Content uploaded by Swen Hutter
Author content
All content in this area was uploaded by Swen Hutter on Sep 09, 2016
Content may be subject to copyright.
1
Chapter 2: Exploring the new cleavage across arenas and public debates:
design and methods
Martin Dolezal, Swen Hutter, and Bruno Wüest
Dolezal, Martin, Swen Hutter, and Bruno Wüest. 2012. "Exploring the new cleavage across arenas and
public debates: design and methods." in Political conflict Western Europe, edited by Hanspeter
Kriesi, Edgar Grande, Martin Dolezal, Marc Helbling, Dominic Hoeglinger, Swen Hutter, and Bruno
Wüest. Cambridge: Cambridge University Press.
Introduction
This volume is a continuation and extension of our previous study on West European politics
in the age of globalization (Kriesi et al. 2008). Aspects of continuity prevail with respect to
the main theoretical questions introduced in Chapter 1, but we now analyze how the
integration-demarcation cleavage manifests itself in various political arenas. We extend the
research design as well as the empirical program to include elections to the European
parliament and non-institutionalized forms of political participation, hence political protest, in
the analysis. Furthermore, we have updated our data on national elections, and scrutinize
public debates about the three issues central to globalization processes: immigration,
economic liberalization, and European integration.
We primarily deal with the programs political and other actors offer, hence with the
supply side; our analysis of citizens’ attitudes is restricted to Chapter 3. As we rely on
secondary data sources and common statistical procedures to explore demand, the chapter at
hand focuses on the most important aspects of how we study the supply side of political
competition. However, the expression of protest transcends this dichotomy somewhat, as our
analysis of political protest is neither wholly supply- nor entirely demand-oriented.
In the following, we explain the various steps of our data collection, from the selection
of countries and time-periods to our method for coding political statements (Dolezal 2008:
2
54f.). More attention is paid to the methodological challenges we faced in analyzing the new
conflict across political arenas and public debates. We also discuss our data aggregation and
statistical procedures. Our previous study and approach received some attention in the
scientific community, so we use the opportunity here to answer some of the objections our
critics have raised.
Selection of countries and time periods
The selection of countries reflects continuity from our previous study, so the integration-
demarcation cleavage is analyzed in six West European countries: Austria, France, Germany,
the Netherlands, Switzerland, and the UK. Since the degree and timing of political change that
we regard as consequences of globalization depend on contextual factors that are nation-
specific, our comparative analysis controls for explanatory factors at the national level. The
six countries selected are similar in many respects, so we follow a most similar systems design
(Przeworski and Teune 1970). All six are stable liberal democracies with consolidated
political institutions and party systems, and all are among the economically most developed
and richest countries in the world. More generally, societal conditions – by which we mean
cleavage structures and economic and cultural contexts – have created broadly similar latent
potentials for manifestations of the new cleavage. Still, the political conditions for mobilizing
these potentials vary considerably from country to country (see Chapter 1).
We excluded East European countries since their democratization began just when
contemporary globalization was accelerating (Dolezal 2008: 54). In the late 1980s and early
1990s, East European party systems were in transition and voter alignments fluid, making
comparison with established liberal democracies difficult. We discussed including Italy in our
sample but eventually decided against it because political change there, meaning the
3
breakdown of the party system of the First Republic during the early 1990s, was attributable
more to national idiosyncrasies than to globalization pressures (e.g., Forno 2003; Newell
2000: 177-178).
1
Other South European countries, notably Spain, only democratized during
the 1970s, again making comparison over time problematic. Scandinavian countries, finally,
were not included in our sample because we lacked the necessary linguistic skill to gather data
from content analyses of print media products.
Our first volume primarily covered the 1990s. The present study focuses on the 2000s,
particularly the political debates over the central issues of globalization. Our longitudinal
comparison of national electoral arenas runs from the 1970s to the mid-2000s, based on our
old data supplemented by the most recent election data for each of our six countries (Chapter
4). For European elections, we compare the election of 1994 with that of 2004 (Chapter 5).
The analysis of political protest events is also longitudinal (mid-1970s to mid-2000s; see
Chapter 6) as is the analysis of the demand side (Chapter 3).
The analyses of debates on issues we regard as central to globalization processes shift
from a combined longitudinal and cross-sectional to a purely cross-sectional perspective.
More specifically, we examine public debates in all six countries, but only during the period
from 2004 to 2006. Because the new cleavage has grown in importance since the 1990s,
choosing a recent time period ensured we could find enough statements on immigration,
economic liberalization, and European integration, the three issues covered in this analysis.
While issue focus and time frame restrict the analysis of public debates, our principal aim is to
form a more comprehensive picture of the structure and constellations in the current
integration-demarcation divide. Looking at public debates will allow for a wider and deeper
focus on other arenas (the public authority, the direct democratic arena) that are not covered
1
Data for parts of the demand analysis was also not available for Italy.
4
in the first part of the present volume but where the integration-demarcation cleavage might
be articulated as well. Exceptional events might distort this overall picture, however, so to test
for this potential weakness, we compare our results from the immigration debate with an
earlier, comparable time period before 9/11 (1999 - 2001). Table 2.1 shows the time periods
investigated in the various arenas and debates as well as the elections observed.
[Table 2.1]
Data collection: sampling and coding newspaper articles
The basic methodological choice was to explore political conflict based on mass media
reports. Research on the transformation of national political competition routinely notes the
importance of the media for political opinion-making and decision-making processes (Bennett
et al. 2004; Ferree et al. 2002; Swanson and Mancini 1996). Political competition is
increasingly transferred “from the backrooms of parliamentary committees and the central
offices of parties and associations to the public sphere” (Kriesi 2004: 184). Hence, virtually
all political actors try to gain public support through the mass media, and political statements
mediated by journalists receive much public attention (Schmidt-Beck and Farrell 2008: 15).
Our data comes from newspaper content rather than from television since newspapers – and
above all the quality press – report more extensively on political matters (e.g., Druckman
2005: 469; Neuman et al. 1992: 50).
Our research strategy therefore differs from the two dominant approaches in the
literature on the programmatic supply of parties: expert surveys (Benoit and Laver 2006;
Castles and Mair 1984; Laver and Hunt 1992) and the analysis of party manifestos (especially
the Comparative Manifestos Project; see Budge et al. 2001; Klingemann et al. 2006). While
both methodological approaches have their merits, we think our strategy has some important
5
advantages for understanding political competition, political conflict and what political parties
supply by way of their statements in mass media.
Selection of newspapers
For each country, we chose one quality newspaper (see Table 2.2). We added the most widely
read tabloid in each country to analyze national election campaigns, but did not do so for
analyzing the debates and the European election campaigns. This decision was not motivated
only by pragmatic considerations regarding the work load involved, but also because quality
newspapers remain the leading medium of political coverage. They are particularly suited for
studying political contests since they both mirror the debates in a detailed manner and
influence the editorial decisions of a wide range of news organizations (e.g., Reinemann
2003; Vliegenthart and Walgrave 2008). We chose newspapers published throughout the
period covered by our study. To update and extend existing datasets on protest politics
particularly for the UK and German protest arenas, we had to draw on additional quality
newspapers (Koopmans 1996; Kriesi et al. 1995).
[Table 2.2]
Comparative content analysis of newspapers involves enormous effort, and
appropriate sampling and coding strategies are crucial. Our coding of election campaigns and
public debates also differed from our coding for protest politics. We applied the core sentence
approach in content analysis to analyze national and European elections as well as debates.
Protest event analysis was used for gathering data on the occurrence and above all the content
6
of protest activities in our six countries.
2
The use of the latter analysis was also dictated by
our updating and extending existing data sets.
The core sentence approach
The identifying of actors and their issue-specific positions was done sentence by sentence,
using a relational method of analysis. This approach is inspired by Wittgenstein (1984 [1921])
and was first transformed into practical coding instructions by Osgood (1959) and later by
Axelrod (1976). It has been adapted for analyzing political competition (Kleinnijenhuis et al.
1997; Kleinnijenhuis and Pennings 2001). The core sentence method is designed to code
every relationship between ‘political objects’ that appears in a text, thus either between two
political actors or between a political actor and a political issue. But in the book at hand, our
interest is exclusively in the relationships between actors and political issues (i.e. ‘actor-issue
sentences’). In this method, each sentence of an article is reduced to its most basic ‘core
sentence’ that contains only the subject (the actor), the object (an issue) and the direction of
the relationship between the two. The number of core sentences in an article does not equal
the number of grammatical sentences, as one sentence can include none, one or several core
sentences. The following example contains two actor-issue core sentences within one
grammatical sentence:
“The pro-European Lib Dems tend to do less well in European elections, but, as with
some smaller parties they may have gained from opposition to the Iraq war.” (The
Times, 8 June 2004, Labour and Tories hit by fringe parties).
2
Survey data is the primary source for political participation research (Chapter 6). But protest event
analysis is far better suited to measure both protest potential and actual protest mobilization as well as the
issues that give rise to protest politics – the most important feature for our present study.
7
The direction between subject and object is always quantified using a scale ranging from -1 to
+1, with three intermediary positions indicating a ‘potential’ or an ambiguous relation. If, for
example, a politician says that in the future he might be in favor of a certain position, we
coded 0.5. Ambiguous relations – no direction at all – were coded 0. In the example above,
the first core sentence refers to the pro-European stance of the Liberal Democrats; the second
to their opposition against the Iraq war. It would therefore be coded as both Liberal
Democrats/+1/European integration and as Liberal Democrats/-1/Iraq War.
By means of this method, we can measure both the positions actors take and the
salience they attribute to certain issues. Positions taken towards a certain category of issues
are calculated by taking the mean value of the direction in all relevant observations. The
salience of a set of issues refers to the number of statements an actor makes to a certain
category relative to all of his or her statements. Core sentences are therefore an inductive
mean for capturing the full complexity of political statements without imposing strong
theoretical expectations, such as a priori categories. Franzosi (2004) and van Atteveldt (2008)
have recently provided theoretical and empirical evidence that relational content analysis is a
useful device for the social sciences in general. Neither examining party manifestos nor
surveying experts yields such precise data both on actors’ position and issue salience.
For analyzing public debates, we examined two additional attributes of the debates:
the frames and the political arenas (the theoretical basis, analytical categories, and concept of
arenas and framing are discussed in detail in the Chapters 8 and 9). By frames, we mean the
problem definition or justification an actor gives when he takes a position. Because other
research indicates actors back their policy positions in public debates using multiple frames
(Lerch and Schwellnus 2006: 307), we allowed for more than one frame per core sentence,
and coded up to five frames for each core sentence. Most studies of media framing focus on
an entire article (de Vreese 2003: 910; Trenz 2005: 849), but we coded frames on a
8
propositional level, and more specifically, only on those statements clearly attributable to an
actor and a position. We are thus able to identify the responsible political actor for each frame
occurring in a newspaper article. The political arena is the general context to which an issue
position refers and in which it is articulated. More specifically, the arena is a site of
structuration whose institutional rules and norms shape the articulation of political conflict.
Examples include an election campaign or a demonstration.
A major disadvantage of a manual content analysis like this is the enormous amount of
work it involves, though a series of computer-based methods have been developed recently to
address this. A common method is to code articles according to only one specific variable,
such as with respect to a left-right ideological scale or a particular set of issues. Methods may
rely on the comparison of relative word frequencies in texts (Hillard et al. 2007; Laver et al.
2003; Zuell and Landmann 2005), the co-occurrence of keywords (Ruigrok and van Atteveldt
2007), or on the presence or absence of a word stem (Hopkins and King 2007). For our
purposes, the disadvantage of these methods is that the type of coding does not fit the
demands of our methodological conceptualisation, especially because they cannot identify
relationships between actors and issues in a single document (Hopkins and King 2007: 5).
More suitable approaches using syntactic parsing and Named Entity Recognition to
automatically code core sentences are still in the experimental stage (van Atteveldt 2008), so
we had to rely on manual coding.
3
Sampling strategies are necessary due to the workload that comes with the core
sentence approach. As in our previous study, the sampling period for national elections was
the two-month period up to the day a national election was held. We selected all articles – but
3
During the data-gathering process, we developed our own coding software framework, including a web
application with centralized data management and intelligent annotation. A software pipeline was built for
the automatic pre-coding of core sentences but it thus far is not at a stage to be applied in practical
research. All the software is open source.
9
no commentaries – related to the electoral contest or politics in general that were published on
Mondays, Wednesdays, and Saturdays in the quality newspapers.
4
As tabloids have far fewer
and smaller articles, we took every day. For European elections, we selected articles again
within the two-month period beforehand, though with the help of an automatic system based
on electronic newspaper archives – which are becoming increasingly available at least for
quality newspapers. Given the specific character of electoral campaigns for the European
parliament, we did not select all articles on politics but used search strings for specific terms
(details are discussed in Chapter 5).
The selection of articles for analyzing the debates on issues related to globalization
was done in a unique manner. First, we identified the relevant events in each country through
yearbooks (Keesing's World Record of Events, Facts on File World News Digest Yearbook)
as well as the annual reviews of the newspapers in our sample. These then formed the basis
for an extensive keyword list for each country, helping us to electronically find potentially
relevant articles in the newspaper databases.
5
The advantage of creating such event lists is that
we knew about the content of the relevant discussions in each country before we engaged in a
keyword search. We could therefore avoid many false negative selections because we could
adapt the keyword lists to the country-specific characteristics.
6
Second, we chronologically
took a random sample of 1,200 articles per country from the list of sub-issues. This helped us
to account for the dynamic of the debate. Time-invariant selection procedures, such as taking
all articles about a political issue published on a certain weekday, fail to capture differences in
4
Dutch newspapers were selected differently, since the Dutch data stem from an affiliated project in
which sampling was not used (Kleinjienhuis et al. 2007). For the newest elections in France and
Switzerland, we drew a chronological sample that yielded approximately the same number of relevant
articles as the selection procedures used for other election campaigns.
5
All newspapers in our sample are accessible via the Factiva or LexisNexis databases; CD-ROMs are
available for some of them (NZZ, Le Monde) as well.
6
While this procedure yielded a manageable number of articles for the debates on immigration and
economic liberalization, we had to run an additional selection for the debate on European integration to
include only the most relevant issues.
10
the intensity of a debate. Our chronological method, by contrast, tracks the frequency
distribution of relevant articles and therefore captures the peaks and troughs in the debate.
Given the time-consuming coding procedure we not only had to sample the articles but
usually did not code the whole articles selected. For national election articles, we coded the
headline, the ‘lead’, if available, and the first paragraph of the article. As tabloid articles are
shorter and fewer in numbers, we decided to code the entire articles. For the European
elections and the debates, we coded at most the first twenty core sentences; this was because
the format of the articles (such as the paragraph structure) is sometimes lost in the electronic
archives we relied on.
Other problems for comparing content are created by the differences between
newspapers and by coder disagreement. There is genuine variation in the way newspapers
report on political competition. Nevertheless, relying on expert judgments, party manifesto
data, and mass survey results for validation suggests our method yields valid data. As
Helbling and Tresch (2009) show, various indicators for party positions are highly correlated.
Moreover, if one uses aggregated measures like average position or relative frequencies as we
do, such differences carry less weight. Coder disagreement is potentially also a serious
problem. In a pretest, we obtained a coefficient of reliability of 0.77
for coder agreement on
what the core sentences were. Inter-coder agreement was 0.88 for the actors and 0.85 for the
issues. Given that the typical level of acceptance for inter-coder reliability is 0.80 (Lombard et
al. 2002), our coder disagreement was within acceptable limits. Additional coder training,
refined coding instructions and a continuous monitoring of the coders during the coding
process were also provided in order to address remaining uncertainties.
7
7
Most of the coding for individual countries was divided among several coders to reduce the influence of
any single coder on the results.
11
Coding protest politics
The study of protest politics, less institutionalized forms of political participation, was also
conducted by quantitative content analysis of media reports. But unlike for election
campaigns and debates, the coding of political protest was done at the level of events. This
methodological choice follows a long-standing tradition of research on social movements and
contentious politics (e.g., Kriesi et al. 1981; Olzak 1989; Tarrow 1989; Tilly 2008; Tilly et al.
1975). Protest event analysis aims at describing protest events so as to allow for cross-
sectional and longitudinal analyses. For these reasons, the definition of a protest event, the
data source, and the sampling strategy are of particular importance.
Our method here follows Kriesi et al. (1995). Research objects are chosen based on a
detailed and broad list of forms of action whose nature is protest. The list includes petitions
and political festivals as well as forms that are demonstrative (mass demonstrations),
confrontational (hunger strikes, occupations), and violent (physical attacks, arson).
8
Strikes,
however, are omitted, as we regard them as the main form of action for industrial conflicts
and not part of what we define as the protest politics arena (e.g., Ebbinghaus and Visser
2000).
The data itself comes from one national quality newspaper per country (see Table 2.2),
where we consulted the Monday edition. This choice was dictated not only by the necessity to
reduce the work of collecting a large number of events over a long period of time, but because
the Monday edition reports on events during the weekend. Since protest activities tend to be
concentrated on the weekend, our dataset includes a high proportion of all protest events
occurring during the period under study. We coded all protest events noted in the Monday
8
For a more detailed discussion of the dataset, see Koopmans (1995).
12
edition, including those that had taken place or took place one week before or after the
publication date. That is why around 25 percent of all coded events occurred during
weekdays.
It can be shown that our strategy yields valid and reliable data for the analyses we
carry out in the present volume. For example, Barranco and Wisler (1999) found that about
half of the public demonstrations in Swiss cities took place either on Saturday or Sunday, and
tests with continuous time data conducted by Koopmans (1995, 1998) for Germany and
Giugni (2004b) for the United States find similar patterns. Germany’s comprehensive ‘Prodat’
dataset that covers two national newspapers, all Monday issues, and all issues of every fourth
week (e.g., Rucht 2003; Rucht and Roth 2008) provides further confirmation. In general, the
results show that the national ebbs and flows of protest mobilization are traced accurately
with our sampling strategy (Koopmans 1998).
Protest event analysis generally, and Kriesi et al.’s (1995) sampling strategy more
specifically, have been the objects of critique in the literature (for reviews, see Earl et al.
2004; Koopmans and Rucht 2002; Ortiz et al. 2005).
9
One main objection is that such analysis
yields a (distorted) communicative rather than a social reality. This is not necessarily
problematic, though, as it is precisely the communicative reality that is relevant for the
population and the political decision-makers. With very few exceptions, both groups learn of
protest largely through the media rather than by personal observation. That fact led Gamson
and Wolfsfeld (1993: 116) to even state that “a demonstration with no media coverage at all is
a non-event”.
Furthermore, some distortion can be empirically assessed and used in interpreting the
results. According to the police, more than a thousand demonstrations take place in Paris
9
For a more detailed discussion based on further empirical tests of the selection bias, see Hutter (in
preparation).
13
every year (Tartakowsky 2004: 14), but only a fraction of them make the national news – or
are ultimately included in our dataset. Many studies have shown that large and violent events
are more likely to be reported than small and peaceful ones (e.g., Fillieule 1997; Hocke 1998;
McCarthy et al. 1996). Furthermore, we agree with Koopmans and Rucht (2002: 247), who
argue that: “for many analytical purposes, it is not so much the actual level of protest but its
composition and trends over time that is of interest”. Countering selection bias by using
additional sources to include more events does not necessarily lead to more reliable results
either. Additional sources, whether another newspaper or from outside the media, may well
not provide any more information than already present in the sample, or they may introduce
their own biases (e.g., Dolezal and Hutter 2007: 342; Myers and Schaefer Caniglia 2004:
536). A further bias is introduced by using Monday editions, since it leads to under-
representing groups such as workers, peasants, and students (Barranco and Wisler 1999) in
the sample. This needs to be taken into account when interpreting our results (see Chapters 6
and 7). Finally, selection bias may also arise because protest events occurring close to the
newspaper’s location are more likely to be reported (e.g., Ortiz et al. 2005). As we are mainly
interested in protests that make the ‘national’ news, we assessed the local bias of our sources,
and observed that our sampling strategy produces efficient and valid results (Hutter and
Giugni 2009: 438).
10
Concluding our brief discussion on the selection bias, we wan to stress that what is
most important for our argument is that the biases are consistent over time. Although some
authors find inconsistent patterns across short periods of a week or a month (e.g., Myers and
Schaefer Caniglia 2004; Swank 2000), most studies show that results tend to be stable
especially within individual newspapers and over longer periods of time (e.g., Barranco and
Wisler 1999; Earl et al. 2004; McCarthy et al. 1996; McCarthy et al. 2008). This is what we
10
Detailed results are available upon request.
14
are interested in and that is why, in the end, we side with Earl et al.’s (2004: 77) argument that
compared to other instruments for retrieving data, such as surveys, the quality standards of
protest event analysis score well.
Our research design resulted in a dataset of 19,182 protest events in our six countries,
from 1975 to 2005, which involved approximately 118 million participants.
11
Some of our
data stems from Kriesi et al.’s (1995) four-country study (covering 1975 to 1989), and we also
rely on updates for Switzerland from 1990 to 1999 (Giugni 2004a) and for the UK from 1975
to 1989 (Koopmans 1996). Most of the data, however, was collected as part of our present
project (see Figure 2.1). While methodology and key variables are the same for the different
datasets, there are three potential sources of difference. One is that the most recent data rely
on a keyword search in electronic archives for selecting relevant articles.
12
We used a very
comprehensive list of keywords to be more efficient, but also to be consistent with the
manually selected datasets. To meet objections (Maney and Oliver 2001), we also performed
comparability tests based on the 1993-1999 time period for Switzerland and for two years in
all the other countries (for the Swiss case, see Hutter and Giugni 2009: 439). Another source
of potential difference is that the datasets differ with respect to certain variables. However, the
main variables used in our analysis are included in all datasets and allow for a longitudinal
analysis from 1975 until 2005. The last source of potential difference is our use of an updated
list of issue categories for our data collection, particularly the inclusion of a new category for
11
As our research focuses on long-term trends and the national level, we excluded all events taking place
on East German and Northern Irish soil. There is also considerable variation by country in the number of
events coded: 5,346 West Germany; 5,107 France; 2,318 Switzerland; 2,063 Great Britain: 1,981 the
Netherlands; and 500 Austria. Where numbers of participants are missing, they have been replaced by the
national median of the number of participants for a given type of event (e.g. a demonstration) in that
country. Following Kriesi et al. (1995), events with more than one million participants are coded as
999,998 (N=13).
12
In case of the UK, we manually selected and coded all legends to pictures, because they were not
systematically included in the electronic text archive of The Guardian (Rootes 2003).
15
mobilization related to the global justice movement.
13
Based on a review of the relevant
literature, we identified and recoded some events in our dataset that belong to this new
category as well.
[Figure 2.1]
Data analysis strategies
Three aspects of our data analysis deserve more specific explanation. The first is our
categorization, and coding, of the universe of actors and issues.
14
The second is the measures
we use for position, salience, and conflict intensity. The third is the specific method we use,
multidimensional scaling. All other statistical procedures employed belong to the standard
approaches political scientists use. In these descriptions, we also answer the critics of our first
volume (van der Brug and van Spanje 2009).
Categorizing actors and issues
To make comparison possible between the six countries as well as over time, it was necessary
to group the many actors into larger categories. A general distinction can be made between
public authority actors, intermediary actors, and non-organized individuals (see Table 2.3).
The first category is characterized by its policy output orientation, and includes both domestic
and foreign state actors as well as supranational and international governmental actors.
13
Some observers draw attention to what they regard as the unique character of the global justice
movement by referring to it as global justice movements or as a ‘movement of movements’. Scholars
struggle to define what it is, and find it difficult to assign specific events to this heterogeneous movement
– or movements (della Porta 2007). We limit ourselves by only subsuming those events that criticize
neoliberal globalization and directly target an international organization or conference (meetings of the
WTO, G-8 and the WEF) under the heading of global justice (see Chapter 5).
14
The definition of frames and arenas, which are key concepts in the analyses of the debates are found in
Chapter 8.
16
Intermediary actors, by contrast, all represent societal or economic interests, hence include
political parties, trade unions, business organizations, and public interest groups. Most studies
of cleavages focus on political parties (Bartolini and Mair 1990), but as Bartolini (2005b: 25-
27) argued more recently, collective groups such as unions, interest groups and social
movements also play important roles. In the category of the non-organized individuals,
experts are the only theoretically relevant and empirically salient group in our data.
[Table 2.3]
Still, this categorization does not fully serve our analytic needs. On the one hand,
public interest groups are far too important to protest politics and the debate on immigration
to be treated as a single category, and hence needed to be disaggregated (see Chapter 8). On
the other hand, the variety of political parties across six countries was too great to disc each
individually, and hence needed to be aggregated into six party families: Communists and Left
Socialists, Social Democrats, Greens, Liberals, Christian Democrats and Conservatives,
Populist and Radical Right. Parties too difficult to categorize and especially certain protest
parties that compete primarily in European elections are classified as ‘other’. Table 2.4 shows
the party classification and lists the most important parties in the six countries observed.
[Table 2.4]
Because we coded issue statements by actors and the main goal of protest events in
great detail, it was also necessary to group the issues into categories. Such categorization is
far more difficult than aggregating actors, and it has a crucial impact on our empirical results.
Here we use the twelve ‘meta-categories’ from our first volume (see Table 2.5) to try to
capture the universe of thematic conflicts articulated in the arenas, countries, and time
periods. While different issues, expressed in various ways, may come onto the political
agenda at any time, they nonetheless refer to a limited set of basic issues and positions. There
17
is also a technical argument: we need enough observations to conduct our analyses. Using the
‘cultural liberalism’ category as an example it becomes clear that no perfect way to regroup
the issues exists: This category primarily refers to the claims by the new social movements
(NSM), i.e. it relates to the new cultural conflicts of the 1970s and 1980s. We did, however,
not include environmental issues in this category because these issues are very salient and
would therefore dominate and perhaps even ‘distort’ the content of a potential ‘NSM
category’, as positions toward environmental protection are relatively centrist and empirically
belong to the economic conflict dimension. And finally, environmental issues are more
institutionalized in the political systems of our countries given the quite important role of
green parties in most cases.
As we specify the positions of actors on the basis of these twelve categories, it is
important that all meta-categories denote a particular direction either in support or in
opposition (see the description column in Table 2.5). More refined categories are used for
analyzing debates about immigration, Europe and economic liberalization, and are introduced
in the respective chapters, as are the modified categories we use to study protest politics.
[Table 2.5]
Measures: position, salience, and conflict intensity
A basic step in the analyses is to calculate the position of actors as well as several saliencies
with respect to the various political issues. The average issue position of an actor is calculated
by taking the mean of all the coded direction measures
over all core sentences (election
campaigns and debates) or events (protest politics) that state a relationship between this actor
and a given issue. As said before, positions range between -1 and 1. By the salience of an
issue for a given actor, we mean the relative frequency with which the actor takes a position
18
on this issue. For the analyses of the debates, we calculate furthermore actor saliencies,
defined as the relative frequency with which an actor is reported in the debate compared to all
actors in the debate.
As regards protest politics, we are not only interested in the salience of an issue for a
specific actor but far more in the salience of an issue in the protest arena as a whole. We use
two measures to establish this. One is to calculate an issue’s share of all coded protest events
in percent. Such relative values are more valid than absolute ones for comparative research
and come closer to our measures for electoral politics and debates (Koopmans 1995, 1998).
The second is to look at the number of participants a certain issue attracts. To compare the
number of participants cross-nationally, we standardized the absolute numbers of participants
by dividing them by the number of inhabitants (Kriesi and Duyvendak 1995: 22; Morales
2009: 143). The second measure is highly influenced by a few large protests.
15
The first
measure, by contrast, gives the same weight to all events and takes into account that some
protest events do not gain attention primarily due to the sheer number of people mobilized
(the ‘logic of numbers’). Rather, it is because they involve damage, or, in its most extreme
form, violence, or because it reflects strong commitment on the part of participants (della
Porta and Diani 2006: 170ff.).
To measure actor positions and issue saliencies, one can rely on documents (from the
media, from party manifestos), or on surveys (judgments by experts) (Keman 2007: 77).
While these sources of information may be valid, complementary approaches for
understanding actor positions and salience, preference for one or another strongly depends on
the research question (Helbling and Tresch 2009). One advantage in our relying on print
media-derived data, we argue, is that it captures the competition among, and confrontation
15
The five percent of all events that involve the most participants account for 77.2 percent of all protest
participants reported in the data set. National figures range from 62.9 (Switzerland) to 87.0 percent (Great
Britain).
19
between, actors better than other data does. The salience of an issue for a party in an electoral
contest, for example, heavily depends on the agenda-setting strategies of other parties, and the
actual strategy a party adopts can therefore deviate from the party’s manifesto. Another
argument is that actor positions on all issues in the political space are best measured by
content analysis, because the range of issues in expert surveys and in at least some party
manifestos often is constrained. Finally, it is almost impossible to find a comparative data set
which encompasses all actors considered in our study. Party manifesto data and expert
surveys, for example, are restricted to parties and ignore most actors in the protest arena.
We use different measures for protest politics on the one hand, and electoral
campaigns as well as public debates on the other to capture the intensity of conflict. For
protest politics, conflict intensity mainly refers to the number of participants and the
characteristics of the forms of action they use. In other words, we focus on how many people
‘enter’ the arena of protest politics to voice their grievances, as well as how radical their
action repertoire is. The average position, which we introduced above, is used as an additional
indicator of conflict intensity in the protest arena. The closer the average position raised in the
protest arena comes to the extreme values +1 or -1, the less contested an issue, because there
are almost no counter-positions voiced in the protest arena itself (see Chapter 6).
16
This does
not preclude that issues may be highly contested across arenas; protest events might address
positions or decisions taken in state or intermediary arenas, for example.
We use a more sophisticated measure of conflict intensity for analyzing election
campaigns and debates that captures the amount of conflict between the parties in a given
country over a specific issue. To do this, we use a measure of polarization (POL_1) of party
or actor positions based on Taylor and Hermann’s (1971) index. This index was originally
16
The reader should not that we here refer to the average issue position for a whole arena and not for a
single actor (e.g. a political party) in a specific arena.
20
designed to measure the degree of left-right polarization in a party system, but it can also be
used to indicate how strongly actor positions differ on any set of issues. For our purposes, it is
defined as
POL_1=
ω
k
x
k
−x
( )
2
k
=
1
K
∑
where
k
ω
is the salience of the issue category for party k,
k
x
is the position of party
k
on the
issue, and
x
is the weighted average position on this scale, where weights are provided by
the party-specific salience of the issue, that is:
K
k k
k
x x
ω
=∑
This is the index used for the polarization of the positions in a given issue category.
We also use this index to measure the degree of polarization on a group
of issues
(POL_2)
. To this end, we take the average of several Taylor and Herman indices, weighting
them by the salience of the corresponding issue category. This polarization index can thus be
defined as:
POL
_2
=
j=1
J
∑
ω
j
ω
kxjk
−
x
j
( )
2
k=1
K
∑
where
j
ω
is the salience of the
j
th
issue category (expressed as a proportion of the total salience
of all
j
issue categories),
k
ω
is the salience of the issue for party
k
,
jk
x
is the position of party
k
on issue
j
, and
x
j
is the average position on issue
j
, computed as
1
K
j k jk
k
x x
ω
=
=
∑
21
As in Taylor and Herman’s original index, the range of values taken depends on the scale of
the issue categories. As actor positions are always measured on scales ranging from -1 to +1,
the distance to the average can range between 0 and 1.
Multidimensional scaling: an empirical response to our critics
We use multidimensional scaling (MDS) techniques, as in our previous study, to uncover the
structure of the partisan space and the party configuration within that space. This will be
primarily done for national elections, though we use this technique for the analyses of actor-
issue relations in European elections and in public debates as well. MDS is a very flexible
method, quite similar to factor analysis, and allows for a graphic representation of similarities
or dissimilarities between pairs of objects. The unfolding technique we use here allows for the
joint representation of actors (e.g., parties) and issues in a common space. In addition, the
variant of MDS called weighted metric multidimensional scaling (WMMDS) enables us to
account simultaneously for similarities between pairs of objects (party positions with respect
to a set of issues in our case) and relationships (the salience of the respective issues for each
party and the salience of the different parties in the party system).
Based on data provided by two expert surveys of party positions on seven issues
17
, van
der Brug and van Spanje (2009) have taken issue with our previous presentation of the
transformation of national partisan spaces. They argue that the national partisan space in West
European countries is essentially one-dimensional, and corresponds to the traditional left-right
scale. They attribute the discrepancy between their findings and our previous results to the
peculiarity of our data sources. Since we base our analysis on the content of newspaper
articles, they maintain our results do not reflect a fundamental restructuring of party
17
Immigration, EU-enlargement, EU integration, taxes, left-right scale, the role of government in the
economy, and the role of government in life choices.
22
competition but only temporary changes in issues that happen to be on the political agenda at
the time of the campaigns.
The analysis in Chapter 4 suggests this reasoning is largely unfounded. It is no
accident that certain issues dominate the political agenda at a given point in time, especially
during the high political mobilization of a national election campaign, and our present
analysis is not dependent on the vicissitudes of single campaigns, at least as far as the 1990s
and 2000s are concerned. Moreover, the kind of data we use offers considerable advantages
over expert surveys, as it is not dependent on two single data points (as are the major surveys
by Laver and Hunt (1992) and by Benoit and Laver (2006)). Rather, our data is based on
actual campaigns where hundreds of issues are publicly debated. Furthermore, we not only
use the direction the party position themselves, but we also weight their issue-specific
positions with respect to the salience of the issues for the different parties in a given
campaign. Estimating the salience of an issue through expert surveys is more or less
impossible.
As it turns out, the dimensionality of the space depends on the inclusion of salience
measures – as calculated using WMMS – and on the number of issue categories included in
the analysis (the results can be found in Appendix A). If we follow the procedure our critics
suggest, meaning using party positions for only seven of our issues – the seven theoretically
most relevant ones – and submit them to an exploratory factor analysis, the space does indeed
turn out to be one-dimensional. But when we submit the reduced set of seven issues to the
MDS procedure we used for our analysis, meaning when we rely not just on the positioning of
the parties but also on the salience they attribute to the different issues, then two-dimensional
solutions turn out to be more consistent with the data, and in every decade. Moreover, when
we apply factor analyses to our full set of twelve meta-categories, the result turns out to be
23
two-dimensional for all three decades. This is true for exploratory as well as confirmatory
factor analyses.
18
Summary
Our basic methodological choice is to analyze the new integration-demarcation cleavage by
conducting content analyses of print media. We rely on secondary survey data to compare the
demand-side of the new cleavage across arenas only in Chapter 3. We not only analyze
national election campaigns (Kriesi et al. 2008) but also integrate analyses of European
elections, political protest and public debates. This extension is promising, as it covers various
forms of institutionalized and non-institutionalized strategies of mobilization; it is also
challenging to conduct such broad research. As in our previous volume, we follow a ‘most
similar systems’ design in analyzing six West European countries: Austria, France, Germany,
the Netherlands, and Switzerland, and the UK. Our study of national electoral and protest
politics extends back to the 1970s, whereas the analyses of European elections (1994 and
2004) and public debates (early 2000s) are restricted to more recent periods.
Our data collection focus is on the content of national newspapers. With respect to
election campaigns and public debates, we use the core sentence approach of media content
analysis: our study of protest politics, by contrast, is firmly in the tradition of social
movement research and relies on protest event analysis. The sampling strategies used for
coding election campaigns and protest events have been discussed in detail elsewhere
(Dolezal 2008; Koopmans 1995), so we highlighted our innovative approach for selecting
articles relevant to public debates. Our description of analysis strategies explained how we
18
The confirmatory factor analysis indicates that the two-dimensional solutions constitute significant
improvements over the one-dimensional solutions. In addition, it also provides theoretically meaningful
solutions, which is not the case for the exploratory solutions.
24
categorized the universe of actors and issues and introduced our most important measures:
position, salience, and conflict intensity. Finally, we presented the technique of multi-
dimensional scaling and we answered our critics who claimed that West European political
spaces essentially are one- and not two-dimensional.
The most challenging task of our project was to establish a research design to ensure
comparability of results across various arenas and debates. The following chapters provide
detailed methodological or technical description only as necessary, or when deviation from
the general design was unavoidable. The next chapter starts our empirical analyses by
updating and extending our previous study on the demand-side of West European politics.
25
References
Axelrod, Robert (1976).
Structure of Decision: The Cognitive Maps of Political Elites
.
Princeton: Princeton University Press.
Barranco, José and Dominique Wisler (1999). ‘Validity and systematicity of newspaper
data in event analysis’.
European Sociological Review
15(3): 301-322.
Bennett, W. Lance, Victor W. Pickard, David P. Iozzi, Carl L. Schroeder, Taso Lagos
and Evans C. Caswell (2004). ‘Managing the Public Sphere: Journalistic
Construction of the Great Globalization Debate’.
Journal of Communication
54(3): 437-455.
Benoit, Kenneth and Michael Laver (2006).
Party Policy in Modern Democracies
.
London: Routledge.
Budge, Ian, Hans-Dieter Klingemann, Andrea Volkens, Judith Bara and Eric
Tanenbaum (2001).
Mapping Policy Preferences. Estimates for Parties,
Electors, and Governments 1945-1998
. Oxford: Oxford University Press.
Castles, Francis G. and Peter Mair (1984). ‘Left-Right Political Scales: Some ‘Expert’
Judgments’.
European Journal of Political Research
12(73-88.
de Vreese, Claes H. (2003).
Framing Europe. Television news and Europen integration
.
Amsterdam: Aksant.
della Porta, Donatella (ed.) (2007).
The Global Justice Movement. Cross-National and
Transnational Perspectives
. Boulder: Paradigm Publishers.
della Porta, Donatella and Mario Diani (2006).
Social Movements. An Introduction
.
Oxford: Blackwell Publishing.
Dolezal, Martin (2008). ‘The design of the study: the distinguishing characteristics of
our approach’, pp. 53-74 in
West European Politics in the Age of Globalization
,
edited by Hanspeter Kriesi, Edgar Grande, Romain Lachat, Martin Dolezal,
Simon Bornschier and Timotheos Frey. Cambridge: Cambridge University
Press.
Dolezal, Martin and Swen Hutter (2007). ‘Konsensdemokratie unter Druck? Politischer
Protest in Österreich, 1975-2005’.
Österreichische Zeitschrift für
Politikwissenschaft
36(3): 338-352.
Druckman, James N. (2005). ‘Media Matter: How Newspapers and Television News
Cover Campaigns and Influence Voters’.
Political Communication
22(4): 463-
481.
Earl, Jennifer, Andrew Martin, John D. McCarthy and Sarah A. Soule (2004). ‘The use
of newspaper data in the study of collective action’.
Annual Review Of Sociology
30(65-80.
Ebbinghaus, Bernhard and Jelle Visser (2000).
Trade Unions in Western Europe since
1945
. London: Palgrave Macmillan.
Ferree, Myra Marx, William A. Gamson, Jürgen Gerhards and Dieter Rucht (2002).
Shaping Abortion Discourse. Democracy and the Public Sphere in Germany and
the United States
. Cambridge: Cambridge University Press.
Fillieule, Olivier (1997).
Strategies de la rue: Les manifestations en France
. Paris:
Presses de Sciences Po.
Franzosi, Roberto (2004).
From Words to Numbers. Narrative, Data, and Social
Science
. Cambridge: Cambridge University Press.
Gamson, William A. and Gadi Wolfsfeld (1993). ‘Movements and Media as Interacting
Systems’.
Annals of the American Academy of Political Science
528(114-125.
Giugni, Marco (2004a). ‘Social Movements: Structures, Processes and Consequences’,
pp. 243-277 in
Handbook of Swiss Politics
, edited by Ulrich Klöti, Peter
26
Knoepfel, Hanspeter Kriesi, Wolf Linder and Yannis Papadopoulos. Zürich:
NZZ Publishing.
(2004b).
Social Protest and Policy Change. Ecology, Antinuclear, and Peace
Movements in Comparative Perspective
. Lanham: Rowman & Littlefield
Publishers.
Helbling, Marc and Anke Tresch (2009). Measuring party positions and issue salience
from media coverage: Discussing and cross-validating new indicators.
Paper
presented at the Annual Meeting of the American Political Science Association
(APSA), Toronto, September 3-6, 2009.
Hillard, Dustin, Stephen Purpura and John Wilkerson (2007). An Active Learning
Framework for Classifiing Political Text.
Paper presented at the annual meeting
of the Midwest Political Science Association, Chicago IL, April 12, 2007.
Hocke, Peter (1998). ‘Determining the Selection Bias in Local and National Newspaper
Reports on Protest Events’, pp. 131-163 in
Acts of Dissent. New Developments
in the Study of Protest
, edited by Dieter Rucht, Ruud Koopmans and Friedhelm
Neidhardt. Berlin: Edition Sigma.
Hopkins, Daniel and Gary King (2007).
Extracting Systematic Social Science Meaning
from Text.
Retrieved Oct. 1, 2007, from
http://people.iq.harvard.edu/~dhopkins/papers.html.
Hutter, Swen and Marco Giugni (2009). ‘Protest Politics in a Changing Political
Context: Switzerland, 1975-2005’.
Swiss Political Science Review
15(3): 427-
461.
Keman, Hans (2007). ‘Experts and manifestos: Different sources - Same results for
comparative research? ’.
Electoral Studies
26(1): 76-89.
Kleinnijenhuis, Jan, Jan A. De Ridder and Ewald M. Rietberg (1997). ‘Reasoning in
Economic Discourse. An Application of the Network Approach to the Dutch
Press’, pp. 191-207 in
Text Analysis for the Social Sciences. Methods for
Drawing Statistical Inferences from Texts and Transcripts
, edited by Carl W.
Roberts. Mahwah: Lawrence Erlbaum Associates.
Kleinnijenhuis, Jan and Paul Pennings (2001). ‘Measurement of party positions on the
basis of party programmes, media coverage and voter perceptions’, pp. 162-182
in
Estimating the Policy Positions of Political Actors
, edited by Michael Laver.
London: Routledge.
Kleinnijenhuis, Jan, Anita M. J. van Hoof, Dirk Oegema and Jan A. de Ridder (2007).
“A Test of Rivaling Approaches to Explain News Effects: News on Issue
Positions of Parties, Real-World Developments, Support and Criticism, and
Success and Failure.“
Journal of Communication
57: 366-384.
Klingemann, Hans-Dieter, Andrea Volkens, Judith Bara, Ian Budge and Michael
McDonald (2006).
Mapping Policy Preferences II. Estimates for Parties,
Electors, and Governments in Eastern Europe, European Union and OECD
1990-2003
. Oxford: Oxford University Press.
Koopmans, Ruud (1995). ‘Appendix: The Newspaper Data’, pp. 253-273 in
New Social
Movements in Western Europe. A Comparative Analysis
, edited by Hanspeter
Kriesi, Ruud Koopmans, Jan Willem Duyvendak and Marco Giugni.
Minneapolis: University of Minnesota Press.
(1996). ‘New social movements and changes in political participation in Western
Europe’.
West European Politics
19(1): 28-50.
(1998). ‘The Use of Protest Event Data in Comparative Research: Cross-National
Comparability, Sampling Methods and Robustness’, pp. 90-110 in
Acts of
27
Dissent. New Developments in the Study of Protest
, edited by Dieter Rucht,
Ruud Koopmans and Friedhelm Neidhardt. Berlin: Edition Sigma.
(2007). ‘Who inhabits the European public sphere? Winners and losers, supporters
and opponents in Europeanised political debates’.
European Journal of Political
Research
46(183-210.
Koopmans, Ruud and Dieter Rucht (2002). ‘Protest Event Analysis’, pp. 231-259 in
Methods of Social Movement Research
, edited by Bert Klandermans and
Suzanne Staggenborg. Minneapolis: University of Minnesota Press.
Kriesi, Hanspeter (2004). ‘Strategic Political Communication: Mobilizing Public
Opinion in 'Audience Democracies'’, in
Comparing Political Communication:
Theories, Cases, and Challenges
, edited by Frank Esser and Barbara Pfetsch.
Cambridge: Cambridge University Press.
Kriesi, Hanspeter and Jan Willem Duyvendak (1995). ‘National Cleavage Structures’,
pp. 3-25 in
New Social Movements in Western Europe. A Comparative Analysis
,
edited by Hanspeter Kriesi, Ruud Koopmans, Jan Willem Duyvendak and Marco
Giugni. Minneapolis: University of Minnesota Press.
Kriesi, Hanspeter, Edgar Grande, Romain Lachat, Martin Dolezal, Simon Bornschier
and Timotheos Frey (2008).
West European Politics in the Age of Globalization
.
Cambridge: Cambridge University Press.
Kriesi, Hanspeter, Ruud Koopmans, Jan Willem Duyvendak and Marco Giugni (1995).
New Social Movements in Western Europe. A Comparative Analysis
.
Minneapolis: University of Minnesota Press.
Kriesi, Hanspeter, René Levy, Gilbert Ganguillet and Heinz Zwicky (ed.) (1981).
Politische Aktivierung in der Schweiz. 1945-1978
. Diessenhofen: Rüegger.
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.
Laver, Michael and W. Ben Hunt (1992).
Policy and party competition
. New York:
Routledge.
Lerch, Marika and Guido Schwellnus (2006). ‘Normative by nature? The role of
coherence in justifying the EU's external rights policy’.
Journal of European
Public Policy
13(2): 304-321.
Lombard, Matthew, Jennifer Snyder-Dutch and Cheryl C. Bracken (2002). ‘Content
Analysis in Mass Communication: Assessment and Reporting of Intercoder
Reliability’.
Human Communication Research
28(4): 587-604.
McCarthy, John D., Clark McPhail and Jackie Smith (1996). ‘Images of Protest:
Dimensions of Selection Bias in Media Coverage of Washington
Demonstrations, 1982 and 1991’.
American Sociological Review
61(3): 478-499.
McCarthy, John D., Larissa Titarenko, Clark McPhail, Patrick S. Rafail and Boguslaw
Augustyn (2008). ‘Assessing Stability in the Patterns of Selection Bias in
Newspaper Coverage of Protest during the Transition from Communism in
Belarus’.
Mobilization
13(2): 127-146.
Morales, Laura (2009).
Joining political organisations. Institutions, mobilisation and
participation in Western democracies
. Colchester: ECPR Press.
Myers, Daniel J. and Beth Schaefer Caniglia (2004). ‘All the Rioting That's Fit to Print:
Selection Effects in National Newspaper Coverage of Civil Disorders, 1968-
1969’.
American Sociological Review
69(4): 519-543.
Neuman, Russell W., Marion R. Just and Ann N. Crigler (1992).
Common Knowledge:
News and the Construction of Political Meaning
. Chicago: Chicago University
Press.
28
Newell, James L. (2000).
Parties and Democracy in Italy
. Aldershot: Ashgate.
Olzak, Susan (1989). ‘Analysis of Events in the Study of Collective Action’.
Annual
Review of Sociology
15(119-141.
Ortiz, David G., Daniel J. Myers, N. Eugene Walls and Maria-Elena D. Diaz (2005).
‘Where do we stand with newspaper data?’
Mobilization
10(3): 397-419.
Osgood, Charles E. (1959). ‘The Representational Model and Relevant Research
Methods’, pp. 33-88 in
Trends in Content Analysis
, edited by Ithiel De Sola
Pool. Urbana: University of Illinois Press.
Przeworski, Adam and Henry Teune (1970).
The Logic of Comparative Social Inquiry
.
New York: John Wiley & Sons.
Reinemann, Carsten (2003).
Medienmacher als Mediennutzer, Kommunikations- und
Einflussstrukturen im politischen Journalismus der Gegenwart
. Köln: Böhlau.
Rootes, Christopher (2003). ‘Britain’, pp. 20-58 in
Environmental Protest in Western
Europe
, edited by Christopher Rootes. Oxford: Oxford University Press.
Rucht, Dieter (2003). ‘The Changing Role of Political Protest Movements’.
West
European Politics
26(4): 153-176.
Rucht, Dieter and Roland Roth (2008). ‘Soziale Bewegungen und Protest - eine
theoretische und empirische Bilanz’, pp. 635-668 in
Die sozialen Bewegungen in
Deutschland seit 1945. Ein Handbuch
, edited by Roland Roth and Dieter Rucht.
Frankfurt/New York: Campus.
Ruigrok, Nel and Wouter van Atteveldt (2007). ‘Global Angling with a Local Angle:
How U.S., British, and Dutch Newspapers Frame Global and Local Terrorist
Attacks’.
The Harvard International Journal of Press/Politics
12(1): 68-90.
Schmidt-Beck, Rüdiger and David Farrell (2008). ‘Introduction: The Age of Non-Party
Actors?’, in
Non-Party Actors in Electoral Politics. The Role of Interest Groups
and Independent Citizens in Contemporary Election Campaigns
, edited by
David Farrell and Rüdiger Schmidt-Beck. Nomos: Baden-Baden.
Swank, Eric (2000). ‘In Newspapers We Trust? Assessing the Credibility of News
Sources that Cover Protest Campaigns’.
Research in Social Movements, Conflict
and Change
22(27-52.
Swanson, David L. and Paolo Mancini (ed.) (1996).
Politics, Media, and Modern
Democracy. An International Study of Innovations in Electoral Campaigning
and Their Consequences
. Westport: Praeger.
Tarrow, Sidney (1989).
Democracy and Disorder: Protest and Politics in Italy 1965-
1974
. Oxford: Oxford University Press.
Tartakowsky, Danielle (2004).
La manif en éclat
. Paris: La dispute.
Taylor, Michael and V. M. Hermann (1971). ‘Party Systems and Government Stability’.
American Political Science Review
65(1): 28-37.
Tilly, Charles (2008).
Contentious Performances
. Cambridge: Cambridge University
Press.
Tilly, Charles, Louise Tilly and Richard Tilly (1975).
The Rebellious Century, 1830-
1930
. Cambridge: Harvard University Press.
Trenz, Hans Jörg (2005).
Europa in den Medien. Die europäische Integration im
Spiegel nationaler Öffentlichkeit
. Frankfurt am Main: Campus Verlag.
van Atteveldt, Wouter (2008).
Semantic Network Analysis. Techniques for Extracting,
Representing, and Querying Media Content
. Charleston: BookSurge.
van der Brug, Wouter and Joost van Spanje (2009). ‘Immigration, Europe and the 'new'
cultural dimension’.
European Journal of Political Research
48(3): 309-334.
29
Vliegenthart, Rens and Stefaan Walgrave (2008). ‘The Contingency of Intermedia
Agenda Setting: A Longitudinal Study in Belgium’.
Journalism and Mass Com-
munication Quarterly
85(4): 860-878.
Wittgenstein, Ludwig (1984 [1921]).
Tractatus Logico-Philosophicus
. Frankfurt am
Main: Suhrkamp.
Zuell, Cornelia and Juliane Landmann (2005). Identifying National and International
Events Using Computer-Assisted Content Analysis.
Paper presented at the first
EASR conference, Barcelona, Spain.
30
Table 2.1: Arena-specific and debate-specific periods of observation
National
elections European
elections Protest
politics Public debates
European
integration Immigration Economic
liberalization
Austria 1975, 1994, 1999,
2002, 2006 1996, 2004
1975-2005
2004-2006
1999-2001, 2004-2006
2004-2006
Britain 1974 (Feb.),
1992,
1997, 2001, 2005 1994, 2004
France 1978, 1988, 1995,
2002, 2007 1994, 2004
Germany 1976, 1994, 1998,
2002, 2005 1994, 2004
Netherlands 1973, 1994, 1998,
2002, 2003, 2006 —
1
Switzerland 1975, 1991, 1995,
1999, 2003, 2007 —
1
As a consequence of organizational problems during the coding procedure, we have no comparative data
for the European election campaigns in the Netherlands.
31
Table 2.2: Selected newspapers for the content analyses
Quality newspapers Tabloids
- National elections
- European elections
- Public debates
- Protest politics
- National elections
Austria Die Presse Die Presse Kronenzeitung
UK The Times The Guardian The Sun
France Le Monde Le Monde Le Parisien
Germany Süddeutsche Zeitung
1
Frankfurter Rundschau Bild
Netherlands NRC Handelsblad NRC Handelsblad Algemeen Dagblad
2
Switzerland Neue Zürcher Zeitung Neue Zürcher Zeitung Blick
1
As there were few articles in the Süddeutsche Zeitung on the European elections, we also drew on the
Frankfurter Allgemeine Zeitung for data.
2
No genuine tabloid exists in the Netherlands (Koopmans 2007), so we used a tabloid-style newspaper
that has a wide circulation instead.
32
Figure 2.1: Combination of protest data sets
33
Table 2.3: Categorization of actors: general categories
Categories Description
Public authority
actors
European Union &
IGO actors supranational and international governmental actors
foreign state actors foreign executive, administrative, and central bank actors
domestic state actors domestic executive, administrative, judiciary, and central bank actors
Intermediary
actors
political parties national and European party actors
trade unions –
business
organizations farmer and employer associations, companies
public interest groups charitable and environmental organizations, religious associations,
and social movement organizations
Non-organized
individuals experts intellectuals and scientists
34
Table 2.4: Categorization of actors: party families (selection of important parties)
Country Communists/
Left Socialists Social Democrats
Greens Liberals Christian
Democrats/
Conservatives
Populist Right Other parties (incl.
EU protest parties)
Austria
Sozialdemokratische
Partei Österreichs –
SPÖ (Social
Democratic Party of
Austria)
Die Grünen/Die Grüne
Alternative (The
Greens/The Green
Alternative)
Liberales Forum – LIF
(Liberal Forum) Österreichische
Volkspartei – ÖVP
(Austrian People’s
Party)
Freiheitliche Partei
Österreichs – FPÖ
(Freedom Party of
Austria)
Liste Hans-Peter Martin
– MARTIN (List of
Hans-Peter Martin)
Britain
Labour Party Liberal Democratic
Party Conservative Party United Kingdom
Independence Party Scottish National Party
France
Parti Communiste
Français – PCF (French
Communist Party)
Parti Socialiste – PSF
(Socialist party) Les Verts (The Greens) Union pour la
Démocratie Française –
UDF (Union for French
Democracy)
Union pour un
Mouvement Populaire –
UMP (Union for a
Popular Movement)
Front National – FN
(National Front) Mouvement des
citoyens – MDC
(citizens’ movement)
Germany
Die Linke (The Left) Sozialdemokratische
Partei Deutschlands –
SPD (Social
Democratic Party of
Germany)
Bündnis 90/Die Grünen
(Alliance 90/The
Greens)
Freie Demokratische
Partei – FDP (Free
Democratic Party)
Christlich-
Demokratische
/Christlich-Soziale
Union – CDU/CSU
(Christian Democratic
/Christian Social Union)
Netherlands
Socialistische Partij –
SP (Socialist Party) Partij van de Arbeid –
PvdA (Labour Party) GroenLinks – GL
(Green Left) Democraten’66 – D’66
(Democrats 66); Volks-
partij voor Vrijheid en
Democratie – VVD
(People’s Party for
Freedom and
Democracy)
Christen-Democratisch
Appel – CDA
(Christian Democratic
Appeal)
Lijst Pim Fortuyn –
LPF (List Pim Fortuyn) Europa Transparant –
EurTrans (Europe
Transparent)
Switzerland
Sozialdemokratische
Partei der Schweiz –
SPS (Social Democratic
Party of Switzerland)
Grüne Partei der
Schweiz – GPS (Green
Party of Switzerland)
Freisinnig-
Demokratische Partei –
FDP (Radical
Democratic Party)
Christlichdemokrati-
sche Volkspartei – CVP
(Christian Democratic
People’s Party)
Schweizerische
Volkspartei - SVP
(Swiss People’s Party)
35
Table 2.5: Categorization of issues
Meta-categories Description
Economic liberalism
Opposition to market regulation; opposition to economic protectionism in agriculture and other sectors of the economy;
support for deregulation, more competition, and privatization.
Anti-immigration Support for a tough immigration and integration policy
Europe Support for European Integration
Welfare Support for an expansion of the welfare state; objection to welfare state retrenchment; support for tax reforms with a
redistributive character; calls for employment and health care programs
Budget Support for a rigid budgetary policy; reduction of the state deficit; cuts on expenditures; reduction of taxes without direct
effects on redistribution
Cultural liberalism Support for the goals of new social movements, with the exception of the environment movement; support for cultural
diversity, international cooperation (excluding the European Union and NATO); support for the United Nations; support for
the right to abortion and euthanasia; opposition to patriotism, calls for national solidarity, defense of tradition, national
sovereignty, and appeals to traditional moral values; support for a liberal drug policy
Culture Support for education, culture, and scientific research
Army Support for the armed forces (including NATO), for a strong national defense, and for nuclear weapons
Security Support for more law and order, for fighting crime, and denouncing political corruption
Environment Opposition to nuclear energy; support for environmental protection
Institutional reform Support for various institutional reforms, including making modifications to the structure of the political system
Infrastructure Support for improving the country’s roads, railways, and other physical infrastructure.
38
38
Appendix A: An empirical response to our critics
Structure of the space for all countries by decade, based on partisan issue-directions
weighted by issue salience, only 7 issues: mds-analyses, stress-values
decade
stress 1970 1990 2000
raw
1-dimension 0.11 0.13 0.15
2-dimensions 0.06 0.08 0.06
stress-I
1-dimension 0.33 0.36 0.39
2-dimensions 0.24 0.29 0.25
39
39
Structure of the space for all six countries, by decade: results of exploratory factor anal-
yses based on mean partisan issue direction
a) seven theoretically most relevant issues
decade
categories 1970 1990 2000
welfare -0.50 -0.70 -0.38
budget 0.72 0.66 0.53
economic liberalism 0.54 0.64 0.65
cultural liberalism -0.75 -0.59 -0.25
Europe 0.14 -0.06 -0.35
anti-immigration -0.17 0.44 0.74
army 0.55 0.82 0.35
EV 2.0 2.6 1.7
proportion 64% 78% 92%
N 25 32 36
b) all twelve issues (rotated solution)
1970 1990 2000
categories Factor1 Factor2 Factor1 Factor2 Factor1 Factor2
welfare -0.37 -0.47 -0.74 0.14 -0.60 0.06
budget 0.57 0.42 0.55 -0.30 0.02 0.74
economic liberalism 0.56 0.19 0.67 -0.07 0.35 0.61
environment 0.01 0.79 -0.39 0.62 -0.51 -0.19
cultural liberalism -0.67 -0.26 -0.40 0.74 -0.56 0.28
Europe -0.12 0.80 0.04 0.43 -0.49 0.06
anti-immigration -0.16 0.13 0.55 -0.01 0.54 0.50
culture -0.41 -0.39 -0.04 0.90 -0.61 -0.03
army 0.83 -0.28 0.73 -0.31 0.25 0.21
security 0.74 -0.07 0.45 -0.14 0.29 0.22
infrastructure 0.23 -0.46 0.51 0.57 -0.43 0.59
instititutional reform -0.34 -0.33 -0.29 -0.11 -0.10 0.68
EV 3.1 2.1 3.6 1.9 2.8 1.8
proportion 39% 32% 39% 32% 42% 40%
n 25 25 32 32 37 37
40
40
Structure of the space for all six countries, by decade: results of confirmatory factor
analyses based on mean partisan issue direction with all the 12 issue categories
1970 1990 2000
categories Factor1 Factor2 Factor1 Factor2 Factor1 Factor2
welfare -0.50 0.23 -0.73 - -0.71 -
budget 0.84 - 0.73 - 0.47 -
economic liberalism 0.39 - 0.65 - 0.62 -
environment - -0.97 -0.48 -0.43 - -0.64
cultural liberalism -0.87 - -0.40 -0.57 - -0.36
Europe - -0.79 - -0.45 - -0.47
anti-immigration . - 0.65 - - 0.96
culture -0.52 - - -0.94 -0.44 -
army 0.99 -0.24 0.77 - - 0.64
security 0.82 - 0.46 - - 0.38
infrastructure - 0.13 0.53 -0.66 0.04 -
instititutional reform - - - - 0.34 -
CHISQ
1)
58.9 88.6 77.4
df 41 50 53
r
1,2
-.21 .30 .82
1)
Goodness of fit one-dimensional solutions: 1970: 93.2, 44 df; 1990: 129.1, 54 df; 2000: 90.4, 54 df
