theory of Executive Policymaking.” American Journal of Political Science 45
(3): 532–550.
Canes-Wrone, Brandice, William Minozzi, and Jessica Bonney Reveley. 2011. “Issue Accountability and the Mass Public.” Legislative Studies Quarterly 36 (1): 5–35.
Carmines, Edward G. 1991. “The Logic of Party Alignments.” Journal of Theoretical Politics
3 (1): 65–80.
Carrubba, Clifford J., Matthew Gabel, Lacey Murrah, Ryan Clough, Elizabeth Montgomery,
and Rebecca Schambach. 2006. “Off the Record: Unrecorded Legislative Votes, Selection
Bias and Roll-Call Vote Analysis.” British Journal of Political Science 36 (04): 691.
Castles, Francs G., and Peter Mair. 1984. “Left-Right Political Scales: Some ‘Expert’ Judgments.” European Journal of Political Research 12 (1): 147–157.
Clarke, Harold D, and Nitish Dutt. 1991. “Measuring Value Change in Western Industrialized Socieites: The Impact of Unemployment.” American Political Science Review 85 (3):
905–920.
Cohen, Jeffrey E. 1995. “Presidential Rhetoric and the Public Agenda.” American Journal of
Political Science 39 (1): 87–107.
Cohen, Jeffrey E. 1999. Presidential Responsiveness and Public Policy-making: The Public and
the Policies that Presidents Choose. Ann Arbor: University of Michigan Press.
Converse, Philip E., and Roy Pierce. 1986. Political Representation in France. Cambridge,
MA: Harvard University Press.
Costello, Rory, and Robert Thomson. 2008. “Election Pledges and their Enactment in
Coalition Governments: A Comparative Analysis of Ireland.” Journal of Elections, Public
Opinion & Parties 18 (3): 239–256.
Dahl, Robert. 1971. Polyarchy. New Haven, CT: Yale University Press.
Dalton, Russell J. 1985. “Political Parties and Political Representation: Party Supporters
and Party Elites in Nine Nations.” Comparative Political Studies 18 (3): 267–299.
Dalton, Russell J. 2008. “Citizenship Norms and the Expansion of Political Participation.”
Political Studies 56 (1): 76–98.
191
bibliography
Davis, Darren W., and Christian Davenport. 1999. “Assessing the Validity of the Postmaterialism Index.” American Political Science Review 93 (3): 649–664.
Debus, Marc. 2009. “Analysing Party Politics in Germany with New Approaches for
Estimating Policy Preferences of Political Actors.” German Politics 18 (3): 281–300.
DeNardo, James. 1985. Power in Numbers. Princeton: Princeton University Press.
Donders, A. Rogier T., Geert J. M. G. van der Heijden, Theo Stijnen, and Karel G. M.
Moons. 2006. “Review: A Gentle Introduction to Imputation of Missing Values.” Journal
of Clinical Epidemiology 59 (10): 1087–1091.
Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper and Row.
Driscoll, John C., and Aart C. Kraay. 1998. “Consistent Covariance Matrix Estimation with
Spatially Dependent Panel Data.” Review of Economics and Statistics 80 (4): 549–560.
Druckman, James N., and Justin W. Holmes. 2004. “Does Presidential Rhetoric Matter?
Priming and Presidential Approval.” Presidential Studies Quarterly 4 (4): 755–778.
Duch, Raymond M., and Jerrold G. Rusk. 1993. “Postmaterialism and the Economic Condition.” American Journal of Political Science 37 (3): 747.
Duverger, Maurice. 1954. Political Parties: Their Organization and Activity in the Modern State.
2nd ed. Paris: Law Book Co of Australasia.
Earl, Jennifer, Sarah A. Soule, and John D. McCarthy. 2003. “Protest under Fire? Explaining
the Policing of Protest.” American Sociological Review 68 (4): 581.
Egan, Patrick J. 2013. Partisan Priorities How Issue Ownership Drives and Distorts American
Politics. Cambridge: Cambridge University Press.
Egan, Patrick J. 2014. ““Do Something” Politics and Double-Peaked Policy Preferences.”
Journal of Politics 76 (02): 333–349.
Eisinger, Peter K. 1973. “The Conditions of Protest Behavior in American Cities.” American
Political Science Review 67 (1): 11–28.
Elling, Richard C. 1982. “Ideological Change in the U. S. Senate: Time and Electoral
Responsiveness.” Legislative Studies Quarterly 7 (1): 75–92.
EPCD Research Group. N.d. “Codebook for European Protest and Coercion Data, 1980
through 1995.”.
Epstein, Leon D. 1980. Political Parties in Western Democracies. New Brunswick, N.J: Transaction Books.
192
bibliography
Erikson, Robert S. 1978. “Constituency Opinion and Congressional Behavior: A Reexamination of the Miller-Stokes Representation Data.” American Journal of Political Science 22
(3): 511–535.
Eulau, Heinz, and Paul D. Karps. 1977. “The Puzzle of Representation: Specifying Components of Responsiveness.” Legislative Studies Quarterly 2 (3): 233–254.
Ezrow, Lawrence. 2010. Linking Citizens and Parties. Oxford: Oxford University Press.
Fearon, James D. 1997. “Signaling Foreign Policy Interests: Tying Hands versus Sinking
Costs.” Journal of Conflict Resolution 41 (1): 68–90.
Fenno, Richard. 1978. Home Style: House Members in Their Districts. Little: Brown.
Fenno, Richard F. Jr. 1977. “U.S. House Members in Their Constituencies: An Exploration.”
American Political Science Review 71 (3): 883–917.
Ferejohn, John A. 1986. “Incumbent Performance and Electoral Control.” Public Choice 50
(1-3).
Ferrara, Frederico F., and J. Timo Weishaupt. 2004. “Get your Act Together: Party Performance in European Parliament Elections.” European Union Politics 5 (3): 283–306.
Francisco, Ronald A. 1995. “The Relationship between Coercion and Protest: An Empirical
Evaluation in Three Coercive States.” Journal of Conflict Resolution 39 (2): 263–282.
Francisco, Ronald A. 1996. “Coercion and Protest: An Empirical Test in Two Democratic
States.” American Journal of Political Science 40 (4): 1179–1204.
Francisco, Ronald A. 2009. Dynamics of Conflict. New York, NY: Springer New York.
Franklin, Mark N. 1993. Parliamentary Questions. Oxford: Clarendon Press.
Fuchs, Dieter, and Robert Rohrschneider. 1998. “Postmaterialism and electoral choice
before and after German unification.” West European Politics 21 (2): 95–116.
Gabel, Matthew, and Kenneth Scheve. 2007. “Estimating the Effect of Elite Communications on Public Opinion Using Instrumental Variables.” American Journal of Political
Science 51 (4): 1013–1028.
Gabel, Matthew, and Simon Hix. 2007. “From Preferences to Behaviour: Comparing MEPs’
Survey Responses and Roll-Call Voting Behaviour.” Paper prepared for presentation at
the Tenth Biennial Conference of the European Union Studies Association, Montreal,
Canada, 17-19 May 2007.
Gamson, William A. 1990. The Strategy of Social Protest. 2nd ed. Belmont: Wadsworth.
Garrett, Geoffrey. 1998. “Global Markets and National Politics: Collision Course or Virtuous Circle?” International Organization 52 (4): 787–824.
193
bibliography
Geddes, Barbara. 1990. “How the Cases You Choose Affect the Answers You Get: Selection
Bias in Comparative Politics.” Political Analysis 2 (1): 131–150.
Gelman, Andrew. 2008. “Scaling Regression Inputs by Dividing by Two Standard Deviations.” Statistics in Medicine 27 (15): 2865–2873.
Gelman, Andrew, Gary King, and Chuanhai Liu. 1998. “Not Asked and Not AnsweredMultiple Imputation for Multiple Surveys.” Journal of the American Statistical Association
93 (443): 846–857.
Gillion, Daniel Q. 2012. “Protest and Congressional Behavior: Assessing Racial and Ethnic
Minority Protests in the District.” Journal of Politics 74 (04): 950–962.
Gillion, Daniel Q. 2013. The Political Power of Protest: Minority Activism and Shifts in Public
Policy. Cambridge, UK: Cambridge University Press.
Gillion, Daniel Q. 2014.
“Representation Congruence and Political Accountability: The Link between Racial Remarks and Policy Decisions.” mimeo, http://
www.sas.upenn.edu/$\sim$dgillion/Representation%20Congruence%20(Web).pdf, retrieved: 13.08.2014.
Giugni, Marco G. 1998. “Was it Worth the Effort? The Outcomes and Consequences of
Social Movements.” Annual Review of Sociology 24 (1): 371–393.
Giugni, Marco G. 2004. Social Protest and Policy Change. Oxford, UK: Rowman & Littlefield
Publishers.
Giugni, Marco G. 2007. “Useless Protest? A Time-Series Analysis of the Policy Outcomes
of Ecology, Antinuclear, and Peace Movements in the United States, 1977-1995.” Mobilization: An International Quarterly 12 (1): 53–77.
Giugni, Marco G., Doug McAdam, and Charles Tilly. 1999. How Social Movements Matter.
Minneapolis: University of Minnesota Press.
Golder, M. 2003. “Explaining Variation In The Success Of Extreme Right Parties In Western
Europe: .” Comparative Political Studies 36 (4): 432–466.
Green-Pedersen, Christoffer, and Peter B. Mortensen. 2010. “Who Sets the Agenda and
who Responds to it in the Danish parliament? A new Model of Issue Competition and
Agenda-setting.” European Journal of Political Research 49 (2): 257–281.
Green-Pedersen, Christoffer, and Rune Stubager. 2010. “The Political Conditionality of
Mass Media Influence: When Do Parties Follow Mass Media Attention?” British Journal
of Political Science 40 (03): 663–677.
Greene, Zachary. 2015. “Competing on the Issues: How Experience in Government and
Economic Conditions Influence the Scope of Parties’ Policy Messages.” Party Politics
p. forthcoming.
194
bibliography
Grose, Christian R., Neil Malhotra, and Robert P. Van Houweling. 2014. “Explaining Explanations: How Legislators Explain their Policy Positions and How Citizens React.”
American Journal of Political Science 00 (0): n/a–n/a.
Gurr, Ted R. 2011. Why Men Rebel. 4th ed. London: Paradigm Publishers.
Hakhverdian, Armèn. 2012. “The Causal Flow between Public Opinion and Policy: Government Responsiveness, Leadership, or Counter Movement?” West European Politics 35
(6): 1386–1406.
Hamann, Kerstin, A. Johnston, and J. Kelly. 2012. “Unions Against Governments: Explaining General Strikes in Western Europe, 1980-2006.” Comparative Political Studies 46 (9):
1030–1057.
Hamann, Kerstin, Alison Johnston, and John Kelly. 2013. “Striking Concessions from
Governments: Explaining the Success of General Strikes in Western Europe, 1980-2009.”
Comparative Politics 46 (1): 23–41.
Hansen, Martin E. 2008. “Back to the Archives? A Critique of the Danish Part of the
Manifesto Dataset.” Scandinavian Political Studies 31 (2): 201–216.
Harmel, Robert. 2011. “The How’s and Why’s of Party Manifestos: Some Thoughts for
a Cross-National Research Agenda.” Paper prepared for the ECPR Joint Sessions of
Workshops, St. Gallen, Switzerland, 12-17 April, 2011.
Harmel, Robert, and John D. Robertson. 1985. “Formation and Success of New Parties A
Cross-National Analysis.” International Political Science Review 6 (4): 501–523.
Harmel, Robert, and Kenneth Janda. 1994. “An Integrated Theory of Party Goals and
Party Change.” Journal of Theoretical Politics 6 (3): 259–287.
Harris, Phil, Donna Fury, and Andrew Lock. 2006. “Do Political Parties and the Press
Influence the Public Agenda? A Content Analysis of Press Coverage of the 2001 UK
General Election.” Journal of Political Marketing 5 (3): 1–28.
Hart, Roderick. 1987. The Sound of Leadership: Political Communication in the Modern Age.
Chicago: Chicago University Press.
Heckman, James J. 1979. “Sample Selection Bias as a Specification Error.” Econometrica 47
(1): 153–161.
Helbling, Marc, and Anke Tresch. 2011. “Measuring Party Positions and Issue Salience
from Media Coverage: Discussing and Cross-Validating New Indicators.” Electoral Studies 30 (1): 174–183.
Hill, Kim Q., and Patricia A. Hurley. 1999. “Dyadic Representation Reappraised.” American
Journal of Political Science 43 (1): 109–137.
195
bibliography
Hinckley, Barbara. 1990. The Symbolic Presidency: How Presidents Portray Themselves. New
York: Routledge.
Hobolt, Sara B., and Robert Klemmensen. 2005. “Responsive Government? Public Opinion
and Government Policy Preferences in Britain and Denmark.” Political Studies 53 (2):
379–402.
Hobolt, Sara B., and Robert Klemmensen. 2008. “Government Responsiveness and Political Competition in Comparative Perspective.” Comparative Political Studies 41 (3): 309–
337.
Hofferbert, Richard I., and Ian Budge. 1992. “The Party Mandate and the Westminster
Model: Election Programmes and Government Spending in Britain, 1948-85.” British
Journal of Political Science 22 (2): 151–182.
Holian, David B. 2004. “He’s Stealing My Issues! Clinton’s Crime Rhetoric and the Dynamics of Issue Ownership.” Political Behavior 26 (2): 95–124.
Holsti, Ole R. 1996. Public Opinion and American Foreign Policy. Cambridge University
Press.
Honaker, James, and Gary King. 2010. “What to do About Missing Values in Time Series
Cross-Section Data.” American Journal of Political Science 54 (2): 561–581.
Hopmann, David N., Christian Elmelund-Praestekaer, Eerik Albaek, Rens Vliegenthart,
and Claes H. de Vreese. 2010. “Party Media Agenda-Setting: How Parties Influence
Election News Coverage.” Party Politics 18 (2): 173–191.
Hug, Simon, and Tobias Schulz. 2007. “Left - Right Positions of Political Parties in Switzerland.” Party Politics 13 (3): 305–330.
Inglehart, Ronald. 1971. “The Silent Revolution in Europe: Intergenerational Change in
Post-Industrial Societies.” American Political Science Review 65 (04): 991–1017.
Inglehart, Ronald. 1977. The Silent Revolution. Changing Values and Political Styles Among
Western Publics. Princeton: Princeton University Press.
Inglehart, Ronald. 2002. “Trends in Political Action: The Developmental Trend and the
Post-Honeymoon Decline.” International Journal of Comparative Sociology 43 (3-5): 300–
316.
Inglehart, Ronald F. 2008. “Changing Values among Western Publics from 1970 to 2006.”
West European Politics 31 (1-2): 130–146.
Inglehart, Ronald, and Hans-Dieter Klingemann. 1976. “Party identification, ideological
preference, and the left-right dimension among western mass publics.” Psychômetrika
29 (1972): 242–273.
196
bibliography
Inglehart, Ronald, and Paul R. Abramson. 1999. “Measuring Postmaterialism.” American
Political Science Review 93 (3): 665–677.
Inglehart, Ronald, and Wayne E. Baker. 2000. “Modernization, Cultural Change, and the
Persistence of Traditional Values.” American Sociological Review 65 (1): 19–51.
Jahn, Detlef, and Sebastian Korolczuk. 2012. “German Exceptionalism: The End of Nuclear
Energy in Germany!” Environmental Politics 21 (1): 159–164.
Jennings, Will, and Christopher Wlezien. 2011. “Distinguishing Between Most Important
Problems and Issues?” Public Opinion Quarterly 75 (3): 545–555.
Jennings, Will, and Christopher Wlezien. 2015. “The Timeline of Election Campaigns: A
Comparative Perspective.” American Journal of Political Science .
Jennings, Will, and Peter John. 2009. “The Dynamics of Political Attention: Public Opinion
and the Queen’s Speech in the United Kingdom.” American Journal of Political Science 53
(4): 838–854.
Jennings, Will, Shaun Bevan, and Peter John. 2011. “The Agenda of British Government:
The Speech from the Throne, 1911-2008.” Political Studies 59 (1): 74–98.
Johnson, Erik W., Jon Agnone, and John D. Mccarthy. 2003. “Where and How Do Movements Matter? The United States Environmental Movement and Congressional Hearings and Laws, 1961-1990.”.
Jones, Bryan D., and Frank R. Baumgartner. 2005. The Politics of Attention - How Government
Prioritizes Problems. Chicago: Chicago University Press.
Kalogeropoulou, Efthalia. 1989. “Election Promises and Government Performance in
Greece: PASOK’s Fulfilment of its 1981 Election Pledges.” European Journal of Political
Research 17 (3): 289–311.
Katz, Richard S., and Peter Mair. 1995. “Changing Models of Party Organization and Party
Democracy.” Party Politics 1 (1): 5–28.
Kavanagh, Dennis. 1981. “The Politics of Manifestos.” Parliamentary Affairs 34 (1): 7–27.
Keman, Hans. 2007. “Experts and Manifestos: Different Sources - Same Results for Comparative Research?” Electoral Studies 26 (1): 76–89.
King, Brayden G., Keith G. Bentele, and Sarah A. Soule. 2007. “Protest and Policymaking:
Explaining Fluctuation in Congressional Attention to Rights Issues, 1960-1986.” Social
Forces 86 (1): 1960–1986.
King, Gary, James Honaker, Anne Joseph, and Kenneth Scheve. 2001. “Analyzing Incomplete Political Science Data.” American Political Science Review 85 (1): 49–69.
197
bibliography
King, Gary, Michael Laver, Richard I. Hofferbert, Ian Budge, and Michael D. McDonald.
1993. “Party Platforms, Mandates, and Government Spending.” American Political Science Review 87 (3): 744–750.
Kirchheimer, Otto. 1990. “The Catch-all Party.” In The West European Party System, ed.
Peter Mair. Oxfrod: Oxford University Press pp. 50–60.
Kitschelt, Herbert. 2005. “Movement Parties.” In Handbook of Party Politics, ed. Richard S.
Katz, and William J. Crotty. London: Sage pp. 278–290.
Kitschelt, Herbert P. 1986. “Political Opportunity Structures and Political Protest: AntiNuclear Movements in Four Democracies.” British Journal of Political Science 16 (1): 57–
85.
Kitschelt, Herbert P. 1993. “Social Movements, Political Parties, and Democratic Theory.”
ANNALS of the American Academy of Political and Social Science 528 (1): 13–29.
Kitschelt, Herbert P. 1994. The Transformation of European Social Democracy. Cambridge, UK:
Cambridge University Press.
Klingemann, Hans-Dieter, Andrea Volkens, Judith L. Bara, Ian Budge, and Michael D.
McDonald. 2006. Mapping Policy Preferences II - Estimates for Parties, Electors, and Governments in Eastern Europe, European Union, and OECD 1990-2003. Oxford, UK: Oxford
University Press.
Kolb, Felix. 2007. Protest and Opportunities - The Political Outcomes of Social Movements.
Frankfurt/New York: Campus Verlag.
Koopmans, Ruud, and Jan W. Duyvendak. 1995. “Political Construction of the Nuclear
Energy Issue and Its Impact on the Mobilization of Anti-Nuclear Movements in Western
Europe.” Social Problems 42 (2): 235–251.
Koopmans, Ruud, and Paul Statham. 1999. “Political Claims Analysis: Integrating Protest
Event and Political Discourse Approaches.” Mobilization: An International Quarterly 4
(1998): 203–221.
Kriesi, Hanspeter. N.d. “Switzerland.” In Phasing-out and Phasing-in: The Comparative Politics and Policies of Nuclear Energy in Europe, ed. Wolfgang C. Müller, and Paul W. Thurner.
Oxford, UK: Oxford University Press p. (unpublished manuscript).
Kriesi, Hanspeter, Edgar Grande, Romain Lachat, Martin Dolezal, Simon Bornschier, and
Timotheos Frey. 2008. West European Politics in the Age of Globalization. Cambridge, UK:
Cambridge University Press.
Kriesi, Hanspeter, Ruud Koopmans, Jan W. Duyvendak, and Marco G. Giugni. 1992. “New
Social Movements and Political Opportunities in Western Europe.” European Journal of
Political Research 22 (2): 219–244.
198
bibliography
Kriesi, Hanspeter, Ruud Koopmans, Jan Willem Duyvendak, and Marco G. Giugni. 1995.
New Social Movements in Western Europe. London: UCL Press.
Kuklinski, James H. 1978. “Representativeness and Elections: A Policy Analysis.” American
Political Science Review 72 (1): 165–177.
Laver, Michael. 2001. Estimating the Policy Positions of Political Actors. London & New York:
Routledge.
Laver, Michael, Kenneth Benoit, and John Garry. 2003. “Extracting Policy Positions from
Political Texts Using Words as Data.” American Political Science Review 97 (2): 311–331.
Lax, Jeffrey R., and Justin H. Phillips. 2009. “Gay Rights in the States: Public Opinion and
Policy Responsiveness.” American Political Science Review 103 (03): 367.
Lax, Jeffrey R., and Justin H. Phillips. 2012. “The Democratic Deficit in the States.” American Journal of Political Science 56 (1): 148–166.
Liang, Kung-yee Y, and Scott L. Zeger. 1986. “Logitudinal Data Analzsis Using Generaliyes
Linear Models.” Biometrics 73 (1): 13–22.
Lipsky, Michael. 1968. “Protest as a Political Resource.” American Political Science Review
62 (4): 1144–1158.
Lohmann, Susanne. 1993. “A Signaling Model of Informative and Manipulative Political
Action.” American Political Science Review 87 (2): 319–333.
Lowe, Will, Kenneth Benoit, Slava Mikhaylov, and Michael Laver. 2011. “Scaling Policy
Preferences from Coded Political Texts.” Legislative Studies Quarterly 36 (1): 123–155.
Lühiste, Maarja, and Laura Morales. 2014. “ResponsiveGov Policy-Related Events
Codebook for the Pilot Study.” ResponsiveGov Project Working Papers http://www.
responsivegov.eu/images/documents/Codebook_web.pdf.
Lühiste, Maarja, Laura Morales, Daniel Bischof, and Luca Bernardi. 2014. “The Effects of
the Fukushima Disaster on Nuclear Energy Debate and Policy: A Comparative Examination.” Paper presented at the MPSA Annual Conference, 3-6 April, Chicago, USA.
Mackuen, Michael B., Robert S. Erikson, and James A. Stimson. 1992. “Peasants or
Bankers? The American Electorate and the U.S. Economy.” American Political Science
Review 86 (3): 597–611.
Mair, Peter. 2001. “Searching for the Positions of Political Actors.” In Estimating the Policy
Positions of Political Actors, ed. Michael Laver. London & New York: Routledge pp. 10–30.
Mansergh, Lucy, and Robert Thomson. 2007. “Election Pledges, Party Competition, and
Policymaking.” Comparative Politics 39 (3): 311–329.
199
bibliography
Manza, Jeff, and Fay L. Cook. 2001. Policy Responsiveness: The State of the Debate. Number
July Institute for Policy Research, Northwestern University.
Marien, Sofie, Marc Hooghe, and Ellen Quintelier. 2010.
“Inequalities in Noninstitutionalised Forms of Political Participation: A Multi-level Analysis of 25 Countries.” Political Studies 58 (1): 187–213.
Marquart,
viele
Maria.
neue
2010.
Fragen.”
“Wasserwerfer-Opfer:
Ein
Video
und
http://www.spiegel.de/politik/deutschland/
wasserwerfer-opfer-ein-video-und-viele-neue-fragen-a-721733.html,
date
accessed: 05/01/2015.
Martin, Shane. 2011. “Parliamentary Questions, the Behaviour of Legislators, and the
Function of Legislatures: An Introduction.” Journal of Legislative Studies 17 (3): 259–270.
Martin, Shane. 2014. “Policy, Office and Votes: The Electoral Value of Ministerial Office.”
British Journal of Political Science (August): 1–16.
Mattila, Mikko, and Tapio Raunio. 2006. “Cautious Voters - Supportive Parties: Opinion
Congruence between Voters and Parties on the EU Dimension.” European Union Politics
7 (4): 427–449.
Mayhew, David. 1974. Congress: The Electoral Connection. New Haven: Yale University
Press.
McAdam, Doug. 1982. Political Process and the Development of Black Insurgency, 1930-1970.
Chicago: University of Chicago Press.
McAdam, Doug, and Yang Su. 2002. “The War at Home: Antiwar Protests and Congressional Voting, 1965 to 1973.” American Sociological Review 67 (5): 696–721.
McCammon, Holly J., and Karen E. Campbell. 2001. “How Movements Win: Gendered
Opportunity Structures and U.S. Women’s Suffrage Movements, 1866 to 1919.” American
Sociological Review 66 (1): 49–70.
McDonald, Michael D., Silvia M. Mendes, and Ian Budge. 2003. “What Are Elections For?
Conferring the Median Mandate.” British Journal of Political Science 34 (01): 1.
McGraw, Kathleen M., Samuel Best, and Richard Timpone. 1995. “"What They Say or What
They Do?" The Impact of Elite Explanation and Policy Outcomes on Public Opinion.”
American Journal of Political Science 39 (1): 53–74.
Meguid, Bonnie M. 2005. “Competition Between Unequals: The Role of Mainstream Party
Strategy in Niche Party Success.” American Political Science Review 99 (3): 347–359.
Meguid, Bonnie M. 2007. Party Competition between Unequals. New York: Cambridge University Press.
200
bibliography
Menaldo, Victor. 2012. “The Middle East and North Africa’s Resilient Monarchs.” Journal
of Politics 74 (03): 707–722.
Meyer, David S. 2004. “Protest and Political Opportunities.” Annual Review of Sociology 30
(1): 125–145.
Meyer, David S., and Suzanne Staggenborg. 1996. “Movements, countermovements, and
the structure of political opportunity.” American Journal of Sociology 101 (6): 1628–1660.
Meyer, Thomas M. 2013. Constraints on Party Policy Change. Colchester, UK: ECPR Press.
Meyer, Thomas M., and Markus Wagner. 2015. “Issue Engagement in Election Campaigns
The Impact of Electoral Incentives and Organizational Constraints.” Political Science Research and Methods p. (forthcoming).
Miller, Warren E., and Donald E. Stokes. 1963. “Constituency Influence in Congress.”
American Political Science Review 57 (1): 45–56.
Miller, Warren E., Roy Pierce, Jacques Thomassen, Richard Herrera, Sören Holmberg, Peter
Esaiasson, and Bernhard Wessels. 1999. Policy Representation in Western Democracies.
Oxford: Oxford University Press.
Morales, Laura. 2014. “ResponsiveGov Project – Key Aspects of the Research Design:
Selection of Policy Case Studies and Measurement of Responsiveness.” http://www.
responsivegov.eu/images/documents/Research%20design.pdf.
Morris, Aldon D. 1993. “Birmingham Confrontation Reconsidered: An Analysis of the
Dynamics and Tactics of Mobilization.” American Sociological Review 58 (5): 621–636.
Mueller, Carol. 1997. “International Press Coverage of East German Protest Events, 1989.”
American Sociological Review 62 (5): 820–832.
Muller, Edward N., and Karl-Dieter Opp. 1986. “Rational Choice and Rebellious Collective
Action.” American Political Science Review 80 (2): 471–488.
Müller-Rommel, Ferdinand. 1984. “Zum Verhältnis von neuen sozialen Bewegungen und
neuen Konfliktdimensionen in den politischen Systemen Westeuropas: Eine empirische
Analyse.” Journal für Sozialforschung 24 (4): 441–454.
Müller-Rommel, Ferdinand. 1985. “New Social Movements and Smaller Parties: A Comparative Perspective.” West European Politics 8 (1): 41–54.
Müller, Wolfgang C. 2000. “Political Parties in Parliamentary Democracies: Making Delegation and Accountability Work.” European Journal of Political Research 37 (3): 309–333.
Müller, Wolfgang C., and Kaare Strøm. 1999. Policy, Office, or Votes? Cambridge: Cambridge University Press.
201
bibliography
Nam, Taehyun. 2006. “What You use Matters: Coding Protest Data.” PS: Political Science
& Politics 39 (2): 281–287.
Nam, Taehyun. 2007. “Rough Days in Democracies: Comparing Protests in Democracies.”
European Journal of Political Research 46 (1): 97–120.
Nohrstedt, Daniel. 2008. “The Politics of Crisis Policymaking: Chernobyl and Swedish
Nuclear Energy Policy.” Policy Studies Journal 36 (2): 257–278.
Norris, Pippa. 1999. Critical Citizens - Global Support for Democratic Governance. Oxford:
Oxford University Press.
Olson, Mancur. 1965. The Logic of Collective Action: Public Goods and the Theory of Groups.
Cambridge: Harvard University Press.
Olzak, Susan, and Sarah A. Soule. 2009. “Cross-Cutting Influences of Environmental
Protest and Legislation.” Social Forces 88 (1): 201–225.
Opp, Karl-Dieter, and Bernhard Kittel. 2009. “The Dynamics of Political Protest: Feedback Effects and Interdependence in the Explanation of Protest Participation.” European
Sociological Review 26 (1): 97–109.
Opp, Karl-Dieter, and Wolfgang Roehl. 1990. “Repression, Micromobilization, and Political Protest.” Social Forces 69 (2): 521–547.
Ostrom, Elinor. 1998. “A behavioral approach to the rational choice theory of collective
action: Presidential address, American Political Science Association, 1997.” American
Political Science Review 92 (1): 1–22.
Page, Benjamin I., and Robert Y. Shapiro. 1983. “Effects of Public Opinion on Policy.”
American Political Science Review 77 (1): 175–190.
Page, Benjamin I., and Robert Y. Shapiro. 1992. The Rational Public - Fifty Years of Trends in
Americans’ Policy Preferences. Chicago and London: The University of Chicago Press.
Pelizzo, Riccardo. 2003. “Party Positions or Party Direction? An Analysis of Party Manifesto Data.” West European Politics 26 (2): 67–89.
Pennings, Paul, and Hans Keman. 2002. “Towards a New Methodology of Estimating
Party Policy Positions.” Quality & Quantity 36 (1): 55–79.
Pétry, François, and Benoît Collette. 2009. “Measuring How Political Parties keep their
Promises: a Positive Perspective from Political Science.” In Do They Walk Like They Talk?,
ed. Louis M. Imbeau. New York: Springer pp. 65–80.
Pierson, Paul. 2000. “Increasing Returns, Path Dependence, and the Study of Politics.”
American Political Science Review 94 (2): 251–267.
202
bibliography
Pitkin, Hanna F. 1967. The Concept of Representation. Berkeley: University of California
Press.
Piven, Frances F., and Richard A. Cloward. 1977. Poor People’s Movement: Why They Succeed,
Why They Fail. New York: Pantheon.
Plümper, Thomas, Vera E. Troeger, and Philip Manow. 2005. “Panel Data Analysis in
Comparative Politics: Linking Method to Theory.” European Journal of Political Research
44 (2): 327–354.
Poguntke, Thomas, and Paul Webb. 2005. The Presidentialization of Politics - A Comparative
Study of Modern Democracies. Oxford: Oxford University Press.
Poole, Keith T., and Howard Rosenthal. 1985. “A Spatial Model for Legislative Roll Call
Analysis.” American Journal of Political Science 29 (2): 357–384.
Powell, G. Bingham Jr. 2000. Elections as Instruments of Democracy: Majoritarian and Proportional Visions. Yale: Yale University Press.
Powell, G. Bingham Jr. 2004. “Political Representation in Comparative Politics.” Annual
Review of Political Science 7 (1): 273–296.
Ragsdale, Lyn. 1984. “The Politics of Presidential Speechmaking, 1949-1980.” American
Political Science Review 78 (4): 971–984.
Rainey, Carlisle. 2014. “Arguing for a Negligible Effect.” American Journal of Political Science
58 (4): 1083–1091.
Ramana, M. V. 2013. “Nuclear Policy Responses to Fukushima: Exit, Voice, and Loyalty.”
Bulletin of the Atomic Scientists 69 (2): 66–76.
Reising, Uwe K. H. 1999. “United in Opposition? A Cross-National Time-Series Analysis of
European Protest in Three Selected Countries, 1980-1995.” Journal of Conflict Resolution
33 (3): 317–342.
Riker, William H. 1996. The Strategy of Rhetorics. New Haven & London: Yale University
Press.
Rokkan, Stein. 1970. Citizens, Elections, Parties: Approaches to the Comparative Study of the
Processes of Development. New York: McKay.
Rose, Richard. 1984. Do Parties Make a Difference? Basingstoke: Macmillan.
Rottinghaus, Brandon. 2008. “Rethinking Presidential Responsiveness: The Public Presidency and Rhetorical Congruency, 1953-2001.” Journal of Politics 68 (03): 720–732.
Rovny, Jan. 2012a. “Where do Radical Right Parties Stand? Position Blurring in Multidimensional Competition.” European Political Science Review 5 (01): 1–26.
203
bibliography
Rovny, Jan. 2012b. “Who Emphasizes and Who Blurs? Party Strategies in Multidimensional Competition.” European Union Politics 13 (2): 269–292.
Rubin, Donald B. 1987. Multiple Imputation for Nonresponse in Surveys. New York: Wiley.
Rucht, Dieter. 1996. “The Impact of National Contexts on Social Movement Structures:
A Cross-Movement and Cross-National Comparison.” In Comparative Perspectives on Social Movements: Political Opportunities, Mobilizing Structures, and Cultural Framings, ed.
Doug McAdam, John D. McCarthy, and Mayer N. Zald. Cambridge, UK: Cambridge
University Press pp. 185–204.
Russo, Federico, and Matti Wiberg. 2010. “Parliamentary Questioning in 17 European
Parliaments: Some Steps towards Comparison.” Journal of Legislative Studies 16 (2): 215–
232.
Saalfeld, Thomas. 2011. “Parliamentary Questions as Instruments of Substantive Representation: Visible Minorities in the UK House of Commons, 2005-2010.” Journal of Legislative
Studies 17 (3): 271–289.
Saalfeld, Thomas, and Daniel Bischof. 2013. “Minority-Ethnic MPs and the Substantive
Representation of Minority Interests in the House of Commons, 2005-2011.” Parliamentary Affairs 66 (2): 305–328.
Samuels, David J. 2002. “Pork Barreling Is Not Credit Claiming or Advertising: Campaign
Finance and the Sources of the Personal Vote in Brazil.” Journal of Politics 64 (03): 845–
863.
Sartori, Giovanni. 1976. Parties and Party Systems: A Framework for Analysis. Cambridge,
UK: Cambridge University Press.
Schattschneider, Elmer E. 1942. Party Government. New York: Rinehart.
Schedler, Andreas. 1998. “The Normative Force of Electoral Promises.” Journal of Theoretical
Politics 10 (2): 191–214.
Schmitt-Beck, Rüdiger. 2003. “Mass Communication, Personal Communication and Vote
Choice: The Filter Hypothesis of Media Influence in Comparative Perspective.” British
Journal of Political Science 33 (02): 233–259.
Schumaker, Paul D. 1975. “Policy Responsiveness to Protest-Group Demands.” Journal of
Politics 37 (2): 488–521.
Schumpeter, Joseph A. 2003. Capitalism, Socialism & Democracy. 5 ed. London & New York:
Routledge.
Searing, Donald H. 1994. Westminster’s World. Understanding Political Roles. Cambridge:
Harvard University Press.
204
bibliography
Senninger, Roman, and Markus Wagner. 2015. “Political Parties and the EU in National
Election Campaigns: Who Talks about Europe, and How?” JCMS: Journal of Common
Market Studies 53 (6): 1336–1351.
Shepsle, Kenneth A., Robert P. Van Houweling, Samuel J. Abrams, and Peter C. Hanson.
2009. “The Senate Electoral Cycle and Bicameral Appropriations Politics.” American
Journal of Political Science 53 (2): 343–359.
Slapin, Jonathan B., and Sven-Oliver Proksch. 2008. “A Scaling Model for Estimating TimeSeries Party Positions from Texts.” American Journal of Political Science 1 (3): 323–344.
Smith, Jackie, John D. McCarthy, Clark McPhail, and Boguslaw Augustyn. 2001. “From
Protest to Agenda Building: Description Bias in Media Coverage of Protest Events in
Washington, D.C.” Social Forces 79 (4): 1397–1423.
Soroka, Stuart N. 2002a. Agenda-Setting Dynamics in Canada. Vancouver, CA: UBC Press.
Soroka, Stuart N. 2002b. “Issue Attributes and Agenda-setting by Media, the Public, and
Policymakers in Canada.” International Journal of Public Opinion Research 14 (3): 265–285.
Soroka, Stuart N. 2012. “The Gatekeeping Function: Distributions of Information in Media
and the Real World.” Journal of Politics 74 (02): 514–528.
Soroka, Stuart N., and Christopher Wlezien. 2004. “Opinion Representation and Policy
Feedback: Canada in Comparative Perspective.” Canadian Journal of Political Science 37
(3): 531–559.
Soroka, Stuart N., and Christopher Wlezien. 2010. Degrees of Democracy - Politics, Public
Opinion, and Policy. Cambridge: Cambridge University Press.
Soroka, Stuart N., and Christopher Wlezien. 2012. “Political Institutions and the OpinionPolicy Link.” West European Politics 35 (6): 1407–1432.
Soule, Sarah A., and Brayden G. King. 2006. “The Stages of the Policy Process and the
Equal Rights Amendment, 1972-1982.” American Journal of Sociology 111 (6): 1871–1909.
Soule, Sarah A., Doug McAdam, John McCarthy, and Yang Su. 1999. “Protest Events:
Cause or Consequence of State Action? The US Women’s Movement and Federal Congressional Activities, 1956-1979.” Mobilization: An International Quarterly 4 (2): 239–256.
Soule, Sarah A., and Susan Olzak. 2004. “When Do Movements Matter? The Politics of
Contingency and the Equal Rights Amendment.” American Sociological Review 69 (4):
473–497.
Spoon, Jae-Jae. 2011. Political Survival of Small Parties in Europe. Ann Arbor: University of
Michigan Press.
205
bibliography
Spoon, Jae-Jae. 2012. “How salient is Europe? An analysis of European election manifestos,
1979-2004.” European Union Politics 13 (4): 558–579.
Spoon, Jae-Jae, and Heike Klüver. 2014. “Do Parties Respond? How Electoral Context
Influences Party Responsiveness.” Electoral Studies 35: 48–60.
Spoon, Jae-Jae, and Heike Klüver. 2015. “Voter Polarisation and Party Responsiveness:
Why Parties Emphasise Divided Issues, but Remain Silent on Unified Issues.” European
Journal of Political Research 54 (2): 343–362.
Spoon, Jae-Jae, Sara B. Hobolt, and Catherine E. de Vries. 2014. “Going Green: Explaining
Issue Competition on the Environment.” European Journal of Political Research 53 (2): 363–
380.
Steenbergen, Marco R., and Bradford S. Jones. 2002. “Modeling Multilevel Data Structures.” American Journal of Political Science 46 (1): 218–237.
Stewart, Mark. 2006. “Maximum Simulated Likelihood Estimation of Random-Effects Dynamic Probit Models with Autocorrelated Errors.” Stata Journal 6 (2): 256–272.
Stimson, James A. 1991. Public Opinion in America: Moods, Cycles, and Swings. Boulder:
Westview.
Stimson, James A., Michael B. Mackuen, and Robert S. Erikson. 1995. “Dynamic Representation.” American Political Science Review 89 (3): 543–565.
Stoll, Heather. 2010. “Elite-Level Conflict Salience and Dimensionality in Western Europe:
Concepts and Empirical Findings.” West European Politics 33 (3): 445–473.
Strøm, Kaare. 1990. “A Behavioral Theory of Competitive Political Parties.” American
Journal of Political Science 34 (2): 565–598.
Strøm, Kaare. 2000. “Delegation and Accountability in Parliamentary Democracies.” European Journal of Political Research 37 (3): 261–289.
Strøm, Kaare, Wolfgang C. Müller, and Torbjörn Bergman. 2003. Delegation and Accountability in Parliamentary Democracies. Oxford, UK: Oxford University Press.
Swyngedouw, Marc. N.d. “Nuclear Energy Politics in Belgium: Big Business and Politics.”
In Phasing-out and Phasing-in: The Comparative Politics and Policies of Nuclear Energy in
Europe, ed. Wolfgang C. Müller, and Paul W. Thurner. Oxford, UK: Oxford University
Press p. (unpublished manuscript).
Tarrow, Sidney. 1994. Power in Movement: Social Movements, Collective Action and Politics.
Cambridge: Cambridge University Press.
Tavits, Margit, and Joshua D. Potter. 2015. “The Effect of Inequality and Social Identity on
Party Strategies.” American Journal of Political Science 59 (3): 744–758.
206
bibliography
Thomas, Martin. 1985. “Election Proximity and Senatorial Roll Call Voting.” American
Journal of Political Science 29 (1): 96–111.
Thomson, Robert. 2001. “The Programme to Policy Linkage: The Fulfilment of Election
Pledges on Socio-Economic Policy in The Netherlands, 1986-1998.” European Journal of
Political Research 40 (2): 171–197.
Tilly, Charles. 1978. From Mobilization to Revolution. New York: Random House.
Tresch, Anke, Jonas Lefevere, and Stefaan Walgrave. 2015. “’Steal Me if You Can!’: The
Impact of Campaign Messages on Associative Issue Ownership.” Party Politics 21 (2):
198–208.
Tsebelis, George. 1995. “Decision Making in Political Systems: Veto Players in Presidentialism, Parliamentarism, Multicameralism and Multipartyism.” British Journal of Political
Science 25 (03): 289–325.
Tsebelis, George. 2002. Veto Player: How Political Institutions Work. Princeton: Princeton
University Press.
Tucker, Joshua A. 2007. “Enough! Electoral Fraud, Collective Action Problems, and PostCommunist Colored Revolutions.” Perspectives on Politics 5 (3): 535–551.
Tufte, Edward R. 1978. Political Control of the Economy. Princeton: Princeton University
Press.
Tulis, Jeffery. 1987. The Rhetorical Presidency. Princeton: Princeton University Press.
Uba, Katrin. 2009. “The Contextual Dependence of Movement Outcomes: A Simplified
Meta-Analysis.” Mobilization: An International Quarterly 14 (4): 433–448.
Van Aelst, Peter, and Stefaan Walgrave. 2001. “Who is that (Wo)Man in the Street? From
the Normalisation of Protest to the Normalisation of the Protester.” European Journal of
Political Research 39 (4): 461–486.
Verhulst, Joris, and Stefaan Walgrave. 2009. “The First Time is the Hardest? A CrossNational and Cross-Issue Comparison of First-Time Protest Participant.” Political Behavior 31 (3): 455–484.
Vieira, Mónica B., and David Runciman. 2008. Representation - Key Concepts. Cambridge:
Polity Press.
Vliegenthart, Rens, Stefaan Walgrave, and Brandon Zicha. 2013. “How Preferences, Information and Institutions Interactively Drive Agenda-Setting: Questions in the Belgian
Parliament, 1993-2000.” European Journal of Political Research 52 (3): 390–418.
Volkens, Andrea, Onawa Lacewell, Pola R. Lehmann, Sven Schultze, and Annika Henrike
Werner. 2012. “The Manifesto Data Collection.” Manifesto Project (MRG/CMP/MARPOR) Berlin: Wissenschaftszentrum Berlin für Sozialforschung (WZB).
207
bibliography
Walgrave, Stefaan. 2008. “Again, the Almighty Mass Media? The Media’s Political AgendaSetting Power According to Politicians and Journalists in Belgium.” Political Communication 25 (4): 445–459.
Walgrave, Stefaan, and Dieter Rucht, eds. 2003. The World Says No to War: Demonstrations
Against the War on Iraq. Minneapolis: University of Minnesota Press.
Walgrave, Stefaan, and Jonas Lefevere. 2013. “Ideology, Salience, and Complexity: Determinants of Policy Issue Incongruence between Voters and Parties.” Journal of Elections,
Public Opinion & Parties 23 (4): 456–483.
Walgrave, Stefaan, Jonas Lefevere, and Anke Tresch. 2012. “The Associative Dimension of
Issue Ownership.” Public Opinion Quarterly 76 (4): 771–782.
Walgrave, Stefaan, Jonas Lefevere, and Michiel Nuytemans. 2009. “Issue Ownership Stability and Change: How Political Parties Claim and Maintain Issues Through Media
Appearances.” Political Communication 26 (2): 153–172.
Walgrave, Stefaan, and Michiel Nuytemans. 2009. “Friction and Party Manifesto Change
in 25 Countries, 1945-98.” American Journal of Political Science 53 (1): 190–206.
Walgrave, Stefaan, and Rens Vliegenthart. 2012. “The Complex Agenda-Setting Power of
Protest: Demonstrations, Media, Parliament, Government, and Legislation in Belgium,
1993-2000.” Mobilization: An International Quarterly 17 (2): 129–156.
Walgrave, Stefaan, Stuart N. Soroka, and Michiel Nuytemans. 2007. “The Mass Media’s
Political Agenda-Setting Power: A Longitudinal Analysis of Media, Parliament, and
Government in Belgium (1993 to 2000).” Comparative Political Studies 41 (6): 814–836.
Ward, Dalston, Jeong H. Kim, Matthew Graham, and Margit Tavits. 2015. “How Economic
Integration Affects Party Issue Emphases.” Comparative Political Studies 48 (10): 1227–
1259.
Weik, Martin. 2001. Computer Science and Communications Dictionary. Dordrecht: Kluwer
Academic Publishers.
Weissberg, Robert. 1978. “Collective vs. Dyadic Representation in Congress.” American
Political Science Review 72 (2): 535–547.
Wessels, Bernhard. 1999. “System Characteristics Matter: Empirical Evidence from Ten
Representation Studies.” In Policy Representation in Western Democracies, ed. Warren E.
Miller, Roy Pierce, Jacques Thomassen, Richard Herrera, Sören Holmberg, Peter Esaiasson, and Bernhard Wessels. Oxford: Oxford University Press pp. 1137–1161.
Wlezien, Christopher. 1995. “The Public as Thermostat: Dynamics of Preferences for
Spending.” American Journal of Political Science 39 (4): 981–1000.
208
bibliography
Woldendorp, Jaap, Hans Keman, and Ian Budge. 2000. Party Government in 48 Democracies.
Composition - Duration - Personnel. Dordrecht etc.: Kluwer Academic Publishers.
Wooldridge, Jeffrey M. 2013. Introductory Econometrics: A Modern Approach. Mason, OH:
South-Western.
Woon, Jonathan. 2012. “Democratic Accountability and Retrospective Voting: A Laboratory Experiment.” American Journal of Political Science 56 (4): 913–930.
Zald, Mayer N., and Roberta Ash. 1966. “Social Movement Organizations: Growth, Decay
and Change.” Social Forces 44 (3): 327–341.
Zaller, John R. 1992. The Nature and Origins of Mass Opinion. New York, NY: Cambridge
University Press.
Zeger, Scott L., Kung-yee Y. Liang, and Paul S. Albert. 1988. “Models for longitudinal data:
A generalized estimating equation approach.” Biometrics 44 (4): 1049–1060.
Zorn, Christopher J.W. 2001. “Generalized Estimating Equation Models for Correlated
Data: A Review with Applications.” American Journal of Political Science 45 (2): 470–490.
209