Content uploaded by Femke Van Esch
Author content
All content in this area was uploaded by Femke Van Esch on Apr 29, 2016
Content may be subject to copyright.
D3.1 Cognitive Mapping Coding Manual
Authors: Femke van E sch, Lieke Brand, Rik Joosen, Sebastiaan Steenman,
Je roen Snellens, Marij Swi nkels
Deli very date: February 2016
2
Contents
Introduction ......................................................................................................................................... 3
Key concepts & the technique of cognitive mapping .................................................................... 3
Ideas/belief systems ............................................................................................................................ 3
Discourses ............................................................................................................................................ 5
Congruence ......................................................................................................................................... 5
Indirect Elicitation .............................................................................................................................. 6
Leaders’ beliefs .................................................................................................................................... 7
Discourses ............................................................................................................................................ 8
Selection criteria: Timing and content ................................................................................................ 9
Direct Elicitation: Citizens’ Beliefs ................................................................................................. 12
Coding .................................................................................................................................................. 14
Coding rules ....................................................................................................................................... 14
Coding complex phrases .................................................................................................................... 16
Coding discourse: some special rules ................................................................................................ 21
Merging and standardisation ............................................................................................................ 21
Inter coder reliability ......................................................................................................................... 23
References........................................................................................................................................... 23
3
Introduction
1
Transcrisis aims to develop a solid understanding of the role of leaders in managing transboundary
crises and the requirements for ensuring an effective and legitimate crisis response. In addition to
material and institutional requirements, leaders’ policy ideas, peoples’ preferences and public
discourses play an important role in this. Establishing the precise nature and effects of these ‘soft’
but influential factors in crisis management is notoriously difficult, for ideas, preferences and
discourses are intangible phenomenon that cannot be observed directly (Fiske & Taylor 1991; Risse-
Kappen 1994). To be able to study these phenomenon and their effects on the effectiveness and
legitimacy of crisis management, Transcrisis will make use of a method that was specifically designed
to analyse such ideational factors: the method of Comparative Cognitive Mapping (CCM; Axelrod
1976; Bougon, Weick & Binkhorst 1977; Curseu, Schalk & Schruijer 2010; Princen & Van Esch
forthcoming; Van Esch 2012; 2014; Young & Schafer 1998).
This coding manual provides a practical methodological guideline that explains how to use
the technique of CCM to uncover ideas and discourses. The manual offers valuable insights for other
scholars resulting from our years of experience in using the technique. However, the practices
described in the manual will be inspired by the research question of central to Work package 3 (WP3)
of the TRANSCRISIS project in particular, and serve as a guide for studying the role and changes in the
ideas and discourses that pertain to the management of the Euro crisis.
Key concepts & the technique of cognitive mapping
Work package 3 will study the interaction between leaders’ policy ideas, dominant expert and public
discourses and citizens’ perceptions of the Euro crisis. In order to do so, the following key concepts
must be defined and operationalised: leaders and citizens’ policy ideas, expert and public discourse
and ideational congruence. The technique of cognitive mapping is especially suited to empirically
measure these ‘ideational’ concepts.
Ideas/belief systems
In the literature, ideas are generally defined as actors' subjective thoughts on how the world works
(Levy 1994). These ideas are not necessarily accurate or rational in the sense of being obtained
through a thorough cost-benefits analysis or process of trial and error. Rather, peoples’ belief
systems emerge and gain strength throughout life and are informed by their experience, education
1
This manual reflects the current state of our discussions on the issue of coding. During the actual research
process, experience may still induce us to divert from the practices described in this manual. The final report of
this Work package will note and account for any differences between plan and practice.
4
and roles. However, while they are not rational, ideas or belief systems are assumed to show some
consistency over time.
Despite their different disciplinary roots and terminology, most studies assume belief
systems or worldviews to consist of convictions on two or three levels of abstraction (cf. Goldstein &
Keohane 1993; Hall 1993; Jervis 2006; Levy 1994; Sabatier & Jenkins-Smith 1993; Van Esch 2007): 1)
diagnostic beliefs that involve thoughts on the state of the world and nature of the circumstances at
hand; 2) instrumental beliefs concerning the means or policies that may provide the intermediary
connection to further the actors’ ends and goals; and 3) principled beliefs that denote the values or
ends central to the actors’ worldviews. These different beliefs are connected through causal and
normative relations to form a belief system or worldview. Many scholars assume belief systems to be
structured hierarchically in the sense that some belief concepts are deemed to be more fundamental
to the actor's worldview than other, more secondary, beliefs. While scholars differ in the kind of
beliefs they deem dominant, the conception of belief hierarchy is theoretically important because
common agreement exists that change in more fundamental beliefs is hard and rare, and if any belief
change occurs ‘it is usually confined to the secondary aspects of belief systems’ (Sabatier & Jenkins-
Smith 1993).
Policy paradigms constitute a particular kind of belief system that is worth mentioning
separately in this regard. Following Hall (1993: 279), a paradigm is defined as a ‘framework of ideas
and standards that specifies not only the goals of policy and the kind of instruments that can be used
to attain them, but also the very nature of the problems they are meant to be addressing’. This
definition overlaps almost completely with the definition of belief system provided above. However,
in contrast to regular belief systems, paradigms are ‘ideas on steroids’ (Baumgartner 2014): Highly
dense and consistent configurations of views, that are incommensurable with rival paradigms (Hall
1993; Princen & ‘t Hart 2014). Moreover, paradigms are seen as an influential determinant of actors’
behavioural patterns and are at the root of grand ideological clashes over policy-making. In the
literature, paradigms are placed at the top of the belief hierarchy and seen to be extremely resistant
to change. Only major crises and the availability of an alternative paradigm may on rare occasions
induce a paradigm shift (Hall 1993). When paradigm shifts do occur they are expected to do so as
sudden and complete ideational U-turns. In this project, however, the hierarchy and stability of ideas
and paradigms are regarded as empirical questions.
5
Discourses
Discourses are systems of thought composed of ideas, attitudes, courses of actions, beliefs and
practices that construct the subjects and the worlds of which they speak (Foucault 1972). While ideas
are in essence individual cognitions, for this project discourses are defined as the way ideas and
paradigms are being shaped and expressed and are therefore social entities. Public discourses are
composed of several frames that focus on and highlight a selection of reality. Entmann (1993)
distinguishes four functions of such discursive frames that tie in seamlessly with the definition of
sense making: The first function is to define what the problem is. A problem definition automatically
bestows certain attributes on a situation like who is to blame for a situation and who the victim. The
second and related function of a frame is to identify the causes of the problem. The third function is
to pass judgment on a situation, its causes and effects and the actors involved. The last function of a
frame is to suggest remedies for the problem. These four functions of discursive frames tie in well
with the different forms of ideas described above and will be used to analyse the public and expert
discourses on the Euro crisis.
Congruence
In addition to providing insight in leaders and citizens’ ideas and public and expert discourses on the
Euro crisis, WP3 aims to explore the congruence between them. To establish the extent to which the
different actors’ and collective sense making efforts overlap, three forms of ideational congruence
are used. The first form of congruence is
issue saliency which indicates whether
similar issues are identified as most
important and pressing (Lindeboom 2012;
Hobolt & Klemmensen 2005). The second
type of congruence is ideological
distance. To establish this, WP3 will study
to what extent actors and discourses take
a similar position on both a left-right and
pro/anti-European scale (Golder &
Stramsky 2010). The third type of
congruence is narrative congruence. This is the most intricate measure that reveals the extent to
which the same arguments are used by leaders and citizens and in the expert and public discourses
(Van Esch, Joosen & Van Zuydam, under review). The three forms of congruence build on one
Narrative
Congruence
Ideological Distance
Issue Saliency
Figure 1: Hier archy o f congruen ce
6
another and are progressive in terms of the intensity of convergence or divergence in sense making
they entail (see figure 1).
Cognitive mapping is a method especially developed for studying intangible ideational factors such as
leaders and citizens’ beliefs and discourses in a structural fashion. The technique rests upon the
premise that ideational factors like ideas and discourses can be represented in graphical form
(Axelrod 1976; Young & Schafer 1998; Clarkson & Hodgkinson 2005). The method is inductive in
terms of the issues that may be studied and can be used to explore the ideas of individuals,
institutions and groups on any subject, as long as these ideas are made known to the researcher
(Axelrod 1976). A cognitive map is constructed by transforming the causal and utility-relations in a
belief-system or discourse into a visual graph in which the concepts are depicted as points and the
relations between these concepts as arrows (Axelrod 1976; Van Esch 2007; Young 1996; Young &
Schafer 1998; see figure 2).
Figure 2: Exce rpt f rom a pre-crisi s cognitive map of Chancellor Angela Merkel
Indirect Elicitation
Cognitive maps can be derived from many different sources as well as directly from the actors under
study (see next section). In WP 3, the belief systems of the political and financial leaders as well as
the national public and transnational expert discourses will be derived indirectly from existing
7
sources. Studies applying the technique have used interviews, speeches, transcripts of decision-
making moments, policy-proposals or communications to derive maps from (Axelrod 1976; Princen &
Van Esch forthcoming; Van Esch 2007; 2012; 2014; Verbeek 1990). Each of these sources has its
strengths and drawbacks and not all texts or communications are suitable, as they need to contain
argumentation and reasoning to be able to build a cognitive maps from. Also a considerable amount
of assertions are needed to achieve a reliable reflection of a belief system or discourse. In addition, in
this study we are interested in comparing leaders and citizens' beliefs and public discourses and track
changes in the cognitive maps over time. In this case, careful selection of sources is needed to
prevent the risk of incomplete source overlap as this would could result in an overestimation of
difference and change and produce unreliable outcomes (Young 1996: 409).
The technique of cognitive mapping thus requires careful preparation. Firstly, careful
selection of sources and the relevant sections within those sources should take place on the basis of
explicit and consistent application of selection-criteria. In addition, the danger of incomplete source-
overlap may be reduced by using several sources over a period of time. This reduces the danger of
overemphasising a specific event with high saliency at a particular moment in time. Finally, use of
broad standardised or merged concepts rather than the specific literal and contingent wording used
in the texts or communications as the basis for analysis and comparison will further increase the
reliability of findings (cf. Heradstveit & Narvesen 1978: 81).
Leaders’ beliefs
The study in Work package three of the TRANSCRISIS project will focus on nine member states and
the EU political level. As such, the research will include the political and financial leaders of those
nine member states: Germany, The Netherlands, Ireland, France, Spain, the UK, Hungary, Slovenia
and Denmark. The country selection represents a representative sample in terms of geography,
exposure to the financial crisis, and types of membership in the Eurozone. As different patterns of
leadership and responsiveness may be expected of elected and non-elected leaders, the study will
include the heads of state and government as well as central bank governors. In addition, we will
study the leaders of selected EU institutions (Commission, Council Secretariat, and ECB).
Most ‘measuring at-a-distance’ methods like cognitive mapping rely on public speeches and
interviews. This raises questions of reliability, for using public sources increases the probability that
the maps represent decision-makers' strategic rather than genuine beliefs. Still, using these sources
is the best option available especially for the current research that focusses on public and political
meaning making rather than private contemplation of leaders. The leaders selected for this study are
8
deemed relevant precisely because of the public role they played in the management of the Euro
crisis.
In addition, public sources may be the only ones available when studying contemporary
events (Hart 1977; Dyson & Raleigh 2014 ), and are widely available, enabling longitudinal study of
leader’s beliefs that may otherwise impossible because this requires a large set of comparable texts
or assertions (Hart 1977: 117). Moreover, use of the more spontaneous and direct source of
interviews to circumvent the interference of speech-writing has its own drawbacks. Rather than a
speech-writers bias, this method may introduce an interviewer-bias. Given the importance of
saliency in CM analysis, building maps from interview-answers may result in outcomes that reflect
the preoccupations of the interviewer rather than the actor of interest (Axelrod 1976b: 257-9).
Finally, some scholars have studied the validity of public assertions in assessing leaders' beliefs
(Marfleet 2000; Renshon 2009; Walker & Schafer 2000; Young 1994). Firstly, these studies caution
against extrapolating affective reactions from public sources. However, Renshon and Young conclude
that analysis of public sources does produce patterns of outcomes very similar to those of private
sources. Nonetheless, to reduce any strategic biases as much as possible, when using public source, it
is advisable to base any single cognitive map on a number of different sources aimed at different
publics and stakeholders over a selected period of time.
As argued before, however, for the current research, these issues are less of a problem as
this study does not aim to uncover leaders’ private thoughts. As, surprisingly, previous research has
indicated that speeches by political leaders on a specific topic like the euro crisis are actually rare
(Van Esch 2014), we will start by collecting all speech-acts we can find on the issue (see below for the
criteria that need to be fulfilled) and if needed randomly select a similar amount for each leader in
order to prevent bias and promote comparability.
Discourses
In addition, Work package three will study both the national public discourses in the nine selected
member states as well as the transboundary expert discourse on the Euro crisis found amongst the
transnational financial elite. The cognitive maps underlying the national public discourses will be
derived from op-ed pieces concerning Euro-crisis management in major broadsheet European
newspapers. In our selection of the newspapers we will rely on established practices within the field
of communication studies and more in specific the practice used by the Reuters Institute for the
Study of Journalism in their study of media coverage of the Euro crisis. This study includes a dataset
of newspaper articles from five of the countries in our study (Picard 2015): France, Germany, The
Netherlands, Spain, United Kingdom. For the remaining four countries in our study, Ireland, Hungary,
9
Slovenia and Denmark, newspapers will be selected using the same criteria as used in the Reuters
study. These criteria were:
1. To select the leading financial/business newspaper in a state.
2. To select two leading newspapers each representative of the conservative or liberal political
ideology in the country.
3. To include the leading tabloid paper.
For the countries that do not have a leading tabloid, a leading centrist paper was used. As a proxy for
the transboundary expert discourse articles from the Financial Times and the Economist will be used
as these are the most consulted newspaper amongst the European financial elite (ComRes/Burson-
Marsteller 2016). As these newspapers do reflect a distinct Anglo-Saxon perspective on the crisis, in
addition, articles from the more continental and well-read EurActiv and EuroObserver will be
included.
2
From these newspapers only Op-ed articles will be used as a first exploration showed
these to be most likely to contain the necessary argumentation to derive cognitive maps from.
The precise number of speeches and media sources that will be included in this study will
have to be determined in practice based on availably and the timeframe of the study. As the time-
frame covered in this study is over five years in which the euro crisis continuously featured as a very
salient topic, it will be impossible to include all articles from these source. Moreover, the technique
of cognitive mapping still relies on hand-coding and is time-demanding. So, even when limited to
only speeches (for leaders) and op-eds (for discourses), we will be unable to analyse all articles that
were published. As such, during the coding process we will assess the amount of media sources
needed to derive a representative map from and select the number of texts randomly from the
population of relevant articles.
Selection criteria: Timing and content
To be able to answer the central question two criteria need to be taken in mind when selecting the
relevant sources: timing and content. Firstly, our study focuses on the sovereign debt crisis that hit
Europe at the end of 2009 onwards and lasted (at least) until the summer of 2015 when the
negotiations on the third Greek bail-out cause much political unrest. For the purpose of our research,
we have divide the Euro crisis into three periods that are characterised by a very different set of
events. This will enable us to study and compare the fluctuation and stability of leaders’ policy beliefs
2
Recently Politico.eu has established itself as a major online news source for EU elites (ComRes/Burson-
Marsteller 2016). However, as it was not in existence during the entire period covered in this study, this source
will not be used.
10
and discourses under different circumstances and provide a contextualised analysis of leaders’
meaning making efforts.
Reviewing the history of the crisis it may be categorized into three time periods. The first
period started at the end of 2009. On October 16th 2009 George Papandreou disclosed the severe
fiscal conditions that the previous government had kept quiet (Raadschelders & Vigoda-Gadot 2015).
On 5 November 2009, the Greek government announced that its budget deficit of national income
was even higher than previously announced. In fact, it turned out to be almost double and would
eventually reach 14% for 2009 (Chackrabortty 2011). As this event set the euro crisis in motion, we
will take 5 November 2009 as the beginning of the first period of the crisis.
This first period may be characterised as a time of uncertainty and confusion as the member
states and the European Institutions were operating unilateral with ‘fire-fighting’ crisis responses
(Hodson & Quaglia 2009; Carmassi & Micossi 2010). On 25 March 2009, the EU Member States
agreed on an emergency plan for Greece which was based on the shared responsibility of a single
currency (Meiers 2015). This was briefly followed by more uncertainty as Greece still refused to ask
for the financial aid and other Member States were publicly discussing the conditions applicable to
the loans (Carmassi & Micossi 2010, p.2). During this time, the Greek sovereign bond yield continued
to climb, with the international financial markets demanding more convincing plans from the EU and
the IMF (Gocaj & Meunier 2013, p.241). This first period of unilateral action and uncertainty ends on
2 May 2010 when the EU presented its first major coordinated response to the crisis by creating a
€110 billion rescue package for Greece (Verdun 2015).
This event also marks the start of the second period. For while the three-year stability
program had the ambition to allow Greece to get its economy back on track (European Commission
2010), it did not calm the markets. Neither did the decision for the establishment of the new
European Semester, nor the decision to turn the temporary facility into the permanent European
Stability Mechanism (ESM). What ultimately did calm the markets was Draghi’s famous ‘whatever it
takes’ speech of 26 July 2012. This is the event that will mark the transition from the second to the
third period in our study. The third period is characterised by a marked different mood than the first
two as the Draghi speech brought about a positive change in the markets, and confidence (partly)
was restored until the outbreak of the political upheaval surrounding the 2015 negotiations on the
Greek third bail-out which will mark the endpoint of our study. For each of these three periods a
similar amount of speeches and media-sources will be selected for all leaders and countries.
11
The second criterion for the selection of speeches is their content. As the work package
focusses on meaning making regarding the euro or sovereign debt crisis, the selected sources will
have to deal with this topic. Belief systems and discourses may be very different for even closely
associated issues, so it is very important to apply clear content criteria. In searching for speeches, all
researchers participating in the data collection will use the same keywords to find those speech acts
that cover the issue of the Euro/sovereign debt crisis. The keywords that will be used are summed up
in the table below but overall it is important to keep in mind that while the euro/sovereign debt crisis
is also often referred to as the economic or financial crisis, for the purpose of this study must be
distinguished from ‘the banking’ as well as the ‘sub-prime mortgage’ crisis. Our study is fore mostly
focussed on governmental policy in the budgetary, monetary and macro-economic domain. The
search for speeches and media sources will be conducted through key word guided searches of Lexus
Nexus, the Eurotopics website (for media sources), government and central bank website as well as
the Google search-engine (for leaders’ speeches). By using these different searches a broader
arrange of speeches and media sources will be found.
Keywords for selecting relevant speech-acts
Europe / EU
Crisis
Euro crisis
Sovereign debt crisis
Financial crisis
As noted above, not only careful selection the sources of data but also the selection of
relevant parts of these sources is important. The unit of analysis to guide this selection are the
sections of the speeches or Op-eds as indicated by the lay-out or format of the original publication.
Only full sections will be coded and all causal and/or utility relations in that section must be included.
However, as we are interested predominantly in how leaders and the public make sense of the euro
crisis (how they diagnose its nature, cause and solutions), only those sections containing at least one
causal relation will be included in the analysis. The numbering of the sections will be noted on the
copy of the text used for coding and will also be registered in the coding software MAPS (and
therefore the final database) during the coding process.
3
3
The numbering will take the form of (page number-section number) whereby only the sections that started on
that page will be counted.
12
All the non-relevant parts in speeches can be safely excluded as they will not contribute to
answering the research questions (Wrightson, 1976: 320). In order to do this consistently, explicit
selection criteria have been set to select only the relevant paragraphs in the speech acts and media.
The sections should contain a direct reference to the euro / sovereign debt crisis or a combination of
a reference to the EU and key words referring to governmental policies in the budgetary, monetary
and macro-economic domain listed in table below. This list is, however, a guideline as sometimes
texts continue on a certain topic in subsequent sections without mentioning it explicitly. A section
referring to the euro crisis which is followed by a section stating that ‘it’ [in the sense of the euro
crisis] leads to unemployment, for instance must be included in the analysis.
Key terms for selecting relevant sections
Euro crisis
Sovereign debt crisis
Europe
European integration / union
European monetary system/EMU
Supranational
Intergovernmental
Any EU Institutions
ECB independence
Price stability/inflation
Sovereign bonds
Financial discipline/SGP
Governmental expenditures/public finances/fiscal policies
Governmental budgetary deficits/public debt
European/international credit facilities
Direct Elicitation: Citizens’ Beliefs
In addition, cognitive maps will be elicited through direct elicitation from 1000 citizens of each of the
nine countries in this study. They will be enlisted to participate in the study through a specialised
polling agency and have to form an accurate representation of the population. Representativeness
with regard to education, prosperity, age and political ideology are particularly important as these
are known to influence peoples’ perceptions of any issue concerning the EU.
The cognitive maps of the citizens will be elicited directly via the web application DART. For
this a freehand approach will be used as this is the most efficient and valid way (Hodgkinson, Maule,
& Bown 2004). Respondents will be asked to select seven concepts out of a list of 50 pre-defined
concepts associated with the Euro crisis. This list of concepts will be assembled after and on the basis
13
of the analysis of leaders’ beliefs and discourses as the aim is to survey peoples’ cognitive maps on
the most salient issues in the debate.
Subsequently, respondents are asked to draw arrows between the concepts to indicate how,
in their eyes, they are linked. The direction of the arrow indicates the direction of the causal effect
(cause → effect) while the colour of the arrow indicates whether the effect is deemed to be positive
(green arrow) or negative (red arrow). Overall this allows respondents a choice of nearly 5000
different relations to compose their cognitive map from. In addition, respondents will be asked to
complete a short survey about their demographic characteristics and political behaviour. This will
allow us to aggregate peoples’ individual cognitive maps on the basis of key characteristics they
share like age, education or party preference.
A note of caution is necessary when using direct elicitation of cognitive maps. In principle,
concepts can be positive (for instance prosperity), negative (recession), or neutral (employment) in
their meaning. Studies on survey research show that double negative language constructions are
more difficult to comprehend and therefore prone to mistakes (De Vaus 2014: 97). This jeopardize
the validity of the study. In the case of CCM, double negative language constructions can be equated
with combining a negative concept with a negative arrow. For instance, the negative relation ‘the
actions of the banks recession’ formally means that the actions of the banks reduced the
Concept
Definition
Data
Method
Leaders belief
system
Diagnostic
Instrumental
Principled
Paradigmatic
Speeches,
(published) interviews
Cognitive Mapping,
Indirect elicitation
Public Discourse
Problems
Causes
Judgements
Remedies
Op-ed pieces in major
national newspapers
Cognitive Mapping,
Indirect elicitation
Expert Discourse
Problems
Causes
Judgements
Remedies
Op-ed pieces in
specialised international
newspapers
Cognitive Mapping,
Indirect elicitation
Citizens belief
system
Diagnostic
Instrumental
Principled
Paradigmatic
Citizens responses
Cognitive Mapping,
Direct (web-based)
elicitation
Congruence
Issue Saliency
Ideological Distance (left-
right, Pro-EU-Eurosceptic)
Narrative Congruence
Leaders’ CMaps
Citizens’ CMaps
CMaps Expert Discourse
CMaps Public Discourses
Comparative
Cognitive Mapping
14
recession, thus improving the economy. However, previous research shows that many respondents
draw this relationship to indicate that the actions of the banks ‘were bad for’ the recession, thus
worsening the recession. In direct elicitation, negative concepts should therefore be avoided (Van
Esch et al 2014).
Coding
Direct elicitation results in the creation of cognitive maps without the direct intervention of
researchers. For leaders’ beliefs as well as the discourses the cognitive maps are elicited indirectly
through coding of text by researchers. As the coding for this project will be done in teams and texts
will be double-coded to increase reliability, strict coding rules are needed to guide this process.
In contrast to content analysis, which is essentially ‘a counting procedure’, the basis of
analysis in a cognitive map is the relationship between concepts, not the concepts themselves.
(Axelrod 1976a: 7). As such, the focus of the text-analysis is not on nouns but on verbs. An analyst
typically looks for the subject-verb-object construction in the text that indicates a causal or quasi-
causal relationship (Young 1996: 397). Any relationship drawn by the decision maker that may be
‘viewed as equivalent, respectively, to the English expressions “is usefully or desirably associated
with” and “is adversely or undesirably associated with”’ is recorded and represented in the map
(Lambert 1966 quoted in: Axelrod 1976a: 8). In general, deriving the causal relations from texts
comes very naturally, and scholars need little training to achieve good rates of inter-coder reliability.
As stated above, to create a cognitive map, all causal and utility relationships between
concepts are manually derived from the text or speech act or texts. Causal relationships follow the
structure of ‘Cause concept – Linkage - Effect concept’. Utility statements are statements to the
effect that something is ‘good’, in ‘someone’s interest’ or ‘in the general benefit’. In addition to
coding concepts, coders need to attribute a value to the relationship between these concepts. Such
relationships (arrows) in the map can be positive (+), negative (-) or non-existent (0) in case an actor
(explicitly) states that a cause concept is not related to an effect or utility concept. Extensive coding
rules exists for deriving cognitive maps from assertions (Bonham & Shapiro 1986; Wrightson 1976;
Gallhofer, Saris & Melman 1986: 8-9).
Coding rules
As argued before, the structure of causal and utility relations in English grammar is Subject-Verb-
Object and in coder terminology the basis structure is ‘Cause concept – Linkage - Effect/Utility
15
concept’ (Wrightson 1976: 292). This structure should be kept in mind by coders at all times, so that
no relationship in a text is overlooked. Also, the concepts should be in the form of variables that can
take on different values: For example, ‘financial stability’, which can be great or small. An entity such
as ‘Rutte’ or ‘The Netherlands’ cannot have different values so it is not a valid concept (Wrightson
1976: 293). When the alternative values of concepts are clearly understood (great or small) they do
not need specification. If this is not the case, however, the alternative values need to be included in
the description of that variable. The coder will come across sentences or phrases that in a literal
sense do not constitute a structural or grammatical relationship. These cases are the most difficult to
code correctly. They are best approached by content analysis where the following question should be
asked: ‘Does the thrust of the phrase, sentence or sentence group imply a relationship?’ (Wrightson
1976: 293). For example, the sentence ‘the most dangerous thought by the Greek people is to exit
the Eurozone’ does not constitute a relationship in a grammatical sense, but in terms of content
implies that exiting the Eurozone is bad for Greece and therefore a (negative utility) relationship
must be coded.
Furthermore, it is important that in the first raw coding, the coder keeps the language as
authentic as possible. Also relations should be included as many times as they are mentioned in the
source. So if the speaker makes a statement twice, the coder must write down that relationship
twice as well. The same goes for contradictory statements: both must be included. However, one
should avoid a statement that is subsequently refuted. When a speaker mentions a relationship but
then follows with a refutation, only the denial or contradiction of a relationship should be recorded.
Only when a relationship is repeated, two relations should be coded. As indicated before, there will
be times in which the coder is forced to paraphrase a sentence to explicate the implied structure of
‘Cause concept – Linkage - Effect/Utility concept’ (see Wrightson 1976: 294 for examples). When
such paraphrasing is required, the original language must still be preserved as much as possible.
Finally, when a statement in a speech or article is attributed to someone else, the coder needs to
judge whether speaker or writer takes on the statement as their own. If so, it should be recorded,
but like in the case of refutation, a statement by someone else followed by the consent of the author
should only be recorded once (Wrightson 1976: 294).
This last example shows that even with a set of standardised, well-developed coding rules
sometimes we need to rely on the judgement and interpretation of the coder. One of the major
pitfalls in coding for cognitive maps, like in all content analysis, is that the coding process will be
influenced by the personal assumptions and biases of the coder (Wrightson 1976: 294). The danger
of this may increase as coders become more familiar with the speaker or writer’s viewpoint and they
16
may start to distinguish relations that objectively are not present simply because of their familiarity
with the speaker or author’s viewpoint. Wrightson suggests this may be solved by adopting the rule
of thumb that relations that transgress the boundaries of a single section are not coded (1976: 295).
As it is unclear why this would solve the problem, in this study we aim to counter this problem by
having all sources coded by at least two coders independently. Any disagreements will be discussed
until agreement exists. If agreement between the two coders cannot be reached, the work package
leader (who in principle will not be amongst the first two coders) will take the final decision. In
addition, we will uphold part of Wrightson’s rule on coding across section boundaries and only code
relations that literally conform to the standard grammatical rule of ‘Cause concept – Linkage -
Effect/Utility concept’ over section boundaries.
The coding scheme for cognitive mapping also contains several specific rules for coding utility
statements. Firstly, the utility concept must only be used when the benefit that is associated with the
cause concept is unspecified (Wrightson 1976: 306). In this study the term ‘benefit of all’ (or variants
like ‘benefit of Germany’) will be used to denote the most general of utility concepts. Utility variables
sometimes consist of a noun with modifier that indicates a positive or negative utility relationship
rather than to shed light on the concept itself. For example, the phrase ‘peril of France’ in the
sentence ‘this economic policy is pursued at the peril of France’ expresses the negative utility of the
said economic policy and is therefore not in (what should be coded as) the ‘benefit of France’
(Wrightson 1976: 306). A good rule of thumb to identify these utility concepts or phrases is that
almost all utility concepts include a noun or a noun clause.
Coding complex phrases
In addition to these basic rules, coders may be confronted with additional complexities. Wrightson
has constructed a total of ten additional rules to help coders deal with these special cases. These
rules are listed below.
Rule 1
Concepts may differ in their level of complexity. Some relationships consist of relatively easy-to-code
concepts such as ‘the banking crisis in the USA caused the European financial markets to crash’ in
which there is one cause concept (the banking crisis in the USA) which caused (+) the effect (the
European financial markets to crash). Other relations may consists of more complex concepts. The
sentence ‘the establishment of the ESM reduced the Portuguese sovereign bond yield as well as
market instability’, for instance includes a dual effect concept (sovereign bond yields and market
17
instability). In other cases the identified cause may consist of a dual concept. All of these complex
dual concepts, whether they are the cause or effect or part of a causal or a utility relationship, are
separated in two concepts only when the individual parts are truly distinguished as two separate
concepts within the text or communication. Whether this is the case may only be derived by reading
through the full text. If the concept are consistently used together in one phrase, they should be
coded as one and the sentence should be coded as one relationship (with a dual concept). If the
concepts are also use individually (as well as collectively), the coder should code them separately as
two concepts. In this case, these sentence should be coded as two different relationships. So the
example above would be coded as follows:
The establishment of ESM / - / Portuguese sovereign bond yields
The establishment of ESM / - / market instability
Rule 2
The coder may also come across ‘either/or’ assertions. For example, ‘either Papandreou implements
more austerity measures or he increases taxes to calm the markets in the Eurozone’. As can been
seen in the example, either/or relationships often contain two relationships. Moreover, concepts in
either/or assertions often need paraphrasing or revision of the original structure to clarify the
concepts and relationships (Wrightson 1976: 300). Finally, these either/or relationships exist
independently of each other and therefore need to be coded as two distinct relationships. However,
this rules does assume that the one relation does not need the other to materialise. If so, the above
mentioned example would be coded as follows:
‘Papandreou implementing more austerity / + / calm the markets in the Eurozone’
‘Papandreou increasing taxes / + / calm the markets in the Eurozone’
Rule 3
In addition to factual relationships, texts may include probability relationships. They may occur in
any kind of sentence and are only different from other sentences because they indicate a probable
rather than definite positive or negative relationship (Wrightson 1976). Key words to look for to
identify a probability relation can be found just preceding the cause or effect concept. Some example
of such words are ‘likelihood’, ‘chance of’, ‘may’, or ‘could’ (Wrightson 1976: 301). In the sentence
‘Draghi’s speech may calm the markets’, the word ‘may’ indicates a probability (positive)
relationship. In contrast to Wrightson’s coding scheme, in this study such probability relations will
not be coded. Only actual relations (in the mind of the actor making the statement) will be coded.
18
Rule 4
Another type of relationship are those including pronoun cause- or pronoun effect concepts like
‘this’, ‘it’, ‘he’, ‘she’, ‘we’ (Wrightson 1976: 303). These pronouns refer to concepts that have been
mentioned earlier in the text. Running into such pronoun concept, the coder must search in the text
for the concept that is represented by the pronoun. Once the coder identifies the substantive
concept that is referred to by the pronoun, it should be included in brackets in the coding of the
phrase. For example, when coming across the sentence ‘he was continuously grumpy and this did not
improve the ambiance in the Eurogroup’, the coder should go back in the text and look for the
concept that is represented by the pronoun ‘he’. If the coder finds out that ‘he’ actually refers back
to ‘minister Dijsselbloem’, the relationship should be coded as follows:
He [minister Dijsselbloem] was continuously grumpy / - / ambiance in the Eurogroup
Rule 5
As indicated above, sentences may lack the formal grammatical structure associated with causal or
utility relationships. When a causal or utility relationship is nonetheless implicated in the content of
the texts, coders should logically derive the relationship and include it in the coding.
4
When relations
are derived from the ‘content analysis’ of a speech or text, coders should make the explicit effort to
question their judgments as in such cases there is an increased risk of reading one’s own perception
of the actor or source into a sentence.
Rule 6
Some sentences contain relationships in which the cause and effect (or utility) concepts are
mentioned in reversed order (Wrightson 1976: 305). This is the case, for example, in the sentence
‘Greece cannot secure debt reduction without Germany’s support’. Following the mere structure of
the sentence without paying attention to the content may suggest that because the phrase ‘securing
a debt reduction’ precedes ‘German support’ it causes it. Closer analysis of the content of the
sentence reveals that in reality the relationship is reversed and Germany’s support is the cause
concept, therefore the example should be coded as follows:
Germany’s support / + / Greece securing a debt reduction
4
For examples see Wrightson (1976: 305)
19
Rule 7
Concepts may also have a dual role in a section and be a cause concept at one point and an effect
concept at another. This may even occur within the same sentence (Wrightson 1976: 315). The figure
below illustrates this type of sentence structure.
An example of a sentence in which this is the case is: ‘Italy’s budgetary policy will make the country
the next target of speculators which would increase its sovereign bond yields and put the country
under an unbearable pressure’. In this case two relations may be distinguished in which ‘Italy being
the next target of speculators’ and ‘increase in Italy’s sovereign bond yields’ function as both an
effect as well as cause concept. As such, the sentence should be coded as:
Italy’s budgetary policy / + / Italy as the next target of speculators
Italy as the next target of speculators / + / increase in Italy’s sovereign bond yields
increase in Italy’s sovereign bond yields / - / [unbearable pressure for Italy] the benefit of Italy
Rule 8
Sometimes the dual role concepts play is less obvious and only implied, because they are part of as a
chain of interdependent events or chronological listing of events. In those cases, relationships may
be deduced from the structure and punctuation of the sentence (Wrightson 1976: 317). An extended
version of the sentence used as an illustration in Rule 7 may, for instance, have appeared as follows
in a text: ‘Italy’s budgetary policy will increase its debt; make the country the next target of
speculators; increase its sovereign bond yields and then put the country under unbearable pressure’.
It is unclear to what extent this sentence represents a chain of events as described under rule 7 or a
20
tree structure in which ‘Italy’s budgetary policy’ is the direct cause of ‘making the country the next
target of speculators’, ‘increase in sovereign bond yields’ as well as ‘putting the country under
pressure’. In such cases, coders again have to analyse the content of the argument to come up with
the proper coding. In this case, economic logic supports the interpretation of the sentences as a
chain of events and the coding should be similar to that under rule 7. The sentence ‘the budgetary
policy of Italy has three effects: 1) it will make the country the next target of speculators; 2) increase
its sovereign bond yields and 3) put the country under unbearable pressure’, however would indicate
a tree structure and should be coded as follows (Wrightson 1976: 318):
‘the budgetary policy of Italy / + / make the country the next target of speculators’
‘the budgetary policy of Italy / + / increase its sovereign bond yields’
‘the budgetary policy of Italy / - / benefit of Italy’
Rule 9
Wrightson also notes that, despite the strict selection criteria used in this study, sections included in
the analysis may contain relations that are irrelevant to the purpose of the analysis. Contrary to
Wrightson’s advice, in the present study this will not be a consideration that warrants the exclusion
of relations. As the topic under study is very complex and many different perspectives of the Euro
crisis exist, this would seriously jeopardize the validity of the findings. Only when in a (introductory or
concluding) section of a speech statements are included regarding the occasion, place or host of the
event where the speech is held, the coder is required to exercise the discretion to omit such
statements from the analysis.
Rule 10
Finally, some sentences seem to include a causal relationship, while actually they do not. For
example, the sentence ‘since the downgrading of Greece’s bonds, the ECB has been unable to calm
the markets’ may be interpreted as including the following causal relationship:
Downgrading of Greece’s bonds / - / ECB ability to calm the markets
On the other hand, in this case it is more plausible that the phrase ‘since the downgrading of
Greece’s bonds’ is meant to denote a mere reference point in time, and not a causal agent. In other
cases, however, the existence of a causal relation may be more plausible. In any case, such
ambiguous assertions should not be coded as causal unless there is clear evidence that the speaker is
21
making a causal assertion (Wrightson 1976). This rule of thumb also applies to statements of
definitions which may take on the appearance of a causal assertion like in the sentence ‘the
monetary policy of the Eurozone is its ability to defend its liquidity’ as this statement is a description
of what monetary policy is about and true by definition, it should not be coded.
Coding discourse: some special rules
The coding of the newspaper articles may raise some difficulties for the regular cognitive mapping
coding rules. Classic cognitive mapping looks only at the causal and utility relations in a text that
reflect on the position of the writer or speaker. Especially in the Op-ed pieces that will be analysed in
this study, several positions and perspectives on an issue are presented. As indicated above, in the
coding of leaders’ speech acts only the position of the leader is coded (see page 16). However, as the
newspaper articles are used as a proxy for the public and expert discourses, not only the position of
the authors but also those they refute are of interest. More specifically, Op-ed pieces may present
the positions of the author, its allies and its opponents which are all part of the discourse. To present
a complete and representative image of the expert and public discourses but still be able to classify
the views of certain newspapers and/or authors coders need to list which voice (self, ally or
opponent) makes the statement with every relation that is coded. The coding software MAPS will be
equipped with a feature to enable coders to do this.
Merging and standardisation
The coding rules described above all concern the first raw coding of the sources. However, to enable
comparison over time and between leaders beliefs and public and expert discourses some
standardisation of the raw concepts in the maps is needed. Also all the concepts included in the
cognitive maps should be unique. As such, in the end all synonyms in the database of cognitive maps
that is being created in the project, should be grouped and merged under a single ‘standardised’
concept. Unfortunately, no mathematically measure exists to establish the proper level of
standardisation (Wang & Laukkanen 2015; Young 1994).
5
In a practical sense, a sufficient high level
5
One should not overemphasize the validity-concerns associated with the process of standardization. In
essence, the more one standardises the concepts in a map, the more one shifts the study from an ideographic
to a nomothetic study. While the former score higher on construct-validity the later perform better in terms of
reliability. Choosing the right level of standardisation is thus in essence a matter of striking the right (or
preferred) trade-off between construct-validity and reliability. The only difference is that in nomothetic studies
the level of standardisation is determined prior to data-collection, in CM research, part of this decision lies
22
of standardisation should be achieved for the coding in terms of standardised concepts to be
possible
6
and a sufficient level of intercoder reliability to be achieved. As such, idiosyncratic events
and concepts should be placed under the label of the broader phenomenon of which such event or
concept is an instance of (Heradstveit & Narvesen 1978: 81). On the other hand, one must be careful
not to eliminate meaningful distinctions that are made in the original sources. All in all, the level of
standardisation must take the middle road to doing justice to the actual meaning of the statements
made and suiting the empirical and theoretical focus of the study at hand.
The right level of standardisation is different for each study and cannot be generalised. The
reason why this is the case may best be explained by referring to the Inuit-words- for-snow effect
7
which says that the more crucial a phenomenon, the more important small differences in concepts
are, and -as such - the more concepts with a similar meaning should be included in the set of
standardised concepts. For instance, in a study on US foreign policy one may probably safely merge
the terms European Union, European integration and European cooperation. However, in a
comparison of French and German policy-preferences concerning political unification such concepts
harbour empirically and theoretically important differences in preferences for the form and level of
integration, and thus should be maintained in the final list of merged concepts.
In order to increase the validity and reliability of the standardising process a set of
procedures should be abided to: Firstly, before analysing a text, the scholar should read through the
entire text to come to grips with the possible importance of subtle distinctions in concepts.
Moreover, in order to make such distinctions, coders should have some knowledge of the policy-area
under study to be able to properly assess the meaning of such distinctions. Secondly, during the text-
analysis phase, transparency is key: the links between standardised and original concepts as used by
the leader should be readily accessible and traced back to the exact place in the text they originate
from (Bonham & Shapiro 1986). Cognitive mapping coding software MAPS guarantees transparency
in this matter. Thirdly, concepts with the lowest saliency and centrality values are the most likely
candidates to be merged into a more dominant akin concept. Finally, the raw coding should be
retained and remain accessible at all times. Since, in work package 3 of the TRANSCRISIS project an
between data-collection and analysis. Naturally, in all CM-studies some theoretical and empirical focus has
already been introduced in the initial choice of the (sections of) sources chosen for analysis.
6
CM coding software MAPS facilitates standardization but there is a limit to the amount of standardized
concepts coders can reliably work with.
7
While the 'Inuit words for snow' example is very a well-known and adequate metaphor to describe the nature
of the standardisation-process, the actual claim that Inuit languages contain more words for snow is false. Like
other languages, most Inuit languages use suffixes to indicate differences in, for instance, types of snow.
23
entire database of cognitive maps will be created, additional merging of concepts may be needed
when all coding is completed to enable the analysis and comparison of the maps.
Inter coder reliability
The formalization of the cognitive map coding rules has significantly increased the reliability of the
coding process (Wrightson 1976; Bonham & Shapiro 1986). In fact, the intercoder reliability of CM
studies is usually fully compatible with the accepted standards of good quantitative work in the social
sciences (Axelrod, 1976, p.84). To reach an optimal level of intercoder reliability, coders should
receive extensive and a careful training (Young and Schafer, 1998). Before the coding process starts
all coders working in the project of Work package 3 will receive such a training.
In addition, all sources will be double-coded and will be submitted to intercoder reliability
checks. To calculate the intercoder reliability of CM coding the three key tasks of the cognitive map
coding process should be taken into account (Axelrod 1976: 85). To determine the intercoder
reliability score the correspondence in the following tasks will be established:
1. Agreement on the identification of the number of codable assertions in each section;
2. Agreement on the identification of each of the three parts of relations that have been
identified (cause concept, effect/utility concept and sign of relation);
3. Agreement on the identification of the proper merged concept for each of the concepts in
the relation.
As the latter two tasks are dependent on the earlier ones, intercoder reliability will only be calculated
for those assertions, concepts and relations that were coded identically in previous steps by both
coders.
References
Axelrod, R. (1976a). The Cognitive Mapping Approach to Decision Making. In R. Axelrod (Eds.),
Structure of Decision: The Cognitive Map of Political Elites. Princeton: Princeton University
Press, 3-17.
Axelrod, R. (1976b). Limitations. In R. Axelrod (Eds.), Structure of Decision: The Cognitive Map of
Political Elites. Princeton: Princeton University Press, 251-265.
Bastian M., Heymann S., & Jacomy M. (2009). Gephi: an open source software for exploring and
manipulating networks. International AAAI Conference on Weblogs and Social Media.
24
Baumgartner F.R. (2014). Ideas, paradigms and confusions. Journal of European Public Policy, 21(3),
475-480.
Bonham, G.M., Shapiro M.J. (1986). Appendix 2: Coding Instructions for the CM Procedure. In. I.N.
Gallhofer, W.E. Saris & M. Melman (Eds.), Different Text Analysis Procedures for the Study of
Decision Making. Amsterdam: Sociometric Research Foundation, 125-139.
Bougon, M., Weick, K., & Binkhorst, D. (1977). Cognition in organizations: An analysis of the Utrecht
Jazz Orchestra. Administrative Science Quarterly, 22(4), 606-639.
Carmassi, J., & Micossi, S. (2010). The role of politicians in inciting financial markets to attack the
eurozone. EuropEos, CEPS, Brussels, 4.
Chakrabortty, A. (2011). Q&A: Greece’s debt crisis. The Guardian, 31 July. Retrieved from:
http://www.theguardian.com/world/2011/jul/31/greece-debt-crisis-how-bad
Clarkson, G.P., & Hodgkinson, G.P. (2005). Introducing Cognizer™: A comprehensive computer
package for the elicitation and analysis of cause maps. Organizational Research Methods, 8(3),
317-341.
Curseu, P.L., Schalk, M.J., & Schruijer, S.G. (2010). The use of cognitive mapping in eliciting and
evaluating group cognitions. Journal of Applied Social Psychology, 40(5), 1258-1291.
De Vaus, D. (2014). Surveys in Social Research: Sixth edition. London and New York: Routledge
Entman, R. M. (1993). Framing: Towards clarification of a fractured paradigm. McQuail's reader in
mass communication theory, 390-397.
European Commission (2010). Economic and Financial Affairs: Euro area and IMF agreement on
financial support programme for Greece. Retrieved from:
http://ec.europa.eu/economy_finance/articles/eu_economic_situation/2010-05-03-
statement-commissioner-rehn-imf-on-greece_en.htm
Fiske, S. T., & Taylor, S. E. (1991). Social cognition. NY: McGraw-Hill.
Foucault, M. (1969). The archaeology of knowledge. Paris: Éditions Gallimard.
Gallhofer, N., Saris W.E., & Melman, M. (1986). Different text analysis procedures for the study of
decision making. Amsterdam: Sociometric Research Foundation.
Goldstein, J. & Keohane, R.O. (1993). Ideas and Foreign Policy: Beliefs, Institutions, and Political
Change. Ithaca: Cornell University Press.
Golder, M. & Stramsky, J. (2010). Ideological congruence and electoral institutions. American Journal
of Political Science, 54(1), 90-106.
Gocaj, L., & Meunier, S. (2013). Time will tell: The EFSF, the ESM, and the euro crisis, Journal of
European Integration, 35(3), 239–253.
25
Hall, P.A. (1993). Policy Paradigms, Social Learning, and the State: The Case of Economic Policymaking
in Britain. Comparative Politics, 25(3), 275-296.
Hart, J.A. (1977). Cognitive Maps of Three Latin American Policy Makers. World Politics, 30(1), 115-
140.
Heradstveit, D., & Narvesen, O. (1978). Psychological constraints on decision-making. A discussion of
cognitive approaches: Operational code and cognitive map. Cooperation and Conflict, 13(2),
77-92.
Hobolt, S.B., & Klemmemsen, R. (2005). Responsive government? Public opinion and government
policy preferences in Britain and Denmark. Political studies, 53, 379-402.
Hodgkinson, G. P., Maule, A. J., & Bown, N. J. (2004). Causal cognitive mapping in the organizational
strategy field: A comparison of alternative elicitation procedures. Organizational Research
Methods, 7(1), 3-26.
Hodson, D., & Quaglia, L. (2009). European Perspectives on the Global Financial Crisis: Introduction.
JCMS volume 47. Number 5. pp.939-953
Jervis, R. (2006). Understanding Beliefs, Political Psychology, 27(5), 641-663.
Laukkanen, M., & Wang, M. M. (2015). Comparative Causal Mapping: The CMAP3 Method. Farnham:
Ashgate Publishing, Ltd.
Levy, J.S. (1994). Learning and Foreign Policy: Sweeping a Conceptual Minefield. International
Organization, 48(2), 279-312.
Lindeboom, G (2012). Public priorities in government’s hands: corresponding policy agendas in the
Netherlands. Acta Politica, 47, 443-467.
Marfleet, B.G. (2000). The operational code of John F. Kennedy during the Cuban missile crisis: A
comparison of public and private rhetoric. Political Psychology, 21(3): 545-558.
Markiczy, L., & Goldberg, J. (1995) A method for eliciting and comparing causal maps. Journal of
Management, 21(2), 305-333.
Meiers, F. (2015). Germany’s Role in the Euro Crisis: Berlin’s Quest for a More Perfect Monetary
Union. New York: Springer.
Picard, R.G. (2015). The Euro Crisis in the Media; Journalistic Coverage of Economic Crisis and
European Institutions. London: I.B Tauris & Co.
Princen, S., & 't Hart, P. (2014). Putting policy paradigms in their place. Journal of European Public
Policy, 21(3), 470-474.
Princen, S.B.M., & Van Esch, F.A.W.J. (forthcoming). Paradigm formation and paradigm change in the
EU’s Stability and Growth Pact. European Political Science Review.
26
Raadschelders, J., & Vigoda-Gadot, E. (2015). Global Dimensions of Public Administration and
Governance: A Comparative Voyage. Hoboken: John Wiley & Sons.
Renshon, J. (2009). When public statements reveal private beliefs: Assessing operational codes at a
distance. Political Psychology, 30(4): 649-661.
Risse-Kappen, T. (1994). Ideas do not float freely: Transnational coalitions, domestic structures, and
the end of the cold war. International Organization, 48(02), 185-214.
Dyson, S. B., & Raleigh, A.L. (2014). Public and private beliefs of political leaders: Saddam Hussein in
front of a crowd and behind closed doors. Research & Politics 1(1).
Sabatier, P.A., & Jenkins-Smith, H.C. (1993). Policy change and learning: An advocacy coalition
framework. Boulder: Westview.
Van Esch, F.A.W.J. (2007) Mapping the road to Maastricht: A comparative study of German and
French pivotal decision makers' preferences concerning the establishment of a European
Monetary Union during the early 1970s and late 1980s. (Unpublished doctoral thesis),
Radboud Universiteit, Nijmegen.
Van Esch, F.A.W.J. (2012). Why Germany wanted EMU. The Role of Helmut Kohl's Belief-System and
the Fall of the Berlin Wall. German Politics, 21(1), 34-52.
Van Esch, F.A.W.J. (2014). Exploring the Keynesian-Ordoliberal Divide. Flexibility and Convergence in
French and German Leaders' Economic Ideas during the Euro-crisis. Journal of Contemporary
European Studies, 22(3), 288-302.
Van Esch, F.A.W.J., Joosen, M.C. & Van Zuydam, S. (under review 2016). Representing the People?:
Comparing the European Cognitive maps of Dutch Political Leaders and their Followers. Politics
& Governance.
Van Esch, F.A.W.J, Steenman, S.C., Snellens, J.F.A. & Bakker, W.E. (2014). Beyond ‘For or Against’ the
EU. Cognitive Mapping as a method to reveal the substantive reasoning behind people’s
perceptions of Europe. Paper presented at the Conference In Search of European Political
Union, Utrecht University, 19-21 June 2014.
Verbeek, B. (1990). Cognitieve Ideeën en Internationale Politiek. Actis Politica, 25(1), 95-116.
Verdun, A. (2015). A Historical Institutionalist Explanation of the EU's Responses to the Euro Area
Financial Crisis. Journal of European Public Policy, 22(2), 219-237,
Walker, S. G., & Schafer, M. (2000). The political universe of Lyndon B. Johnson and his advisors:
Diagnostic and strategic propensities in their operational codes. Political Psychology 21(3),
529-543.
Wrightson, M.T. (1976). The Documentary Coding Method. In R. Axelrod (Eds.), Structure of Decision.
The Cognitive Maps of Political Elites. Princeton: Princeton University Press, 291-332.
27
Young, M. D. (1994). Foreign policy problem representation and President Carter (Doctoral
dissertation, The Ohio State University).
Young, M.D. (1996). Cognitive Mapping Meets Semantic Networks. A special issue: Political
Psychology and the Work of Alexander L. George. The Journal of Conflict Resolution, 40(3), 395-
414.
Young, M.D., & Schafer, M. (1998). Is there Method in our Madness?: Ways of Assessing Cognition in
International Relations. Mershon International Studies Review, 42(1), 63-96.
