Roel Pieter Bolsius

0459259

Leiden, June 2012

THE COMPLEX RELATIONSHIP BETWEEN PRESS FREEDOM

AND

CORRUPTION

Master of Science thesis Comparative Politics and Democracy

Institute of Political Science, faculty of Social and Behavioural Sciences
Leiden University

Supervisor:
Dr. Daniela Stockmann

Second reader:
Dr. Maria Spirova
2
TABLE OF CONTENTS

ACKNOWLEDGMENTS............................................................................................................................ 5

CHAPTER 1: INTRODUCTION ............................................................................................................... 7

CHAPTER 2: LITERATURE REVIEW ................................................................................................... 9

CORRUPTION ............................................................................................................................................ 10
Defining Corruption............................................................................................................................ 10
The Political and Economical Consequences of Corruption .............................................................. 14
THE RELATIONSHIP BETWEEN PRESS FREEDOM AND CORRUPTION ............................................................ 16
Mechanisms ........................................................................................................................................ 17
Criticisms ............................................................................................................................................ 20

CHAPTER 3: THE PUZZLE AND RESEARCH QUESTION ............................................................. 23

VARIABLES AND DATA ............................................................................................................................. 23

CASE SELECTION ...................................................................................................................................... 26
ANALYSIS ................................................................................................................................................. 26
FINDINGS AND RESEARCH QUESTION ....................................................................................................... 28

CHAPTER 4: THEORY AND HYPOTHESIS........................................................................................ 31

CHANGING ATTITUDES TO FIGHT CORRUPTION........................................................................................ 31

ATTITUDE CHANGE BY THE MEDIA .......................................................................................................... 32
HYPOTHESIS ............................................................................................................................................. 36

CHAPTER 5: POLITICAL AWARENESS AND CORRUPTION ....................................................... 39

MEASURING POLITICAL AWARENESS ....................................................................................................... 39

VARIABLES AND DATA ............................................................................................................................. 41
ANALYSIS ................................................................................................................................................. 43
FINDINGS .................................................................................................................................................. 45

CONCLUSION AND DISCUSSION ........................................................................................................ 47

BIBLIOGRAPHY ...................................................................................................................................... 51

APPENDIX 1: REPORTERS WITHOUT BORDERS 2010.................................................................. 57

APPENDIX 2: FREEDOM HOUSE PRESS FREEDOM ...................................................................... 58

3
APPENDIX 3: TRANSPARENCY INTERNATIONAL’S CPI 2010 .................................................... 59

APPENDIX 4: FREEDOM HOUSE DEMOCRACY ............................................................................. 61

APPENDIX 5: POLITICAL AWARENESS EES 2009 .......................................................................... 65

APPENDIX 6: POLITICAL AWARENESS CSES MODULE 3 ........................................................... 67

APPENDIX 7: IMPERFECTIONS CSES MODULE 3.......................................................................... 69

4
ACKNOWLEDGMENTS

“Yes, to dance beneath the diamond sky with one hand waving free.”
Mr. Tambourine Man, Bob Dylan

Until a year ago I had never thought about actually graduating in Political Science. This
thesis however shows that I changed my mind somewhere along the way. I am happy
that I did. Especially this last year was extremely satisfying. Therefore I would like to
thank everyone that taught me.
In particular I am thankful to dr. Daniela Stockmann for supervising me. Without
her comments and assistance this thesis would have looked differently—and not for the
better.
I am also grateful to dr. Maria Spirova, who happens to be the second reader of
this thesis. She taught me to think of corruption as an interesting study object.
Furthermore, I would like to convey thanks to Bart Hoving, Remy Koolschijn and
Rianne Simons for commenting on drafts. Also Jessica Kelder and Roelof Lammers made
great comments, but I am even more thankful to them for the many discussions we
have had about this thesis and thesis writing in general.

5
6
CHAPTER 1: INTRODUCTION

The ongoing financial crisis in Europe has caused many sleepless nights among the
heads of government. While they are imposing unpopular (financial) measures on their
citizens in order to maintain the economical stability, corruption is fuelling this crisis
(Transparency International, 2012). That corruption is not only causing problems in the
“far-away” non-democracies but also rants in the “nearby” European democracies,
shows that attempts to destroy corruption have not always been effective, not even in
well established democracies. Despite the general idea that more democracy, openness
and transparency would lead to lower levels of corruption, the crisis shows that
corruption is still causing harm in these relatively democratic, open and transparent
societies. In order to free all societies from the consequences of corruption, more
research is necessary on the mechanisms that cause corruption and specific effects of
these processes on corruption.
In this thesis the focus is on the relation between press freedom and corruption.
Because of the hidden nature of corruption, free and independent media are believed to
be important in the battle against corruption by uncovering corruption. So far, one body
of the literature focuses on why the media is capable of fighting corruption, whereas the
other body is busy demonstrating why the found negative association between press
freedom and corruption is overstated or caused by other factors than the media. In this
thesis I steer a middle course: I do acknowledge the corruption curbing capabilities of
the press, although I argue that this relationship is more complex than it appears to be
at first sight. As is demonstrated in chapter 3, the marginal effects of the press differ
across the levels of corruption in democracies; news media are more effectively curbing
corruption as the corruption level of the country is lower. Thus, it is the context in which
the media operate that the effectiveness of the media can be on corruption, not the
mere existence of free media itself.

7
 This begs the question what exactly makes that the effectiveness of free press to
curb corruption is stronger as the corruption in a country is lower. I aim to answer this
question in this thesis by reflecting a cross-national comparative study I have
conducted. In order to be able to present the analysis it needs to be established what
corruption actually encompasses and why it is such a problem. It is also necessary to
illustrate the current knowledge about the negative relationship between the press and
corruption in order to understand how this thesis contributes to this field of research.
Therefore this thesis is continued with an extensive review of the literature (chapter 2).
As already mentioned, I acknowledge the negative relationship between free press and
corruption, but I consider this relationship to be complex. This is shown in chapter 3.
The analysis itself is build around a theoretical framework that explains the factors that
are influencing the capabilities of media to curb corruption. This model is discussed in
chapter 4. Chapter 5 provides for the empirical illustration that this model indeed
explains the differences in the marginal effects of the press on corruption between
different countries. In the final chapter of this thesis the findings are discussed as well as
the implications of these findings on the existing literature.

8
CHAPTER 2: LITERATURE REVIEW

In order to analyze the relation between press freedom and corruption, it is necessary
to work with a clear, unambiguous and measurable definition of corruption that holds
for every country. This is important, because any research effort that involves
corruption is heavily influenced by how it is defined (Kurer, 2005: 2002). Firstly, because
the definition for corruption determines what exactly is tried to capture when
measuring corruption. Secondly, any definition of corruption specifies what is
understood with corruption and, as a consequence, which strategies are qualified to
curb corruption (Andersson & Heywood, 2009: 750). Therefore, in the first section of
this chapter I discuss which different strategies there are to define corruption and what
implications these different strategies have on comparable research on corruption. This
section concludes with the argumentation for using Transparency International’s
definition for corruption in this thesis. This is followed by a section that shows that
corruption is a problem for political and economical reasons and that research to
corruption is therefore justified and even necessary in order to be able to actively fight
corruption.
After corruption is defined, I turn the focus to the relationship between freedom
of news media and corruption—the subject of the empirical analysis in this thesis. In
that section it is illustrated, using the current literature, that the corruption curbing
effects of the press are most significant when the press freedom is high. The alleged
mechanisms are discussed, just as the critiques on the empirical evidence of this
relationship. The chapter is concluded with a review of that body of the literature that is
less convinced about the causality of the relationship between press freedom and
corruption. I conclude with this, because although I adhere to the position that the free
media are indeed an effective weapon in the fight against corruption, I also
acknowledge that the relationship should not be overstated.

9
Corruption

Defining Corruption

Most of the definitions of corruption that are used by various scholars originate from
Nye’s (1967) definition. He defines corruption as:

“[B]ehaviour which deviates from normal duties of public role because of private-
regarding (family, close private clique), pecuniary or status gains; or violates rules
against the exercise of certain types of private-regarding influence. This includes
such behaviour as bribery (use of rewards to pervert the judgment of a person in a
position of trust); nepotism (bestowal of patronage by reason of ascriptive
relationship rather than merit); and misappropriation (illegal appropriation of
public resources for private-regarding uses)” (Nye, 1967: 966).

This definition and, therefore, most follow-up definitions rest upon three important
requirements: firstly, the emphasis on the public role an official fulfils when he conducts
his abusive behaviour; secondly, the requirement of personal gain; and thirdly, officials’
behaviour can only be corrupt when the behaviour “deviates from normal duties”. In
order to demonstrate the difficulties for comparable research one runs into when an
attempt of defining corruption is made, I discuss the implications of the three
requirements separately.
The first requirement—behaviour has to be expressed in the public role of the
official—is little debated in the literature. Obviously, an official can also conduct
questionable behaviour in his private life, such as beating his wife or stealing from a
shop, but this is not corruption. This requirement also separates corruption from other
but related kinds of misbehaviour that are conducted by private organisations, such as
business corruption and fraud, (Gardiner, 2002: 28). The second requirement already
leads to more debate, because it is unclear how far the requirement for private-
regarding gain reaches. Clearly, if the official personally benefits from his public
misbehaviour this has to be considered as private-regarding. If the misconduct leads to
10
beneficial effects for the direct family, then most scholars also consider this as private-
regarding gains. However, when misbehaviour of the official is mainly benefiting a
specific constituency, political or ethnical group it is less clear whether the requirement
of private-regarding gain is met. This dispute is largely related to the debate concerning
the third requirement, that corruption is behaviour that deviates from “normal”
responsibilities. Especially this condition contributes majorly to the variations there exist
in what is considered as a corrupt act between countries. Moreover, it can be that
within the same country at various points in time there are differences in what is
considered to be a normal duty for an official. And to make it even more complex, even
within one society there can be strong individual differences in perception of what is
considered as normal; what for some people might be normal, is perceived as abnormal
and immoral behaviour by others.
It is this dispute about what has to be considered as normal behaviour that
makes it hard to define corruption in such a way that a definition holds in a comparative
research setting. The different strategies to determine “normal duties” underlie the
categorization of the definitions for corruption. The different categories Gardiner (2002)
identifies are the legal definitions, the public interest definitions and the public opinion
definitions. All these types of definitions come with their own difficulties and
complexities for the use in a comparative research framework.
The legal definitions use the law as main criterion and are the most tangible
method to define corruption. If behaviour is prohibited by the nation’s laws and,
obviously, also meets the requirement of the public role and private benefits, then this
misbehaviour can be considered as corruption. However, if the behaviour is not
prohibited by the law then it cannot be labelled as corruption. Scott (1972: 5) rightfully
sees problems in using only legal definitions. Firstly, not everything that is legal is
necessarily ethical. Thereby, countries have different laws, and consequently in different
countries the same behaviour is judged differently. A comparative study to corruption
would be nearly impossible when the law determines what the definition of corruption

11
is, because there is variation in what kind of behaviour have to be considered as
corruption. This makes it hard to interpret the results of such a study.
Using the public interest in order to determine whether behaviour is corrupt
comes down to the following: “if an act is harmful to the public interest, it is corrupt
even if it is legal; if it is beneficial to the public, it is not corrupt even if it violates the
law” (Gardiner, 2002: 32). To use this strategy in a comparable framework it has to be
assumed that the public interest is equal across the planet, otherwise the same difficulty
arises that arise when the legal definition is applied for corruption. In addition, even if
the public interest around the world would be similar, the question is how public
interest can be measured and which values should be included in these measurements.
Suppose that parliaments are be able to express the “general will” and therefore be able
to represent the public interest, this “public interest” strategy would be equal to the
legal strategies of defining normal duties. However, when it is assumed that the
legislature cannot always express the public interest, then this would mean that it would
be ethical to conduct illegal behaviour sometimes. This would create an internal
ambiguity in such a definition for corruption; corruption is not acceptable behaviour,
but at the same time it is acceptable. Thus, defining corruption purely as a result of the
public interest also gives complexities.
The last strategy for determining normal public officials’ behaviour that is
considered by Gardiner (2002: 32) is to use public opinion to determine the moral
standards. However, also this method leads to variation between countries, because the
people in different countries might have different moral standards. This makes
comparisons between countries problematical. Even more challenging is that even
within the same country there might also exist discrepancies about what is acceptable
behaviour and what is not. Therefore using a public opinion for corruption is also not
completely satisfactory in a comparable framework.
Despite the difficulties all strategies bring, it is necessary to select one definition.
In this thesis the Transparency International’s definition for corruption will be used. This
NGO’s mission is to stop corruption and to promote transparency around the world. The

12
data about corruption that is used in this thesis is mainly obtained from this NGO,
therefore the definition of corruption that is used in this thesis is equal to Transparency
International’s definition for corruption: “the abuse of entrusted power for private gain”
(Transparency International, 2009: 14). The reason for the selection of this definition is
that Transparency International was the first to provide for a yearly systematic attempt
to quantify corruption around the world. Their enormous amount of data is a major
source. Consequently, by using this data I am doing research to the phenomenon
corruption as is defined by that NGO. However, it is not only the availability of data that
underlies this selection. Because of the pioneering role of Transparency International in
measuring corruption and the use of their data by various scholars, they have shaped
the research field. Much of what is known about the factors involved with corruption
directly or indirectly derives from Transparency International: “Without the CPI, it is
doubtful whether many secondary studies which seek to identify the causes of
corruption would have been undertaken, since the index offers an ideal large-n basis for
analysis” (Andersson & Heywood, 2009: 57).
Nevertheless, also Transparency International’s definition leaves room for
variation among countries. Even more so, because Transparency International draws
there information upon surveys held among country experts, journalists and mainly
Western business leaders by various organisations. Thus, it is the aggregate of how all
the individual respondents define corruption that actually should be the definition that
is used in this thesis. This means that, although Transparency International gives
guidance by promoting to use their formal definition, the measure reflects the
interpretation of corruption by the respondents. It can therefore never be completely
sure what the nature of the phenomenon is that is studied. Although, the definition
leaves some room for interpretation and Transparency International provides data on
corruption that might be slightly off from their definition for corruption, many scholars
on corruption have advocated to let Transparency International’s data have a role in
comparative research (for example Johnston, 2000; Senior, 2006; Rose-Ackerman,
2008). Therefore, I am confident enough to use their definition and their data.

13
Nevertheless, considering all the possible pitfalls, careful interpretation of any results is
required.

The Political and Economical Consequences of Corruption

The great tenacity by those who try to fight corruption is grounded in the harmful
consequences of corruption. Firstly, corruption has harmful effects on the economy and,
secondly, corruption is conflicting with the basic principles of democracy. The harmful
effects on the economy are well described by Mauro (1995). He shows that corruption
lowers investments and are therefore weaken the economic growth1. These negative
consequences for the economy are even more a problem for countries that need
economical development. For example, the governmental actions in poor states are
fundamental for their economic development. This developmental process is extremely
fragile and therefore it requires “[…] single-minded hard work from all holders of public
office. If the top political elite […] consumes its time and energy in trying to get rich by
corrupt means, it is not likely that the developmental plans will be fulfilled” (Leys, 2002:
70). It is also imaginable that the improperly obtained wealth of the political elite is
secured by putting in foreign bank accounts in stead of investing it in their countries,
which will have a negative effect on the economy. Another negative effect on the
economy can be that donor countries decrease their investments, because the morality
of the elite is not in agreement with those of the countries. In rich countries the
negative consequences on the economy caused by corruption have less impact.
Nevertheless, also these countries’ societies are vulnerable for the harmful effects of
corruption. Although low levels corruption has less impact individually, its aggregates
might have an enormous negative effect on the (economic) society.
Although for the (mainly rich) democracies, corruption may have less impact on
the economy, this kind of behaviour is still harmful; especially for those who adhere to

1
A more extensive discussion on the exact numerical effects on the economy can be found in Mauro
(2002).

14
the normative position that democracy is desirable. “Corruption […] breaks the link
between collective decision making and people’s power to influence collective decisions
through speaking and voting, the very link that defines democracy” (Warren, 2004: 328).
There are other political phenomena that exclude people from their equal share to
power—indeed, a prime-minister has more power than a student—but in a true
democracy those positions are not excluded for parts of the society beforehand.
Therefore these phenomena are not necessarily conflicting with democratic norms.
But it is not only exclusion as a result of corruption that is a problem for a
democracy. Corruption lowers the trust in political institutions and social trust too
(Rothstein, 2000; Bjrønskov, 2004). If the corrupt patterns of the political elite are well-
known, this is likely to rob them of their authority (Philp, 2001). The authority challenge
that results from corruption is one that may occur in every state, both rich and poor,
that is faced by corrupt political elites. Corrupt elites lack the authority to prevent the
“common man” to do the same. Thereby, social trust is related to economic equality
(Rothstein & Uslaner, 2005). “This implies that with a high level of social trust, the best
of both worlds is possible – economic prosperity goes together with a reasonably fair
distribution of resources and democracy” (Rothstein & Eek, 2009: 82). Also on the
individual level social trust correlates with many aspects that are considered to be good,
such as education, personal happiness, health and tolerance towards minorities
(Uslaner, 2002).
Despite the acknowledgement by most scholars that corruption can cause great
harm, some argue that corruption in the right context might not always be harmful and
might sometimes even have beneficial consequences. Gardiner (2002: 37), for example,
gives several examples of relatively harmless or alleged healthy effects of corruption:
bribes can make the system work despite the inefficiencies of the legal, official system;
corruption can divert the scarce resources of a country and may erode the status and
influence of the authoritarian elite; and the prospect of gains from corruption might also
attract officials to public offices that otherwise would have been empty or filled with
less capable people, because of the low salaries. Rose-Ackerman (2002: 353-371)

15
evaluates in more detail under what conditions corruption is harmful and when not.
One of the would-be positive aspects of corruption she has analyzed is when people
arrange themselves with corruption to avoid (or ease) inefficient rules. How attractive
and positive the consequence of corruption might seem, investors cannot be trusted
that their only aim is to avoid inefficient burdens, “they will, instead, want to reduce the
impact of all state-imposed burdens, justified or not” (Rose-Ackerman, 2002: 367). It is,
therefore, highly questionable whether all the alleged positive consequences of
corruption, are so positive after all. Thereby, it is uncomfortable, at least, from the
perspective of a Rechtsstaat or the rule of law, that it is a firms’ judgment whether some
bribes are justified. This can do serious harm to the authority and sovereignty of the
state, especially in those societies that are struggling building a viable state.
In the previous paragraphs I have tried to demonstrate that corruption can have
harming consequences for both the political and economic arena of a society. Despite
the great damage it can cause, however, it is still a worldwide existing problem. It is
commonly believed that free and independent press play an important role in the battle
against corruption. “if corruption involves harms caused by exclusion, a key means for
fighting corruption will involve empowering those harmed to protect themselves by
democratic means: with information, arguments, organization, and votes” (Warren,
2004: 341). Nevertheless, if free and independent media are indeed the Holy Grail, then
why is corruption still so widespread? Therefore the next section is devoted to the
relationship between medium freedom and corruption.

The relationship between press freedom and corruption

Hgh levels of press freedom are associated with low levels of corruption (see for
example Ahrend, 2002; Brunetti & Weder, 2003; Chowdhury, 2004). Ahrend (2002)
shows that this association cannot be attributed to spurious variables and that the
direction of this negative relationship runs from press freedom to corruption. Others
(for example Freille et al, 2007) believe that this relationship “simply picks up wealth
effects and the institutional environment more generally” (Freille et al, 2007: 839). In
16
order to better understand the relationship between free press and corruption, the
mechanisms are discussed that are believed to explain why the free press is capable to
curb corruption. Consequently, criticisms on this relation will be discussed. This section
concludes with my point of view on this debate.

Mechanisms

Reduction of corruption can be the result of two things: either the corrupt officials
themselves decide not to be corrupt anymore or the public officials are captured and
punished if they conduct corrupt behaviour. The media are a contributing factor to both
corruption curbing explanations. That the media can contribute to the first mechanism
is illustrated by Ahrend’s (2002) model of the influence of the press on corruption.
Similar to Treisman (2000) he argues that whether an official will behave corrupt is a
rational decision based on the expected profit against the expected costs2. This theory
sais that if the expected costs outbalance the expected benefits, then an official will not
behave corrupt. The major expected cost for a public official conducting corrupt
behaviour is being caught and being punished. A strengthening of the media enhances

2
The model Ahrend (2002: 7) provides is that public official’s utility (U) is equal to the official's wage (w)
plus the bribe that is expected minus the fine (F) that is expected. This is represented in the following
formula:

U(b, h, F, w) = w + B(b, hB)(1-PD(hB, b, M(hM))) – PD(hB, b, M(hm))*F

The rent the official extracts (B) is a function of the bribe rate (b) and the official’s human capital (hB).
Human capital is everything that a person has intellectually collected and gained during his professional
life and what gives him the ability to perform labour and brought him to the position holds. The human
capital corresponds with the value an official has to offer to the bribe payer. The probability that the
corrupt official will be caught and sentenced (PD) is depending on the rent he extracts and the functioning
of the media (M). The functioning of the media is also depending on the journalist’s human capital and
abilities (hm). For the situation in which the abilities of the media increase and every other variable
remains the same, it can be seen from Ahrend's (2002) utility equation that it becomes less likely for an
official that corruption is beneficial.
The rent an official may extracts from corrupt behaviour does not have to be financially. From the widely
used definition of corruption that corruption is “the abuse of public office for personal gain” (for example
Warren, 2004), it can be concluded that any personal gain from the abuse of public office has to be
considered as corruptive rent. Thus, Ahrend's (2002: 6-9) model can be well-translated to any form of
corruption.

17
the official’s believe that this actually happens3. An increase in the functioning of the
media, for example assigning more freedom to news media, results in a shift in the
balance between the costs and the benefits in the direction of the costs. Consequently,
enlarging the freedom of the media leads to less corruption. What, then, are the
media’s characteristics that make them increase the probability of detection and
punishment?
The work of Brunetti and Weder (2003) gives insight in the answer to this
question. They identify two types of corruption: extortive and collusive corruption. The
mechanisms of the free media on these types of corruption are different. Extortive
corruption is the kind of corruptive behaviour in which a public official seeks to extract
rent from a private actor by only providing a certain service he is designated to provide
when he gets a rent. For example, a public official only hands out a driving license after
the successful examinee gives a certain amount of money, a bribe. Faced with this kind
of blackmailing a private actor can decide to try to fight the corruption or to provide the
rent to the public official. The private actor also makes a rational decision of which
strategies leave him with more utility. In many circumstances the costs of paying the
bribe outbalance the costs of fighting the corruption. The free and independent media
act as additional channels of external control and are therefore reduce the costs of
fighting extortive corruption (Brunetti & Weder, 2003: 1804-1805). For extortive
corruption it applies that without corrupt behaviour of the public official, the utility for
the private individual would have been higher.
The mechanism for collusive corruption is different. Collusive corruption is the
kind of corruption from which both the public and the private party gain. For example, a
certain person does not want to take driving lessons, but when he gives the public
official a certain amount of money he will get the driving license without taking any

3
The fact that the official believe that the probability of corruption increases, Is obviously a result of the
fact that the probability will increases. However, whether an official behaves corrupt is decision based on
the official’s perception.

18
lessons or examination. Thus, for the private party there is no incentive to fight the
corruption or make it public. Therefore, it is much more difficult to fight this kind of
corruption. According to Brunetti and Weder independent journalists have “incentives
to actively investigate any wrongdoing” (2003: 1805). The possibility exists, however,
that the journalists can be bought and become corrupt themselves. Brunetti's and
Weder's argue that this only increases the incentives for other journalists to detect
these kinds of corrupt arrangements. “The more involved a corrupt arrangement the
more fame an investigative journalist can earn by uncovering it“ (Brunetti & Weder,
2003: 1805). Thus, forming an effective cartel of corrupt journalists is impossible when
entry to journalism open for everyone.
Another characteristic that makes the media an effective corruption curbing tool
is related to political accountability. Transactions with representatives of the state, the
government or other public officials, always implies some asymmetry of information
between these representatives and the citizens. Because of the existence of rent
corruption arises spontaneously in this context (Lederman et al., 2005: 3). Privatising
these state transactions and, consequently, let the market resolves this information
problem, would not be a solution to corruption, because the state intervenes precisely
in those situations in which the market fails (Banerjee, 2007). Among the possibilities to
enhance political accountability is to reduce the informational problem citizens they are
faced with in transactional situation with representatives of the state. The media can
fulfil a role in providing the information.
This mechanism is even more effective in a democracy with healthy electoral
competition. (Persson and Tabellini, 2000; Chowdhury, 2004). “The presence of press
freedom brings public corruption cases to the voters while voters in a democracy in turn
punish corrupt politicians by ousting them from public offices. Hence, elected politicians
react to the voters by reducing corruption” (Chowdhury, 2004: 93-94). That this
pathway actually works is shown by Peters and Welch (1980). They showed that the
support declined for candidates between 1968 and 1978 for the United States Congress
who were accused of corruption.

19
 And finally, the press can change people’s believe what corruption is and when it
should be considered as a problem. If the citizens see corruption as a greater problem
they are more likely to report such behaviour. Thereby, voters who consider corruption
as a problem may be more likely to elect officials who consider corruption also as a
problem, consequently throwing out the “rascals” (Gardiner, 2002: 33). “values about
corruption are likely to affect how they behave themselves—whether they will offer
bribes or will abide by the requirements of the law. For all of these reasons,
understanding public opinion about corruption will provide a basic for effective law
enforcement efforts” (Gardiner, 2002: 33). All these effects contribute to corruption
curbing abilities of the news media.

Criticisms

Although few scholars disagree that high press freedom is associated with low levels of
corruption, not everyone is convinced that this is the result of a causal relationship. As
Graber points out rightfully: “The importance of a free press has been so axiomatic that
its presumed benefits have seldom been questioned, at least not publicly” (1986: 257).
In his research he found several instances of press investigation in which it was highly
questionable whether the benefits did outweigh the costs of detecting public
wrongdoings. He therefore concludes that the effects of the free media on corruption
should not be overstated and that the effects might even increase corruption. That this
can be the case sometimes is illustrated in the work of Vaidya (2005) and Čábelková and
Hanousek (2004). Vaidya demonstrates using a rational choice model that the media
might release false campaigns and accusations against the authorities driven by sales
numbers’ and public attention’s considerations. Thereby, he argues that the media
might be corrupted themselves and might therefore decide not to publish about
corruption.
Vaidya (2005) mentions a scenario in which the news that is brought by the
media does not result in less corruption, because the information in the news is
incorrect. Čábelková and Hanousek (2004) describe a situation in which there is nothing

20
wrong with the message, but in which the effects of the message do not lead to the
desired effect. They argue that in societies in which the perception of corruption is
high—and this is fed by media reports about corruption—that the public can have the
feeling that they have to bribe too in order to get what they want. “Believing that
everybody takes bribes, officials lose the fear of being punished for receiving them. If
everyone believes everyone else is corrupt, corruption becomes a pillar of the culture in
which bureaucrats go about their business” (Čábelková and Hanousek, 2004: 383)4.
Another point for criticism is the alleged spuriousness of the relationship of the free
media and corruption. It can be that this relationship “simply picks up wealth effects
and the institutional environment more generally” (Freille et al., 2007 839). For
example, rich countries can afford a free press or education. Moreover, the detection of
corruption is likely to be related to the quality of the legal system (for example Andvig &
Moene, 1990), salaries paid to public official (Andvig & Moene, 1990: 66), electoral
competition (Persson et al., 2003) and other factors related to the level of development.
Naturally, the level of corruption in a country is affected by other factors than
just the media. However, given the evidence I acknowledge that the free media can play
an important role in diminishing corruption. But at the same time I support Vaidya
(2005) in his critique that the effects of the press should be not overstated. As is shown
in the following chapter (chapter 3) there are complexities in the relationship between
the free press and corruption. In that chapter provides for the evidence that the
marginal effects of the media on corruption differ across the levels of corruption; the
corruption curbing abilities of the media are stronger as the corruption is lower.

4
Čábelková and Hanousek (2004) have empirically shown that this argument holds in Ukraine. In this
thesis only democracies are taken into account. Ukraine is not a democracy and therefore this effect does
not necessarily have to hold for democracies too. However, the theoretically argument is elegant enough
to take it into consideration in this thesis.

21
22
CHAPTER 3: THE PUZZLE AND RESEARCH QUESTION

In the literature review I demonstrates that the freedom of the press is associated with
the levels of corruption. I also consider this association to be a causal relationship,
although I support Vaidya (2005) in his statement that the positive effects should not be
overstated. I found that the relationship between media freedom and corruption is a
complex one: the relation between the press freedom and corruption is weaker for
highly corrupt countries than for mildly corrupt countries. This chapter will provide for
the empirical evidence of this finding.
In order to do so, firstly the case selection is discussed. Thereafter the
independent variable—press freedom—and the dependent variable—corruption—are
operationalised and the reliability and validity of the different data sources are
discussed. Then the results of the analyses are presented. This analysis results in a
puzzle and a question which are discussed in the final section of this chapter.

Variables and Data

The freedom of the press is considered to be the independent variable in the analysis.
This degree of press freedom in a country is expressed in a score on the Free Press Index
provided by Reporters Without Borders. This is a France-based non-governmental
organisation which promotes press freedom and freedom of information around the
world. Every year they present a Press Freedom Index. In order to compile an index,
Reporters prepared a questionnaire that asses the degree of press freedom in a country.
These questions are about violations directly affecting journalists (such threats,
murders, imprisonments) and the news media (such as censorship). The questionnaire
also took into consideration the level of self-censorship, financial pressure,
independence of the media et cetera. The survey was held among many journalists and
journalist organisations around the world. The index therefore not only is a reflection of
23
the freedom felt by domestic journalists, but also by foreign the foreign press. The
data that is used in the analysis here is from the year 2010. For this particular year the
index runs from 0 (highest press freedom) to 105 (the score of the worst scoring country
on the list).
To test the robustness of the analysis also data from Freedom House’ Freedom
of the Press 2010 has served as representation of the independent variable. Freedom
House is a non-governmental organisation which supports democratic change and
monitors freedom. They too compile a press freedom index based on a survey
assessment about three broad categories: the legal environment, the political
environment and the economic environment. This index runs from 0 (maximum press
freedom) to 100 (minimum press freedom). The main difference between this index and
Reporters Without Border’s index is that the survey Freedom House held to compile the
index is also answered by other people than journalist of journalist organisation, such as
country experts, consultants, international travellers and human right organisations.
Both indices share several deficiencies. For example, the press freedom is
determined based on the perception of the press freedom by individuals. The amount of
precise questions limits the possibility of misconceptualisation of press freedom by the
different respondents. However, the respondents might experience the degree of press
freedom in the various aspects differently. Thereby, both indices falsely imply accuracy,
especially because score are showed in decimals. Nevertheless, there are no real
alternatives to these indices5 and despite the imperfections, the indices provide for a
failry good impression on the degree of press freedom in the different countries. For the
analysis it does not matter for example whether Finland (0 on the Reporters Without
Borders index) experience more press freedom than Austria (0.50 on the same index);
what matters is that they are both countries with a very high degree of press freedom.

5
There are many other press freedom indices, but they are both not so widely used in the literature or do
not taking into account all democracies. For example, the IREX’ Media Sustainability Index does not
include most European Union countries and other Western democracies.

24
 The dependent variable in this analysis is corruption. This is represented by the
Transparency International’s Corruption Perception Index 2010 (from now on CPI). This
index is a compound index using data from different other institutions6. The index
ranges from 0 (extremely corrupt) to 10 (highly clean) in a 100 steps. There are also
some drawbacks to the use of this index. Partially, these are similar to the imperfections
of the free press index. For example, the perceptions of corruption can be different due
to the different definitions of corruption the different data sources apply if they
measure corruption. Therefore the index might not exactly measure what Transparency
International self defines to be corruption7. For example, it is likely that the index focus
too much on business transactions, because the composite CPI does mainly drawn upon
surveys held among business leaders and country experts. It can happen that no
respondents are drawn from the country that is in question (Andersson & Heywood,
2009). Also similar to the press freedom index is that the CPI falsely implies accuracy.
This is in particular a problem because the index is presented in the form of league
table, giving the countries in addition to a CPI score also a ranking. This may have as
consequence that a small improvement in the CPI score can result in a large
improvement on the ranking. Nevertheless, whether a country scores 2.4 or 2.8 does
not make that much of a difference, both scores show that that particular country is
very corrupt.
Notwithstanding the imperfections of the CPI, given the hidden nature of
corruption it is hard to find other ways of measuring corruption other than using
perceptions. Thereby it is widely used for academic purposes and therefore critically
reviewed by many political scientists (for example Chowdhury, 2004). The index might
not be perfectly accurate, but it gives at least an indication of the degree of corruption.

6
These data sources vary from year to year. For the 2010 index the following institutions provided for the
data: Asian Development Bank; African Development Bank; Bertelsmann Foundation; World Bank;
Economist Intelligent Unit; Freedom House; Global Insight; IMD International, Switzerland, World
Competitiveness Center; Political & Economic Risk Consultancy.
7
Transparency International (2009, 14) defines corruption as: “the abuse of entrusted power for private
gain”.

25
Case Selection

As is discussed in the literature review (chapter 2), one of the mechanisms why the free
press is capable of curbing corruption is because it gives the people information, so that
they can vote wisely at the next elections. Thereby, the levels of corruption are affected
by the degree of political and electoral competition. In order to control for the fact that
press in democracies can be more influential than in another regime environment, even
if the functioning of the press is qualitative equal, and in order to control for the degree
of political and electoral competition, only liberal democracies are taken into account
for an analysis. In principle, every country that is considered to be a democracy is taken
into account. Only the availability of data provided by Reporters Without Border,
Freedom House of Transparency International further limits the case selection.
Whether a country can be considered as a democracy is a consequence of their
Freedom House 2010 scores on the political rights index and the civil liberties index. This
is in line with the analysis to democratic transitions and consolidations by Stepan and
Linz (1996: 445). They use a score of 2 or lower on the political rights index and scores of
3 or lower on the civil liberal index as threshold to consider a country to be a
democracy. In order to prevent that electoral and political competition still influence the
analysis, because of the variance in scores on the indices, I use a stricter threshold; a
score of 2 or lower for both indices results in a case selection. These countries and their
scores are displayed in Appendix 4.

Analysis

Now the independent and dependent variables are conceptualised and operationalised,
the focus can be turned to the actual analysis. The aim here is not the show whether
there is a relationship between press freedom and corruption, but to show that there
are complexities in this relation. As is discussed in the literature review (chapter 2), the
relationship and the direction of the relationship are already established by Ahrend
(2002), Brunetti & Weder (2003) and Chowdhury (2004). Therefore it can be assumed

26
here that press freedom is the independent variable and corruption the dependent. In
order to show the complexities three ordinary least square analyses are executed. The
democracies are divided in a mildly-corrupt category (50% highest CPI scores) and a
highly-corrupt category (50% lowest CPI scores), because the point that I want to
illustrate is that for the negative relationship between press freedom and corruption is
stronger as the countries are less corrupt. I do not want to specify the strength of this
relation and explanation power of the free press in values, I just want to show that there
is a difference in the relation between the countries that associated with their degree of
corruption. The median is therefore a good cut-off point, because it divides the cases
into two equal groups. This makes it possible to use simple statistical methods—OLS
regression analyses between only an independent and dependent variable—to show the
complexity. The results of the various OLS regression analyses are displayed in table 1.
For all analyses the press freedom index has been considered as the sole
predictor for the level of corruption. The rows 1 to 3 show the results using data from
Reporters Without Borders as independent variable. The first OLS regression analysis
included all democracies that made the case selection (ALL DEMOCRACIES RWB). The
other two analyses only included the 50 percent most corrupt countries (LOW CPI RWB)
or the 50 percent least corrupt countries (HIGH CPI RWB) according to CPI. In order to
check whether the results are robust, the analysis is also conducted using the press
freedom index by Freedom House (row 4 to 6). Also for this data three different OLS
analysis have been executed, including every democracy (ALL DEMOCRACIES FH), only
the corrupt countries (LOW CPI FH) or the least corrupt countries (HIGH CPI FH).
Table 1: results of various regression analyses considering press freedom to be the
predictor for corruption.
B Std. error p R2adj
1 ALL DEMOCRACIES RWB -0.196 0.024 0.000 0.538
2 LOW CPI RWB -0.070 0.019 0.001 0.328
3 HIGH CPI RWB -0.150 0.023 0.000 0.616
4 ALL DEMOCARIES FH -0.175 0.017 0.000 0.627
5 LOW CPI FH -0.058 0.016 0.001 0.293
6 HIGH CPI FH -0.137 0.028 0.000 0.439

The results for the OLS regression analyses of which the results are shown in row
27
1 to 3 are from the same indices and, consequently, the analyses were established on
the same scales (Reporters Without Borders press freedom index and CPI). This makes it
possible to compare the unstandarized regression coefficients. The regression analysis
including all democracies (row 1 ALL DEMOCRACIES RWB) is to show the basis relation
for which complexities are determined. If all democracies are taken into account than
the negative relation between press freedom and corruption is the strongest (B =- .196).
However, the predicting power of press freedom for the degree of corruption is largest
when the only the highest 50% on the CPI index (least corrupt countries) are taken into
account (R2adj = 0.616). This is higher than when all countries are taken into account
(R2adj = 0.538) or higher than if only the 50% most corrupt countries were included in the
analysis (R2adj = 0.328). Also from the results it becomes clear that the negative
relationship between press freedom and corruption is stronger in least corrupt
countries (B = -0.150) than in the highly corrupt countries (B = -0.070). The regression
analyses using the data about press freedom from Freedom House (row 4 to 6) are
showing similar patterns: in corrupt countries the negative relationship between press
freedom and corruption is stronger than in mildly-corrupt countries. Thereby in mildly-
corrupt countries press freedom is a better predictor for corruption than in highly-
corrupt countries.

Findings and Research Question

The results presented in this chapter confirm the association between press freedom
and corruption in democracies. Considering the literature (in particular Ahrend 2002;
Brunetti & Weder, 2003; and Chowdhury, 2004) I have assumed that this relation is
based on causality and that the negative relation runs from press freedom to
corruption. This assumption makes it possible to carry out OLS regression analyses. The
results show that the negative relationship between press freedom and corruption is
stronger for the 50% least corrupt democracies than for the 50% most corrupt
countries. Moreover, the free press is also a better predictor for corruption in the least
corrupt countries than for corruption in the 50% most corrupt countries.
28
 The median of the democracies ranked by their CPI score is used as cut-off point
to divide the sample of democracies into a mildly- and a highly-corrupt group. This is
rather an arbitrary cut-off point. The aim is however not to specify the strength and the
predictability of the relationship exactly, but to show that there are differences. Dividing
the sample into two groups has been proven to be helpful in showing this difference. As
a result, it has been made plausible that marginal effects of media differ across levels of
corruption in democracies; the ability of the news media to reduce corruption is greater
as the level of corruption is lower.
In this chapter the complexity in the relationship between press freedom and
corruption is demonstrated. This is already a contribution to the current knowledge
about corruption. This complexity means that simply enlarging the press freedom would
not automatically leads to a dramatic reduction of corruption. This knowledge is even
more of interest, because in especially those countries in which it is more important to
fight corruption—the countries that are face by a lot of corruption—the media does not
seem to be effective. In order to understand how these countries still can be helped by
the media, it needs to be established why the media is not as effective in curbing
corruption in corrupt countries as in less corrupt countries. To answer this question a
theory is developed and presented in the next chapter (chapter 4).

29
30
CHAPTER 4: THEORY AND HYPOTHESIS

In the previous chapter the empirical data confirmed that there is a negative
relationship between press freedom and corruption, but that the marginal effects differ
across the levels of corruption. Higher levels of corruption are associated with a weaker
and less predictable relationship between the variables. The complexity in this relation
can only be explained partially by other factors such as the functioning of the judicial
system and wealth levels. In this chapter I will argue that this complexity is a result of a
factor that is related to the press itself. I will lay out an argumentation that explains that
the press is less effective in changing peoples’ attitudes in highly corrupt countries than
in mildly-corrupt countries. Firstly, it is explained why attitude change is important in
order to curb corruption. Secondly, Zaller’s (1992) model for attitude change is
explained and it is argued how this model can be applied to the puzzle of this thesis. The
chapter is concluded with a hypothesis that results from the discussed theoretical
model.

Changing Attitudes to Fight Corruption

As is discussed in the literature review (chapter 2), corruption reduces when either the
public official decides not to be corrupt anymore, or when public officials that are
corrupt, are punished or ousted from their public office. Different pathways are
discussed that may illustrate the mechanisms behind one of these corruption curbing
factors. All these mechanisms have in common that the role of the media in the fight
against corruption, can be explained by the media’s capabilities of changing the attitude
of the people that are involved with corruption. Firstly, increasing the functioning of the
press increases the official’s perception of the expected costs of being corrupt, because
information brought by the media about corruption increases the official’s probability of
being caught and punished. As a result of the increased expected costs is that the
31
balance now leans more towards the expected costs side and consequently it can now
be that being corrupt is not a valuable strategy anymore. Secondly, information about
corruption brought by the press reduces the information a-symmetry between the
public officials and the citizens. The public can use this information to oust corrupt
public officials from office or elect those politicians that actively fight corruption among
the non-elected public officials. It is this change in voting that leads to less corruption.
The free press only is an effective instrument for the fight against corruption
when news brought by them results in a change of people’s behaviour or thinking about
corruption; or in other words, because of its capability to change people’s attitudes, the
free press is able to fight corruption. The empirical evidence that underlies the puzzle of
this thesis (chapter 3) shows that the relationship between press freedom and
corruption is weaker in corrupt countries. In the previous paragraphs of this chapter I
have reasoned that the attitude changing capabilities of the media are the reason that
the press is effective in curbing corruption. This suggests that the press is less effectively
changing people’s attitudes in corrupter countries. What, then, makes the media in
corrupt countries less effective in changing attitudes?

Attitude Change by the Media

In order to answer the question posed in the last paragraph it needs to be explained
how the process of attitude change works. This first requires explaining what an attitude
is. Attitudes are the evaluations and associated beliefs and behaviours towards some
object (McGuire et al., 1985). These attitudes are receptive for change when the context
changes or a person receives new information about the object on which it can base his
evaluations. The process of persuasion, or attitude change, by receiving information or
messages can be separated in series of discrete steps with their own mediators
(Hovland, 1963; Zaller, 1996: 21).
The first step of this chain reaction is the exposure to persuasive messages.
Second is the step of reception of the message; this is the mere fact that people are
actually “taking in” the message to which they are exposed. The final step in the
32
persuasion chain reaction is the acceptance of the message; the fact that persons
change their evaluations towards the object because of the newly received information.
Thus, if the news media in corrupt are less effective in curbing corruption than their
colleagues in mildly-corrupt countries the problem probably can be found in a factor
that is diminishing the effectiveness of one or more of the steps in the persuasion chain
reaction.
Zaller (1992; 1996) argues that knowledge about exposure to media cannot lead
to any predictions about the likeliness of attitude change by a citizen. Although it is
relatively straightforward to measure exposure—someone has been exposed or not—it
does not say anything about the actual intake of information. The mere fact that people
have a physical proximity to a persuasive message does not say anything about the
attention they pay to it. “Even when someone is watching, television viewing is often a
secondary activity-second to an astonishing variety of other activities, such as eating,
conversation, dancing, sorting wash, playing Monopoly, scolding children, and reading,
to mention only some of the mentionable ones” (O’Sears & Kosterman, 1994: 258)8. This
argument also holds for listening to the radio. Reading the newspaper and being on the
internet somewhat escapes this argument. However, even for these forms of media the
exposure cannot predict the probability of attitude change.
Although a message cannot lead to attitude change without the individual being
exposed to it, it is the actual intake of a message that is a better predictor for the
probability of attitude change. Whether reception of the message has taken place is,
however, much more difficult to ascertain. Nevertheless, because it is one of the steps
in the chain of persuasion, it is important to understand reception before attitude
change can be explained. To use an analogy, if a researcher wants to know whether a

8
Although this quote might be a bit jolly and I do not understand why he mentions Monopoly (the most
boring game there exists, I just do not believe anyone ever plays it), the point should be well received. A
lot of the time media use is a secondary activity. But even when this “secondary” exposure is not
considered as exposure, it is likely that people do receive some of the information brought be the media.
Therefore using exposure would lead to serious measurement problems.

33
certain drug has an effect on a particular disease it is not enough to know whether the
patient took the drug from the drug store, but the researcher has to know whether the
patient actually took the drug (Zaller, 1996: 22). If the researcher finds that the patient
took the drug, he can also safely assume that drug was taken from the drugstore. This
also holds for the persuasive message; it is not said that when a person is exposed to
persuasive information he actually takes in the information, but if there has been
reception the researcher can assume that the person has been exposed to the message.
Both the exposure to and the reception of political information is related to the
attention people pay to those kinds of messages. People who pay most attention to
these messages are those people that know a lot about public affairs and are, therefore,
able to recognise political information. At the other end of the attention spectrum are
those who do not know a lot about public affairs. (Delli Carpini & Keeter, 1996: 219) It is
a combination of disinterest, but also no ability to understand political messages, that
leads to low intake of political information. Most citizens fall in between both extremes.
It is this combination of attention to news about public affairs and the ability to retain a
message about public affairs that determines the, what Zaller (1992) calls, “political
awareness”. “Political awareness denotes intellectual or cognitive engagement with
public affairs as against emotional or affective engagement or no engagement at all”
(Zaller, 1992: 21). Zaller (1992; 1996) shows that political awareness is the key factor in
whether press messages are able to effectively change an individual’s attitude or not.
The probability of a long-term attitude change is the result of the combination of
probability of reception of the message after exposure and the probability of
acceptance upon reception. For both the probability of acceptance and the probability
of acceptance upon reception the individual’s political awareness is the most important
determinant. If an individual’s political awareness is low, the probability of reception of
the persuasive information is also low. Indeed, if the individual does not have the
“language” or cognitive ability to understand the information, it is most likely that this
person does not take in the information, although he is exposed to it. However, if such
an individual does take in the message, it is likely to have an effect on the attitude of

34
that person. The person does not have much knowledge yet about that particular topic
covered by the message. Thus every bit of persuasive information can be that piece of
information that tips the balance of knowledge towards the different attitude.
At the other extreme—when the individual’s political awareness is high—the
logic works in the opposite direction. This individual does have the “language” and
attention for this new information and will likely take in the information. Thus, if a
citizen is politically aware then the probability of reception of the message is high.
However, because that person already has a lot of previous knowledge about that
particular topic, new information is not very likely going to tip the balance toward a
different attitude. Indeed, there is a strong correlation between political knowledge and
stability of attitudes (Delli Carpini, 1996: 232-234; Clawson & Oxley, 2008) Therefore it is
not probable low that a message that attacks a person’s attitude will lead to attitude
change. The probability that an individual experiences attitude change is therefore not
monotonic, but highest with the middle levels of political awareness. This is because the
probability of change is the result of a multiplication of the reception rate and the
acceptance rate. This is supported by a model created by Zaller (1992: 123). In table 1
this is shown by showing hypothetical data. The model Zaller proposes is for attitude
change after a message that is intended to persuade the receiver.

Table 2: Attitude change in response to a hypothetical message (from Zaller, 1992: 123).
Level of awareness
Low Middle High
Prob(Reception) 0.10 0.50 0.90
Prob(Accept│Reception) 0.90 0.50 0.10
Change (Reception X Acceptance) 0.09 0.25 0.09

In order to explain the different effects the press have on corruption between
the countries, it is necessary to translate this individual model to the macro-level. It
would, however, be incorrect to simply take the average awareness of all people in one
country and then argue that media in countries with an average of medium awareness
35
are the most effective in fighting corruption. For this reason that the relation between
the political awareness and the probability of attitude change is nonmonotonic and non-
linear: it are the people that are in the middle level group of political awareness that
have the highest probability to change their attitudes after receiving messages, whereas
the political unaware and the political highly aware are the people that are least likely to
change their attitudes after receiving new information.
Thus, this means that the country average cannot be taken as an independent
variable. Consider the following example: assume a situation in which the country’s
political awareness average is 0.5 on a scale from 0 to 1. The boundary situations from
which this average can be obtained are that either half of the population is maximum
political aware (1 on a 0 to 1 scale) and the other half minimum political aware (0 on the
0 to 1 scale), or that the entire population is exactly middle political aware (0.5 on the 0
to 1 scale). How the people are distributed among this scale has great implications for
the size of the group that experiences attitude change after receiving persuasive
information. When everyone is on the edges of the political awareness spectrum—
either being a 0 or a 1 on the 0 to 1 scale—the total amount of people that will change
their attitudes after receiving new information will be lower than when everyone is
exactly at the middle of this spectrum. Thus, in a situation in which the middle political
awareness is the most effective political awareness for attitude change, then it is the
proportion of the group of the medium awareness that will be able to predict the
amount of people that will change their attitudes in a country.

Hypothesis

Concluding, the two main arguments of this theory chapter are: firstly, the press needs
to change the attitude of people toward corruption in order to be an effective tool in
the fight against corruption; and secondly, it has been argued, using an extension on
Zaller’s model for attitude change, that in societies with larger proportions of middle
political awareness people the media is more effective in changing the attitudes of the
people than in countries with smaller proportions of medium people. This thesis is build
36
around the observation that in corrupt countries the press is less effective in curbing
corruption (see chapter 3). This observation combined with these two arguments lead
to the following hypothesis:
Hypothesis: The proportion of people that are middle political aware negatively
correlates with the level of corruption.

37
38
CHAPTER 5: POLITICAL AWARENESS AND CORRUPTION

In the previous chapter I have hypothesised that there is a positive relationship between
the proportion of the people that have middle political awareness in a country and
corruption. This chapter is devoted to the give to empirical evidence to confirm this
expectation. In order to do so, firstly different measure for political awareness will be
discussed. At the end of that section the choice for the measure for political awareness
is explained. Thereafter the variables are presented and the data sources from which
the data is extracted. In the analysis section the actual analysis is presented as well as
the results. The chapter is concluded with discussion on the findings.

Measuring Political Awareness

In the quest for a measure for political awareness one needs to be aware of the
multidimensionality it encompasses. As is discussed in the theory chapter (chapter 4),
political awareness is the combination of the attention people pay the certain news
messages and the ability to understand these messages. Without attention and ability to
understand messages brought by the press it is impossible that this information can
contribute to the receiver’s knowledge and, consequently, this information cannot lead
to attitude change. In the literature several suggestions are found as measures for
political awareness. A justified measure for political awareness for this thesis should
both be precise and it should hold in the comparative framework. Therefore a
discussion about the different alleged measures is included in this work.
Traditionally, education levels in combination with self-reported media were
used as main indicators for political awareness (Price & Zaller, 1993: 138). The argument
for using education as indicator is that better educated people tend to learn faster from
new information than lower educated people (Tichenor et al., 1970). Thereby, higher
educated people are believed to be more socialized to pay attention to political affairs
39
(Price & Zaller, 1993: 138). Indeed, correlations have been found between education
and learning from news (Robinson & Levi, 1986). However, it has not been said that fast-
learning from the media necessarily means that high-educated people pay attention to
political affairs. Only after the intake of the information they are better trained to learn
from it, but there is a great variance between citizens in the amount of attention they
pay to public affairs, also under the high-educated people (Luskin, 1990). And, although
higher educated people have a better ability to understand politics because they started
with more knowledge about politics, it is the attention to news that keeps them up-to-
date. The political context is changing over time, thus in order to hold the ability to
understand the political news, the attention dimension of political awareness have to be
fulfilled to. That therefore education levels cannot be the sole measure for political
awareness is supported by empirical findings that political knowledge have not been
increased over the years despite an increase in education levels (Elkin & Soltan, 1999).
This attention level might be measured by the (self-reported) media use. This is,
however, also not a useful measure in the context of the research in this thesis. Firstly,
because different kind of media influence individual’s political awareness differently9.
This problem might be overcome to distinguish between exposure to “high-brow” and
“low-brow” media10, as Price and Zaller (1990) argue, but even within these different
categories of media the influence of particular instances of the media is different. For
example, what is more contributing to political awareness, the local news or the
national radio? Making distinctions between different kinds of media would result in
serious measurement issues. The second reason, one that may be even more important,
why the use of exposure measures is not justified is that they do not take into account

9
For example, Aarts and Semetko (2003) found that in the Dutch context that regularly watching the
public service channels on television watching had positive effects on the political knowledge, whereas
watching the commercial channels has a negative effect.
10
Price and Zaller (1990b: 4) consider news to be low-brow if the medium from which it is brought is
“locally oriented” and “although, they carry national and international news, [they] carry it in a very
abbreviated form, and [they] tend to focus on crime, human interest, sports and celebrity events”. High-
brow media, on the other hand, are “nationally oriented” and “although, they carry crime and human
interest stories, [they] focus more on national politics and international affairs”.

40
the actual intake of the information. As is been discussed in the chapter 4, measuring
the actual intake of information is important in order to be able to make statements
about attitude change. Despite the fact that exposure to information is necessary for
the actual intake, and therefore is probably correlating, exposure measures do not catch
the actual intake.
The best measure is to test the political knowledge (Zaller, 1992: 335; Delli
Carpini & Keeter, 1993: 1180). In a latter work Zaller (1996) even treats the terms
“political awareness” and “political knowledge” as being the same. The advantage of
this measure is that it also captures political learning that already has occurred over
time, or, in the words of Zaller (1992: 335): “political ideas that the individual has
encountered, understood, and stored in his head”. Therefore, testing political
knowledge automatically accounts for the persons ability to learn about political affairs
from the mass media and the use of the media to gain the knowledge. Price and Zaller’s
(1993) analysis confirms that the likelihood to learn from political knowledge can be
best predicted by the person’s pre-existing knowledge. Therefore, political knowledge
will serve as the representation of political awareness in this thesis.

Variables and Data

In the previous section is argued that political knowledge is a precise measure for
political awareness, but the second demand for a justified measure is that it also holds
in a cross-national comparative setting. This is easier said than done, because of the
difficulties that arise when people from different countries are asked the same question.
For example, a Dutchman is expected to be able to name his own prime-minister easier
than to name the Spanish prime-minister. For Spaniards the opposite argumentation
holds. Difficulties might arise however, when similar institutions have a different role in
society. For instance, the president of France is much more involved in daily politics than
the German president. Therefore it can be expected, considering the average political
knowledge is equal in both countries, that more French are able to name their president
correctly. This makes asking country specific questions but about similar institutions not
41
automatically useful in a comparative setting. Therefore, measuring political knowledge
has to be done with great prudence if used in a comparative setting.
What are the options then? An opportunity to obtain comparable data about
political knowledge are surveys around European Elections. The EES 2009—a survey
held among voters of the, at that time, 27 European Union countries—includes 4
questions about the institutional functioning of European politics and 3 about national
political actors with a different degree of difficulty. According to Fraile (2010: 9-10), this
survey provides a unique opportunity to use these results in a comparative research
framework, because 4 of the 7 questions asked in this survey are equal in all countries. A
drawback of using this method is that the questions are mainly about politics at a
supranational level. The question is whether this construct can actually be used as
representation for the political awareness to their domestic public affaires. The measure
would be mirroring this when the saliency of European politics is equal among the
different European countries. However, this is not in reality (see for instance Schuck et
al, 2011; and De Vreese & Boomgaarden, 2006). This method more or less measures the
political awareness for European matters, whereas I am looking for a way to measure
political awareness for domestic affaires, because corruption is mainly a domestic issue.
Most likely European political awareness and domestic awareness strongly correlates.
Intuitively it makes sense to believe that there would not be many people that know
everything about European politics and nothing about their domestic politics11. Still,
using only this measure to draw strong conclusions would be incorrect.
An alternative might be found in the data from the CSES modules. This database
includes results of surveys that are held around national elections in many countries
around the world. These surveys also include political knowledge questions. The

11
Delli Carpini and Keeter (1996: 157) found that people who know a lot about a certain aspect of
domestic politics are also likely to know a lot about other aspects. Although this still not empirically prove
that people who do know a lot about European politics do know a lot about domestic politics too or the
other way around. Nevertheless, it strengthen the idea to political knowledge reflects a comprehensive
set of knowledge and not just the aggregate of the political knowledge of different topics.

42
questions that are asked are country specific and are therefore not adequate to show
exact difference of average political knowledge between countries. Despite their
argument against the use to show differences in political awareness between countries,
Grönlund and Milner (2006) show that the CSES modules can be used in comparative
research. In there study they look at which factors contribute to individual political
knowledge. This is a different use than the cross-national approach that I use. However,
I am not looking for a measure to compare the exact average political awareness of a
country, but I am looking for a way to test whether to size of the group of middle
political aware individuals is related to corruption levels. Comparing countries is just
done as means to an end. Therefore, as long as the set of questions in every country has
somewhat similar difficulty, it should be able to make a comparison between the levels
of corruption in a country and the size of the medium group.
Thus, for the analysis between political awareness and corruption data from the
EES 2009 and CSES Module 3 serve as independent variable. The dependent variable—
corruption—is drawn from Transparency International’s Corruption Perception Index
2010 (CPI). A extensive discussion on the validity and reliability of this index can be
found in chapter 3.

Analysis

In order to analyse the relationship between the proportion of the people that is middle
political aware and corruption, the political knowledge questions in the EES 2009 and
the CSES Module 3 serve as representation of the political awareness. In the EES 2009
four question included in the EES 2009 were about the European Union institutions.
Another three country specific questions were included too, but—although the
questions were claimed to be standardized—in order to avoid problems that are
involved with comparing the results between countries, only the questions about the EU
are used in the analysis. The CSES Module 3 includes the results for 3 country specific
questions. The cut-off points for the low, middle and high political aware scoring groups
are determined by the quartiles of the scale of how many correct answers the
43
respondents were able to give. These cut-offs are constructed in such a way that if
political awareness would be distributed equally among the society that half of the
population would fall in between the two cut-off points.
For the CSES Module 3 these cut-off points do not cause any troubles, because
this data set included only 3 questions. People were able to answer 0, 1, 2 or 3
questions correctly (results available in appendix 6), therefore, people who were able to
answer no question correctly were placed in the low political aware group; answering 1
or 2 questions correctly places that person in the middle political aware category; and
answering all questions correctly leads to a high political aware label. The same quartiles
are used to determine the cut-off points for the data of the EES 2009. However, because
the people had to respond to 4 questions, the scale of correct answers counts 5
possibilities. In order to prevent a skewed scale, I decide that answering zero questions
correctly leads to a low political aware label. At the other end of the spectrum are the
people who answered every question correctly; they are considered to be high political
aware. The others, those who answered 1 to 3 questions correctly are considered to be
middle political aware.
Simple bivariate correlation techniques are used to establish the relationship
between the proportion of people that are middle political aware and CPI. Using the EES
2009 dataset the empirical result show a positive correlation (rs[27] = .534, p = .004).
Since the CPI is coded from 0 (maximum corruption) to 10 (no corruption), this result
means that large proportions of people that are middle political aware are associated
with low levels of corruption. To test these results for robustness, the same analysis
have been executed, but then using the data from CSES Module 3 as representation for
the political knowledge. Simple bivariate correlation techniques are used to show a
relation as well, although less significant (rs[19] = .432, p = .065)12.

12
Simple bivariate correlation techniques analyzing all countries that are in the dataset did not show a
significant relationship (rs = .179, p = .438). Especially Sweden and Norway did not seem to fit the
expectations. However, a look at greater detail at the dataset and the results of the survey led me to
believe that the data for these countries are not reliable as a consequence of too difficult political
…
44
Findings

The results in this chapter confirm the hypothesis that there is a correlation between
the proportion of people in a country that is middle political aware and corruption. In
order to show this correlation two datasets have provided for the data to calculate for
the proportion of people that are middle political aware. The EES 2009 only includes
questions about the European Union. The advantage of asking the same questions in
every country is that the results can be easily compared. On the contrary, the
disadvantage is that the difference between countries that are revealed are not
necessarily a result of the difference in political awareness. It might be that the
measured differences are a result of a difference in saliency of the European Union in.
This might bias the results. De Vreese and Boomgaarden (2006) do show that there are
differences in the level of saliency in Europe. However, they showed that these
differences could be explained by the attention for a major European event was
concerning some of the countries at that particular. The EES 2009, however, includes
results from a survey that is held around the European election, a major European event
concerning all the countries. Because this event concerns every country to a particular
extent, I believe that the saliency is somewhat equalized. Nevertheless, the difference in
saliency might lead to mistakes when the results are interpreted.
Therefore, also another source is consulted in order to determine the political
awareness. The advantage of the CSES Module 3 is that this survey includes country
specific questions that are believed to have equal difficulty level. Because it is assumed
that the degree of difficulty is the same, the results are comparable. It remains,
however, unclear how it can be tested whether the questions really have the same
difficulty across the countries. In appendix 7 I have argued that there are, indeed,
differences in the degree of difficulty of the questions. Although the difficulty of the

…
knowledge questions. A complete argumentation for this is included in appendix 7: Imperfections CSES
Module 3.

45
questions are important for the classifying people to their levels of political awareness, I
am mainly interested in the proportion of group of the people that are middle political
aware. This means that as long as the degree of difficulty of questions are at least
somewhat similar it is good enough to extract the proportion of this middle group.
Despite the difficulties there are in measuring political awareness in a
comparative setting, I consider the findings in this chapter as valid because of the
reasons I have mentioned above. Even more so, because the results of both analyses,
using different sources with their own advantages and disadvantages, point out in the
same direction.

46
CONCLUSION AND DISCUSSION

In this thesis is confirmed that there is a significant relation between press freedom and
corruption in democracies. However, I gave the evidence that this relation is complex. In
order to demonstrate these complexities, I have classified the democracies into a highly
corrupt category—the lowest 50% scoring democracies on Transparency International’s
Corruption Perception Index 2010—and into a mildly-corrupt category—the highest 50%
on that index. In chapter 3 is shown that the relationship between the press freedom
and corruption for the highly corrupt countries is weaker than for the mildly corrupt
countries. Thereby, for the highly corrupt countries the press freedom is a worse
predictor for corruption than for the mildly-corrupt countries. These results imply that
the marginal effects of the press vary across the levels of corruption in democracy; the
press has a stronger corruption curbing effect as the level of corruption is lower. These
new findings on the relationship between press freedom and the levels of corruption
have provided for the puzzle and research question of this thesis. The question that was
aimed to answer is what causes these patterns. Why is the effectiveness of free press to
curb corruption stronger as the corruption in a country is lower?
In order to answer this question I have dived into the literature. The mechanisms
that are attributed to news media to curb corruption (described in chapter 2) all have in
common that diminish corruption after information brought by the press has resulted in
a change of attitude by the receiver of this information. Zaller’s (1992) model for
attitude change is used to understand what kind of people are most influential by
messages by the press. As is demonstrated in chapter 4, the people who are middle
political aware are most susceptible for information by the media. For an entire societiy
this would mean that the susceptibility is higher as the proportion of people that are
middle political aware is larger. This argument in combination with the understanding
that the media are effective in curbing corruption because of there capability to change

47
attitudes and the finding that the marginal effects of the press on corruption levels
differ across the levels of corruption, led to the following hypothesis: the proportion of
people that are middle political aware negatively correlates with the level of corruption.
In chapter 5 it is actually empirically demonstrated this hypothesis holds: the proportion
of people that are middle political aware correlates with the level of corruption. This
means that the effectiveness of the free press to curb corruption is stronger as the
corruption is lower, because the people who are most susceptible for media influence—
those that are middle political aware—negatively correlates with the level of corruption.
The findings in this thesis have implications for both academic as well as
real world reasons. Both implications go hand in hand, because any knowledge about
how the mechanisms on corruption work, are useful to understand how to fight
corruption. This thesis provides for two new findings. Firstly, the marginal effects of the
press on corruption differ across the levels of corruption; and secondly, these different
patterns can be explained by the fact that there is a correlation between the proportion
of the people that are middle political aware—the group that is most likely to
experiences attitude change by media messages—and corruption. The first finding
implies that the two opposing bodies of the literature can be brought together. Yes,
enhancing the press freedom in any country will result in lower corruption, but in
corrupt societies this effect in weaker.
For the fight against corruption this means that in the countries in which fighting
corruption is most necessary—the corrupt countries—the media is less effective. I am
not saying that freedom of media does not have any effect, but much more effort is
necessary. It is simply not enough to advocate transparency and media freedom in
corrupt countries. A broader, more comprehensive package of measures have to be
brought at the same time to curb corruption. The problem however is that it are mainly
the countries that are not facing a lot of corruption that are advocating good
governance around the world. The non-corrupt countries have to understand that what
works in their country is not necessarily effective in other countries.

48
 I have demonstrated that there is a correlation between the proportion of the
people that have middle political awareness and corruption and I have argued why this
correlation would exist. However, it has just been the first attempt to understand the
nature and causes of the complexities. Although, I am confident that my explanations
and theoretical framework are valuable contributions to the current literature, I suggest
for more research. Especially, because the political awareness is determined by general
political knowledge. Although it is likely that political knowledge about different topics
strongly correlate, I keep the possibility open that knowledge about corruption fall
outside these correlations for that reason that the nature of corruption is perceived
differently in every country. This would make the results less valid.
I also suggest to do more research to the comparability of political awareness
and political knowledge. I understand that many scholars have been working on this
topic for many years, but improvements have to be made. I can imagine that it is helpful
to include more political knowledge questions in surveys. This would have been
extremely helpful for this thesis, because for this thesis the distributions of political
knowledge were required. Because of the low amount of political knowledge questions
in each dataset, coincidence could play a too important role. It would also be helpful to
establish correlations between domestic political knowledge questions and
supranational political questions. If it is known how the knowledge about these different
levels correlate within countries, these correlations can be used to compare the
distributions between countries. Although differences between countries will come to
light, standardisation becomes possible. Obviously, I do these suggestions without
having paid a lot of effort trying this. I believe, however, that more on this should be
possible. If one will succeed to construct these correlations, then these correlations will
prove to be very fruitful for comparable research in various ways; this is not least in the
interest of topic studied in this thesis.

49
50
BIBLIOGRAPHY

Aarts, K. & Semetko, H.A. (2003) “The Divided Electorate: Media Use and Political
Involvement”, Journal of Politics 65(3): 759-784.
Ahrend, R. (2002) “Press Freedom, Human Capital and Corruption”, Delta Working Paper
No. 2002-11, available [Link]
Andersson, S. & Heywood, P.M. (2009) “The Politics of Perception: Use and Abuse of
Transparency International’s Approach to Measuring Corruption”, Political Studies
57(4): 746-767.
Andvig, J.C. & Moene, K.O. (1990) “How corruption may corrupt”, Journal of Economic
Behavior and Organization 13(1): 63-76.
Banerjee, A. (1997) “A theory of misgovernance”, Quarterly Journal of Economics,
112(4): 1289-1332.
Bjørnskov, C. (2004) Social Capital, Political Competition, and Corruption, Aarhus: Aarhus
School of Business, Aarhus University.
Brunetti, A. & Weder, B (2003) “A free press is bad news for corruption”, Journal of
Public Economics 83(7-8): 1801-1824.
Čábelková, I. & Hanousek, J. (2004) “The power of negative thinking: corruption,
perception and willingness to bribe in Ukraine”, Applied Economics 36(4): 383-397.
Chowdhury, S.K. (2004) “The effect of democracy and press freedom on corruption: an
empirical test”, Economic Letters 85(1): 93-101.
Converse, P. (1964) “The nature of belief systems in mass publics” in Apter, D. (ed.),
Ideology and Discontent (pp. 206-261). New York: Free Press.
De Vreese, C.H. & Boomgaarden, H. (2006) “News, Political Knowledge and
Participation: The Differential Effect of News Media Exposure on Political Knowledge
and Participation”, Acta Politica 41(4): 317-341.

51
Delli Carpini, M.X. & Keeter, S. (1993) “Measuring Political Knowledge: Putting First
Things First”, American Journal of Political Science, 37(4):1179-1206.
Delli Carpini, M.X., & Keeter, S. (1996) What Americans Know About Politics and Why It
Matters, New Haven: Yale University Press.
Elkin, S.K. & Soltan K.E. (1999) Citizen Competence and Democratic Institutions,
University Park: Penn. State Univ. Press.
Egmond, M.H. van, Sapir, E.V., Brug W. van der, Hobolt, S.B. & Franklin, M.N. (2010) EES
2009 Voter Study Advance Release Notes. Amsterdam: University of Amsterdam.
Fraile, M. (2010) A comparative study of political knowledge in the 2009 European
Election. Unpublished paper presented at the PIREDEU Final Conference, 18-20
November 2010, Brussels.
Freille, S., Haque, M.E. & Kneller, R. (2007) “A contribution to the empirics of press
freedom and corruption”, European Journal of Political Economy 23(4): 838-862.
Gardiner, J.A. (2002) “Defining Corruption”, in Heidenheimer, A.J. & Johnston, M. (eds.)
Political Corruption: Concepts and Contexts, Piscataway, NJ: Transaction Publishers.
Graber, D. (1986) “Press freedom and the general welfare”, Political Science Quarterly
101(2): 257-275.
Grönlund, K. & Milner, H. (2006) “The Determinants of Political Knowledge in
Comparative Perspective”, Scandinavian Political Studies 29(4): 386-406.
Johnston, Michael. (2000a) "The New Corruption Rankings: Implications for Analysis and
Reform", Paper presented at the International Political Science Association, World
Congress, Quebec City, Canada.
Kurer, O. (2005) “Corruption: An alternative Approach to Its Definition and
Measurement”, Political Studies 53(1): 222-239.
Lederman, D., Loayza, N.V. & Soares, R.R. (2005) “Accountability and Corruption:
Political Institutions Matter”, Economics & Politics 17(1): 1-35.
Leys, C. (2002) “What is the Problem about Corruption?”, in Heidenheimer, A.J. &
Johnston, M. (eds.) Political Corruption: Concepts and Contexts, Piscataway, NJ:
Transaction Publishers.

52
Luskin, R. (1990) “Explaining political sophistication”, Political Behavior 12(4): 331-361.
Mauro, P. (1995) “Corruption and Growth”, The Quarterly Journal of Economics (1995)
110 (3): 681-712.
Mauro, P. (2002) “The Effects of Corruption on Growth and Public Expenditure”, in
Heidenheimer, A.J. & Johnston, M. (eds.) Political Corruption: Concepts and Contexts,
Piscataway, NJ: Transaction Publishers.
McGuire, W., Lindzey, G., & Aronson, E. (1985) “Attitudes and attitude change”,
Handbook of social psychology: Special fields and applications, 2, 233–346.
Nye, J.S. (1967) "Corruption and Political Development: A Cost-Benefit Analysis", The
American Political Science Review, 61(2):417-427.
O’Sears, D. & Kosterman, R. (1994) “Mass Media and Political Persuasion”, in Shavitt, S.
& Brock, T.C. (eds.) Persuasion: Psychological insights and perspectives, Needham
Heights, MA: Allyn & Bacon, pp. 251-278.
Persson, T. & Tabellini (2002) Political Economics: Explaining Economical Policy,
Cambridge: MIT Press.
Persson, T., Tabbellinni, G. & Trebbi, F (2003) “Electoral Rules and Corruption”, Journal
of the European Economic Association 1(4): 958-989.
Peters, J.G. & Welch, S. (1980) “The Effects of Charges of Corruption on Voting Behavior
in Congressional Elections”, The American Political Science Review 74(3): 697-708.
Philp, M. (2001) “Access, accountability and authority: Corruption and the democratic
process”, Crime, Law and Social Change 36(4): 357-377.
Price, V., & Zaller, J. (1990) “Evaluation of media exposure items in the 1989 NES pilot
survey (Tech. Rep. to the National Elections Studies Board of Overseers). Ann Arbor:
Institute for Social Research.
Price, V. & Zaller, J. (1993) "Who gets the News? Alternative Measures of News
Reception and Their Implications for Research", Public Opinion Quarterly 57(2): 133-
164.
Robinson, J.P. & Levy, M.R. (1986) "Interpersonal Communication and News
Comprehension", Public Opinion Quarterly 50(2): 160-175.

53
Rose-Ackerman, S. (2002) “When is Corruption Harmful?”, in Heidenheimer, A.J. &
Johnston, M. (eds.) Political Corruption: Concepts and Contexts, Piscataway, NJ:
Transaction Publishers.
Rose-Ackerman, S. (2008) “Corruption and Government”, International Peacekeeping
15(3): 328-343.
Rothstein, B. (2000) “Trust, Social Dilemmas and Collective Memories”, Journal of
Theoretical Politics 12(4): 477–503.
Rothstein, B. & Eek, D. (2009) “Political Corruption and Social Trust: An Experimental
Approach”, Rationality and Society 21(1): 81-112.
Rothstein, B. & Uslaner, E.M. (2005) “All for All. Equality, Corruption and Social Trust”,
World Politics 58(3): 41–73
Schuck, A.R.T., Xezonakis, G., Elenbaas, M., Banducci, S.A. & Vreese, C.H. de (2011)
“Party contestation and Europe on the news agenda: The 2009 European
Parliamentary Elections”, Electoral Studies 30(1): 41-52.
Scott, J.C. (1972) Comparative Political Corruption, Englewoord Cliffs: Prentice-Hall.
Senior, I. (2006) Corruption – TheWorld’s Big C, London: Institute of Economic Affairs.
Stappenhurst, R. (2000) The Media’s Role in Curbing Corruption. Washington DC: World
Bank Institute.
Stepan, A. & Linz, J.J (1996) Problems of Democratic Transition and Consolidation:
Southern Europe, South America, and Post-Communist Europe, Baltimore: The John
Hopkins University Press.
Tichenor, P.J., Donohue, G.A. & Olien, C.N. (1972) "Mass Media Flow and Differential
Growth in Knowledge", Public Opinion Quarterly 34(2): 159-170.
Transparency International (2003) Global Corruption Report 2003, Berlin: Transparency
International.
Transparency International (2009) The Anti-Corruption Plain Language Guide, Berlin:
Transparency International.
Transparency International (2012) Money, Politics, Power: Corruption Risks in Europe,
Berlin: Transparency International.

54
Treisman, D. (2000) “The causes of corruption: a cross-national study”, Journal of Public
Economics 76(3): 399-457.
Uslaner, E.M. (2002) The Moral Foundations of Trust, Cambridge: Cambridge University
Press.
Vaidya, S. (2005) “Corruption in the media’s gaze”, European Journal of Political
Economy 21(3): 667-687.
Warren, M.E. (2004) “What does Corruption Mean in a Democracy?”, American Journal
of Political Science, 48(2): 328-343.
Zaller, J.R. (1992) The Nature and Origins of Mass Opinion, Cambridge: Cambridge
University Press.
Zaller, J.R. (1996) “The Myth of Massive Media Impact Revived: New Support for a
Discredited Idea” in Mutz, D.C., Sniderman, P.M. & Brody, R.A. (eds.) Political
Persuasion and Attitude Change (pp 17-29), Ann Arbor: University of Michigan Press.

55
56
APPENDIX 1: REPORTERS WITHOUT BORDERS 2010

Press freedom Press freedom

country score country score
Australia 5.38 Lithuania 2.50
Austria 0.50 Luxembourg 4.00
Belgium 4.00 Malta 4.00
Benin 19.00 Mauritius 18.00
Brazil 16.60 Mongolia 19.42
Bulgaria 19.00 Namibia 7.00
Canada 7.00 Netherlands 0.00
Cape Verde 8.00 New Zealand 1.50
Chile 10.50 Norway 0.00
Costa Rica 8.08 Panama 21.83
Croatia 17.50 Poland 8.88
Cyprus 13.40 Portugal 12.36
Czech Republic 7.50 Romania 16.00
Denmark 2.50 Samoa 33.00
Dominican Republic 26.13 Serbia 23.00
Estonia 2.00 Slovakia 11.50
Finland 0.00 Slovenia 13.44
France 13.38 South Africa 12.00
Germany 4.25 South Korea 13.33
Ghana 8.00 Spain 12.25
Greece 19.00 Sweden 0.00
Hungary 7.50 Switzerland 0.00
Iceland 0.00 Taiwan 14.50
Ireland 2.00 Trinidad and
Israel 23.25 Tobago 8.50
Italy 15.00 United Kingdom 6.00
Japan 2.50 United States 6.75
Latvia 8.50 Uruguay 11.75

57
APPENDIX 2: FREEDOM HOUSE PRESS FREEDOM

Press freedom Press freedom

country score country score
Argentina 49 Lithuania 21
Australia 22 Luxembourg 12
Austria 21 Malta 22
Barbados 19 Mauritius 27
Belgium 12 Mongolia 39
Benin 33 Namibia 34
Brazil 43 Netherlands 14
Bulgaria 34 New Zealand 14
Canada 19 Norway 10
Cape Verde 28 Panama 44
Chile 30 Poland 24
Costa Rica 19 Portugal 16
Croatia 40 Romania 43
Cyprus 22 Samoa 29
Czech Republic 18 Sao Tome and
Denmark 11 Principe 28
Dominica 23 Serbia 35
Dominican Republic 39 Slovakia 23
Estonia 17 Slovenia 25
Finland 10 South Africa 32
France 23 South Korea 30
Germany 17 Spain 24
Ghana 26 Sweden 10
Greece 29 Switzerland 13
Hungary 23 Taiwan 24
Iceland 10 Trinidad and
Ireland 15 Tobago 23
Israel 29 United Kingdom 19
Italy 33 United States 18
Japan 21 Uruguay 25
Kiribati 27 Vanuatu 23
Latvia 26

58
APPENDIX 3: TRANSPARENCY INTERNATIONAL’S CPI 2010

CPI CPI
country score country score
Argentina 2.90 Hungary 4.70
Australia 8.70 Iceland 8.50
Austria 7.90 Ireland 8.00
Belgium 7.10 Israel 6.10
Benin 2.80 Italy 3.90
Brazil 3.70 Japan 7.80
Bulgaria 3.60 Latvia 4.30
Canada 8.90 Lithuania 5.00
Cape Verde 5.10 Luxembourg 8.50
Chile 7.20 Malta 5.60
Costa Rica 5.30 Mauritius 5.40
Croatia 4.10 Mongolia 2.70
Cyprus 6.30 Namibia 4.40
Czech Republic 4.60 Netherlands 8.80
Denmark 9.30 New Zealand 9.30
Dominican Norway 8.60
Republic 3.00 Panama 3.60
Estonia 6.50 Poland 5.30
Finland 9.20 Portugal 6.00
France 6.80 Romania 3.70
Germany 7.90 Samoa 4.10
Ghana 4.10 Serbia 3.50
Greece 3.50 Slovakia 4.30

59
 CPI CPI
country score country score
Slovenia 6.40 Taiwan 5.80
South Africa 4.50 Trinidad and
South Korea 5.40 Tobago 3.60
Spain 6.10 United Kingdom 7.60
Sweden 9.20 United States 7.10
Switzerland 8.70 Uruguay 6.90

60
APPENDIX 4: FREEDOM HOUSE DEMOCRACY

Country PR CL Country PR CL
Andorra 1 1 Luxembourg* 1 1
Australia 1 1 Malta* 1 1
Austria 1 1 Marshall Islands* 1 1
Bahamas 1 1 Micronesia* 1 1
Barbados 1 1 Nauru* 1 1
Belgium 1 1 Netherlands* 1 1
Canada 1 1 New Zealand* 1 1
Cape Verde 1 1 Norway* 1 1
Chile 1 1 Palau* 1 1
Costa Rica 1 1 Poland* 1 1
Cyprus 1 1 Portugal* 1 1
Czech Republic 1 1 Saint Kitts and Nevis* 1 1
Denmark* 1 1 Saint Lucia* 1 1
Dominica* 1 1 San Marino* 1 1
Estonia* 1 1 Slovakia* 1 1
Finland* 1 1 Slovenia* 1 1
France* 1 1 Spain* 1 1
Germany* 1 1 Sweden* 1 1
Hungary* 1 1 Switzerland* 1 1
Iceland* 1 1 Tuvalu* 1 1
Ireland* 1 1 United Kingdom* 1 1
Kiribati* 1 1 United States* 1 1
Liechtenstein* 1 1 Uruguay* 1 1
Lithuania* 1 1 Belize* 1 2

61
Country PR CL Country PR CL
Croatia* 1 2 Trinidad and Tobago* 2 2
Ghana* 1 2 Vanuatu* 2 2
Greece* 1 2 Antigua and Barbuda* 3 2
Grenada* 1 2 Botswana* 3 2
Israel* 1 2 El Salvador* 2 3
Italy* 1 2 Guyana* 2 3
Japan* 1 2 India* 2 3
Latvia* 2 1 Indonesia* 2 3
Mauritius* 1 2 Jamaica* 2 3
Monaco* 2 1 Mali* 2 3
Panama* 1 2 Mexico* 2 3
Saint Vincent and Montenegro* 3 2
Grenadines* 2 1 Peru* 2 3
South Korea* 1 2 Ukraine* 3 2
Taiwan* 1 2 Albania* 3 3
Argentina* 2 2 Bolivia* 3 3
Benin* 2 2 Ecuador* 3 3
Brazil* 2 2 Lesotho* 3 3
Bulgaria* 2 2 Macedonia* 3 3
Dominican Republic* 2 2 Paraguay* 3 3
Mongolia* 2 2 Senegal* 3 3
Namibia* 2 2 Seychelles* 3 3
Romania* 2 2 Sierra Leone* 3 3
Samoa* 2 2 Turkey* 3 3
Sao Tome and Principe* 2 2 Bangladesh* 3 4
Serbia* 2 2 Bosnia-Herzegovina* 4 3
South Africa* 2 2 Colombia* 3 4
Suriname* 2 2
62
Country PR CL Country PR CL
Comoros* 3 4 Kosovo 5 4
East Timor* 3 4 Morocco 5 4
Liberia* 3 4 Niger 5 4
Malawi* 3 4 Nigeria 5 4
Maldives* 3 4 Pakistan 4 5
Moldova* 3 4 Singapore 5 4
Mozambique 4 3 Thailand 5 4
Papua New Guinea* 4 3 Togo 5 4
Philippines 4 3 Uganda 5 4
Solomon Islands 4 3 Venezuela 5 4
Tanzania 4 3 Armenia 6 4
Zambia* 3 4 Central African Republic 5 5
Burkina Faso 5 3 Djibouti 5 5
Georgia 4 4 Ethiopia 5 5
Guatemala* 4 4 Fiji 6 4
Guinea-Bissau* 4 4 Madagascar 6 4
Honduras 4 4 The Gambia 5 5
Kenya 4 4 Algeria 6 5
Kuwait 4 4 Angola 6 5
Lebanon 5 3 Azerbaijan 6 5
Malaysia 4 4 Bahrain 6 5
Nepal 4 4 Brunei 6 5
Nicaragua* 4 4 Cambodia 6 5
Sri Lanka* 4 4 Congo (Brazzaville ) 6 5
Tonga 5 3 Cote d’Ivoire 6 5
Bhutan 4 5 Egypt 6 5
Burundi* 4 5 Gabon 6 5
Haiti* 4 5 Iraq 5 6
63
Country PR CL Country PR CL
Jordan 6 5 Zimbabwe 6 6
Kazakhstan 6 5 Belarus 7 6
Kyrgyzstan 6 5 Chad 7 6
Mauritania 6 5 China 7 6
Oman 6 5 Cuba 7 6
Qatar 6 5 Guinea 7 6
Russia 6 5 Laos 7 6
Rwanda 6 5 Saudi Arabia 7 6
Tajikistan 6 5 Syria 7 6
United Arab Emirates 6 5 Burma 7 7
Yemen 6 5 Equatorial Guinea 7 7
Afghanistan 6 6 Eritrea 7 7
Cameroon 6 6 Libya 7 7
Congo (Kinshasa) 6 6 North Korea 7 7
Iran 6 6 Somalia 7 7
Swaziland 7 5 Sudan 7 7
Tunisia 7 5 Turkmenistan 7 7
Vietnam 7 5 Uzbekistan 7 7

64
APPENDIX 5: POLITICAL AWARENESS EES 2009

country Correct answers out of 4

0 1 2 3 4 Total

Austria Count 108 491 365 28 8 1000

% within country 10,8% 49,1% 36,5% 2,8% ,8% 100,0%
Belgium Count 212 383 265 111 31 1002
% within country 21,2% 38,2% 26,4% 11,1% 3,1% 100,0%
Bulgaria Count 319 402 192 68 19 1000
% within country 31,9% 40,2% 19,2% 6,8% 1,9% 100,0%
Cyprus Count 128 392 311 123 46 1000
% within country 12,8% 39,2% 31,1% 12,3% 4,6% 100,0%
Czech Republic Count 145 429 288 112 46 1020
% within country 14,2% 42,1% 28,2% 11,0% 4,5% 100,0%
Denmark Count 172 428 336 55 9 1000
% within country 17,2% 42,8% 33,6% 5,5% ,9% 100,0%
Estonia Count 290 389 225 82 21 1007
% within country 28,8% 38,6% 22,3% 8,1% 2,1% 100,0%
Finland Count 185 431 308 59 17 1000
% within country 18,5% 43,1% 30,8% 5,9% 1,7% 100,0%
France Count 126 499 297 69 9 1000
% within country 12,6% 49,9% 29,7% 6,9% ,9% 100,0%
Germany Count 175 479 289 49 12 1004
% within country 17,4% 47,7% 28,8% 4,9% 1,2% 100,0%
Greece Count 104 370 350 130 46 1000
% within country 10,4% 37,0% 35,0% 13,0% 4,6% 100,0%
Hungary Count 248 411 255 65 26 1005
% within country 24,7% 40,9% 25,4% 6,5% 2,6% 100,0%
Ireland Count 91 366 386 127 31 1001
% within country 9,1% 36,6% 38,6% 12,7% 3,1% 100,0%
Italy Count 323 349 241 70 17 1000
% within country 32,3% 34,9% 24,1% 7,0% 1,7% 100,0%
Latvia Count 273 383 229 84 32 1001
% within country 27,3% 38,3% 22,9% 8,4% 3,2% 100,0%
Lithuania Count 293 348 245 79 35 1000
% within country 29,3% 34,8% 24,5% 7,9% 3,5% 100,0%
Luxembourg Count 137 533 278 44 9 1001
% within country 13,7% 53,2% 27,8% 4,4% ,9% 100,0%
Malta Count 351 410 183 39 17 1000
% within country 35,1% 41,0% 18,3% 3,9% 1,7% 100,0%
The Count 214 476 255 51 9 1005
Netherlands % within country 21,3% 47,4% 25,4% 5,1% ,9% 100,0%
Poland Count 211 372 283 91 45 1002
% within country 21,1% 37,1% 28,2% 9,1% 4,5% 100,0%
Portugal Count 326 470 152 46 6 1000
% within country 32,6% 47,0% 15,2% 4,6% ,6% 100,0%
Romania Count 423 295 187 71 27 1003
% within country 42,2% 29,4% 18,6% 7,1% 2,7% 100,0%
Slovakia Count 133 340 347 134 62 1016
% within country 13,1% 33,5% 34,2% 13,2% 6,1% 100,0%
Slovenia Count 65 509 340 69 17 1000
% within country 6,5% 50,9% 34,0% 6,9% 1,7% 100,0%
Spain Count 167 354 341 120 18 1000
% within country 16,7% 35,4% 34,1% 12,0% 1,8% 100,0%
Sweden Count 132 439 355 73 3 1002
% within country 13,2% 43,8% 35,4% 7,3% ,3% 100,0%
United Count 227 367 300 90 16 1000
Kingdom % within country 22,7% 36,7% 30,0% 9,0% 1,6% 100,0%

Total Count 5578 11115 7603 2139 634 27069

% within Country 20,6% 41,1% 28,1% 7,9% 2,3% 100,0%

66
APPENDIX 6: POLITICAL AWARENESS CSES MODULE 3

country Correct answers out of 3

0 1 2 3 Total

Australia Count 248 536 834 254 1872

% within country 13,2% 28,6% 44,6% 13,6% 100,0%
Austria Count 165 247 405 348 1165
% within country 14,2% 21,2% 34,8% 29,9% 100,0%
Brazil Count 71 308 408 213 1000
% within country 7,1% 30,8% 40,8% 21,3% 100,0%
Czech Count 452 888 1059 765 3164
Republic % within country 14,3% 28,1% 33,5% 24,2% 100,0%
Switzerland Count 278 541 696 487 2002
% within country 13,9% 27,0% 34,8% 24,3% 100,0%
Deutschland Count 409 487 844 355 2095
% within country 19,5% 23,2% 40,3% 16,9% 100,0%
Finland Count 36 243 568 436 1283
% within country 2,8% 18,9% 44,3% 34,0% 100,0%
France Count 85 492 791 632 2000
% within country 4,3% 24,6% 39,6% 31,6% 100,0%
Croatia Count 118 320 287 279 1004
% within country 11,8% 31,9% 28,6% 27,8% 100,0%
Ireland Count 282 435 551 167 1435
% within country 19,7% 30,3% 38,4% 11,6% 100,0%
Iceland Count 215 213 651 306 1385
% within country 15,5% 15,4% 47,0% 22,1% 100,0%
Israel Count 174 307 459 260 1200
% within country 14,5% 25,6% 38,3% 21,7% 100,0%
Japan Count 415 559 323 76 1373
% within country 30,2% 40,7% 23,5% 5,5% 100,0%
Korea (South) Count 38 85 374 503 1000
% within country 3,8% 8,5% 37,4% 50,3% 100,0%
The Count 376 645 911 427 2359
Netherlands % within country 15,9% 27,3% 38,6% 18,1% 100,0%

67
Norway Count 960 574 299 179 2012
% within country 47,7% 28,5% 14,9% 8,9% 100,0%
New Zealand Count 167 347 465 170 1149
% within country 14,5% 30,2% 40,5% 14,8% 100,0%
Poland Count 163 230 586 838 1817
% within country 9,0% 12,7% 32,3% 46,1% 100,0%
Portugal Count 281 373 358 304 1316
% within country 21,4% 28,3% 27,2% 23,1% 100,0%
Sweden Count 764 364 294 125 1547
% within country 49,4% 23,5% 19,0% 8,1% 100,0%
Taiwan Count 451 414 589 451 1905
% within country 23,7% 21,7% 30,9% 23,7% 100,0%

Total Count 7975 10051 13484 8588 40098

% within country 19,9% 25,1% 33,6% 21,4% 100,0%

68
APPENDIX 7: IMPERFECTIONS CSES MODULE 3

Including all countries from the CSES Module 3 into account, do not result in empirical
evidence for a relationship between the proportion of people that are middle political
aware and corruption (rs[21] = .179, p = .438). A scatter plot of this relation is displayed
in figure XX. It is expected that high CPI score—thus countries with low corruption—are
associated with large proportions of people that are middle political aware. Especially
Sweden and Norway do not fit these expectations.

relationship proportion of the people that are middle political aware and corruption

10.00
Sweden
9.00
Norway
8.00

7.00

6.00
CPI

5.00

4.00

3.00

2.00

1.00

0.00
0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00
proportion of the people that are middle political aware (%)

Figure 1: Scatter plot of the relationship between the proportion of the people that are middle aware
and corruption.

Looking at the dataset, however, leads me to believe that there are problems
with these data. It seems that the political knowledge questions in these countries were
more difficult than in other countries given the high percentage of people that could not
69
answer any question correctly (Norway 47.7% and Sweden 49.4%). The questions were
constructed in such a way that approximately 2/3 would be able to answer the first
question correctly, 1/2 the second and only 1/3 the last question (Grönlund & Milner,
2006: 389). Theoretically it is possible to hold a survey without a single person having
every question incorrect. This is when the 1/3 of people who answer the first question
incorrectly are exactly that 1/3 who answer the last question correctly, no matter what
they respond to the second question. If, however, being able to answer a question is
transitive over the difficultness rate—that an individual cannot answer a question that is
more difficult than another question he answered incorrectly—then this would result in
a maximum of people that cannot answer a single question correctly. The percentage of
people that cannot answer any question correctly is then equal to the people who could
not give a correct answer to the easiest question. Thus, for the questions in the CSES
module this would be maximally 1/3 of the respondents. This reasoning is shown
schematically in figure 5.

Figure 2: Figure 6: q1, q2 and q3 are questions for which the percentage that is able to answer the
questions correctly is respectively 2/3, 1/2 and 1/3. The letters A through F are the respondents. A cross
represents a incorrect answer and a checkmark represents a correct answer. The left figure shows that
with these correctness rates it is possible that no individual answers every question incorrect. It does
not even matter what the persons answer for q2. The right figure shows a situation for which the group
of political unaware people—those who cannot answer any question correctly—is largest. From the
figure it becomes clear that the maximum amount of people that does not answer any question
correctly cannot be larger than the incorrectness rate of the easiest question. In the CSES module 3 this
is suppose to be around 33.33%.
70
If the fact whether an individual is able to answer a question is not transitive and only
related to the percentage of correctness to a question, then the percentage of people
that cannot answer a single question correctly in this survey is around 11 %
(=1/3*1/2*2/3). Since, being able to answer is probably not completely transitive, but
also not completely random the expectation is that the percentage of political unaware
people—those who are not able to answer a question correctly—is between 11.11% and
33.33%. Scores lower are not likely, but this cannot be ruled out. Meanwhile political
unawareness rates higher than 33.33 are impossible if the questions have the correct
difficultness. Sweden and Norway show low political awareness rates that are well
above 33.33%; 49.4% and 47.7 respectively. Therefore I do not consider the data from
those to countries to be reliable.
When the same analysis is executed, but without the data for Sweden and
Norway the analysis shows that it is, indeed, more likely, that there is a relationship
(rs[19]=0.432, p=0.065) between the medium aware group is calculated from the CSES
module and corruption as represented by a CPI value. A scatter plot of this relationship
including all countries but Sweden and Norway is shown in figure 6.

71
 relationship proportion of the people that are middle political aware and corruption

10.00

9.00

8.00

7.00

6.00
CPI

5.00

4.00

3.00

2.00

1.00

0.00
0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00
proportion of the people that are middle political aware (%)

Figure 3: Scatter plot of the relationship between the proportion of the people that are middle aware
and corruption. Sweden and Norway were excluded from the analysis.

72
73