Elecdem Training Network in Electoral Democracy

SEVENTH FRAMEWORK PROGRAMME THE PEOPLE PROGRAMME MARIE CURIE ACTIONS – NETWORKS FOR INITIAL TRAINING (ITN)

ELECDEM TRAINING NETWORK IN ELECTORAL DEMOCRACY

GRANT AGREEMENT NUMBER: 238607

Deliverable D15.1 – Dynamics of Political Communication Final Report

Early Stage Research fellow (ESR)

Thomas Vitiello

Host Institution Sabanci University

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Abstract

The traditional systems of political communication in Western democracies are being destabilized by changes in late modern society. On the one hand, the channels by which political communication comes to be are multiplying in a process that is becoming more diverse, fragmented, and complex. And on the other hand, power relations among key message providers and receivers are being rearranged, the culture of political journalism is being transformed; and conventional meanings of ‘democracy’, ‘citizenship’ and of the role of the media as democratic actors are being questioned and rethought. Fundamental questions centre on citizen participation in elections and whether vote choices and electoral outcomes reflect the underlying preferences of citizens. For example, can elections reflect the preferences of voters if parties do not mobilize voters or the media do not cover salient policy issues? From the late 1990s onwards, democratic elections have transformed into ‘two- screen’ media events, with the Internet playing an increasingly important role in providing political information that can help voters to make meaningful distinctions between policy positions of parties and candidates. This report aims at defining a framework of analysis for systematically examining links between media systems, party systems and voter behaviour. The report specifically focuses on the spread of Web-based political communication of non- party actors in electoral politics, through the development of online applications, which seek to help voters to pick the party or candidate that is closet to their own policy preferences.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

TABLE OF CONTENTS

TABLE OF CONTENTS...... 3 TABLE OF FIGURES...... 5 TABLE OF TABLES ...... 5

EXECUTIVE SUMMARY...... 6

THE CHANGING DYNAMICS OF POLITICAL COMMUNICATION ...... 9

Chapter 1. OBJECTIVES...... 9 1.1. Evaluating Electoral Democracy...... 9 1.2. Key Methodologies and Data...... 11

CHAPTER 2. TRENDS IN POLITICAL COMMUNICATION ...... 15 2.1. Modernization, Secularization, Globalization...... 16 2.2. Commercialization...... 19 2.3. Personalization & Professionalization...... 21 2.4. Comparing Political Communication...... 22

CHAPTER 3. A FRAMEWORK OF ANALYSIS FOR ADVANCED MEDIA SYSTEMS...... 25 3.1. The Four Dimensions of a Media System ...... 27 3.2. Three Models of Media Systems...... 32

CHAPTER 4. THE STUDY OF ELECTORAL CAMPAINGS ...... 39 4.1 Do political campaigns matter?...... 40 4.2. A Brief Summary of Campaign Effects ...... 43 4.3 How to Capture Campaign Effects?...... 46

CHAPTER 5. THE RISE OF THE INTERNET: NEW MEDIA – NEW RESEARCH ?...... 49

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

5.1 The Internet, Politics, and Democracy...... 49 5.2 A Changing Campaign Environment: The increasing role of non-party actors in electoral politics...... 52 5.3 The World of Voter Engagement Applications: Campaign Actor or Research Tool? ...... 53

CHAPTER 6. LOCATING PARTIES AND CANDIDATES IN A COMMON POLITICAL SPACE, A GUIDE TO VEA’S METHODOLOGY...... 61 6.1. Selection of relevant parties or candidates...... 63 6.2. Selecting and analysing relevant party documents...... 64 6.3. Issue identification and selection: Determining what’s at stake in the election ...... 66 6.4. Framing and ‘scaling’ the issue propositions ...... 67 6.5. Placing the parties: Step 1, first round of placements and generation of text snippet database...... 72 6.6. Contacting the parties for self-placement: Reaching agreement...... 74 6.7. Comparing Voter and Party Positions: Agreement or proximity approach? 76 6.8. Developing a multidimensional ‘map’ of party competition ...... 78

CHAPTER 7. MEASURING PUBLIC OPINOIN ONLINE: AN ASSESSMENT OF THE FIRST TURKISH ONLINE VOTER ENGAGEMENT APPLICATION...... 81 7.2. The Turkish Media System and Political Parallelism ...... 83 7.3. A Profiling of Oypusulasi users’: who was plugged in? ...... 86 7.4. ‘Looking it up’: Through which web channel did the users access oypusulasi.org? ...... 89 7.5. The Oypusulasi ‘mini-campaign’ and the double-screen effect...... 91 7.6. Press-party parallelism and audience segmentation: what output for a VEA? ...... 97

CHAPTER 8. CONCLUSIONS...... 102

REFERENCES...... 105

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

TABLE OF FIGURES

Figure 1 Centrality model of communication content ...... 13 Figure 2 Relations of Individual Cases to the Three Models ...... 37 Figure 3 The 'Recalculte your position' button allows users to select which issues areas are considered in their output...... 70 Figure 4 Examples of 'recalculate position' for the user and the candidates according to the issues selected ...... 71 Figure 5 Output displaying text justifying a single issue coding ...... 73 Figure 6 An example of single issue output ...... 80 Figure 7 Proportion of Women among Users ...... 88 Figure 8 Median Income of the Users ...... 88 Figure 9 Proportion of 15-24 years old Users per Region ...... 89 Figure 10 Hourly Distribution of Connections to the Oypusulasi Website between May 15th and June 12th ...... 93 Figure 11 Number of Connections and Publication of Newspapers Items ...... 96 Figure 12 Evolution of the PTVs (scale 0-10) expressed on the Oypusulasi website ...... 99 Figure 13 Distribution of the Advices Given in favour of BDP and CHP by the Oypusulasi Website ...... 100 Figure 14 Distribution of the Advices Given in favor of AKP and SP by the Oypusulasi Website ...... 101 Figure 15 Distribution of the Advices Given in favor of BBP and MHP by hte Oypusulasi Website ...... 101

TABLE OF TABLES Table 1 Newspaper Sales per 1,000 Adult Population, 2000 ...... 28 Table 2 Pattern of Variation in Four Media System Dimensions ...... 38 Table 3 Channel of Access to the Oypusulası site ...... 90 Table 4 Media Items as Interruptions for the MITS Analysis ...... 92 Table 5 Newspaper Items Covering the Oypusulasi website ...... 93 Table 6 Results of the MITS Analysis ...... 94

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

EXECUTIVE SUMMARY

Within the ELECDEM network, the work package 15 on the Dynamics of Political Communication aimed at systematically examine links between media systems, party systems and voter behaviour. Among those three areas, the media sphere has sparkled great interested among consultants and researchers in order to explain the interaction between political parties and voters during election times, and in some cases to explain election outcomes. However, in order to grasp and understand the impact of media systems on elections it is important to acknowledge the plurality of existing media systems within advanced and less advanced democracies. At the same time, the forces of globalization and commercialization are pushing media systems toward a greater degree of homogenization. Therefore, studying the dynamics of political communication implies:

- To review the trends undergone since several decades in electoral democracies. Whether this means to appraise the development proper to the media sphere such as the commercialization of media outlets and the personalization of politics due to the central place of television for political communication; it also involves the evaluation of phenomena involving the political and social spheres of society, such as the globalization, secularization, and modernization of societies in electoral democracy.

- To analyze the context in which political communication occurs. Political systems and institutional settings have a long standing impact on the relationship between elites and citizens, however it is also important to set a framework of analysis which also includes media systems and which is allowing for comparison and differentiation between electoral democracies with regard to the media structure.

- To assess the state of the art of the study of electoral campaigns. In modern democracies, where elections are the key institution, political campaigns are the

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

most intense time for interaction between political elites and citizens over matters of public policy. Therefore, how to capture campaign effect is central to the study of political communication.

- To acknowledge that the dynamics of political communication are not written in stone, but are changing and nowadays more than ever. The rise of the Internet as channel for political communication and the rise of non-party actors in electoral politics have made the sphere of political communication more fragmented and saturated than ever.

The notion of electoral democracy refers to the role of elections in promoting accountability, representation and legitimacy in democracies. Fundamental questions centre on citizen participation in elections and whether vote choices and electoral outcomes reflect the underlying preferences of citizens. For example, can elections reflect the preferences of voters if parties do not mobilize voters or the media do not cover salient policy issues? In a fragmented and saturated communication sphere, it is becoming increasingly harder for citizens to know precisely how parties and candidates stand on specific public policy issues. Hence, some online ‘vote helpers’ have emerged and have encountered a great success among the increasing volatile electorate.

- Online Voting Engagement Applications (VEA) provide a platform on which stances of parties and candidates on issues can be compared between each other and with the stances of users.

- The nature of the impact of VEA sites on the electorate is of different kind: it can increase turnout, increase the political information available to the users, and help the users to decide for whom to cast their vote.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

- In order to reach a substantive number of users, VEAs need to establish media partnership. That is, one or several media regularly promote in their outlets the existence of the online VEA, and in exchange the team of researchers building the VEA site provides the media stories with substantive content on the political landscape and on the issues preferences expressed by the users, among other possible topics.

- In the fragmented media environment of the 21 st century, a diversification of the media partners is desirable. On the one hand, there should be a diversification among traditional media sources that is television, national newspapers, regional newspapers, free newspapers, radio. Choices should be made according to the media structure of the country in which the VEA is implemented. And on the other hand, the new media that Web-based political communication represents should not be neglected, since it bears the potential for being as effective as traditional media in reaching users.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

THE CHANGING DYNAMICS OF POLITICAL COMMUNICATION

Chapter 1. OBJECTIVES

1.1. Evaluating Electoral Democracy

These changes in the landscape of the political communication echo socio-cultural and political transformations that the traditional structure of modern societies have undergone since several decades. With regard to electoral politics, declining levels of party attachment of voters, and the withdrawal of parties from society, have weakened the core vote for traditional political parties (Inglehart, 1997; Dalton and Wattenberg, 2000). As a consequence of social differentiation and of the deregulation of the media markets, parties can no longer count on the support of affiliated newspapers to get their messages across; instead they compete for coverage through the increasingly differentiated broadcasting media. By emphasizing the end of key building blocks of electoral politics made out of parties, organized interests and media outlets, and by drawing power away from the formal political arena, these processes have led to discussions on the poor health of democracy (Dalhgren, 2005).

However, other trends also related to those processes, such as the increase in political

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 voices and new modes of political engagement, may suggest that dispersions and openings can lead to democratic gains. A peculiar characteristic of the “postmodern” campaign environment (Norris, 2000) is the emergence of “non-party actors” in electoral politics (Farell and Schmitt-Beck, 2008). In addition to the fragmentation and diversification of the media landscape, an increasing numbers of interest groups and of ordinary and unorganized citizens are voicing their interests and concerns in the political arena. As a result of those interrelated trends, political campaigns have regained interested among political scientist. As a matter of fact, volatility has become one of the main characteristics of Western electorates, and an increasing number of voters are deciding for whom to cast their ballot during the election campaign (Ihl, 2002; Cautrès and Muxel, 2009).

In our modern democracies, where elections are the key institution, political campaigns are the most intense time for interaction between political elites and citizens over matters of public policy (Manin, 1997; Jerit, 2004; Kriesi, 2008). However, modern voters often struggle to define their own ideological orientations as well as being uncertain of the policy stances of the parties who compete for their votes. At the same time, research shows that issues and policy considerations have become more important for voters in the same period they have been de-emphasized by political parties in Europe and beyond (Franklin, 2003). The voter’s task is made even more difficult by the proliferation of new parties across the European party systems and, the corresponding decline in the popularity of traditional parties, which has transformed the national sphere in many countries into a fragmented and multi-faceted political landscape. This growth in the numbers of electorally viable parties leaves voters facing a problem of information overload. Yet, in the era of the spread of Web- based communication, where facing this extensive reservoir of election-related information, sorting, selecting, and processing information has turned into an even more pressing problem for voters than in the past. All of these factors mean that modern Europe’s voters face a challenging decision: ‘which party should I vote for?’, with few reliable transparent resources on which to base their decision.

In this context, it is not surprising that during campaign periods actors of different kinds make to increase the quality of the electoral process their mission. On this new terrain, in many European countries, agencies of different kinds (universities, government agencies, media organization) have over the course of recent elections developed “vote helpers” – Web-

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 base software tools that may assist citizens in taming the flood of political information at elections, also know as Voter Advice Applications (VAAs), and more recently as Voter Engagement Applications (VEAs). The capacity of the electorate to make meaningful distinctions between competing party policy positions is essential to the functioning of modern democracies, and voters can only make meaningful distinctions on the basis of reliable political information. The core goal of VEA sites is to provide voters with such information in a structured manner. In an age of voter dealignment, these VEAs are seen as offering a useful resource for undecided voters seeking authoritative information on the issue positions of parties, at least in multiparty systems where such applications mostly emerged (Farell and Schmitt-Beck, 2008).

Therefore, we can note that:

- VEAs are a transnational phenomena, as exemplefied by the EU Profiler VEA, which was made possible for the technological forces behind globalization, and mostly the Internet.

- VEAs have the potential for becoming a key actor in the dynamics of political communication during elections, VEA sites interact with the three main actors of the public domain: the parties, the citizens/voters, and the media.

- The succes of VEAs is not only related to the nature of the party system in which it is developed, fragmented party systems suggest that voters may need more for information in order to decide for whom to cast their vote, but also to the nature of the elections (primary versus second-order election) and of the media system of the country, as it will be shown in Chapter 7.

1.2. Key Methodologies and Data

1.2.1. Content Analysis

One of the key methods used for this research is content analysis of party manifestos and candidate platforms running for office. Content analysis is a technique for quantitative

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 analysis of extensive texts within the framework of a communication model. As Riffe et al. (1998) define it:

“Quantitative content analysis is the systematic and replicable examination of symbols of communication, which have been assigned numeric values according to valid measurement rules, and the analysis of relationships involving those values using statistical methods, in order to describe the communication, draw inferences about its meaning, or infer from the communication to its context, both of production and consumption.” (p. 20)

Thus, content analysts are interested in the linguistic context of a text, especially the quantitative aspect. Besides this approach to language, there are two ideas: the idea that the texts “hide” something and the idea that the analysis must be systematic in one way or another, i.e., the analysis must explain the text both in part and in whole, both the elements and the structure, and no textual elements must be excluded, unexplained, or vague (Lindkvist, 1981). The fundamental assumption of content analysis links the interest of the text producer and the quantitative profile of the text. The text “hides” the interest of the producer, but it can be revealed by quantitatively measuring the text. The manifest text is coded, but when relating the measured result to a general communication model the character of different textual elements can be explained. Content analysis views communication content as an antecedent condition, and has presented possible consequences of exposure to content that may range from attitude change to the cognitive images people learn from it (Riffe et al., 1998). But content is itself the consequence of another antecedent processes that have led to or shaped its construction. Many of the symbols that show up in communication messages that contain particular images, ideas, or themes reflect the importance – and clearly antecedent – values of the culture, or its leaders. Thus, as simply summarized by the content- centred model of communication process in Figure 1, content analysis can be an important tool in theory building about both communication effects and processes.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

ANTECEDENT CONDITIONS (a) Individual psychological/professional (b) Social, political, economic, cultural or other contextual factors

That are assumed or demonstrated to affect,

COMMUNICATION CONTENT

Which is an antecedent/correlate of,

(a) assumed or demonstrated

(b) immediate or delayed

EFFECTS

Figure 1 Centrality model of communication content Source: Riffe et al., 1998: 9

1.2.2. Electoral Survey Design

Since one decade, the Internet is having a profound effect on survey research. Similarly to the different point of views existing on the impact of the Internet for electoral democracy, some argue that soon Internet surveys will replace traditional methods of survey data collection. While others are urging caution or even voicing skepticism about the future role of Web surveys will play. In an article wrote in 2000, Couper underlines that a few years earlier he was predicting that “the rapid spread of electronic data-collection methods such as the Internet would produce a bifurcation in the survey industry between high-quality surveys based on probability samples and using traditional data collection methods, on the one hand, and surveys focused more on low cost and rapid turnaround than on representativeness and accuracy on the other” (p.466). But in 2000, he admitted that he was wrong and that instead the impact of Web surveys was “more of a fragmentation than a bifurcation, with vendors trying to find or creat a niche for their particular approach or product”. Thus, there is a wide array of approaches representing varying levels of quality and cost.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

One of these approaches has led to the development of Web surveys incorporated into VEAs. VEA sites utilise both issues positioning statements, usually about 30, and traditional socio-demographics and propensity to vote questions. And sometimes, before the users leave the VEA sites, they are asked to optionally answer additional survey questions on voting behaviour and political socialisation. Therefore making the development of such ‘vote helpers’ a task busy with electoral survey design, as we will see in Chapter 6.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

CHAPTER 2. TRENDS IN POLITICAL COMMUNICATION

There has always been an intimate connection between mass communication and the conduct of politics, in whatever kind of regime. In totalitarian or authoritarian societies, ruling elites use their control of the media to ensure conformity and compliance and to stifle dissent by one means or another. In democracies, the media have a complex relationship with sources of power and the political system. On the one hand, they usually find their raison d’être in their service to their audiences, to whom they provide information and views according to judgements of interest and need. In order to perform this service they need to be independent of the state and of powerful interests. On the other hand, they also provide channels by which the state and powerful interests address the people, as well as platforms for the views of political parties and other interest groups. They also promote the circulation of news and opinion within the politically interested public. (Manin, 1995; Norris, 2000, 2004; McQuail, 2005)

This general view of the neutral and mediating role of the media in politics has to be modified to take into account of variant forms, especially one in which particular media choose to play a partisan role on behalf of a party or interest, or are closely allied with some powerful economic interest of ideological bloc. There is a third possibility where the state and the parties have considerable effective power over nominally free media and uses this power for its own advantage. In global terms, the situation is not at all unusual (McQuail, 2005).

Contemporarily, political systems and political communications systems are dynamic, constantly evolving, never settled. “Just when we think we understand how it all works, things change” (Swanson, 2004: 45). Sometimes the changes seem to be evolutionary, steps along a path that leads to a destination we can foresee. At other times, the familiar path turns in new and unexpected directions. In his influential book on the principles of representative government , the term he uses for the form of government of Western liberal democracies, Bernard Manin (1997: 247-303) argues that the democratic systems of government are changing profoundly. According to Manin, after the classical parliamentarism of the

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 nineteenth century and the party democracy that was established at the beginning of the twentieth century, representative government currently takes the form of an “ audience democracy ”. The characteristics of this new form of government include personalization of elections and the rise of experts in political communication, increasing vagueness of political offers, the omnipresence of public opinion, and the transfer of the political debate from the backrooms of parliamentary committees and the central offices of parties and association to the public sphere.

Similarly, party researchers point to the decline of the ideologically oriented and structurally rooted mass party and the rise of the “electoral professional party” (Panebianco, 1988) or the “cartel party” (Mair, 1997). This transformation has led, on the one hand, to the declining importance of the traditional party apparatus and of party militants, and, on the other hand, it has reinforced the importance of the party leaders and of the much more independent electoral audience (McAllister, 2007; Karvonen, 2010). Media researchers note that political communication is no longer focused on parties but on the media (Swanson and Mancini, 1996). They observe the increasing independence of the mass media from the political parties. Finally, they begin to speculate about the arrival of a third age of political communication (Blumler and Kavanagh, 1999), where the public possesses greater autonomy with regard to the media. Characteristics of this new style of political communication include the multiplication of the means of communication, an affluence of communication channels, increasing commercialization, and further acceleration of the speed with which political information becomes accessible for a significant part of the public.

This chapter aims to provide an overview of the trends characteristic of political communication in modern societies.

2.1. Modernization, Secularization, Globalization A powerful trend is clearly underway shaping the structure of the public spheres. Media systems across the world are becoming increasingly alike. Political systems, meanwhile, are becoming increasingly similar in the patterns of communication they incorporate.

Much of the literature on homogenization sees it in terms of Americanization or 16

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 globalization: that is, in terms of forces external to the national social and political systems in which media systems previously were rooted. Other explanations focus on changes internal to these national systems. An important distinction can also be made between media centric perspectives, for which changes in media systems are autonomous developments that then influence political and social systems, and those that see social and political changes as causally prior to media system change (Hallin and Mancini, 2004).

It is reasonable to say that homogenization is to a significant degree a convergence of world media toward forms that first evolved in the United States. First being an American peculiarity, commercial broadcasting is now becoming the norm worldwide. The personalized, media-centred forms of election campaigning, using techniques similar to consumer-product marketing, which again were pioneered in the United States, similarly are becoming more and more common in European politics (Poguntke and Webb, 2005). Their diffusion around the globe cannot, however, be attributed to the action of a single agent. It has not been an unilateral process. Where European countries have borrowed American innovations, they have done so for reasons rooted in their own economic and political processes, often modifying them in significant ways (Farrell and Webb, 2000). Two important elements of globalization clearly rooted within Europe should be noted here. One is European integration: with the Television without Frontiers Directive of 1989, the European Union embarked deliberately on an attempt to create a common broadcasting market. Closely related is a strong trend toward internationalization of media ownership. The search for ever greater amounts of capital to invest in new technologies and to compete in liberalized international markets has produced a strong trend toward the development of multinational media corporations (Herman and McChesney, 1997).

The term modernization has often been proposed as an alternative to Americanization. It reveals an effort to stress that changes in political communication in Europe are not created purely by exogenous forces, but are rooted in a process of social change endogenous to European society. The term modernization is problematic. It carries an evolutionist connotation, an implicit assumption that change is to be seen as “progress”, necessary and unilinear. It also amalgamates together many dimensions of change – technological, cultural, political, and economic – that need to be distinguished analytically if we are to be clear about the forces at work, even if we conclude in the end that these different dimensions are

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 interrelated. (Hallin and Mancini, 2004)

One important component of the modernization perspective is the idea that the importance of group solidarity and the centrality of organized social groups are giving way to greater individualism (Giddens, 1990). According to this view, the European political order used to be organized around social institutions, such as political parties, trade unions, and churches, among others, rooted in ideological commitments and group loyalties related to broad social divisions, especially those of social class and religion. The ties of individuals to these groups were central both to their identity and to their material well-being, and the institutions connected with these groups were central to the organization of the public sphere. If political communication is being transformed, this cannot be understood without reference to the collapse of this old political order, and its displacement by a more fragmented and individualistic society. This is sometimes interpreted as the “decline of party”, though some analysts dispute this interpretation, arguing that professional electoral parties are actually more effective than earlier mass parties at conquering and wielding political power (Manin, 1997; Mair, 1997). It does seem to be correct, however, that the stable psychological and sociological bonds that once existed between parties and citizens have been weakened in this transformation. Party membership has declined (Dalton and Wattenberg, 2000), as have church and trade union membership.

It is clear that the mass media play an important role in this process of political change; indeed, the increasing centrality of the mass media to the process of political communication is central to the very definition of Americanization or modernization in most discussions of political change. Most accounts of political change in Europe list media system change as a significant and independent factor:

“…new technologies and…changes in the mass media…have enabled party leaders to appeal directly to voters and thereby undermined the need for organizational networks…”(Mair, 1997: 39).

“The mass media are assuming many of the information functions that political parties once controlled. Instead of learning about an election at a

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

campaign rally or from party canvassers, the mass media have become the primary source of campaign information. Furthermore, the political parties have apparently changed their behaviour in response to the expansion of mass media. There has been a tendency for political parties to decrease their investments in neighbourhood canvassing, rallies, and other direct contact activities, and devote more attention to campaigning through the media” (Dalton and Wattenberg, 2000: 11-12).

The case of politics provides fairly clear evidence of the adaptation of a social institution to the rise of mass media, especially given the fact that the media have become the main source of information and opinion for the public. The challenge to politics from the growing centrality of the mass media and the rise of ‘media logic’ have taken several forms, such as the commercialization of media content or the personalization of political communication.

2.2. Commercialization Hallin and Mancini (2004) argue that the most powerful force of homogenization and globalization within the media system is commercialization. While at one level the term ‘commercialism’ may refer objectively to particular free-market arrangements, it has also come to imply consequences for the type of media content which is mass produced and ‘marketed’ as a commodity, and for the relations between the suppliers and the consumers of media (McQuail, 2005). The term ‘commercial’, applied as an adjective to some types of media provision, identifies correlates of the competitive pursuit of large markets (Bogart, 1995).

Commercialization has transformed both print and electronic media in Europe, though the change is especially dramatic in the later case. In the case of the print media, the post- World War II period is characterized by a gradual decline of the party press and general separation of newspapers from their earlier rooting in the world of politics. As party papers declined, commercial newspapers have grow in strength; these newspapers, similar to their American counterparts, tend to be catch-all papers, defining themselves as politically neutral 19

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

(though generally liberal and centrist in ideological orientations) and committed to an informational model of journalism (Hallin and Mancini, 2004). Picard (2004: 61) argues that nowadays the primary content of newspapers is commercialized news and designed to appeal to broad audiences, to entertain, to be cost effective and whose attention can be sold to advertisers. The result is that stories that may offend are ignored in favour of those more acceptable and entertaining to a larger number of readers, that stories that are costly to cover are downplayed or ignored and that stories creating financial risks are ignored. This leads to homogenization of newspaper content, to coverage of safe issues and to a diminution of the range of opinion and ideas expressed (Bourdieu, 1998; McQuail, 2007; Picard, 2004).

The most dramatic changes, however, has clearly been the commercialization of European broadcasting (Hultén, 2007). Despite the growing global flow of messages, products, and institutional forms, mainly coming from the United States, public service broadcasting was regulated by norms and values firmly rooted in the distinct cultural and political paradigms that prevailed in the different nation states of Europe. “Sustaining and renewing the society’s characteristic cultural capital and cement” was indeed one of the central missions of public service broadcasting (Blumler, 1992: 11). In important ways the public service system limited the social and political impact of television, creating continuity between television culture and the established culture of wider society (Hallin and Mancini, 2004).

Since the 1980s, the commercial “deluge”, as many discussions have characterized it, is dramatically undercutting this system, disrupting the connection between broadcasting and national systems. There has been much comment on the “tabloidization” of newspapers as they compete for readers. The equivalent process in television has led to many new forms of ‘reality’ television that deal in all kinds of ‘human interest’ and dramatic topics in a variety of formats (Bird, 1998; Connell, 1998).

Many of the characteristics commonly attributed to television in discussions of the transformation of political communication – personalization, for example, and the tendency to focus on the experience and perspective of the “common citizen” (Neveu, 1999) – are characteristics of commercial media, more than of television as a technology, and were developed only to a limited extent under the public service system (Hallin and Mancini,

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

1984). Langer (2003) shows that those transformations in the media are a question of access (who gets in the news) and of representation (how they are depicted). The new format characterized by infotainment, dramatization and personalization does make news more accessible to the many, but also led to a trivialization of what people learn from news (McQuail, 2005).

2.3. Personalization & Professionalization Following this perspective, media like to personalize abstract topics to make them more concrete and interesting to the audience. There is a general tendency to look for well- know people, especially politicians and celebrities, around which to construct news (McQuail, 2005: 312). This trend can be traced back to the growth of television from the 1950s and 1960s onwards as main source of information and opinion for the public. In the early years of television’s development, relatively few resources were devoted to the coverage of politics, which was seen to be not well suited to the new medium (Patterson, 1994). That view changed rapidly as the potential of television to market politics to voters became apparent (Schudson, 2002; McAllister, 2007). By the late 1960s and early 1970s the television coverage of politics – and especially political leaders – had been established, and in turn television began to influence the way in which voters viewed their leaders, making television an indispensable tool for modern election campaigning in all the established democracy (Mughan, 2000; Norris, 2000; Pogtunke and Webb, 2005).

One indication of the profound nature of the impact of television on political leaders has been the increasing importance of televised leaders’ debates during election campaigns (Pogtunke and Webb, 2005). In most established democracies, the debate has become an established and formal part of the election campaign. While it is tempting to see television as the prime mover behind the personalization of politics, political parties also play a key role in the process. Parties find it easier to market political choices to voters through “a familiar personality, who can promote the party’s policies much more effectively to voters when compared to the simple dissemination of a press release or through the publication of a policy document” (McAllister, 2007: 8).

Thus, the media and particularly the television confer particular salience and

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 vividness to the individuality of the candidates, at least the ones whom fit the new medium and that we can call “media figures” (Manin, 1997). But the growing role of personalities at the expense of platforms in the way how politics is communicated to the citizens is also a response to the new conditions under which elected officials exercise their power. Nowadays, it is more difficult for candidates to make detailed promises: such platforms would become unwieldy and unreadable, because the government intervenes in a whole series of areas. Facing those new circumstances, the voters are increasingly relying on the personal trust inspired by the candidate as matter of selection, rather than evaluation the plans for future actions. (Manin, 1997)

The widespread of television as the main mean of political communication and the weakening of voting based on social classes deeply affected the nature of campaigning. As noted earlier, Manin (1997) describes an “audience democracy”, in which in place of, or in addition to, traditional campaign practices such as rallies of the party, parties rely on the sophisticated use of the mass media to persuade voters – the consumers of political communication – to support them at election time, and they offered campaigns that featured the appealing personalities of the parties (Franklin, 1994; Kavanagh, 1995; Negrine, 1996; Swanson, 2004). The modern model of campaigning has been identified with some important and fundamental changes in political parties and their relationship to voters. The media- intensive modern model has brought the professionalization of political campaigning, as technical experts in using mass media, opinion polling, and marketing techniques have been brought into political parties – sometimes as consultants, sometimes as party employees or officials. As political campaigning has been professionalized, the content of campaigns also has changed. “The modern model is, above all, a model for how to win elections” (Swanson, 2004: 49). It leads to catch-all political parties and catch-all political campaigns, in which parties try to appeal to the broadest range of political opinions and, thereby, the greatest number of votes. At the core of the modern, catch-all political campaign is, often, the personality of the party leader, whose appealing qualities are featured to catch the attention and support of voters (Manin, 1997).

2.4. Comparing Political Communication Scholars in comparative politics were never really interested in the mass media and

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 political communication. In communication science on the other hand, political communication has always been a central subject; though it was believed for a long time that it would suffice to describe singular phenomena in the realm of national politics or to subscribe to historical studies. Thus, until the early 1990s communication research lacked an international orientation comparable to that of political science (Pfetsch and Esser, 2004). Comparative research in political communication deserves more attention because it enables us to inspect the findings critically by using the examination made by the community of researchers, and only by doing so enables us to reach conclusions with an extensive claim to validity (Pfetsch and Esser, 2004).

Moreover, as developed in the previous sections, in the twenty-first century we are confronted with developments in the realm of politics and mass communication that rule out the conception of political communication as a phenomenon that could be defined within singular national, cultural, or linguistic boundaries. In fact, the challenge since a couple of decades is to face the developments and consequences arising from the modernization and globalization of political processes. The fundamental transformation of the media systems of the Western world, caused by the changes in information technology and communication infrastructure and by the global media economy and diffusion of news (Pfetsch and Esser, 2004), also belongs to the driving forces behind comparative research. A clear sign of the globalisation of media is the growth and concentration of internationally active media conglomerates (Meir, 2007). This development has had significant repercussions for national media systems. In almost all European countries there has been a reorientation of media policy with respect to deregulation and the opening up of media markets.

Thus, the comparative study of political communication is an essential supplement to the nationally focused studies on which most of our knowledge of the subject has been built. They have brought to light transnational trends that otherwise would be difficult to notice, and these trends have greatly advanced the appreciation of how the relationship between political institutions and media institutions shapes political communication everywhere. In addition, we now know that the structures and processes of media development and communications do systematically impact the development of democracy, the legitimization of political power, and the participation in politics (Norris, 2004; McQuail, 2005), making

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 comparative political communication a central aim of comparative research.

If we assume that the specific structures, norms, and values in political systems shape the political communication roles and behaviours, we observe that comparative research is often designed in such a way that the countries studied are selected with regard to the contextual conditions of the object of research. Thus the crucial questions to be answered are 1) What always applies regardless of the contextual conditions? 2) How does the object of investigation ‘behave’ under the influence of different contextual conditions? Gurevitch and Blumler (2004) stress that comparative research should be designed to realize “double value”. It should aim to shed light not only on the particular phenomena being studied but also on the different systems in which they are being examined (Pfetsch and Esser, 2004).

However, the fundamental problem of comparative research in the social sciences lies with the establishing of functional equivalence. In order to establish such equivalence in the literature scholars refer to the “most similar” and “most different systems design” (Przeworski and Teune, 1970). Studies that are based on a “most similar design” make it possible to study the cultural differences in most similar systems. Studies that are based on a “most different design” reveal the similarities in the systems that differ the most. The difficulty of ascertaining functional equivalent also lies down in the selection and construction of indicators and methods in such a way that it does not amount to contortions and the interpretation of measurement artefacts as differences.

Besides the need for pursuing comparative research, this chapter has stressed out two comprehensives themes of comparative research in political communication. On the one hand, fears concerning the homogenization of media, media contents, and political communication processes as a result of technological, social, and political change led to the debate of concepts of convergence such as Americanization globalization, and modernization. On the other hand, the suspicion that the media would dominate the modern political publicity process with the implication of dysfunctional effects on modern democracies provoked an exhaustive preoccupation with the structures, actors, media contents, and effects of political communication. In order to take into account both those themes, the next chapter will propose a metatheoretical approach to building a framework of study of media systems.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

CHAPTER 3. A FRAMEWORK OF ANALYSIS FOR ADVANCED MEDIA SYSTEMS

According to the conventional view, modern political communication is shaped by some transnational trends that have led to structurally similar but by no means identical consequences in each country, and especially in countries that have advanced media systems. The underlying process is one of adaptation, where national institutions and practices shape in locally appropriate ways the manner in which transnational trends become manifest in each country. National studies reveal, among other things, how the practices of political communication always reflect particular political cultures, institutions, actors, histories, and circumstances. This chapter presents a framework of research in comparing political communication and media systems as developed by Hallin and Mancini (2004b) in their book Comparing Media Systems. Three Models of Media and Politics .

Their study is based on a “most similar systems” design. As Lijphart (1971) stresses, one of the greatest problems in comparative analysis is the problem of “many variables, few cases.” One of the principal means of solving that problem, he notes, is to focus on a set of relatively comparable cases, in which the number of relevant variables will be reduced. This approach will reduce the number of cases, which is a benefit in a field such as communication research where the literature and the data available have been rather limited until recently. And it is also a benefit in the sense that it is impossible for analysts to handle competently more than a limited number of cases. Hallin and Mancini (2004b: 6) limit themselves to North America and Western Europe, in a way that they are dealing with systems that have relatively comparable levels of economic development and much common culture and political history.

The authors build their work on the agenda set out by Siebert, Peterson and Schramm in Four Theories of the Press (1956) where the authors defend the thesis that “the press always takes on the form and coloration of the social and political structures within which it operates” (pp.1-2). Thus, the scope of Hallin and Mancini (2004b: 8) is to attempt to show how different media models are rooted in broader differences of political and economic structure. “They argue that one cannot understand the news media without understanding the nature of the state, the system of political parties, the pattern of relations between economic

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 and political interests, and the development of civil society, among other elements of social structure.” However, they wish to distinguish themselves from Siebert et al. on one specific point. While in Four Theories of the Press , Siebert et al. seem to assume that the media will always be the “dependent variable” in relation to the “system of social control”, which it “reflects”. Hallin and Mancini argue that if in many cases it may be reasonable to assume that the media system “reflects” other aspects of social structure – the party system, for example – there is good evidence that media institutions have an impact of their own on other social structures.

This position takes up on the belief that the media, as introduced in Chapter 2, have become an important exogenous variable affecting other political institutions. It is also one of the reasons why scholars in comparative politics have begun to pay attention to media institutions they previously ignored. Recently, this is beginning to change, due in part to a growing feeling that the media are less “reflective” than they once were. Sometimes this change may actually be exaggerated. For instance, media scholars often tend to have a professional bias toward overstressing the independent influence of media. Much speculation in comparative politics about “videocracy” (Mazzoleni, 1995) might be an example here.

Hallin and Macini’s work is an attempt to replace the four theories with “a new set of models, better-grounded empirically but sharing something of the parsimony of the originals” (p.10). Their ideal type models, which will be further elaborated in the following sections, are the Liberal Model, which prevails across Britain, Ireland and North America; the Democratic Corporatist Model, which prevails across northern continental Europe; and the Polarized Pluralist Model, which prevails in the Mediterranean countries of southern Europe. The Liberal Model is characterized by a relative dominance of market mechanisms and of commercial media; the Democratic Corporatist Model by a historical coexistence of commercial media and media tied to organized social and political groups, and by a relatively active but legally limited role of the state; and the Polarized Pluralist Model by integration of the media into party politics, weaker historical development of commercial media, and a strong role of the state. It should be stressed that the primary purpose of those ideal types is not the classification of individual systems, but the identification of patterns of relationship between system characteristics.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

3.1. The Four Dimensions of a Media System

The framework they use to compare media systems is structured around four major dimensions according to which media systems in Western Europe and North America can usefully be compared: (1) the development of media markets, with particular emphasis on the strong or weak development of a mass circulation press; (2) political parallelism, that is the degree and nature of the links between the media and political parties or, more broadly, the extent to which the media system reflects the major political divisions in society; (3) the development of journalistic professionalism; and (4) the degree and the nature of state intervention in the media system.

3.1.1. The Structure of Media Markets

One of the most obvious differences among media systems has to do with the development of the mass circulation press. In some countries mass circulation newspapers developed in the late nineteenth and early twentieth century. In other they did not. That historical difference is reflected today in sharply different rates of newspaper circulation, from a high of 720 per thousand adult population in Norway to a low of 78 per thousand in Greece. As can be seen in Table 1, high rates of newspaper circulation are characteristic of Scandinavia and other parts of Northern Europe, and low rates characteristic of Southern Europe.

“The distinction here is not only one of quantity. It is also a distinction in the nature of the newspaper, its relation to its audience and its role in the wider process of social and political communication. The newspapers of Southern Europe are addressed to a small elite – mainly urban, well-educated, and politically active. They are both sophisticated and politicized in their content, and can be said to be involved in a horizontal process of debate and negotiation among elite factions. The newspapers of Northern Europe and North America, by contrast, ten to be addressed to a mass public not necessarily engaged in the political world. They are, in this sense, involved in a vertical process of communication,

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

mediating between political elites and the ordinary citizen [...].” (Hallin and Mancini, 2004b: 22)

Norway 719.7 Finland 545.2 Sweden 541.1 Switzerland 453.7 United Kingdom 408.5 Germany 375.2 Austria 374.3 Denmark 347.1 Netherlands 345.9 United States 263.6 Canada 205.7 Ireland 191 France 190 Belgium 186.5 Spain 129.4 Italy 121.4 Portugal 82.7 Greece 77.5 Table 1 Newspaper Sales per 1,000 Adult Population, 2000 Source : Hallin and Mancini (2004b: 23)

The newspapers in Southern Europe, with their relatively low circulation, have not historically been profitable business enterprises, and have often been subsidized by political actors. The high-circulation newspaper markets of Northern Europe, on the other hand, have sustained strong commercial media enterprises. One interesting manifestation of this difference in patterns of development of the press is the fact that there are large gender differences in newspaper readership in Southern Europe, while these differences are small or nonexistent in the other regions covered by the book of Hallin and Mancini. Because the media were closely tied to the political world in Southern Europe, and because women were historically excluded from that sphere, the habit of newspaper reading never developed among women there.

The differential development of mass circulation newspapers is naturally accompanied by differences in the relative roles of print and electronic media. “In countries

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 where mass circulation newspapers are absent, the mass public relies heavily on electronic media for information about political affairs” (p.24). The presence or absence of a mass circulation press has deep implications for the development of the media as political institutions.

3.1.2. Political Parallelism

“One of the most obvious differences among media systems lies in the fact that media in some countries have distinct political orientations, while media in other countries do not” (p.27). Even though the true party press has almost disappeared, and even if the political tendencies of European newspapers are fuzzier today than they were a generation ago, distinct political tendencies persist, more in some countries than in others. This distinction is expressed by the concept of party-press parallelism , proposed by Seymour-Ure (1974). By party-press parallelism, Seymour-Ure meant the degree to which the structure of the media system paralleled that of the party system. “It exists in its strongest form when each news organization is aligned with a particular party, whose views it represents in the public sphere” (p.27). This kind of one-to-one connection between media and political parties is less and less common nowadays; they more often are associated not with particular parties, but with general political tendencies. It is why, Hallin and Mancini prefer to use the more general term of political parallelism , while recognizing that party-press parallelism in the stricter sense does in some cases persist.

Political parallelism has a number of different components, and there are a number of indicators that can be used to assess how strongly it is present in a media system (Patterson and Donsbach, 1993). Perhaps most basically, it refers to media content – “the extent to which the different media reflect distinct political orientations in their news and current affairs reporting, and sometimes also their entertainment content” (Hallin and Mancini, 2004b: 28). Another of the most important components of political parallelism is organizational connections between media and political parties or other kinds of organizations. A closely related component of political parallelism is the tendency for media personnel to be active in political life , often serving in party or public offices. This is also much less common today. Somewhat more common is a tendency in some systems for the career paths of journalists and other media personnel to be shaped by their political

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 affiliations , in the sense that they work for media organizations whose politics coincide with their own. Political parallelism is also often manifested in the partisanship of media audiences , with supporters of different parties or tendencies buying different newspapers or watching different TV channels. Finally, it is manifested in journalistic role orientations and practices , in some historical periods journalists retain more of the “publicist” role that is, an orientation toward influencing public opinion, while in some other periods journalists are more likely to see themselves as providers of neutral information or entertainment.

In short, in systems where political parallelism is strong, the culture and discursive style of journalism is closely related to that of politics.

3.1.3. Professionalization

The ideal type of professionalization is based on the history of the classic “liberal” professions, above all medicine and law. Journalism departs substantially from that ideal type. One of the central criteria of this model is that the practice of a profession is “based on systematic knowledge or doctrine acquired only through long prescribed training” (Wilensky, 1964: 138). Journalism has no such systematic body of knowledge or doctrine. As much as it departs from the ideal type of the liberal professions, journalism has come to share important characteristics with them, in order to compare media systems in terms of the degree and form of professionalization of journalism, Hallin and Mancini (2004b: 34-37) focus primarily on three related dimensions of professionalization:

(1) Autonomy has always been a central part of the definition of professionalism. This is one of the key reasons why many occupations try to “professionalize” themselves, to justify greater control over their work process. Control of the work process in journalism is to a significant extent collegial, in the sense that authority over journalists is exercised primarily by fellow journalists. The authors stress the fact that the autonomy here mentioned is not necessarily the autonomy of individual journalists, but of the corps of journalists taken as a whole.

(2) Distinct professional norms , a set of shared norms is common too many professions. In the case of journalism, these norms can include ethical

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

principles such as the obligation to protect confidential sources or to maintain a separation between advertising and editorial content, as well as practical routines and criteria for judging excellence in professional practice and allocating professional prestige. Naturally, the existence of distinct professional norms is related to autonomy, in the sense that such norms could not govern the practice of journalism if that practice were controlled by outside actors.

(3) Public service orientation . Another important element of the concept of “professionalism” is the notion that professions are oriented toward an ethic of public service. It is a historically specific conception of the journalist’s role in society with important consequences for the practice of journalism and the relation of the media to other social institutions. Since journalism lacks esoteric knowledge, journalists’ claims to autonomy and authority are dependent to a particularly great extent on their claim to serve the public interest.

Professionalization can be threatened either by political instrumentalization or by commercialization, and in many cases by both.

3.1.4. The Role of the State

The state plays a significant role in shaping the media system in any society. But there are considerable differences in the extent of state intervention as well as in the forms it takes. The most important form of state intervention is surely public service broadcasting, which has been present in every country in Western Europe and North America except in the smallest country, i.e. Luxembourg, and in most countries has until recently been the only or the primary form of broadcasting. There has been a strong shift toward commercial broadcasting in recent years, but public service broadcasting remains quite significant in most of the countries that Hallin and Mancini cover in their study.

Public broadcasting has been the most important form of state ownership of media. However, in many countries the state has also owned news agencies, newspapers, or other media-related enterprises, either directly or through state-owned enterprises. Press subsidies

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 have also been present in most of the countries covered by the authors. The state, and in many cases state-owned enterprises, are also advertisers. Finally, subsidies for the film industry are also very common.

In the broadest terms, “a distinction can be made between relatively liberal media systems, in which state intervention is limited and the media are left primarily to market forces, and systems in which social democratic or dirigiste traditions are manifested in a larger state role in the ownership, funding, and regulation of media” (p.44). Systems also vary in the effectiveness of media regulation: a weaker state role can result either from a deliberate policy favoring market forces or from failure of the political system to establish and enforce media policy. Apart from issues of media ownership, funding, and regulation, the state always plays an important role as a source of information and “primary definer” of news (Hall et al., 1978).

According to Hallin and Mancini (2004b), the four dimensions outlined so far cover most of the major variables relevant to comparing the media systems of Western Europe and North America, at least from the point of view of media and politics. The following section presents more in details the three ideal types of media systems that the authors identify out of the four dimensions of analysis.

3.2. Three Models of Media Systems

3.2.1. The Mediterranean or Polarized Pluralist Model

What distinguishes Southern Europe (Greece, Italy, Portugal, and Spain) – and to a lesser degree France – from the rest of Western Europe and from North America is most basically the fact that liberal institutions, including both capitalist industrialism and political democracy, developed later. One important legacy of the long-lasting forces of the ancien régime – the landholding aristocracy, the absolutist state, and the Catholic or Orthodox Church – is the fact that the political spectrum remained wider and political differences sharper in Southern Europe than in Northern Europe or North America. Political scientists

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 refer to it as polarized pluralism (Sartori, 1976).

This heritage and the long and conflicted transition to capitalism and bourgeois democracy in Southern Europe produced a media system closely tied to the world of politics. The weakness of liberal social and economic institutions limited the development of the mass circulation press, and limited development of the market economy restricted both the resources available to commercial newspapers and the need for the kind of information- oriented content that was crucial to their social function in a market economy (Hallin and Mancini, 2004b: 128-129). Therefore the media developed in Southern Europe as an institution of the political and literary worlds more than of the market, and which main purpose was the expression of ideas. Another characteristic worth noting of the press market in the Mediterranean countries is that gender differences in newspaper readership are quite large, reflecting the closeness of the press to the world of politics and the traditional exclusion of women from the latter, as well as historically high rates of female illiteracy (Vincent, 2000).

Whether under dictatorships in Southern Europe the tradition of a pluralistic and politically engaged press was cut off, once democracy was consolidated, a high degree of political parallelism re-emerged, with the media serving to represent the wide range of political forces that contended for influence. The commercial press did not develop as strongly as in the Liberal or Democratic Corporatist systems. Newspaper circulation remained relatively low and electronic media correspondingly central. Broadcasting too has tended to be party-politicized, with France moving away from that pattern in the 1980s. One of the most characteristic patterns of the Mediterranean region is the use of the media by various actors as tools to intervene in the political world, leading to high level of media instrumentalization. A correlate to intrumentalization is, compared to Liberal or Democratic Corporatist systems, the relatively low level of journalistic autonomy and professionalism, with political loyalties often superseding commitments to common professional norms and institutions. The state has also tended to play an interventionist role in many ways, though clientelism and political polarization have often undercut its effectiveness as a regulator, except in France. (Hallin and Mancini, 2004b: 113-124 and 138-142)

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

3.2.2. The Democratic Corporatist Model

What the authors call the Democratic Corporatist Model developed in Northern and Central Europe – in Scandinavia, the Low Countries, Germany, Austria and Switzerland. A strongly intertwined historical evolution has affected the communication system across the all region. The interaction of these countries has often been conflictual in character. But conflict too is a social relationship, and has often meant exportation or mutual influence of cultural models. Luther’s challenge to the Church played a particularly important role in creating a common culture and a common public sphere in Northern and Central Europe.

The common history of the countries of this region has meant that, despite many differences among them, their media systems share important common characteristics. Hallin and Mancini (2004b: 143-145) summarize those characteristics in terms of three “coexistences” that they identify as distinctive to the Democratic Corporatist countries. In the first place, a high degree of political parallelism has coexisted with a strongly developed mass-circulation press. Political partisanship has weakened substantially over the last generation. Nevertheless, the experience of a strong advocacy press not only characterizes the history of the media in Northern and Central Europe, but in important ways still affects journalism, media structures, and the way these interact with other social forces. At the same time strong commercial media markets developed in all these countries and the Democratic Corporatist countries remain distinctive for their high levels of newspaper circulation.

The second “coexistence” is closely related to the first: a high level of political parallelism in the media has coexisted with a high level of journalistic professionalization, including a high degree of consensus on professional standards of conduct, a notion of commitment to a common public interest, and a high level of autonomy from other social powers. Again the former characteristic has weakened in relation to the latter. The media in Democratic Corporatist countries have traditionally reflected the divisions and diversity of society, yet have functioned as members of a profession with strong institutional coherence, consensus on its own rules of conduct, and substantial autonomy from other social institutions. The experience of those countries suggests that other forms of journalistic professionalism can exist, apart from the Liberal Model of neutral professionalism.

The third “coexistence” has to do with the role of the state. In the Democratic

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Corporatist countries – except in Austria and Germany – liberal institutions were consolidated early. In this sense, there is a strong tradition of limits on state power. On the other hand, strong welfare state policies and other forms of active state intervention developed in those countries in the twentieth century, and these tendencies are manifested in important forms of public-sector involvement in the media sphere that distinguish the Democratic Corporatist from the Liberal countries. This is reflected “in the media field by a strong consensus that the state must play a positive role as the guarantor of equal opportunities of communication for all the organized social voices in pursuit of the “common good”” (p.197).

3.2.3. The Liberal Model

The Liberal or Anglo-American model of the mass media is the only model that has really been analyzed in media studies as a coherent model. In fact, the early consolidation of liberal institutions in Britain and its former colonies – here Canada, Ireland, and the United States – together with a cluster of social and political characteristics related to this history – early industrialization, limited government, strong rational-legal authority, moderate and individualized pluralism and majoritarianism – are connected with a distinctive pattern of media-system characteristics (pp.198-248). These include the strong development of a commercial press which expanded with relatively little state involvement, and became overwhelmingly dominant, marginalizing party, trade union, religious, and other kinds of noncommercial media. Ireland being the exception, commercial broadcasting played a larger role than in most of continental Europe. In each of those countries, an informational style of journalism has become dominant emphasizing a relatively strong professionalization of journalism, the development of a strong tradition of “fact-centered” reporting, and the strength of the objectivity norm, and traditions of political neutrality – though with a very important exception in the British press. In general, “media have been institutionally separate from political parties and other organized social groups, for the most part, since the late nineteenth century. Finally, state intervention in the media sector has been limited by comparison with the Democratic Corporatist or Polarized Pluralist systems” (p.246).

Nevertheless, there are significant differences among the four countries. The British and to a lesser extent the Irish and Canadian systems share important characteristics in common with continental European systems both in their political institutions and cultures

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 and in their media systems. The strength of public broadcasting and the persistence of party- press parallelism in the British press is one of those shared characteristics. It also suggests that the common assumption that commercialization automatically leads to the development of politically neutral media is incorrect. Public service broadcasting also occupies a different role within the Liberal Model countries. In the United States it has always been marginal, while in the three other countries it has played a central role in media history.

The Liberal Model of media system is commonly taken around the world as the normative ideal of the neutral independent watchdogs. The authors underline that if “the Liberal countries have long and strong traditions of press freedom, very successful cultural industries, and the BBC can be said to deserve its reputation as a model public broadcasting system, with both relatively strong political independence and a good balance of responsiveness to public taste and a public service orientation, other characteristics of the Liberal systems are less attractive, however” (p.247). The British press is characterized by partisan imbalance and a fairly high degree of instrumentalization and the U.S. press by a lack of diversity. Both the British press and American television are characterized by high degrees of commercialization that strain journalistic ethics and raise questions about how well the public interest is served.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Figure 2 Relations of Individual Cases to the Three Models Sources : Hallin and Mancini, 2004b: 70

In their book, Hallin and Mancini (2004b) have argued that there are important connections between the patterns of development of media systems, based on the four dimensions, underlined earlier in this chapter, and certain keys characteristics of the political system. Usually, however, the connections between media system and political system variables cannot be interpreted as a mechanistic, one-to-one correspondence. Elements of political structure interact, for one thing, with other kinds of factors, including technological and economic factors. Nor do these connections arise from one-way causal relationships. Media systems have their own effects on the political system in many cases; and the process Hallin and Mancini are describing is really one of co-evolution of media and political institutions within particular historical contexts. The patterns of variation of the characteristics of the three models can be summarized by the Table 2.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Polarized Democratic Liberal Pluralist Corporatist Development of Mass Press Low High High

Political Parallelism High High Low

Professionalization Low High High

State Intervention High High Low

Table 2 Pattern of Variation in Four Media System Dimensions Source : Hallin and Mancini, 2004b: 299

“The Liberal Model has dominated media studies and has served as the principal normative model against which other media systems have traditionally been measured” (p.306). However, the authors suggest that it is probably the Polarized Pluralist Model, more than the other two outlined in their book, that is most widely applicable to other systems as an empirical model of the relation between media and political systems. They suspect that scholars in many parts of the world will find the different aspects of analysis of Southern Europe relevant to their own cases, such as the role of clientelism, the strong role of the state, the role of media as an instrument of political struggle, the limited development the mass circulation press, and the relative weakness of common professional norms. As we will see in Chapter 7, the case of Turkey encompasses several of those characteristics.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

CHAPTER 4. THE STUDY OF ELECTORAL CAMPAINGS

A common subject of comparative analysis is communication in election campaigns. It seems sensible to study comparable objects in the countries with advanced media system, for example television advertisement spots by political parties, which appear sufficiently similar for their verbal and nonverbal messages to be evaluated and their contents analyzed. Although the political systems of Western countries appear to be relatively different (party system, election rights), a sizeable similarity is recognizable in the political party broadcasts says Kleinsteuber (2004) in a comparative study carry out in France, Germany, Italy and the United States. The comparison “of the content, style, and effects of exposure to televised political advertising show some striking similarities across cultures.” The results show, among others, that the spots concentrated on issues, that messages are usually positively formulated and emotional arguments are put before logical ones (Kaid and Holtz-Bacha, 1995: 221-222).

Another comparative analysis concentrates on the formulation of campaign agendas. It compares the roles of parties and the media. In the struggle to control the mass media agenda differences between countries should observed, some of which are the strength of party systems, public service versus commercial in the television networks, the method of courting media consumers, varying degrees of election campaign professionalization and cultural differences, as put forward in Chapter 2. Another approach is used in cross-country study that compare the influence of interpersonal and mass communication on voting decisions (Holtz-Bacha, 2004). Some studies also address the question of whether a kind of Americanization might have taken place in the procedure of election campaign in Western countries (Swanson and Mancini, 1996).

Whether the interest in comparing political campaigns and their effect is relatively recent, the study of national political campaign goes back to the early work in the United States of the Columbia School in the 1940s and 1950s. Therefore, keeping as connecting threat the question if political campaigns matter, we will begin this chapter by introducing the main studies of the field, first in the United States and then in Europe, and by briefly presenting the literature on campaign effects and on the methods traditionally used to

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 measure campaign effects.

4.1 Do political campaigns matter? This question, like so many in political sciences, seems natural, important and straightforward. Yet the study of electoral campaigns has not always received the focus it may deserve. To present an initial framework on campaign effects, we begin by reviewing some of the important book-length treatments of the subject.

The prevailing scholarly consensus on campaigns is that they have minimal effects. Minimal effects mean in essence minimal persuasion. Because of the existing information and prejudices that voters possess, campaigns rarely change their minds. This view is said to originate with the early studies of the Columbia School (Berelson, Lazarsfeld, and McPhee, 1954; Lazarsfeld, Berelson, and Gaudet, 1948). For example, Berelson et al. (1954) found that during the 1940 presidential campaign over half of their sample had decided on a candidate three months before Election Day and only 8 percent actually switched their intention from one candidate to the other during the campaign. However, those studies also argue that even early deciders benefited from the reinforcing effect of campaign discourse: “political communication served the important purposes of preserving prior decisions instead of initiating new decisions. It kept the partisans ‘in line’ by reassuring them in their vote decision; it reduced defection from the ranks” (Lazarsfeld et al., 1948, p.87). Both works also emphasize how campaigns can “activate” preferences. In this perspective, the net result of the campaign stimulus is the reconstitution of a pre-existing party coalition.

Following the “minimal findings” of Voting (Berelson et al., 1954) and The People’s Choice (Lazarsfeld et al., 1948), American political scientists’ analyses of National Election Studies (NES) data tended to emphasize long-standing and fairly immutable factors such as party identification (Campbell et al., 1960). However, scholars who looked beyond presidential elections began to conclude differently. Jacobson’s (1983) work on congressional elections demonstrated the powerful effects of candidate spending on outcomes. Moreover, Bartels’s (1987, 1988) work on presidential primaries demonstrated that, in an arena where predispositions like party identification are useless, campaigns have substantial effects.

As suggested above, the activation hypothesis usually refers to the activation of party

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 identification. Party identification may serve or appear as a basis for mobilization in some cases, but it cannot explain significant changes in party support. Jenkins (2006) argues that there are a numbers of problems with the notion of activating predispositions. The first is that predispositions in and of themselves are not linked to parties, so voters must also know or learn where the parties stand on the issues. And the second problem is that people have more than one predisposition, so either some or all of a person’s attitudes will be activated. In contrast to the ‘minimal effects’ model, Johnston et al. (1992) have championed the idea that the campaign can change people’s minds or the dimension of choice. Studying the 1988 Canadian campaign with a rolling cross section design, they could track the selection of issues by candidates had direct consequences on voters. They also track the impact of a seminal debate about midway through the campaign. According to those arguments, campaign events and issues offer strategic, dynamic opportunities for a new coalition or the reconstitution of long-term forces.

Campaign events play an even greater role in Thomas Holbrook’s (1996) book on American presidential campaigns. By and large, Holbrook finds that events do shift opinion. These shifts are manifestations of persuasion – or at least of voters’ shifting from indecision to preference – and constitute on their face a refutation of the minimal effects model. However, these effects are dwarfed by the overall impact of national and economic conditions, and thus campaigns can only be influential when these conditions do not overwhelmingly favor one side. Thus for Holbrook, we might say that campaigns matter but in limited ways. Campbell (2000) arrives at much the same conclusion in his study of American presidential campaigns. Whether presidential campaigns by and large have predictable effects because of several systematic conditions, he emphasizes how campaigns bring about these effects through unpredictable and unsystematic campaign effects. His conclusion echoes Holbrook’s: “Perhaps the best characterization of campaign effects is that they are neither large nor minimal in an absolute sense, but sometimes large enough to be politically important” (1996: 188).

With the growing importance of television during the 1960s the media also became the subject of research in Europe. In 1961, Joseph Trenaman and Denis McQuail published Television and the Political Image , which presented the findings of their study about the British parliamentary election in 1959. While television did not have much effect on images,

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 it proved its influential role by improving voters’ campaign knowledge. By asking how people used the media during the campaign, the follow-up study on the occasion of the British election in 1964, Television in Politics , by Jay Blumler and McQuail (1968) can be regarded as an early example of the uses and gratifications perspective. Some time later Elisabeth Noelle-Neumann published the first articles on her concept of the Spiral of Silence, which propagated the “return to the concept of powerful media” (1973) and also prompted German communication research to turn into elections. The international perspective first arose with the publication of La télévision fait-elle l’élection? by Jay Blumler, Roland Cayrol, and Gabriel Thoveron (1978), which studied the role of television for the electorate and the factors influencing the interest in the election by comparing the 1974 electoral campaigns in Belgium, France, and Great Britain.

One of the earliest European studies that compared various countries, was a study initiated by Jay Blumler on the occasion of the first direct election of the European Parliament in 1979 (Blumler, 1983). Although high symbolic relevance for the integration process was attributed to the direct election of the European Parliament, the campaigns in the individual countries proved to be rather different and concentrated more on national than European aspects. It was not so much the common event that determined how campaigns were led. Instead, the new situation was dealt with according to the traditional patterns of existing national campaign models, in particular on the part of the broadcasting stations. These findings have confirmed the powerful influence of the systemic variables. However, besides the national differences and particularities the study revealed some similarities between the nine countries in which the first election of the European Parliament took place in 1979. Holtz-Bacha (2004: 218) summarizes the main results of the first European Election study as follows:

“[…] these common features lie in a similar journalistic approach to political events. Regardless of what the media were offered by the political actors, their reporting concentrated more on aspects of the campaign itself than on political issues. […] Moreover, the European Election study assessed the victorious advance of television as a campaign channel – in all countries and in almost all socio-demographic groups of the electorate. Finally, cross-national

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

findings supported the passive learning model of mass media effect with television playing a central role”

It is worth emphasis that European elections provide a unique chance for internationally comparative research in political campaigning. The event takes place simultaneously in all European Union member states. The campaigns unfold at more or less the same time, thus keeping certain external conditions constant. But the fact that national members of the European Parliament are elected and thus the campaign remains a national campaign instead of being transnational, the election allows for comparing how the individual countries deal with the election (Holtz-Bacha, 2004).

In addition to this latter difficulty, the event itself poses problems. The European Elections is usually regarded as a “second order” election (Reif and Schmitt, 1980). The electorate is barely interested and their knowledge about issues and candidates is very limited. The parties prefer to spend their money on elections where power is at stake – national elections being regarded as the “primary order” election – and thus conduct the European campaign with less intensity. The media also treat the European Election as a matter of lesser importance. Those respective perceptions and behaviors can be explained by the perception of the European Parliament as a symbolic body, which does not elect a government or a prime minister. Comparative research on European elections therefore cannot lead to general conclusions about modern campaign in general. Transnational comparisons, however, can deliver findings about similarities and differences at the macro- level and about the influence of systemic variables such as, for instance, political culture or the media system, on the political communication process. (Holtz-Bacha, 2004)

4.2. A Brief Summary of Campaign Effects Since the pioneer works of Lazarsfeld et al. (1948), the effect of media on citizens’ political attitudes and vote choice has been illustrated with mixed empirical evidence (Zaller, 2002). When considering political campaigns, De Vreese and Semetko (2004) distinguish two types of effects: effects of direct and personal campaign experiences, i.e. the extent to which voters engage in discussions about politics; and effects of indirect and mediated campaign experiences, i.e. the exposure to news coverage. In general, citizens experience the campaign

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 in a mediated way through news reports of campaign events, policy statements, and the horserace. Brady, Johnson and Sides (2006) create a straightforward typology of the mediated effects of political campaigns, these are: persuading, priming, informing and mobilizing, and altering strategic considerations.

The elites’ effort to persuade their audience of the trustfulness of their claims stands at the core of political campaigns. In political science, the mobilization model of electoral participation claims that people vote because they are mobilized to do so by the elites (Caprara, 2008). As such a campaign or a specific campaign event may change a voter’s mind, leading them to vote for a party that they would not otherwise have voted for. Zaller (1996, p.36) finds that “very large campaign effects – effects of mass communication – do occur” in his study of congressional races. However, strong campaign effects are rare in the literature. Bartels (1993) argues that the media exposure only occasionally produces strong and unidirectional opinion changes. Accordingly, “the apparent effects of the media exposure will often be modest in magnitude, not because the media cannot be persuasive but because opinions at the beginning of a typical campaign are already strongly held and because media messages during the course of a campaign are, in any case, only occasionally inconsistent with those preexisting opinions” (p.275).

Nevertheless, even if ultimately vote choice does not change, campaigns can certainly shift other sorts of attitudes, such as voters’ issue positions, where they locate parties and candidates on these issues, and how they evaluate candidates’ characteristics and traits – all of which will still be related to the probability that a voter will support a given candidate. As a matter of fact, a growing list of studies demonstrates that there are media effects on turnout, party support, candidate choice, and political attitudes toward an issue during a campaign (Norris, 2000; De Vreese and Semetko, 2002, 2004; Peter, 2004; De Vreese and Boomgarden, 2006a, 2006b; Maier and Rittberger, 2008; Gerber, Karlan and Bergan, 2009). In some cases campaigns can shape public opinion by making certain issues or considerations salient to voters, this action of priming can occur with or without a related change in vote choice. Ansolabehere (2006) describes how economic voting might be won over by public information provided by the campaign, and Bartels (1988) provides evidence that campaigns in fact prime economic considerations.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Particularly, the literature suggests that news media play an important role in shaping the attitudes of an electorate toward an ongoing campaign. How do the mass media affect political attitudes? Three major processes in relation to attitude formation and change are employed in communication research: agenda-setting, priming, and framing (see Scheufele and Tewksbury, 2007, for overviews). Whereas agenda-setting increases the accessibility of an issue and therefore shapes the importance we assign to an issue, framing influences the audiences’ understanding of a subject, and priming influences the standards by which individuals evaluate the campaign. Framing and priming function by (re-)organizing and connecting beliefs, which render these beliefs more likely to be integrated into subsequent judgments (Entnam, 1993). The overall idea is that the way in which news about politics is framed or primed shapes voter’s evaluation of issues and of the parties’ stance on these.

Since it is mostly during elections that citizens and political elites interact over matters of public policy (Manin, 1997; Kriesi, 2008), campaigns provide voters with the opportunity to gather information about parties and candidates. Voters begin the campaign in a state of comparative ignorance. Throughout the course of the campaign, individuals become more informed about parties’ positions on various issues. From a normative perspective, it is hoped that the campaign will capture their attention and inspire them to learn about and cogitate on the choice before them. Informing is related to other effects: growing interest, attention to campaign news, and perhaps most importantly propensity to vote, that is electoral mobilization.

Information that can emerge in political campaigns are, indeed, factual knowledge (Popkin, 1991) but also rather strategic information such as electoral viability (i.e. how much vote a given party is likely to get, this is particularly important in majoritarian political systems, or in case of a threshold requirement under proportional representation) and coalitional signals, which is most pertinent under proportional representation (i.e. which parties will a given party be most/least likely to form a government coalition with in parliament).

Overall, Brady et al. (2006) conclude that campaign studies should move from the dichotomous persuasion/no persuasion operationalisation that characterizes the minimal effects account of campaign effects. Rather, campaign effects are better understood as a set of

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 continuous variables – including priming, informing, and mobilizing, which depend on an array of historical and circumstantial factors, as well as the predispositions of the candidates and voters involved in a given election. The three components that characterize the exposure to and the acquisition of political information are ability, motivation and opportunity. In short, people need sufficient skill, drive, and means to obtain and retain political information (Delli Carpini & Keeter 1996; Luskin 1990).

4.3 How to Capture Campaign Effects? In order to assess such effects, studies of media and political behavior have usually relied on two methodological tools: experimentation and observational studies through survey of the correlation between news consumption and political attitudes and behavior. Each of these approaches can be subdivided into two different sorts of research design: laboratory experiments and field experiments for experimentation, and panel design and rolling cross- section design for survey research. Each of these methods has its own advantages and shortcomings.

Experiments are particularly useful in establishing causality as long as the design possess internal validity; experiments eliminate competing causal explanations via random assignment and thereby isolate the causal impact of some treatment. However, it may be hard to establish a laboratory context that is credible for subjects. Even credible setups, there remains the problem of external validity. Does “looking real” mean “being real”? At most the laboratory can establish potential, rather than actual, impact. Several studies have proposed to identify the effects of media through field and natural experiments, such as measuring the impact of exposure to bias news coverage, the arbitrary boundaries of media coverage, or agenda-setting strategies of particular news sources.

Gerber, Kaplan and Bergan (2009) provided subjects with free newspaper subscriptions during the 2005 Virginia gubernatorial election, the newspapers being the Washington Post and the Washington Times; the first is generally viewed as a liberal newspaper and the second is widely viewed as a conservative paper. Furthermore, the Post had endorsed the Democratic candidate for the Virginia gubernatorial election and the Times had endorsed the Republican. Thus, they have the opportunity to compare the effects of

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 exposure on voting behavior in the Virginia gubernatorial election of two apparently very different news sources where the framing of stories, priming on issues, and persuasive efforts are arguably biased toward different candidates. They found that those assigned to the Post group were eight percentage points more likely to vote for the Democratic candidate for governor than those not assigned a free newspaper.

Della Vigna and Kaplan (2007) use the introduction of Fox News Channel in some cable systems but not others to measure the effects of the conservative channel’s news coverage on percentage of the Republican Party vote share in Presidential elections. They estimate that half a percentage point shift in favor of Bush in the 2000 Presidential election was due to exposure to Fox News. Similarly, Fox News convinced three to eight percent of their viewers to vote Republican.

However, both those researches had little control over what is reported in the media outlets about the election, whether subjects are exposed to content about the election, and to some extent whether their access to the media source is manipulated. A number of studies have measured the effect of media exposure using laboratory experiments. Since the media exposure is randomly assigned, these experimental studies are not vulnerable to the problem of unmeasured differences across the exposed and unexposed group.

Maier and Rittberger (2008) analyzed how media exposure affects public attitudes to EU enlargement through a three-factor experimental pre-test/post-test design to make within- and between-condition comparisons. They exposed respondents to fictitious positive or negative articles over the accession of Macedonia to the EU. They found that the experimental groups differed significantly in their views on whether or not to admit Macedonia to the EU between the pre-test and the post-test. The laboratory method avoids the selection bias of observational studies by randomizing the groups’ exposure to media sources, but the generality of the results is unclear as the laboratory artificially compels individuals to consume certain types of media.

There is extensive observational research linking attitudes and behavior to media exposure. The most common approach is to ask survey respondents about their media exposure and their political views and behaviors. Specifically, to capture campaign effects, panel survey are used. Panel design seems at first sight ideal for analyzing campaign-induced

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 change in that it generates observations that permits to characterized individual trajectories, like a “treatment”. They allow interviewing the same respondents at different times during the campaign and to set up control groups. Then, the content of the media is coded and the effects of exposure in the light of the content that citizens reported having been exposed to in the different news outlets can be assessed. Hence, differences in attitudes and reported behavior are ascribed to media exposure (DeVreese and Boomgarden, 2006a, 2006b).

However, several validity and reliability concerns rise out of such studies. The biggest issue with panel studies may be the gap between waves (Brady et al., 2006). The greater the gap the harder it is to identify campaign effects as competing explanations accumulate. Even when a single event could explain shifts, as with a debate, identifying its impact necessitates conditioning on cross-sectional differences, for example in exposure. In addition, such investigations are unlikely to provide an accurate measurement of a causal effect, since a person’s choice of which shows and how much to watch are correlated with the subject’s political attitudes. The tendency for individuals to seek out information that agrees with their pre-existing views has been documented (Prior, 2007). Also, the validity of self-report media exposure measures is discussed. Self-reports of regular news exposure are reliable measures of how much news people think they watch. As measures of people’s actual news, they lack validity. Apparent media effects may arise not because of differences in exposure, but because of differences in the accuracy of reporting exposure (Prior, 2009).

The rolling cross-section addresses some of these concerns. Because it typically is composed of daily samples, the design does not pre-commit researchers to timing, to events, or to a theory of dynamic form. A major disadvantage of the rolling cross-section is that the daily sample is small. Describing aggregate trends over time require smoothing to separate real change from sampling noise. And to interpret dynamic trends persuasively, contextual information about media coverage and events is necessary, which is particularly time and labor consuming.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

CHAPTER 5. THE RISE OF THE INTERNET: NEW MEDIA – NEW RESEARCH ?

The global political communication landscape and opportunities for political communication research continue to be shaped by developments in new technology. With the arrival of the Internet in the 1990s, the new opportunities for political communication and political communication research provided by the availability of new and old media sources online continue to evolve in surprising ways. Certain blogs have become the primary news venue for many of the politically sophisticated, and create opportunities for political communication in politics and the media to challenge traditional news media in a variety of previously unimagined ways (Oates, Owen and Gibson, 2006).

With the ability to transmit information instantaneously around the globe, new media technology brings governments and citizens potentially closer together than ever before while at the same time providing greater opportunity to drive them apart (Davis and Owen, 1998; Dahlgren, 2005). Despite global media abundance, citizens in advanced industrial societies have become “distrustful of politicians, sceptical about democratic institution and disillusioned about how the democratic process functions” (Dalton, 2004: 1). While citizens in some low-income “societies in transition” have found ways to use the internet to promote open society, most citizens of the world do not experience the media abundance brought by new technology. The global reality is in fact a stark digital divide, both between information rich and poor countries and among publics within those countries (Norris, 2001). Political communication and political communication research is not unaffected by this digital divide.

5.1 The Internet, Politics, and Democracy

The earlier mass media of press and broadcasting were widely seen as beneficial (even necessary) for the conduct of democratic politics. The benefit stemmed from the flow of information about public events to all citizens and the exposure of politicians and government to the public gaze and critique. However, negative effects were also perceived, because of the dominance of channels by a few voices, the predominance of a ‘vertical flow’, and the heightened commercialism of the media market, leading to neglect of democratic 49

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 communication roles. The typical organization and forms of mass communication limit access and discourage active participation and dialogue.

The new electronic media have been widely hailed as “a potential way of escape from the oppressive ‘top-down’ politics of mass democracies in which tightly organized, political parties make policy unilaterally and mobilize support behind them with minimal negotiation and grass-roots input” (McQuail, 2005: 151). They provide the means for highly differentiated provision of political information and ideas, almost unlimited access in theory for all voices, and much feedback and negotiation between leaders and followers. They promise new forums for the development of interest groups and formation of opinion. Not least important as Coleman (1999) points out, is the “role of new media in the subversive service of free expression under conditions of authoritarian control of the means of communication”. It is certainly not easy for governments to control access to and use of the Internet by dissident citizens, but also not impossible.

An important argument that accompanied the rise of the Internet is that politics might work better and more democratically with new tools of campaigning and electronic polling. The ideas concerning the public sphere and civil society have stimulated the notion that new media are ideally suited to occupy the space of civil society between the private domain and that of state activity. The ideal of a public sphere seems open to fulfilment by way of forms of communication, and in that case the Internet would allow citizens to express their views and communicate with each other and their political leaders without leaving their homes.

Several approaches have emerged toward the rise of this new media. Dahlberg (2001) describes three basic camps or models. First, there is the model of ‘cyber-libertarianism’ that wants an approach to politics based on the model of the consumer market. Secondly, there is a ‘communitarian’ view that expects the benefits to come from greater grass-roots participation and input and the strengthening of local political communities. Thirdly, there is a perceived benefit to ‘deliberative democracy’ made possible by improved technology for interaction and for exchange of ideas in the public sphere.

Bentivegna (2002) has summarized the potential benefits of the Internet to politics in terms of six main attributes: interactivity as opposed to one-way flow; co-presence of vertical and horizontal communication, promoting equality; disintermediation, meaning a reduced

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 role for journalism to mediate the relationship between citizen and politicians; low costs for senders and receivers; speed greater than with traditional media; absence of boundaries. She also describes the main limitations and the obstacles which have so far prevented any democratic transformation. In her view “the gap between the political realm and citizens has apparently not been reduced, participation in political life remained…stable” (p.56). Among other reasons to it, she cites ‘the glut of information’ that limits the effective use that can be made of it; the cacophony of voices that impedes serious discussion; the difficulties for many in using the Internet. In addition, there is the now much demonstrated fact that the new media tend to be used mainly by the small minority that is already politically interested and involved (Norris, 2000; Tewkesbury, 2003).

Following the first hopes of the Internet as a revitalizing democratic tool, there has been a growing tendency to downplay the probably benefits for the public sphere. Scheufele and Nisbet’s (2002) inquiry into the Internet and citizenship concluded that there was a “very limited role for the Internet in promoting feelings of efficacy, knowledge and participation”. The existing political organizations have generally failed to make use of the potential of the Internet, but rather turned it into another branch of the propaganda machine (Gibson et al., 2003). Stromer-Galley (2000) found, for instance, that campaign managers did not really want interaction which is risky, problematic and burdensome. Whether over time, the candidates have started to familiarize themselves with the use of blogs and campaign sites, they have learnt to use more sophisticated presentation technologies to convey their message, but less advanced interactive innovations because these latter options interfere with the candidate’s message (Druckman, Kifer, and Parkin, 2007).

In more recent years, specifically since the 2008 American Presidential Elections campaign and with a different emphasis since the use of web-based communication in social movements all around the world, the Internet has been praised again for its democratic virtues. The Internet is again emphasized as a source of political information for the citizens, of their involvement in politics, but also as an area which experienced unprecedented advances in candidate strategies and tactics, and in the news coverage of the campaign (Kaid, 2009; Kersting, 2012; Smith, 2009).

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

5.2 A Changing Campaign Environment: The increasing role of non-party actors in electoral politics

The technological innovation of the Internet has also paralleled and accelerated the involvement of non-party actors in the ‘postmodern’ campaign environment (Norris, 2000). In such new context of electoral politics, parties can no longer count on the support of affiliated newspapers to get their messages across; instead, they compete for coverage through the increasingly differentiated and intrusive broadcasting media. While such media- related challenges have been closely scrutinized by researchers, other significant changes have so far found far less attention. Prominent among these are the ever more vocal organized and individual actors who operate in the campaign environment independently of parties and candidates. Little is know about their role in seeking to influence the electoral process by launching ‘outside campaigns’ (Magleby, 2000).

Lasting coalitions between political parties and interest groups have constituted one of the key building blocks of traditional cleavage politics in established democracies. However, as a consequence of social differentiation, the landscape of organized interests has been transformed over recent decades. The number of organization seeking influence on policies has risen enormously, and the new organization represent interests that are often more specific than the traditional ones. In recent years, there has been a significant “depillarization” – an erosion of the ties between established parties and traditional organized interests. (Webb, Farrell and Holliday, 2002)

In the era of cleavage politics organized interests were an important asset on which parties could count at elections. The main function of campaigns was to mobilize parties’ core support groups (Rohrschneider, 2002), and interest organization leaders played an important role in this by reminding their clientele groups of the party-political camp to which they belonged. Nowadays, not only have individual voters ‘begun to choose’ (Rose and McAllister, 1986), so have the organized interests.

Another important development is paralleling this trend in contemporary democracies – their mutation into ‘public relations democracies’ (Davis, 2002), a type of political system in which participation in public communications is increasingly seen as a vital prerequisite of political success not only by political parties, but also by interest groups. As the public space

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 becomes more crowded, each of them must engage in ever more sophisticated and intensified strategic communications in order to make itself heard and its interests known.

To the extent that interest groups engage in such activities, they contribute yet another degree of complexity to ‘postmodern’ contemporary election campaigns. In essence, parties lose some of their control over the communication flows of campaigns, and other aspects of the electoral process. In recent years there have also been indications of a growing personal involvement of ordinary, unorganized citizens in the electoral process, autonomously seeking to voice their demands and take part in campaign discourse (Farell and Schmitt-Beck, 2008a).

The exponential spread of weblogs and other forms of Internet-based political communications may signal a tendency for unorganized voters to engage more strongly than before, and above all, outside the routes reserved for them in the parties’ own ‘ground war’ where they try to stimulate and channel the activities of grassroots volunteers towards their overall campaign goals, launcing their own ‘mini-campaigns’ (Abold, 2008).

Some non-party actors may also interfere for other motives than gaining the best for themselves from future policy-makers. During campaign periods, associations make it their mission to increase the quality of the electoral process itself, for instance by scrutinizing the candidates of all parties according to certain general normative criteria, or by seeking to enhance the information base on which voters generally draw when casting their votes. The Internet provides not only individual citizens with unprecedent opportunities for entering public political debate and making their voices heard. Its technical potential allows also for new ways of feeding impartial information into the campaign process, both in quantitative and qualitative terms.

5.3 The World of Voter Engagement Applications: Campaign Actor or Research Tool?

5.3.1 The 21 st campaign environment: a fertile ground for VEAs Most mass media have entered the Web where they sustain extensive reservoirs of election-related information. Yet, against this background of information abundance sorting, selecting, and processing information has turned into an even more pressing problem for

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 voters than in the past. Democratic theorists underline the need for citizens to be informed over matters of public debate in order to fulfil the democratic ideal. And the more citizens participate in it, the better (Habermas, 1962). But for that purpose, citizens – or at least most of them - have to be familiar with the current affairs of their society, and when the time of election comes citizens need to be aware of the positions of the ones running for office on those issues. In the past, television, radio, newspapers and magazines have taken care of current-affairs information (Gans, 2003; Luhmann, 1971; Schudson, 1995). They have narrowed down the infinite number of the issues a society could worry about to the most urgent ones and spread them to a majority of citizens. Nowadays, the Internet proposes an abundant information environment which on one side allows citizens to access primary sources of information without anyone stepping in (Negroponte, 2005; Gilmor, 2004), but on the other side it is feared that this is true only for people who actively seek that type of knowledge (Prior, 2007).

Here a new terrain for non-party actors has opened up. In many European democracies agencies of different kinds (Universities, gov agencies, media organizations) have over the course of recent elections developed ‘vote helpers’ – Web-based software tools that may assist citizens in taming the flood of political information at elections (Farrell and Schmitt-Beck, 2008b). In an age of voter dealignment, these VEAs are seen as offering a useful resource for undecided voters seeking authoritative information on the issue positions of parties, at least in multiparty systems where such applications mostly emerged. In the Western world, VEA sites have become popular with voters. VEAs have reached millions of voters during election campaigns in most Western countries (Walgrave, van Aelst and Nuytemans, 2008). For instance the Dutch StemWijzer and Kieskompas sites generated 4.7 and 3.4 million advices respectively in the 2006 campaign, the 2009 German Wahl-O-Mat site reached over 6 million users, the 2010 Canadian Vote Compass site was visited by over 4 million users, and the 2012 French Boussole Présidentielle reached 1,5 million users. In 2009, the EU Profiler launched the first trans-national VEA covering 33 different countries for the European Parliamentary elections. Recently, VEAs appeared also in Turkey and in some of the Arab Spring countries of northern Africa. Hence, VEA sites are assuming important roles at the interconnection of the relationship between citizens and politics.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

5.3.2 The DNA of VEAs: Issue Voting

Within the context of the spread of Internet-based political communication, online Voting Engagement Applications (VEA) have proliferated and their impact is rapidly increasing (Cedroni and Garzia, 2010; Walgrave et al., 2008). Those ‘vote helpers’ might bear “the potential for a substantive turn in campaign communications, as they provide information on parties stance on a set of issues, that is the substance of politics” (Schmitt- Beck and Farell, 2008b: 17). In fact, VEAs help users to compare the stances of parties and candidates on major political issues with their own political preferences.

Particularly, VEAs sites are providing useful information to voters on parties’ stances on salient issues, usually around 30 issues, by comparing the users’ positions with the ones of the parties. This process considerably decreases the cost of information seeking for voters and may also stimulate the users to reach for more information since extract and link toward party platforms are often provided on VEAs sites. The methodology behind VEAs clearly emphasizes issue voting as a bulk of modern politics.

Issue voting has its intellectual roots in the Downsian tradition (Downs, 1957). It presumes that issue voting is the final result of a sophisticated decision calculus; that it represents a reasoned and thoughtful attempt by voters to use policy preferences to guide their electoral decision. Candidates and parties announce positions on issues in order to win votes, and voters choose the alternatives that best represent their interests on those issues. Key (1966: 7) provided supporting evidence for issue voting underlining “the parallelism of policy preferences and the direction of the vote”, and concluded that “the electorate behaves about as rationally and responsibly as we should expect, given the clarity of the alternatives presented to it and the character of the information available to it”.

However, the vast majority of research on the topic finds weak or no evidence that issue preferences explain the vote (Ansolabehere, Rodden and Snider, 2008). Survey data over the past several decades echoes the findings of The American Voter (Campbell et al., 1960): voters rely on their party identification and impressions about candidate image when deciding how to vote, and ideology or opinions on specific policies play a modest role at best. Converse (1964) famously concluded that the mass of the electorate holds nonattitudes on

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 important policy questions.

But, Ansolabehere et al. (2008) emphasized that those studies rely almost entirely on single items to measure voters’ policy preferences. A small minority construct scales, and that branch of the literature finds substantively important effects of issues (Carmines and Stimson, 1980; Erikson and Tedin, 2007). Issue scales constructed from questions in the World Values Survey have also been used to understand vote choice and other forms of political behavior in cross-national studies such as Inglehart (1990). Contrary to the bulk of the literature, Ansolabehere et al. (2008) find that issue voting may rival party in explaining the vote. The combined effects of issue preferences are about as large as party identification in the multivariate analyses predicting the vote.

5.3.3. A Theoretical Framework of VEA Proliferation and its Effects

Studies of the dynamics of political communication emphasize the triadic relationship between the political, media and public domains. Political parties and the mass media play a crucial role as they set the terms of the political debate. Especially for issues that are characterized by high levels of complexity and uncertainty, political parties and the media play an essential role in informing the public (Popkin, 1991; Page and Shapiro, 1992). Researchers have focused both on how the attempt of political parties to set the agenda and to frame an issue, is filtered through media coverage, and on how the relationship could be the inverse (Asp, 1983; Brandenburg, 2003; Hopmann et al., 2012; Walgrave and van Aelst, 2006). Other studies have focused on the fact that the media play a crucial role in increasing public awareness of issues and/or contributing to public issue perceptions in terms of salience and content. Lastly, the interaction between public opinion and party behaviour is a major theme within political science (Hellström, 2008). Empirical evidence demonstrates that two types of linkages exist: a top-down linkage in which public opinion is receptive to political parties and a bottom-up linkage whereby parties adjust their policies to public opinion (Wlezien, 1995; Stimson, 2004). What is clear from all this, interactions between media, politicians and voters are not unidirectional.

The proliferation of the Internet has transformed channels of political communication. As already underlined, the party/candidate profiling websites - Voter Engagement

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Applications - that help voters to determine which party or candidate is closest to their own political preferences, are relatively new online phenomena. In making such VEA’s, developers make crucial choices on the number and type of issues included, the selection of parties and candidates, statements formulation, the method to position the parties/candidates on those statements, the output that provides the result to the voter and the manner in which developers report on the usage of the tool (Krouwel, Wall and Vitiello, 2012). All these aspects locate the tool at the centre of the relationship between parties, citizens, and media. As some VEA developers closely collaborate with media-partners, the effects of the tool may become amplified.

Hence, VEAs may have direct effects on parties, citizens and media, potentially locating such a Web-based tool at the core of the dynamics of political communication in the “post-modern campaign” environment (Norris, 2000). The VEAs have a direct effect on parties because during the interactive process with parties on positioning them on the statements of the tool, the content analysis of the platforms may reveal ambiguous or vague propositions in parties’ policy stance; therefore ‘forcing’ the political parties to clarify their positions. The VEAs affect voters in different ways. They provide voters with additional information on the party’s stances, they may increase electoral turnout, and they even can have an impact on vote choice (Marschall and Schmidt, 2010; Ruusuvirta and Rosema, 2009; Wall, Krouwel and Vitiello, 2012). Finally, the VEAs produce a media effect by the fact that VEA builders need media partnership in order to promote the tool to the public. This cooperation between media and VEA makers may result in increased media coverage of substantive campaign issues through the content that the VEAs provide to the media partners.

Out of those preliminary observations, we see that VEAs bear the potential for being a linkage arena between parties, citizens and media. Parties, and particularly small parties, may use VEAs to clarify their stances on issues, voters may found in VEAs a useful time-saver for comparing parties’ positions to their own, and the media could use VEAs as a tool to produce original and substantive content for their printed, visual or radio outlets and as a marketing device to attract web-users on their websites.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

5.3.4 Effects of VEAs on Voter Behaviour

While early studies in voting behaviour, as presented in Chapter 4, emphasized social groups and partisan identification as main determinants of electoral behaviour and as strong providers of political cues (Campbell et al., 1960; Lazarsfeld, Berelson, and Gaudet, 1948), by the 1970s dramatic changes began to affect Western societies and the stable social cleavages on which they were based. The major consequence on voting behaviour has been a sharply reduced ability of social cleavages to structure individual voting choice (Franklin, Mackie and Valen, 1992). Social modernization, rising levels of education and the spread of television as a source of political information lead to a process of partisan de-alignment (Inglehart, 1977; Dalton and Wattenberg, 2000). Voters became increasingly able to orient themselves in the complexities of politics thanks to stronger cognitive skills as well as less costly information.

In this context of changes in the sources of political and electoral cues, there has been an increasing focus on short-term determinants of voting behaviour, such as the role of candidates and issues (Franklin et al., 1992; McAllister, 2007). Walgrave et al. (2008) argue that the potential for VEAs to influence their users’ political behaviour lies in their informative effect. VEAs intent to significantly reduce the cognitive cost needed for a voter to engage in informed issue voting, which can be prohibitively costly due to the time and effort required in a fragmented and saturated media environment.

Despite their increasing popularity among voters, VEA sites have only recently started to receive attention from political scientists. While providing transparent information on where political parties stand on a variety of issues relevant to the ongoing campaigns, such an extensive usage of VEAs necessarily raises a few fundamental questions: How do they function? Who visits them? And, what sort of advice do they produce? (Fivaz and Schwarz 2007; Hooghe and Teepe, 2007; Walgrave, Nuytemans and Peepermans, 2009; Wall et al., 2009). More recently, the question of their informative versus persuasive effect on voting behavior has also been discussed (Ladner, Fivaz and Pianzola, 2010; Ramonaite, 2009; Walgrave et al., 2009; Wall et. al., 2012).

Analyses that focus on user demographics have consistently shown that VEA-users are more male, more urban, and more educated than national populations as a whole

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

(Çarko ğlu, Vitiello and Moral, 2012; Hooghe and Teepe, 2007; Ruusuvirta and Rosema, 2009; Wall et al., 2009). However, there are indications that the gap between VEA users and the rest of the population is narrowing over time (Fivaz and Nadig, 2010).

In their research on the mobilization effect of the German VEA Walh-O-Mat , Marschall and Schmidt (2010) found through a Web survey that nearly 70 percent of the sample users contended that they were going to talk about the tool with others; almost 60 percent of users said that they had been stimulated by the Wahl-O-Mat to seek additional political information, and about 11 percent were motivated to go to the ballot boxes by their VEA experience. In a post-electoral survey of Smartvote users, a Swiss VEA, Ladner et al. (2010) found that 86.5 percent of users declared that Smartvote was important or rather important for getting information about parties and candidates during the election campaign. In the same survey, online media, national TV channel, national newspapers, and advertisement of parties and candidates in the letterbox were important or rather important for respectively 68.4, 61.4, 61.2, and 19.5 percent of Smartvote ’s users.

The use of VEAs is already per se revealing a certain political interest in the ongoing political campaign. Then the information provided by VEAs can be used by viewers in different ways. First of all, it can be ignored or acknowledged as being valid information. Once acknowledged, the received information can be discarded as considered not useful, accepted as confirming already existing preferences, accepted as helping in the decision- making process, and accepted as leading to belief change. The range of the effect can be about the personal importance of a particular political issue, a realization that a political party is the closest to one’s preferences or even a conclusion that the VEA is positively evaluated and worth revisiting (Lupia & Philpot, 2005).

Estimates of percentages of users who feel that their eventual decision was influenced by their visit to a VEA vary from a low of 6 percent (Marschall, 2005) to a high of 74 percent (Fivaz and Schwarz, 2007). In a post-electoral survey of the Dutch VEA Kieskompas , Wall et al. (2012) found that when asked to describe the influence that their visit on kieskompas.nl had on their vote choice 9.2 percent of users said that their VEA experience helped them to choose among several parties, and 1.1 percent said that it led them to vote for a party they had not previously considered; while 30.2 percent said that it confirmed their preference and 29

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 percent that it had no effect on their vote choice, suggesting that the effect of VEAs on vote choice relies mostly on their confirmation dimension.

In short, VEAs effect on users can be of three kind (Garzia, 2010): (1) by changing individuals’ information-seeking behaviour, that is motivating users to gather further information about politics and political parties; (2) by motivating people to turnout, even if they had not intended to do so before using the website; (3) and lastly, by affecting individuals’ vote intentions, that is helping voters to choose between two or more parties that they were contemplating or, less frequently, to convince already decided voters to change their political preference.

Last but not least, VEAs are a relatively cheap way of collecting data on users’ political positions, but also on their evaluation of candidates’ traits and on their propensity to vote for each parties or candidates. Moreover, when visiting VEA sites, on an optional base users are also asked to answers typical questions that we found on elections studies such as items on political interest, voting behaviour in past elections, socio-demographics characteristics, exposure to media and campaign events, and so on. Some users also accept to leave their email address in order to be contacted later on for further research. This particular feature allows pushing the analysis one step further by making the research able to trace individual trajectories, like in a panel study, of some VEAs’ users.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

CHAPTER 6. LOCATING PARTIES AND CANDIDATES IN A COMMON POLITICAL SPACE, A GUIDE TO VEA’S METHODOLOGY

If any reference is made to this chapter, please quote as follows: Krouwel, A., Vitiello, T., and Wall, M., 2012. ‘The Practicalities of Issuing Vote Advice: A new methodology for profiling and matching.’ International Journal of Electronic Governance , forthcoming.

The popularity and impact of the online VEA sites necessitate that VEA-developers and other scholars investigate the quality, reliability and the validity of the ‘advice’ they give to visitors. In Belgium, Walgrave et al. (2009) have shown that the specific selection of statements and framing propositions has a considerable impact on the ‘voting advice’ that is produced: specific configurations favour certain parties. In the Netherlands, Kleinnijenhuis and Krouwel (2007) have shown that the manner in which the sites compare voters and party attitudes differs substantially. By aggregating the same answer pattern in both ‘agreement’- based and ‘proximity’-based configurations, it was shown that matching user’s responses with parties’ positions by means of a ranked list of parties, based on an additive scale of voter-party similarity, results in different outcomes compared to mapping parties and voters in a multidimensional political landscape. Therefore, it is important to investigate the process of political mapping that is undergone in order to compare users’ and parties’ issue positions and their aggregation into deeper lying dimensions of party competition.

This chapter extensively describes a method that was developed at the Dutch VEA Kieskompas in 2006 to calibrate parties on issues and position them in a multidimensional space (Kleinnijenhuis et al., 2007; Krouwel and Fiers, 2008); the EU Profiler VEA has adopted some of those characteristics (Treschel and Mair, 2011). This approach combines existing methodologies (such as the systematic coding of party policy materials) with innovative features, including the use of automated text-analysis to extract salient issues; the practices of specifying a ‘hierarchy’ of party policy sources; furnishing the specific text (or texts) that are used to justify each party’s placement on each issue by the expert; and a process of ‘authorisation’ whereby the parties and the site coding team interact (often over several ‘rounds’ of contact) to produce a final coding decision. Since its first application in the 2006 Dutch legislative election, this method has subsequently been applied in many other

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 party profiling projects (Trechsel and Mair, 2011; Arian et al., 2010; Çarko ğlu, Vitiello and Moral, 2012; Costa Lobo et al., 2010; Jadot et al., 2012).

Kieskompas.nl is a VEA that was first developed for the 2006 Dutch parliamentary election using a novel approach to calibrate political parties on salient issues and issue- dimensions by analysing their official policy documentation. The purpose of this chapter is to describe the Kieskompas methodology, which involves eight key steps: 1. Selecting the most relevant political parties or candidates to be included. 2. Selecting authoritative sources – party platforms, policy documents, websites, debate transcripts and campaign material - from which party positions can be extracted, as well as creating a hierarchical order among different types of sources (should any two sources provide alternative positioning) for party placement. 3. Identifying the salient issues in a specific election year, i.e., selecting the issues on which political parties will be positioned, and on which voters will be compared to parties. 4. Framing propositions about these salient issues that give ‘ideological’ direction to the proposition and determine it’s ‘loading’ on one of the deeper cleavage dimensions that demarcate the political landscape. 5. Searching the selected sources (which were identified in step 2) and extracting specific text snippets that justify the calibrations of parties or candidates on an issue. Having selected these snippets, a party coding on each issue is decided by the coding team. Each specific coding decision at this stage is thus accompanied by the text snippets upon which that decision is based. 6. A process of party authorisation of coding decisions. This process begins with sending out the blank questionnaire to parties to collect their self-placements. The VEA team then compares their hand-coded placements on issues with the self-placements of the political parties. The team then undertakes a process of interaction with the parties until the maximum possible amount of agreement is reached over the coded position of the party (though the coding team retain discretion over the final coding decisions). 7. Comparing voters’ opinions to issue positions of parties, i.e. , summarising the voter- party comparison rules that drive the site’s output.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

8. Developing a multidimensional ‘map’ of party competition (determining the salient cleavages underpinning party competition, and how the individual issue items load on these cleavages).

In the subsequent paragraphs, each of these steps are discussed, and some of the advantages and trade-offs inherent in this methodology are outlined.

6.1. Selection of relevant parties or candidates

VEAs can include all parties or candidates that participate in an election, however more typically they include only a selection of the most relevant parties. In multi-party systems with large numbers of parties registering for an election, filtering out micro parties that do not stand a chance of gaining parliamentary representation reduces the complexity of output for users. To avoid exclusion of electorally relevant parties, the Kieskompas method uses two key criteria for inclusion: (1) all parties with one or more seats in parliament that enter the elections are automatically included; (2) parties that do not have any seats in the outgoing parliament but which consistently poll at least one seat in a number of opinion polls are included in the analyses. In candidate-based elections such as the US presidential elections (see www.electoralcompass.com) candidates were included that consistently ended up in opinion polls as serious candidates with a genuine chance of winning the party’s nomination or – in adapting Sartori’s (1976) notion of ‘blackmail potential’ - could take away the nomination from another candidate. In the French presidential elections, for instance, candidates included polled at least one percent in opinion polls, while in Mexico all four candidates could be included. In countries with no electoral history or reliable opinion polling, such as in the founding elections of Egypt and Tunisia, where a wide variety of new parties exist, criteria for party selection must be adjusted. Common criteria in such contexts include: having formal recognition and legal status, fielding candidates in a sufficiently large number of constituencies to be nationally relevant and being considered a viable contender by experts. In general terms, the trade-off in party selection is between the risk of over-inclusion (leading to an artificially crowded political landscape) and over-exclusion (leading to potentially relevant parties being excluded). As the latter is a more troubling outcome, the

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 exclusion rules employed by Kieskompas only seek to exclude parties who are highly unlikely to receive meaningful parliamentary representation.

6.2. Selecting and analysing relevant party documents

Once the relevant parties or candidates are selected, VEA developers need to decide which sources provide authoritative information on issue positions and reveal the most salient issues on which parties compete. The most stable cross-time source for this is official documentation that parties or candidates themselves produce in order to communicate their policy positions (Budge, 2000; Laver and Garry, 2000; Marks et al., 2007; Mikhaylov, Laver and Benoit, 2008). In most advanced democracies, such positions are formally documented in a manifesto that is especially written for the upcoming election. However, manifestos also have several drawbacks for VEA designers, and political scientists in general: they are often strategic (i.e., designed to please large groups of voters, rather than explain a party’s true policy stances); they differ in terms of length and topics addressed across parties (important issues are not always addressed in all manifestos as parties tend to de-emphasise issues that can hurt them electorally); and the manifestos of government parties systematically differ from those of non-government parties, the latter being less constrained by recent ‘track records’ and possible future (coalition) government participation (Marks et al., 2007, pp.26- 27). In addition, new parties may not even have a formal manifesto and even if they have a document, this may not reveal a policy position on all relevant issues. Indeed, some parties focus heavily on a single or a very limited number of issues. Even well-established parties which have agreed positions on a range of issues may only release their formal manifestos late in the electoral campaign, often too late to be useful for VEA designers – as VEAs are typically launched at the start of the formal campaign. Thus, party manifestos are important, but they are not the sole source of party stances on issues. Party websites usually provide more extensive information on policies and additional information, including statements by leaders, press releases, motions officially accepted at party conferences and even full policy documents on specific topics, all of which helps VEA developers to pin down a party when an issue position is either ambiguous or not

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 mentioned in the party manifesto. Additionally, the reactions of party leaders and representatives in the media can be a source for party positioning on issues. Party leaders’ policy statements can be carefully planned speeches that are timed and targeted at specific voter groups or to get media attention. In addition, debates on radio and TV often serve to introduce new positions, to clarify stances taken or to sharpen the distinction between competitors. If VEA developers use multiple sources, they need to determine which sources are relatively most authoritative and establish a hierarchy among them. This hierarchy may differ from country to country and even from party to party in the same country. Within traditional mass parties, the manifesto will outweigh any policy statement made by an individual party representative, whereas a declaration by the leader of a populist party may override the official party programme. A general rule for the relative weight of sources is that more recent sources should be given precedence; so current manifestos outweigh previous manifestos, recent speeches by the leader outweigh those made in the past, and so on. This hierarchy of documents should be formalised ( i.e. , a rank ordering of sources should be written up for the coding team) and integrated into the coding template. With certain sources, some prudence is required. In parliamentary debates, for example, representatives of government parties may not express the official position of the party, but the government’s position on the issues, or may formulate their contribution to the debate in a manner that is acceptable to coalition partners. Even more caution is required when statements of party leaders are used when such leaders are also government ministers. VEA developers should be very careful in distinguishing in what capacity and setting political statements are made. Even during debates, parties that are often in government will make declarations with current and future coalition bargaining in the back of their heads. As a consequence of their coalition potential, such ‘governmental’ parties are much more difficult to position on issues than minor, more extreme and permanent opposition parties. This is why it is advisable to enter into an authorisation process with the parties themselves, which will be discussed below.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

6.3. Issue identification and selection: Determining what’s at stake in the election

The selection of the issues to be included in the VEA is the most crucial choice to be made in the construction of the application. Walgrave et al. (2009) ran a large number of simulations in which a subset of issue statements was taken from a larger pool of items on which the positions of parties and users were known. They found that the statistical dispersion of the matches observed between party and voter is relatively narrow, but they also underline that in real political terms these differences are really substantial, concluding that issue-selection dramatically affects site output. At the moment, a major problem for VEA designers is that there is no current consensus among political scientists as to how salient issues can be identified. In order to limit political mismatches, issue identification should be the result of a process involving several methods and actors. Although extant methods for statement development differ substantially, most party profilers start by drawing up a list of relevant issues, and subsequently narrow down this list to a final selection (Krouwel and Fiers, 2008, pp.7-10; Walgrave et al., 2009). Statements can be extracted from the party programmes or websites, using automated text analysis or careful reading, but can also be drawn up by country experts and/or journalists. The Kieskompas method combines a computer assisted text analysis with expert opinion. Kieskompas collects the relevant party platforms and website content and then uses large corpus text analysis methods to extract salient issues in terms of frequency. This computer-generated issue-lists are complemented with two further issue lists, one produced independently by the country experts and one by journalists. Kieskompas asks academic country experts to carefully read the party programmes and other relevant text sources to identify salient issues, not simply by their frequency, but on a scale of ‘urgency’ or ‘strength’ of the issue. This approach prevents the exclusion of issues that are mentioned less frequently but are phrased in very strong terms, and can therefore be considered important for a party. Journalists compile a list of the most relevant issues based on their day-to-day observation of national politics. Both experts and journalists are asked to indicate the core of the political dispute at stake, information that is used in framing the proposition. All issue lists are then compared and the most frequently mentioned and the most strongly emphasized issues are included in a list of roughly 35 to 55 issues, drawn up in

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 multiple sessions with the country experts. Although Kieskompas normally uses 30 issues, parties and candidates are positioned on a wider range of by the coding team in order to allow experts exclude issues on which party positions cluster too much to usefully discriminate positional differences. The selected 30 issues need to be salient for political competition, address a wide range of relevant policy fields in order to prevent a bias towards specific parties that ‘own’ certain issues and, to make the final cut, they must be able to differentiate between parties. Issues on which parties broadly have shared goals should be avoided. These ‘valence issues’ or 'emtpy signifiers' touch on such broad and primary issues where parties only differ on the ways to achieve the same goal (Walgrave et al., 2008). The involvement of numerous experts for selecting salient issues combined with computer-assisted content analysis is designed to minimise bias in the issue selection and is a substantive empirical improvement over a priori selection of issues and bipolar policy dimensions. However, the absence of a universally agreed process for statement selection remains troubling for VEA practitioners, particularly in the light of Walgrave et al.’s (2008) research findings.

6.4. Framing and ‘scaling’ the issue propositions

One of the most contested issues in VEA design is how to frame the issue propositions in such a manner that they can be used to elicit voter and party positions without introducing bias. VEAs differ considerably in their approaches to the development of specific propositions. Some VEAs take the statements directly from the party platforms or invite party representatives to co-develop the statements during workshops, while others only involve experts in the formulation process. The framing of these statements also depends on the answer categories and whether their propositions need to be scaled to deeper-lying issue dimensions. While many VEAs use binary answer categories (yes/no, agree/disagree), Kieskompas uses Likert items with five response levels, which measures attitudes in terms of level of agreement or disagreement on propositions. The answer categories can vary, so that also propositions can be framed towards a five-point scale ranging from strong agree to strongly disagree; from much more to much less, from much lower to much higher, etc . The Likert item format was adopted after studies found that Dutch VEA Stemwijzer ’s binary answer

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 categories (agree versus disagree, as well as a ‘don’t know’) denied parties and voters the opportunity to adopt nuanced positions and resulted in incorrect (self-) positioning of parties on multiple propositions, which distorted the voting advices given (Groot, 2003, pp.23-24). Dichotomous answer categories also reduce the discriminatory power between parties belonging to a similar political orientation as well as of centre parties (Groot, 2003, pp.24-27; Arendsen, 2003; Kleinnijenhuis and Krouwel, 2007, 2008). With Likert items, parties and users can indicate the direction of their attitude, as well as the intensity to which they (dis)agree with the proposition. This allows for a more nuanced differentiation between parties, even when they belong to the same political ‘camp’. The labelled five point scales also offer users and parties a centrist position. For political parties this is important, as the centre is a real existing political position. When VEAs like Stemwijzer and Smartvote do not use the middle category, they force parties and voters to take sides, even when they are actually centrist or neutral on the issue. Nonetheless, the decision to use Likert response scales, and especially the provision of a ‘neutral’ midpoint, implies some tradeoffs, especially for particularly controversial political issues where a ‘neutral’ response may be a safe haven for satisfying voters, or for parties who do not wish to engage with an issue (Gemenis, 2011). How the middle category is interpreted by users, which statements drive respondents to the middle category, whether the midpoint indicates an non-attitude or a centrist attitude and what the best label is for this mid-point is still fiercely debated (Triga, Baka and Figgou, 2012). One obvious advantage of using Likert items over binary answer categories is that issue propositions can be considered items of more profound political dimensions, such as the left-right or conservative-progressive divide. Locating parties or candidates in a common political space brings us to the problem that there is no consensus over the number and type of salient cleavage dimensions, let alone on which issues belong to which deeper-lying ideological dimension. When issues are scaled towards cleavage dimensions, two problems need to be addressed. First, determining to which dimension an issue needs to be scaled and secondly the weight of each issue on this dimension. To avoid bias towards one ideological orientation, party or candidate, propositions in a VEA need to be framed in different ‘political directions’. The entire set of propositional statements must be ‘balanced’ in terms of subjective framing. Some VEAs balance the propositions by taking an equal number from all

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 party platforms. Within the Kieskompas method, roughly a quarter of the questions will have ‘rightist’ framings, a quarter ‘leftist’, a quarter ‘progressive’ and a quarter are formulated in the ‘conservative’ direction. As such, each proposition is scaled towards one of the four poles of the political landscape. The total set of propositions is also grouped into broader issue clusters or topics, such as healthcare, environmental policy, welfare state, foreign policy, etc. Within such broader topics, several propositions are thus ‘framed’ in a variety of ‘ideological directions’ in order to reduce the danger of acquiescence bias, ‘pushing’ the voter towards a certain response (Evans, Heath and Lalljee, 1996). Also, clustering multiple propositions reduces the effect of the selection of specific issues and allows for more subtle levels of differentiation between parties, particularly when they are in the same political block or in the crowded political centre. Both the selection and directional framing of issue propositions are vital to the functionality and fairness of the tool – however, we must acknowledge that issue selection and proposition framing are human work. Even if it is done in the most rigorous manner possible, it remains a qualitative and interpretive process. However, bias can be minimised by using a range of diverse sources and multiple experts who balance the issue selection and framing to prevent design pitfalls of proposition formulation. A further difficulty of dimensional framing is that occasionally questions are framed in a double-barrelled manner, as observed by Gemenis (2011). However, in many instances, such phrasing is necessary to convey the interdependency of dimensional issues – for instance, the necessity to pay for promised increases in public spending with increases in government tax take. For example, while the statement ‘Social programmes should be maintained even at the cost of higher taxes’ from the 2009 EU Profiler site may be double barrelled, this type of formulation is necessary to convey the trade-off between taxation and social spending.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Figure 3 The 'Recalculte your position' button allows users to select which issues areas are considered in their output. Example from the 2012 French Presidential Vote Compass: www.laboussolepresidentielle.fr

With regard to weighting, VEA-developers could opt to simply weigh all issues as 1, or assign a relative weight to each issue. For this relative weighting, as well as determining if the issue is scaled to the correct dimension, data can be used from previous iterations or from elections studies and surveys that included questions on the concerned issues. While VEA designers have typically used apriori dimensional assignment (according to the judgement of country experts), both the dimension to which an issue scales ad the weight of its loading to the deeper-lying dimension can be determined through factor analysis (Kleinnijenhuis and Krouwel, 2008). VEA developers could also opt to scale an issue to two dimensions. Experiments are currently being carried out (van der Linden et al., 2012) employing inductive dimensional analysis of public opinion data to construct weightings that are based on the extent to which individual items scale on a given factor. This represents a promising approach as Gemenis (2011) provides evidence that inductive dimensional analysis often points out that often items that are scaled to a dimension apriori by coding teams do not scale well empirically. An alternative approach allows voters to exclude issue categories from their output calculation (i.e., giving items that fall into such issue areas a zero weighting) an example of this approach can be seen in Figure 3 above. 70

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Figure 4 Examples of 'recalculate position' for the user and the candidates according to the issues selected Those two examples keep the issues of the socio-economic dimension constant (all of them are selected as important issues) and differ on the progressive-conservative dimension. On the top output ‘immigration and integration’ ( immigration et integration ) is selected as important issue, while on the bottom output ‘values’ ( questions de société ) is selected. (Example from the 2012 French Presidential Vote Compass: www.laboussolepresidentielle.fr )

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

6.5. Placing the parties: Step 1, first round of placements and generation of text snippet database

Political scientists devote considerable time, effort and expense to the study of parties’ policy stances and they have developed several methods to determine the ideological or policy positions adopted by parties. Examples of such methods include: expert surveys, surveys of placement of parties by voters, surveys of the political parties themselves, roll-call behaviour of politicians and content analysis of party manifestos (for a detailed discussion, see Benoit and Laver, 2006, pp.123-153). VEA developers use various methods to calibrate parties on the issues (Gemenis, 2011). One method (used for example by Dutch VEA Stemwijzer ) is simply to allow parties to position themselves on the issues. A second method is to let experts position parties or candidates on the issues, either with or without analysing party documents. A third method is automated coding of the political texts by computer programs to determine party positions (Benoit and Laver, 2003, 2007). There are, however, major methodological problems with automated text-analysis. When ‘word-based’ techniques are used and the distribution and patterns of words in text are analysed, this does not tap into the connotation of words in their context – crucial in political communication – and they chunk texts into words, negating semantic, grammatical and discursive structures and frames that give meaning. Moreover, automatic text analyses are incapable distinguishing positive or negative direction in a text, which is crucial for identifying relative party positions on issues. Considering the importance of VEAs we have to accept that hand coding is an interpretive approach, which is very labour-intensive, expensive to replicate or change and sensitive to individual errors of interpretation and low inter-coder reliability (Budge, Robertson and Hearl, 1987; Klingemann, Hofferbert and Budge, 1994).

Letting parties position themselves is a method that is susceptible to manipulation by political parties, while experts can also be biased if no control mechanism is included. In 2006, Kieskompas developed a method that combines expert judgements based on text- analysis with self-placement of parties, including an interactive process with the parties. An important improvement on previous experts’ surveys and safeguard against bias is that experts base their judgements on thorough text analysis (instead of general interpretations) and fully document and justify each positioning on an issue with text snippets from official party documentation. Thus, unlike most expert surveys and VEAs – Vote Compass is 72

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 completely transparent with regard to source texts and coding. In order to maximise transparency, all text extracts underpinning the party calibrations are made available to each user with a simple mouse-click, allowing voters to also enter into a debate on the position of the parties on these issues. Improving further on expert surveys, the Kieskompas method fully documents and justifies each positioning of parties or candidates with text snippets from official party documentation. Thus, unlike traditional expert-surveys, the Kieskompas method leads to a database of party positions on salient issues in which each position attributed to the party is justified by a reference and hyperlink to a text snippet from an authoritative party document. All text snippets that are used as evidence for the calibration of parties’ positions are added to the Vote Compass website and made accessible with hyperlinks, so that the positioning of parties and candidates is completely transparent, also to the user. In addition, Vote Compasses provide hyperlinks directly to the original document that was used to position the party. The manner in which this information is displayed in Figure 5, and it is important to note that each text snippet is accompanied by a hyperlink that allows the user to examine the original source document (and to check whether the snippets may have been taken out of context).

Figure 5 Output displaying text justifying a single issue coding (note hyperlink to source document at the bottom of this text). Example taken from the French 2012 Vote Compass: http://www.laboussolepresidentielle.fr

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Notwithstanding elaborate safeguards in the various methods applied, pure objective party calibration will remain impossible as language production and interpretation is subjective. The calibration of parties and candidates is also complicated by other problems. Sometimes no position can be found in official texts for some parties. If then the party does not want to identify its stance on the issues by self-placement (see next paragraph), the party is coded as ‘no opinion’, for want of a better solution. Another problem encountered by VEA developers is that parties are not always unitary actors. When parties are internally divided into factions, it may happen that individual members or different documents of the same party state dissimilar policy positions. In such cases, Vote Compass usually opts to show all the different positions, including the text-snippets, in order of the ‘official’ or ‘dominant’ position first, followed by the deviant positions. Thus, there may be controversy over which and how many text sections should be included, as well as where to cut off the text snippet. In short, there are many reasons why positioning parties or candidates may be less than straightforward. In the next section we describe the process of authorisation, which involves contacting the parties or candidates and locking them into a clarification of their stances.

6.6. Contacting the parties for self-placement: Reaching agreement

Kieskompas brings an important innovation to party profiling methods in that it adds a self-placement by each party or candidate to the text-based coding by experts. Country experts contact actors within each party who are authorised to assess that party’s stance on a given issue. For some parties, the party leader will fill out the self-placement questionnaire, while in others party communications authorised party spokesmen or campaign leaders will fill out the form. In some countries, parties now even assign a team with the task of reacting to the various VEA-requests. Self-assessment of their issue position gives parties the opportunity to challenge expert-coding decisions. Each party is provided with the blank propositions at an early stage and are also asked to position themselves and provide appropriate text snippets from their manifestos or other official documents to substantiate and justify their self-placement. Self-placements are then compared with the hand-coded positioning data and well-considered adjustments can be made. Discrepancies between the self-placement and coder positioning are communicated to the parties, as well as the text-

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 snippets that the coders used as a justification. Parties are subsequently asked – in extensive direct communication - to indicate what they consider to be incorrect about the disputed expert judgement and are allowed to provide alternative text-snippets on order to substantiate their self-placement. This interactive process with parties is overall (with a few important exceptions) constructive and leads to clarification where issue positions were unclear. For example in the Dutch elections of 2006 the hand-coded calibrations of Kieskompas corresponded with 82 per cent of the party self placements, while after two rounds of contact this level of agreement increased to 92 and 99 per cent respectively (Krouwel and Fiers, 2008, p.22). Indeed, most of the disagreement is related to the intensity of the calibration, not of the direction. Usually, after a first round of discussion the agreement-level increases to over 90 per cent, while multiple rounds usually result in only very few propositions remaining disputed between the expert and the party itself. Willingness to participate in the self-placement procedure varies across countries (Wagner and Ruusuvirta, 2011, pp.5-6), and depends on the authority of the expert(s) contacting the party and the number of iterations and success of previous VEAs. Kieskompas usually achieves compliance from the large majority of parties, while a cross-national project during the EU elections in 2009 showed lower levels of participation (Krouwel and Fiers, 2008; Trechsel and Mair, 2011). A major difficulty with asking parties to self-place is that parties face strategic incentives to misrepresent their issue positions in order to ‘game’ popular VEA sites (Wagner and Ruusvirta, 2011). However, the methodology described above seeks to control against much manipulation firstly by requiring parties to provide formal texts justifying their self- placement, and secondly by leaving the final coding decision in the hands of the VEA team for unresolved coding disputes with parties. Gemenis (2011) points out that expert coding is often low in terms of inter-coder reliability. His proposed approach of checking inter-coder reliability to avoid inter-coder dynamics leading to distorted codings is a valid one, and should be considered for future VEA iteractions (resources permitting, of course). However, we feel that the process of engagement with parties is highly beneficial as they are the actors who find their political positions being assessed by VEA designers. The process is designed to deal with arbitrary expert judgments and allows parties to respond to potential miscoding without handing them complete discretion over their placement.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

6.7. Comparing Voter and Party Positions: Agreement or proximity approach?

While there are many ways to compare voters and party attitudes (Wall, Krouwel and Kleinnijenhuis, 2011), most VEA models have typically fallen into two categories: ‘agreement’-based comparisons and ‘proximity’-based comparisons. Most party profilers use the ‘agreement’ approach, in which the user’s responses to the propositions generate a ranked list of parties, based on an additive scale of voter-party similarity. On each issue, the user is compared to each party’s position, leading to an overall ‘agreement score’ and agreement level per party. The Dutch ‘ Stemwijzer ’ and the Irish ‘Votomatic’ VEA sites employed this type of approach. The rational assumption is that citizens will vote for the party with which they have the highest level of agreement. This approach is appealing by its straightforwardness of advice given to the users, but its major shortcoming lies in the lack of information provided on where parties and users agree. Users are often not given information whether they ‘match’ with the party in the realm of economics, welfare, justice, or another aspect of the political competition. In his work on spatial voting theory, Downs assumes that ‘the political parties in any society can be ordered from left to right in a manner agreed upon by all voters’ (1957, p.142). Therefore, it becomes essential to take into account in the advice given to the users the n-dimension existing in a political space. Hence, ‘proximity based’ profiling sites show voter positions relative to parties or candidates in a multi-dimensional policy space, whereby all the issues included are aggregated over two dimensions. In the debate concerning which method has to be used to compare voters and parties’ positions, Louwerse and Rosema (2011) indicate that the advice given depends strongly on the spatial model adopted (Wall, Krouwel and Kleinnijenhuis, 2012). By using the data produced by Stemwijzer , a VEA proposing 3 answer items to its users and using a one- dimensional space (left-right) to position voters and parties, they test the data with 7 other spatial models to demonstrate that the choice for a particular spatial framework affects the outcome at both the individual and the party levels. In order to deal with those discrepancies, they underline the need to focus on two aspects of VEAs: the way how the result is presented to the users, and with which spatial model in mind the statements are selected and framed. On the former, they suggest that using a two-dimensional spatial model in which user and parties are represented is a more appropriate way to present the advice than merely providing the 76

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 name of a single party. A clear advice for a single-party “may suggest more precision and objectivity than what can be substantiated” (Louwerse and Rosema, 2011, p.23). And on the latter, they argue that statements should be selected and framed according to which spatial model VEA makers are going to use for representing the positions of users and parties. In case the VEA team opts for a spatial model, such as in ‘proximity’-based VEAs, “it may preferable to use a more fine-grained mechanism for capturing answers to statements, such as the four- or five-point scales” (p.24). The Kieskompas method addresses both issues raised by Louwerse and Rosema (2011). In determining a common political space the most important decision is to extract the number and content of relevant issue-dimensions (Benoit and Laver, forthcoming ). Journalists and commentators often describe party politics as a one-dimensional game, yet research shows that parties and candidates in most advanced democracies operate in a multidimensional political space. There are several methods to estimate the latent multidimensional ideological space that represents how people think about politics and how this abstract space is linked to party positions on salient issues (Çarko ğlu and Hinich, 2006; Evans, 2004). Salient issue-dimensions differ per country and can be adapted to the national political cleavage structure. When constructing a multi-dimensional space, Vote Compass applications use academic studies on salient issue-dimensions. In most countries, Vote Compass utilises a ‘material’ socio-economic Left-Right dimension and a moral-cultural, non-material dimension. The first dimension refers to the traditionally dominant economic aspects of party competition and voters’ concern: unemployment, wages, pensions, taxes, the welfare and public finance. The other dimension incorporates issues that conceptually do not fit in the economic left-right axis, purporting instead to matters such as values, identity, the environment and foreign policy, among others. These axes aggregate the topics, which Marks et al. (2007) have denominated as Green-Alternative-Libertarian (GAL) versus Traditional- Authority-Nationalist (TAN). The use of a two-dimensional political space according to those two axes makes it possible to represent modern polities in the face of the changes undergone since several decades (Inglehart, 1999). Note that the computation of averaged, or summated positions on the two dimensions depends on a priori considerations, both with respect to the question to which of the two dimensions an issue belongs, and with respect to which end of a dimension it leans. Since the

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Kieskompas method assigns all to only one dimension, it means that for each issue we need to decide whether it is material or moral, and whether the most positive score (totally agree) indicates the left or right stance, or the progressive or conservative stance. The averaged positions for parties and VEA users with respect to a set of specific propositions function as the coordinates of parties and voters on the political map (Kleinnijenhuis and Krouwel, 2007). Visualization of the political landscape is thus an important part of the tool, and this is made possible by the aggregation of multiple issues over two underlying dimensions. The proximity approach shows to the site users not just how close they are to each party, it also shows where they stand in terms of major structural dimensions of political competition in their country. This method assumes that, ‘other things being equal, the voter will choose the party which, in policy terms, is least distant from them’ (Evans, 2004, p.100). We discuss the process by which ‘maps’ are generated in the next subsection.

6.8. Developing a multidimensional ‘map’ of party competition

Using confirmatory factor analysis and Mokken scaling analysis, we check whether each issue can be scaled to the overarching dimensions. If an issue does not ‘scale’ on either dimension, it can be excluded from the algorithm used to generate the political ‘map’ and users can be encouraged to consider this issue in isolation. The formally and mathematical Vote Compass calibration procedure computes the summated positions Σk P ojk of each party o and averaged positions of the voter Σk I jk on the two dimensions j1 and j2, before distances along each of the axes are computed. The averaged positions of the parties and the user are presented graphically along the two dimensions, thus giving rise to Euclidean distances between a voter and a party.

2 ½ Ao = – [ Σj | Σk P ojk – Σk I jk | ]

The site also visualises the extent to which users’ spatial position is made uncertain due to ‘ideological inconsistency’ (i.e., the extent to which their answers deviate from the logical pattern anticipated by our scaling procedure) via an ellipse which is drawn around the

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 user’s issue position – based on the standard deviation of their answers on the x and y axes. This is a deliberate addition, and a critical component of the placement of VEA users in a political space. The ‘position’ of the user is not one single point, but an approximate area which will be wide for ideologically inconsistent voters, whose answers are a mixture of left and right, progressive and conservative, and narrow for voters whose answers point consistently in the same ideological direction. The visual representation of an approximate area also allows the users to observe their relative position within the overall dimensions represented. This is a crucial element of the tool because the distance between the voter and parties signifies different things according to which theory is being tested. Whether in the ‘agreement’-based comparison equal distance between two parties means an equal propensity to support a party, the ‘proximity’-based comparison implies that it does not. What matters, is the relative position of party and voter on the dimensions in question, and the saliency that each user attributes to the different issues (Evans, 2004). Such as the ‘directional’ theory integrates an affective element to the cognitive process of spatial models of voting (Rabinowitz and Macdonald, 1989), having an area of collocation in the output produced by the VEA allows the users to position themselves on one side or another of the ellipse according to the strength of the affective intensity which they feel on a dimension or issue. While both the ‘agreement’ and the ‘proximity’ approaches have their pros and cons, Vote Compass nevertheless opted for the proximity approach as this reduces the distorting effect of proposition selection. The aggregation of responses into two deep-level ideological dimensions – under the assumption that most issues can be scaled on either the material or the moral dimension – allows us to assume that deleting or adding issues would not dramatically affect the overall positioning of either users or parties. Thus, in order to counterbalance the problematic nature of issue selection, user and the party’s positions should be aggregated in a manner that is assumed to be robust to changes to the specific set of selected issues. However, any system used to generate an overall ‘result’ will inevitably be a simplification of the complex realities of political competition within a country. It is also important to point out that well-designed VEAs facilitate multiple outputs – and their central idea is to encourage voters to examine party policy positions. As well as the political ‘map’ and ‘closeness’ outputs, Kieskompas sites also allow users to compare their responses with party positions on

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 each individual issue measured by the application. Thus, while producing a 2 dimensional summary of the political space, the Kieskompas design also seeks to cater for users who are interested in a given single issue or in only a subset of issues. Figure 6 displays this issue-by- issue output.

Figure 6 An example of single issue output on the question ‘The state should provide a minimum income subsidy to all needy families’ (translation); with user ‘Tend to disagree’. Example taken from the Turkish 2011 Vote Compass: http://www.oypusulasi.org

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

CHAPTER 7. MEASURING PUBLIC OPINOIN ONLINE: AN ASSESSMENT OF THE FIRST TURKISH ONLINE VOTER ENGAGEMENT APPLICATION

If any reference is made to this chapter please quote as follows:

Çarko ğlu A., Vitiello, T., and Moral M. 2012. ‘Voter Advice Application in Practice: Answers to Some Key Questions from Turkey’, Journal of International Electronic Governance , forthcoming.

On June 12 th , 2011 Turkey had the third general election wherein the conservative Justice and Development Party [Adalet ve Kalkınma Partisi -AKP] dominated the election polls with 49.8% of the votes. The AKP’s first election victory in 2002 came in with 34.3%, to be followed by 46.6% in the 2007 general elections, and so this third election victory was a sign of continued success. As the electoral appeal of the AKP widened, the nature of the Turkish party system and behaviour of key players therein slowly changed, a process which in turn generated diverse consequences for government practices. One result of the AKP’s establishment is that The Turkish party system has become a predominant party system where uncertainty about election results, and especially about the winner, has considerably diminished, and the opposition is in considerable disarray 1. A prominent feature of predominant party systems in general is the increasing complacency of the dominant party in power 2. Election successes are typically taken for granted and real competition between parties is replaced by competition between fractions within the dominant party. An immediate implication of this development for election campaigns is a focus, not on alternative issue positions and policy differences but rather on inner party struggles and rhetorical differences.

1 See Esen and Ciddi (2011) and Çarko ğlu (2011) for the recent claims of predominant party system in Turkey. See Özbudun (2000) and Kalaycıo ğlu (2005) for concise reviews of modern Turkish politics. See Çarko ğlu (1998), Rubin and Heper (2002), Sayarı (2002, 2007) and Tezcür (2012) on the historical characteristics of the Turkish party system.

2 Ware (1996) states that a dominant party should “usually” win. See Sartori (1976, 171-172) for a more commonly adopted definition of electoral success over time.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

The establishment of such an atmosphere in the campaign, characterized by low information content, was also due to a 10% threshold of electoral support for gaining representation in the Turkish Grand National Assembly [Türkiye Büyük Millet Meclisi - TBMM]. In this system, unless a political party receives 10% of the nationwide votes, it cannot receive any seats in the TBMM. Accordingly, commentators in the June 2011 election focused on whether or not the Nationalist Action Party [Milliyetçi Hareket Partisi -MHP] was going to remain above or below the 10% nationwide electoral threshold. Thus, a great deal of attention was directed toward the horse-race, polarizing attention on the upcoming seat shares in the TBMM. In addition, since the 1990s, ethnic and ideological cleavages have proved to be particularly powerful in structuring the Turkish electorate (Çarko ğlu and Kalaycıo ğlu, 2009; Kalaycıo ğlu, 1999). This feature has also led to the priming of a nationalist and polarized frame in public discourse, pushing aside policy stances. For the June 2011 election, the widening electoral power of the AKP was clearly evident in how the election campaign was conducted by the political parties as well as how it was covered in the media. As the AKP confidently moved towards its third consecutive election victory, the media coverage appeared increasingly polarized and biased. More importantly, the information content of the election debates and discussions appeared to sink even lower. The web based voter advice application (VEA) ( www.oypusulasi.org ) that was launched during the last four weeks of the campaign preceding the June 12 th 2011 election can hence be considered as a natural response to this growing partisan alignment of the Turkish media. VEAs are web-based software tools that typically provide substantive information on issues, instead of the horse race upon which the election campaigns often tend to concentrate. VEAs are seen as a valuable resource for undecided voters seeking authoritative information on the issue positions of parties (Farrell and Schmitt-Beck, 2008). From a normative perspective, VEAs aim to deepen the knowledge of the electorate in a modern democracy by providing objective information, thus allowing voters to compare parties and leaders with regards to their stances on policy issues. However, as we will demonstrate, such applications are critically dependent on media promotion and become most effective when and if traditional media supports VEAs. Despite the very limited support the Oypusulasi application received from traditional

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 media sources, it was able to reach approximately 190,000 respondents 3. As we will illustrate below, the application proves to be a valuable additional information source for election analysts in providing crucial information to interested voters about the issue positions of the major parties in an election. We show below that the sample of respondents reached via this application exhibit intriguing patterns. We also demonstrate the effectiveness of media support in promoting such applications. In conclusion, we argue that effective partnership with media is essential in reaching a large and well-balanced sample of the electorate.

7.2. The Turkish Media System and Political Parallelism Since the AKP’s coming to power following the 2002 general election, an increasing concern has been voiced about the freedom of media, pressures upon the press, and the emergence of an advocate/partisan’s [yanda ş] media (Çarko ğlu, 2010; Kaya and Çakmur, 2010; Kaymas, 2011). According to Freedom House’s reports, Turkey was ranked in 2011 as 116 th out of 206 countries in the Freedom of the Press Index with a status of “partly free”. Similarly, the 2011 Press Freedom Index released by Reporters without Borders ranked Turkey as 138 th among 179 nations, with increasing restraints on press especially in 2003, 2004, and 2009-2011 4. Media systems are shaped by the wider context of political structure and thus, the influence of the political system on media institutions is not unexpected. As Hallin and Mancini (2004, pp.46-65) note, the role of state, interest groups, the structure of clientelism and the party system all influence the structure of a media system. Thus, in Turkey’s case, these developments are hardly surprising within the rising predominant party system in the country. The conceptual framework of the press–party parallelism (PPP) developed by Seymour-Ure (1974), which is defined as the partisanship or the alignment of a newspaper to a particular party, provides plenty of room for analysis especially when the expected role of

3 190.000 advices given represents less than 1% of the Turkish electorate, usually the number of advices given by VEAs varies from about 5 to 35% of the electorate, but those numbers are reached with the support of national newspapers and televisions promoting the web application. This relatively small percentage of the Turkish electorate that has visited www.oypusulasi.org is due to the fact there was no such partnership in Turkey.

4 For details see: http://en.rsf.org/ .

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 the press in providing a non-partisan coverage of the news agenda is weak. Organizational ties, media content, personal affiliations of journalists, and the political characteristics of a newspaper’s readership are typically used as the basis in diagnosing the alignment between a party and a media outlet. Hallin and Mancini (2004) prefer to use the term ‘political parallelism’, and note that PPP in a strict sense is in decline, but political parallelism is still common in the form of media organizations’, associations not only with particular parties but also with general political tendencies. Kaya and Çakmur (2010) and Bayram (2010) show that political parallelism in Turkish media history is nothing new. However, since the liberalization of the media markets in the late 1980s and early 1990s, a new competitive environment with increasing conglomerate control of media institutions has emerged. Kaya and Çakmur (2010) identify three characteristics of this new context. Firstly, the Turkish mainstream media heavily promotes the culture of commercialization by a showing preference to sports, scandal, and popular entertainment. The incorporation of entertainment in informative programs has led to the rise of ‘infotainment’ (Brants, 1998). Secondly, “in Turkey, unlike many European countries, deregulation and development of commercialization did not decrease the degree of political parallelism in the media institution. Instead, it increased the instrumentalization of the Turkish media by business interests” (Kaya and Çakmur, 2010, pp.528-529). Thirdly, a distinctive change in the Turkish media landscape is the rapid development of the conservative/Islamist media, which easily rivals the mainstream commercial media. Recent empirical studies reveal that the level of parallelism in the 1990s reached a level unseen since the one-party era (1925-1945) (Bayram, 2010). From a different angle, Çarko ğlu and Yavuz (2010) show that during the first term of the AKP, from 2002 to 2007, internal pluralism decreased in the major newspapers of the country. Internal pluralism refers to a situation when individual media outlets contain multiple viewpoints and provide a fair share to all political views. While on the other hand, external pluralism is system wide pluralism, in which every political current in the system finds an outlet for the expression of its views, and is able to make its voice heard (Hallin and Mancini, 2004, pp.26-30). Therefore, a decreasing internal pluralism indicates an increasing bias in individual media outlets towards particular political parties as well as an increasing polarization in Turkish media between 2002 and 2007.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Due to the combination of infotainment, political parallelism, and polarization, Turkish media appeared to be unable to focus their attention on issues relevant for the voters and to engage the political parties in an informative debate during the 2011 election campaign. The campaign appeared to focus on non-policy issues, harsh rhetoric, and the horse race. Parties gave hints of their specific policies on rare occasions. Though, such debates accomplished little for the undecided, but interested voters, since issues were framed in terms of personal conflicts between leaders instead of policy content. This, in turn, increased the perception of factionalism within parties. However, a characteristic of the ‘post-modern campaign’ environment (Norris, 2000) is that, in response to the erosion of the ties between parties and organized interests, new actors have emerged (Dalton and Wattenberg, 2000). Farrell and Schmitt-Beck (2008) describe this trend as ‘the era of non-party actors in politics’. On one side, this trend is characterized by the increasing number of organizations seeking influence on policies; on the other, the change implies a growing personal involvement of ordinary, unorganized citizens in the electoral process. The exponential growth of Internet-based political communications may indicate a tendency for unorganized voters to engage more strongly than before. Some of those non-party actors may also pursue objectives unrelated to future policies but rather to the quality of the electoral process itself, such as the accountability of the candidates or the transparency of parties’ policy propositions. By guiding voters on how to match their policy preferences to particular parties, VEAs comfortably fit into this definition of a ‘post-modern campaign’ environment. Given this political background, the project team started to work for a VEA in Turkey approximately 18 months prior to the general elections in June 2011 5. From inception, the objective was to find a media partner that could help the VEA to reach as wide a national political spectrum of voters as possible. Accordingly, large mainstream media groups were contacted. Nevertheless, eventually no institutional support could be obtained from any Turkish media group. Our analysis will show that despite reluctance to systematically disseminate the content and findings from the VEA by sponsoring it, media attention from prominent columnists and news agencies remained supportive.

5 Ali Çarko ğlu of Koç University-Istanbul headed the Turkish project team. André Krouwel of the Vrije University-Amsterdam and founder of kieskompas.nl was a partner and consultant in the project. 85

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

7.3. A Profiling of Oypusulasi users’: who was plugged in?

Despite their increasing popularity among voters, VEA sites have only recently started to receive attention from political scientists. During a campaign, the number of connections to VEAs varies from hundreds of thousands to millions across many countries (Cedroni and Garcia, 2010; Walgrave, van Aelst and Nuytemans, 2008). While providing transparent information on where political parties stand on a variety of issues relevant to the ongoing campaigns, such an extensive usage of VEAs necessarily raises a few fundamental questions: How do they function? Who visits them? And, what sort of advice do they produce? (Fivaz and Schwarz 2007; Hooghe and Teepe, 2007; Walgrave, Nuytemans and Peepermans, 2009; Wall et al., 2009). More recently, the question of its informative versus persuasive effect on voting behaviour has also been discussed (Ladner and Pianzola, 2010; Ramonaite, 2009; Walgrave et al., 2008; Wall, Krouwel and Vitiello, forthcoming). Since this VEA site was the first of its kind to emerge in Turkey, we first focused on the socio-demographic characteristics of Oypusulasi users. In addition, gathering information on the geographical location of each connection to the website allowed us to look at the geographical distribution of the connections and at diverse characteristics of users by regions. On the Oypusulasi website, users were not only asked to position themselves on thirty statements in order to receive a political recommendation but also to provide information on such demographics as gender, year of birth, and level of education; answering these questions is not required for the tool to work. The location of connection was identified using a GEO-IP database. This database checks the users’ Internet protocol (IP) address against a large database and returns the city if the users’ IP is in a known IP-city range. Since personal IP addresses cannot be stored due to privacy concerns, this procedure had to be used for compiling city-level data, which then were aggregated first into provinces and then regions. As of April 2011, 42.9% of households in Turkey had access to the Internet. Despite this relatively low level of Internet penetration, the interest in political content is high among Turkish Internet users: 72.7% of them indicate that they read or download online news and newspapers, and 50.8% declare they read and post opinions on public or political issues via social network websites. Therefore, the Internet appears to be a communication tool mostly

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 used for access to political information. In terms of Internet access, the country is also characterized by a salient digital divide between male and female and urban and rural areas. While 54.9% of males use the Internet, only 35.3% of females do; this divide is persistent among all age groups. In urban areas, 53.2% of the population uses the Internet, while in rural areas only 25.7% of the population does 6. Therefore, consistent with other VEA users, Oypusulasi users were predominantly young, male, with a university degree and urban. 75.2% of the users were male, 67.3% were below the age of 34, 87.5% attended or graduated from a university, and 76.2% connected to the VEA from the provinces of Istanbul, Ankara, and Izmir, the three largest urban centres in Turkey. The connections to the Oypusulasi web site are mostly clustered in western regions, particularly in Istanbul, Western Anatolia, and the Aegean. While 18% of the Turkish population resides in the Istanbul region, 52% of connections to Oypusulasi came from there. Figures 7 to 9 show other relevant regional characteristics of the website users. Most female users also came from the western regions, where the median income and the proportion of 15- 24 years old users is the highest. The south eastern regions have the lowest proportion of female users, the lowest median income per users, and the lowest percentage of 15-24 years old users among all the regions. Such regional divide conforms our expectations and is in line with previous research on the regional characteristics of Turkish voters (Çarko ğlu, 2000, 2003).

6 Data from the Information and Communication Technology Survey in Households 2011, accessed on January 27, 2012 on www.turkstat.gov.tr.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Figure 7 Proportion of Women among Users

Figure 8 Median Income of the Users

Thus, the socio-demographics and geographical data collected on the users of the Oypusulasi site veer in the same direction of the data on Internet users, suggesting that two significant digital divides exist in Turkey: a divide between males and females, and a divide between urban and rural areas. With regards to technology and Internet usage, studies in western countries underline the generational divide among possible others. Oypusulasi users

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 have a median age of 30 years, in comparison to the median of the total population in 2010 at 29.2 years. Thus, the fact that the majority of the VEA users in Turkey are below 34 represents the national population demographic of a youthful society. Therefore, rather than a generational divide as a consequence of a web based survey, our data seem to primarily reflect the age structure of the country.

Figure 9 Proportion of 15-24 years old Users per Region

7.4. ‘Looking it up’: Through which web channel did the users access oypusulasi.org?

When media partners are involved, hyperlinks toward the VEA site are available to large reader groups; therefore, estimating the importance of each of the channels of access to the website is possible. Lacking those partnerships, we had to rely on another tool for gathering this information. A ‘referrer’ variable was inserted in the tool; similar to the service provided by Google Analytics. We were able, via this feature, to keep track of the webpage through which the users accessed the Oypusulasi website. Table 3 displays a comparison of diverse channels of access to the website between the users who connected from Turkey and the users who connected from other countries. Through the same procedure described in the previous section, using a GEO-IP database, we 89

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 were able to identify the countries from which the users connected to the website. The comparison between connections from Turkey and connections from abroad is relevant for us because Al-Jazeera was an international partner of the project, and thus relayed the existence of such a website that provided substantive information on the political platforms of the main parties running in the Turkish election of 2011. Among the connections from Turkey, 84% of the users directly entered the name of the site in the address bar of their web browser, 7.9% of the users were redirected to oypusulasi.org from the Al-Jazeera website, 3.3% from the Do ğan Haber Ajansı website, and 2.9% from Google Search. Several other items record a flux inferior to 1% of the connection to the Oypusulasi website. Comparing those data with the connections from abroad, the most striking difference regards the fact that the connections emerging the Al-Jazeera website, from abroad, are almost as numerous as those going directly to the Oypusulasi website, 47.4% and 48.1% respectively. Clearly, the existence of effective media creates an important difference in the way users are attracted to a website.

Channel of Access Connections from Turkey (%) Connections from Abroad (%) Oypusulasi 84 48,1 Al-Jazeera 7,9 47,4 Do ğan Haber Press Agency 3,3 1,0 Google Search 2,9 1,0 Eksi Sozluk 0,9 0,3 Facebook 0,4 0,3 Twitter 0,1 0,1 Le Monde 0,05 0,5 Kamil Pasha 0,03 0,2 Other 0,42 1,1 100 100 N 57105 14462 Table 3 Channel of Access to the Oypusulası site Note: Do ğan Haber Ajansı is a Turkish press agency; Ek şi Sözlük is a Turkish collaborative hypertext dictionary; Le Monde is the online version of a French newspaper; Kamil Pasha is the blog of Jenny White, professor of Anthropology at Boston University.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

7.5. The Oypusulasi ‘mini-campaign’ and the double-screen effect

In the absence of a media partner, the team behind the project had to launch its own ‘mini-campaign’ (Abold, 2008) in order to reach out to the voters. This ‘mini-campaign’ can be described as a ‘double-screen campaign’ since it involved both traditional media events such as promotion in television and coverage by newspaper columnists, and Internet-based promotions, such as the use of social networks. Another important but difficult to measure method of promoting the VEA was the massive number of emails to university students, academics, and informal networks of the project team members via social networks. Those mailings are likely to impacted the early days of the campaign and also likely to have generated an inflow in the later stages; however this consequence is hard to demonstrate. The number of connections that the Oypusulasi website received during the last four weeks preceding the election was about 190,000. However, after cleaning the data for users that did not answer any questions and did not position themselves on any issue statements, the dataset includes 73,041 observations. Among these, 58,579 come from Turkey. Since the media events of the ‘double-screen campaign’ occurred in Turkey, only those latter domestic connections are kept for the analysis in this section. In order to measure the effect of the ‘double-screen campaign’, a multiple-interrupted time series (MITS) analysis is applied to the dataset 7. In order to be able to conduct a MITS analysis, the dataset was recoded from daily to hourly basis (28 days to 672 hours) so that both the short and long term effects of each interruption could be measured. The interruptions are of three kinds: the promotion of the Oypusulasi website on Twitter, the newspaper articles and columns mentioning the tool, and lastly the television programs in which the website was discussed. Table 4 and Table 5 list different media items taken into analysis as interruptions in the following MITS model:

Y = βₒ + β₁Time Trend +∑ λi Hours i ∑ µi Promotion Events i +ε

7 See Lewis-Beck (1980) for an accessible exposition of the methods of MITS.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Where Y indicates the hourly number of connections, βₒ indicates the reference level of the dependent variable for the dependent variable when no promotion takes place at the very beginning of the analysis period. Then for every promotion event for which we have an exact time of diffusion, we coded a dummy variable for that event at the time of the event and zero otherwise. Since newspapers items are run on a daily basis, they are included into the hourly analysis by coding every hour from 7 AM to 12 PM as positive in order to cover the hourly exposure of a newspaper piece or a column; we can also incorporate into the analysis both the Twitter items and the television items by taking their exact hour of diffusion. Although the official Twitter account of Oypusulasi was retweeted several times, we observed the most salient changes in the slope when ‘celebrities’ or opinion leaders, are mentioning or retweeting the address of the VEA website. Therefore, the Twitter items included in the analysis only represent the retweets of journalists and academics. These retweets took place on May 16 th , 17 th and 26 th . In addition, we included eight television news programs in which the Oypusulasi website was promoted. These programs were also included as interruptions on May 17 th , 18 th , 21 st , 23 rd , 24 th , 28 th , 29 th and 31 st . Figure 10 displays the hourly frequency of connections which forms the basis of our analysis. The results of the MITS analysis are shown in Table 6.

Interruption Interruption Media Name of Twitter Account/TV Date Type Program 16.05.2011 Retweet Twitter Retweet by Deniz Ülke Arıbo ğan 17.05.2011 News Program Ntv “Haber Merkezi” 17.05.2011 Retweet Twitter Retweet by Kanat Atkaya 18.05.2011 News Program Ntv “Güne Ba şlarken” 21.05.2011 News Program A Haber “Nasıl Yani?” 23.05.2011 News Program CNNTürk “Güne Merhaba” 24.05.2011 News Program CNNTürk “Be ş N Bir K” 26.05.2011 Retweet Twitter Retweet by Koray Peközkay 28.05.2011 News Program TRTHaber “Haber Tadında” 29.05.2011 Tv Program CNNTürk “E ğrisi Do ğrusu” 31.05.2011 News Program CNNTürk “Güne Merhaba” Table 4 Media Items as Interruptions for the MITS Analysis

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Interruption Date Interruption Type Media Name of Columnist

18.05.2011 Column Vatan Ru şen Çakır 19.05.2011 Column Radikal Koray Çalı şkan 24.05.2011 Column Bursa Gerçek Hakan Yusuf Ekren 31.05.2011 Column Hürriyet İsmet Berkan 01.06.2011 Column Ak şam Ali Saydam 01.06.2011 Column Milliyet Mehve ş Evin 07.06.2011 Column Zaman Şahin Alpay Table 5 Newspaper Items Covering the Oypusulasi website

Figure 10 Hourly Distribution of Connections to the Oypusulasi Website between May 15th and June 12th

Several patterns are noteworthy. First, we see that the time trend has a significant and negative coefficient indicating that the amount of respondents to web questionnaires were higher at the beginning of our mini-campaign and started to decline towards the end of the campaign as Election Day approached. There was a decline of 10 respondents for every ten hours compared to the reference point. Early hours of a day typically demonstrate lower levels of connections to the application web site.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Retweets by Kanat Atkaya and Deniz Ülkü Arıbo ğan, two opinion leaders 8, provoked the largest hourly increase compared to any other promotion activity, on the 17 th and 16 th of May respectively. Then, the third largest impact is triggered by the TRT television news program Haber Tadında [Flavoured News] on the 28 th of May. It is noteworthy that the retweets brought in a larger hourly following according to our model than did the TV programs broadcasted later in the campaign. Except for the CNN-Türk ’s morning program, Güne Merhaba [Hello Today] on the 23 rd of May, no other TV program appears to have brought in a significant group of respondents after checking the hour of the day and the long term trend.

Unstandardized Significance Coefficients level Std. B Error (Constant) 124,7 14,9 ,00 Overall trend (t) -,1 ,0 ,00 May 17th Retweet - Kanat Atkaya 593,1 75,7 ,00 May 16th Retweet - Deniz Ülke Arıbo ğan 457,9 75,7 ,00 May 28th TRTHaber - Haber Tadında 457,7 75,5 ,00 May 23rd CNNTürk - Güne Merhaba 278,9 75,5 ,00 June 7th Zaman - Şahin Alpay 180,2 18,6 ,00 May 31st Hürriyet - İsmet Berkan 169,4 18,9 ,00 May 31st CNNTürk - Güne Merhaba 106,5 77,8 ,17 May 19th Radikal - Koray Çalı şkan 67,5 18,9 ,00 May 24th CNNTürk - Be ş N Bir K 64,2 75,6 ,40 May 18th Vatan - Ru şen Çakır 63,8 18,9 ,00 May 17th NTV - Haber Merkezi 54,6 75,6 ,47 Ali Saydam (Ak şam) June 1st 29,4 18,4 ,11 Mehve ş Evin (Milliyet) May 29th CNNTürk - E ğrisi Do ğrusu 17,6 75,5 ,82 May 26th Retweet - Koray Peközkay -29,5 75,5 ,70 May 21st A Haber - Nasıl Yani? -36,7 75,6 ,63 *Dummy variables for hours of the day are not reported. Adjusted R-Squared= 0,47 Table 6 Results of the MITS Analysis

8 Kanat Atkaya is a well-known columnist in the newspaper Hürriyet , and Deniz Ülkü Arıbo ğan is a professor of political science in İstanbul Bilgi University, Istanbul.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Note: Each item name includes the names of the source (television, newspapers, or Twitter) and of the actor mentioning Oypusulasi (opinion leaders, TV programs, or columnists). A Haber, CNNTürk, NTV, and TRTHaber are televisions. Ak şam, Hürriyet, Milliyet, Radikal, Vatan, and Zaman are newspapers.

The newspaper coverage of Oypusulasi by columnists Ru şen Çakır ( Vatan [Homeland] daily) on the 18 th , by Koray Çalı şkan ( Radikal daily) on the 19 th , and lastly by İsmet Berkan ( Hürriyet [Freedom] daily) on the 31 st of May, all appear to have significantly increased the inflow of respondents. However, among all columnists the most significant one in the MITS model we employed appears to be Şahin Alpay’s article in the Zaman [Time] daily on the 7 th of June. It should also be noted that the relaying of Şahin Alpay’s article on news websites Moralhaber.net and Timeturk.com generated a triple source impact that brought in approximately 3,000 people over a time span of 17 hours in the above specification. Similarly, İsmet Berkan’s column appears to have attracted about 2,900, Koray Çalı şkan’s 1,150 and Ru şen Çakır’s 1,080 respondents above the overall time trend and the day of the week they appeared during the promotion process. It is important to note at this juncture that Şahin Alpay’s article appeared in the largest selling conservative leaning newspaper Zaman , and thus reached a reader group that is likely to be distinct from mainstream outlets. Our promotion campaign appears to have had trouble in reaching the conservative constituencies in earlier campaign efforts. Therefore, Şahin Alpay’s large impact may be due on one hand to his appeal to a conservative segment of respondents, and on the other hand to the in-depth coverage of the logic and usefulness of the application in his column. Short and relatively simplistic coverage of the application’s content appears to yield fewer respondents. Furthermore, such coverage usually does not help pushing the respondents above the underlying trends controlled for in our model. It would also appear that over time, the new columns’ coverage of our application brought in more respondents. Momentum for the campaign seemed to build as a result of retweets and TV appearances. In addition to the above analysis on the impact of newspapers upon connections, Figure 11 reflects a clear pattern between the number of connections to the Oypusulasi website and the days in which newspapers items were published (see Table 5 for a detailed list). One can reasonably argue that newspaper coverage matches with the registered peak in

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 the number of connections to the Oypusulasi website. Thus, the findings of the MITS analysis and the data displayed in Table 6 and Figure 11 reveal that the ‘mini-campaign’ launched in order to promote the Oypusulasi website resulted in a ‘double-screen’ effect on the number of connections. The users that connected to the VEA website appear to have been attracted to it through multiple sources of information: from traditional media sources, such as television and newspapers, but also from Internet- based communication, such as social networks and online media websites. Even with the lack of national media partners, significant media attention has been generated toward the tool. Despite the fact that the campaign mostly focused on the horse-race, this finding indeed suggests there was still room for a more substantive debate.

Figure 11 Number of Connections and Publication of Newspapers Items

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

7.6. Press-party parallelism and audience segmentation: what output for a VEA?

In an electoral campaign, the main objective of a VEA is to provide its users with an implicit electoral advice reflecting basic policy preferences. The logic behind it is that “other things being equal, the voter will choose the party which, in policy terms, is least distant from them” (Evans, 2004, p.100). We call this “implicit” advice since at no juncture does the Oypusulasi ever explicitly state that the voter is closer to a given party and thus he or she should consider voting for that party. All implicit advice does is to graphically show where the respondent stands in relation to the parties included in the VEA analysis. Before providing an implicit advice to its users, the Oypusulasi website also asked them to estimate the likelihood that they will ever vote for each of the main parties competing in the election. By offering this question before the final graphical summary result is given, the estimated propensity to vote for each party is not ‘affected’ by the advice of the VEA. Eijk et al. (2006) demonstrate that this ‘propensity to vote’ (PTV) survey item is a convenient way to operationalize the electoral utility that voters attach to each party competing for their votes at the election. Since, as noted earlier, the Turkish media system is characterized by a high level of press-party parallelism, we can evaluate the heterogeneity of the PTVs expressed in light of the media events mentioned in the previous section. For instance, among the newspapers that discussed the tool in one of their columns, Zaman is identified as having a conservative stance and pro-AKP, while Hürriyet , is perceived as adopting a more neutral stance in the media landscape 9. Therefore, we would expect their respective audiences to express PTVs toward different political parties. When observing Figure 7.6, we see that the users generally expressed consistently higher PTVs for CHP than for any other party. This finding clearly indicates a partisan bias in the sample reached by our VEA, since CHP received only one quarter of the total vote in the elections while AKP received approximately 50%. However, what is relevant to our analysis here is that, on several occasions, the incoming sample of users appears to be more heterogeneous including supporters of parties other than the CHP as well: particularly on May 28 th and on June 7 th . On both dates, we see sharp increases in the PTVs expressed for the

9 See Çarko ğlu and Yavuz (2010) for the partisan leanings of different newspapers in Turkey.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 following parties: AKP, the Grand Union Party ( Büyük Birlik Partisi -BBP), and Felicity Party (Saadet Partisi -SP). AKP and SP are conservative, pro-Islamist parties, while BBP has a more nationalist right-wing party identity 10 . When crossing those two dates with Tables 7.2 and 7.4, we see that on May 28 th the television channel TRTHaber mentioned the Oypusulasi website in one of its news programs, and on June 7 th , the newspaper Zaman discussed the tool in one of its columns. Both media institutions are known for their rather conservative audience (Çarko ğlu and Yavuz, 2010) and therefore may have attracted towards the VEA users likely to express PTVs in favour of those parties. In other words, as Oypusulasi received coverage in a relatively more conservative media, the inflow of respondents to our website immediately reflected the predominantly right-wing conservative party choices of the readers of these newspapers. As long as diverse media support our application, and in the case of the Turkish media system we believe that it is necessary in order to reach different ideological groups, our tool clearly demonstrates a natural openness to all ideological orientations.

10 For ideological differentiation of political parties in Turkey see the summary exposition in Çarko ğlu (2003), p. 124.

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Mean 3

ay n n n n Ju Ju Ju .May M .May .May . .Jun . .Jun . 6 8.May 2 6 8.May 3 9 1 1 20. 2 24.May 2 2 30.May 01 0 05.Ju 07 0 11

AKP BBP BDP CHP MHP SP

Figure 12 Evolution of the PTVs (scale 0-10) expressed on the Oypusulasi website

Note: The parties displayed are the Justice and Development Party ( Adalet ve Kalkınma Partisi -AKP), the Grand Union Party ( Büyük Birlik Partisi -BBP), the Peace and Democracy Party ( Barı ş ve Demokrasi Partisi-BDP), the Republican People’s Party ( Cumhuriyet Halk Partisi -CHP), the Nationalist Action Party ( Milliyetçi Hareket Partisi-MHP) and the Felicity Party ( Saadet Partisi -SP). To keep the figure simple, several minor parties (the Democrat Party-Demokrat Parti-DP, the Left Democratic Party-Demokratik Sol Parti -DSP, and the People’s Voice Party-Halkin Sesi Partisi -HSP) are not included in the figure.

On Figure 13 to 15, we can see for each of the parties in Figure 12 were implicitly advised to the users during the four weeks of the campaign. Overall, every distribution has a four-peak structure due to the distribution of the connections to the website which is itself four-peak shaped (see Figure 11). However, three observations emerge out of those distributions. Firstly, on May 19 th when the newspaper Radikal mentioned the tool in one of its columns, we see a high concentration of respondents advised towards BDP (Figure 13). We believe that this increase is largely attributable to the liberal left-wing orientation of the Radikal daily newspaper and its core reader constituency which apparently was more open to the policy preferences of the BDP in 2011. Secondly, on May 28 th , when the television

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

TRTHaber mentioned the VEA in one of its news programs, we observe a fifth peak emerging in the advices of conservative and nationalist parties, such as AKP, BBP, MHP, and SP (Figure 14 and 15). And lastly, on June 7 th , there is a particularly high concentration of AKP, BBP and SP’s advice to users on the day in which the newspaper Zaman discussed the tool. In short, connections to the Oypusulasi website on any given day, as well as the political tendencies of users. appear to be largely dependent on which media outlet had mentioned the VEA.

18% 16% 14% 12% 10% 8% 6% 4% 2% 0%

n ay ay ay Jun Ju Jun Jun M May M .May M May . . . 6. 8.May 0. 2. 4 6. 8.May 0. 1 1 2 2 2 2 2 3 01 03.Jun 05. 07 09.Jun 11

BDP CHP

Figure 13 Distribution of the Advices Given in favour of BDP and CHP by the Oypusulasi Website

Note: Peace and Democracy Party ( Barı ş ve Demokrasi Partisi-BDP), Republican People’s Party ( Cumhuriyet Halk Partisi -CHP)

100

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

18% 16% 14% 12% 10% 8% 6% 4% 2% 0%

n ay ay ay Jun Ju Jun Jun M May M .May M May . . . 6. 8.May 0. 2. 4 6. 8.May 0. 1 1 2 2 2 2 2 3 01 03.Jun 05. 07 09.Jun 11

AKP SP

Figure 14 Distribution of the Advices Given in favor of AKP and SP by the Oypusulasi Website Note: Justice and Development Party ( Adalet ve Kalkınma Partisi -AKP), Felicity Party (Saadet Partisi -SP)

16% 14% 12% 10% 8% 6% 4% 2% 0%

n ay ay ay Jun Ju Jun Jun M May M .May M May . . . 6. 8.May 0. 2. 4 6. 8.May 0. 1 1 2 2 2 2 2 3 01 03.Jun 05. 07 09.Jun 11

BBP MHP

Figure 15 Distribution of the Advices Given in favor of BBP and MHP by hte Oypusulasi Website Note: Grand Union Party ( Büyük Birlik Partisi -BBP), Nationalist Action Party ( Milliyetçi Hareket Partisi-MHP)

101

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

CHAPTER 8. CONCLUSIONS

To our knowledge, the Oypusulasi website was among the very first VEAs ever run during an election campaign in Turkey. As such, the data collected through the tool and the analysis presented in Chapter 7 assume the status of a ‘health check’ on the users’ profile, the diffusion of the website, and its impact in the Turkish context. The development of a VEA in Turkey was a response to the growing partisan alignment of the media in the country. Unfortunately, this political parallelism is also what made promoting the VEA more difficult. The focus on substantive issues proposed by a VEA was not likely to provoke an interest within a media environment characterized by political parallelism, rhetorical games, and political pressures on press bodies. As a matter of fact, no national media partner offered to sponsor the website. Nevertheless, the data presented in Table 11 suggest that when a media partner promotes the website, a substantive proportion of users are attracted to the application by the media partner. Even though no national media sponsored the website, the fact that some media outlets still discussed the VEA in their news programs and columns led to a relative promotion of the application through a ‘mini-campaign’ (Abold, 2008). This ‘mini-campaign’ was characterized by a ‘double-screen effect’: on the one side, traditional media outlets such as television and newspapers, which considerably increased the number of users, and on the other side, the Oypusulasi website, which was relayed through social networks and other Internet-based communications. Our analyses further confirm our expectations that the diversity of the media discussing the VEA is essential to order to reach a heterogeneous group of users. Since media audiences in Turkey are politically segmented, it is only when, for instance, conservative media mentioned the website that users with conservative opinions came to use the VEA. Hence, it is not surprising that the political parallelism of the Turkish media is salient in our findings. The simultaneous increase in ‘recommendations’ in favour of conservative parties and the presence of the VEA in the conservative media also tells us about the tool itself. The VEA measured a parallel evolution of the PTVs expressed for conservative parties and of the ‘recommendations’ in favour of those same parties. Thus, the tool was successful in capturing this group of users. At first sight, it suggests that the statements selected in order to position

102

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 parties and voters on political dimensions are valid. Of course, this preliminary finding should be investigated further. The data collected through the VEA website will provide an assessment of the issue dimensionalities expressed by the Oypusulasi users, and check for the validity of the issues selected in a comparative fashion. The Turkish case can provide an intriguing ground for comparison both with data collected through VEAs in western countries, and with data issued from the VEAs that recently went online in some of the Arab Spring countries, such as Egypt and Tunisia. Moreover, the political parallelism observed in the data indirectly suggests that VEAs could, and should, be integrated into broader research designs in order to use their full potential to measure campaign effects. Such online applications should be parallel with panel survey studies, so that the data generated by VEAs can be assessed and compared with the data compiled by panel studies. Also, the PTVs expressed and the ‘recommendations’ given on the website would gain informative power if they could be aggregated with a content analysis of media items. While at first, political scientists focused their analysis on ‘recommended’ parties, now there is a shift of our interest in the field towards the linkages between media coverage of election campaigns and the VEA users’ ideological orientations. VEAs are observed and studied as tools offering voters with structured information on party positions. Therefore, on one hand, the interactions between VEAs and its users and on the other hand, between the VEA users and media content are acquiring an increasing saliency in the debate. To conclude with some general recommendations on how to develop and implement VEAs, it is crucial to establish media partnership in order to reach a substantive number of users, and a wide range of users’ profile. The idea is that one or several media regularly promote in their outlets the existence of the online VEA, and in exchange the team of researchers building the VEA site provides the media with stories with substantive content on the political landscape and on the issues preferences expressed by the users, among other topics. In addition, in the fragmented media environment of the 21 st century, a diversification of the media partners is desirable. On the one hand, there should be a diversification among traditional media sources that is television, national newspapers, regional newspapers, free newspapers, radio. And on the other hand, if necessary there should be a diversification in the political profile of the media such as the case of the Turkish media system has illustrated.

103

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Choices should be made according to the media structure of the country in which the VEA is implemented. Lastly, the new media that Web-based political communication represents should not be neglected, since it bears the potential for being as effective as traditional media in reaching users.

104

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

REFERENCES

Abold, R., 2008. ‘1000 Mini Election Campaigns: The Utilization of Private Weblogs in the Run-up to the 2005 German Election’, in Farrell, D.M., and Schmitt-Beck, R. (Eds.), Non- Party Actors in Electoral Politics. The Role of Interest Groups and Independent Citizens in Contemporary Election Campaigns . Nomos, Baden-Baden, pp.209-235.

Ansolabehere, S., 2006. ‘The Paradox of Minimal Effects’. In Brady H. E. and Johnston R. (Eds.), Capturing Campaign Effects . Ann Arbor: The University of Michigan Press, pp. 29- 44.

Ansolabehere, S., Rodden, J., and Snyder, J. M., 2008. ‘The Strength of Issues: Using Multiple Measures to Gauge Preference Stability, Ideological Constraint, and Issue Voting’. American Political Science Review 102(2): 215-232.

Arendsen, R., 2003. ‘Partijen zonder mening winnen op Stemwijzer.nl’. In: Politiekdigitaal (online). Available at: http://www.politiekdigitaal.nl/nieuwedemocratie/partijen_zonder_mening_winnen_op_stemwijzer.nl

Asp, K., 1983. ‘The Struggle for the Agenda’, Communication Research , 10: 333–55.

Bartels, L.M., 1987. "Candidate Choice and the Dynamics of the Presidential Nomination Process." American Journal of Political Science 31:1-30.

Bartels, L.M., 1988. Presidential Primaries and the Dynamics of Public Choice. Princeton: Princeton University Press.

Bartels, L. M., 1993. ‘Messages Received: The Political Impact of Media Exposure’. The American Political Science Review , 87 (2): 267-285.

Bayram, S., 2010. ‘Political Parallelism in the Turkish Press, a Historical Interpretation’, Turkish Studies , 11(4): 579-611.

Benoit, K., and Laver, M., 2007. ‘ Estimating party policy positions: Comparing expert surveys and hand coded content analysis’ , Electoral Studies , 26: 90 - 107.

Benoit, K., and Laver, M., 2003. ‘ Estimating Irish Party Positions Using Computer Wordscoring: The 2002 Elections ’, Irish Political Studies , 18(1): 97 - 107.

Benoit, K., and Laver. M., 2006. Party Policy in Modern Democracies , London: Routledge.

Benoit, K. and Laver, M. ( forthcoming ). “The Dimensionality Of Political Space: Epistemological And Methodological Considerations.” European Union Politics .

Bentivegna, S., 2002. ‘Politics and New Media’, in Lievrouw, L., and Livingston, S. (Eds.), The Handbook of New Media , London: Sage Publications, pp. 50-61. 105

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Berelson, B. R., Lazarsfeld, P. F., and McPhee, W. N., 1954. Voting-. A Study of Opinion Formation in a Presidential Campaign. Chicago: University of Chicago Press.

Bird, S. E., 1998. ‘An Audience Perspective on the Tabloidisation of News’, The Public , Vol. 5, N. 3, pp.33 – 50.

Blumler, J. G., 1983. Communicating to Voters. Television in the First European Parliamentary Elections . London: Sage.

Blumler, H., (Ed.) 1992. Television and the Public Interest . London: Sage.

Blumler, J. G., Cayrol, R., and Thoveron, G., 1978. La Télévision Fait-Elle l’Élection? Paris: Presses de la Fondation National des Sciences Politiques.

Blumber, J.G., and Gurevitch, M., 2000.’Rethinking the study of political communication’. In J. Curran and M. Gurevitch (Eds.), Mass Media and Society (3rd edi., pp. 155-172). London: Arnold.

Blumler, J. G., and Kavanagh, D., 1999. ‘The Third Age of Political Communication: Infuences and Features’, Political Communication , Vol. 16, pp.209 - 230.

Blumler, J. G., and McQuail, D., 1968. Television in Politics. Its Uses and Influence. London: Faber and Faber.

Bogart, L., 1995. Commercial Culture . New York: Oxford University Press.

Bourdieu, P., 1998. On Television and Journalism , London: Pluto.

Brady, H. E., Johnston R., and Sides, J., 2006. ‘The Study of Political Campaigns’. In Brady H. E. and Johnston R. (Eds.), Capturing Campaign Effects (pp. 1-27). Ann Arbor: The University of Michigan Press.

Brandenburg, H., 2002. ‘Who Follows Whom? The Impact of Parties on Media Agenda Formation in the 1997 British General Election Campaign’, Harvard International Journal of Press/Politics 7: 34–54.

Brants, K., 1998. ‘Who’s Afraid of Infotainment’ European Journal of Communication , 13(3): 315-335.

Budge, I., 2000. ‘Validating Party Policy Positions’, British Journal of Political Science , 31(1): 210-223.

Budge, I., Robertson, D. and Hearl, D. (Eds.), 1987. Ideology, Strategy and Party Change: Spatial analysis of post-war election programmes in 19 democracies, Cambridge: Cambridge University Press.

106

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Campbell, A., Converse, P., Miller, W., Stokes, D. (1960) The American Voter . New York: John Wiley & Sons, Inc.

Campbell, J. E., 2000. The American Campaign-. U.S. Presidential Campaigns and the National Vote. College Station: Texas A&M Press.

Caprara, G. V., 2008. Will Democracy Win?. Journal of Social Issues , 64 (3), 639-659.

Çarko ğlu, A., 2011. ‘Turkey’s 2011 General Elections: Towards a Dominant Party System?’, Insight Turkey , 13(3): 43-62.

Çarko ğlu, A., 2010. ‘Introduction’, Turkish Studies , 11(4): 514-519.

Çarko ğlu, A., 2009. ‘Turkey’s Local Elections of 2009: Winners and Losers’, Insight Turkey , 1(2): 1-18.

Çarko ğlu, A., 2003. “The Rise of the New Generation Pro-Islamists in Turkey: The Justice and Development Party Phenomenon in the November 2002 Elections in Turkey”, South European Society & Politics , 7(3): 123-156.

Çarko ğlu, A., 2000. ‘Geography of April 1999 Turkish Elections’, Turkish Studies , 1(1): 149- 171.

Çarko ğlu, A., 1998. ‘Turkish Party System in Transition: The Connection Between Macro- Level Party Competition and Agenda Change in Election Manifestoes’, Political Studies , XLVI: 544-571.

Çarko ğlu, A. and Hinich, M. J., 2006. ‘A spatial analysis of Turkish party preferences”’ Electoral Studies , 25(2): 369 - 392.

Çarko ğlu, A. and Kalaycıo ğlu, E., 2009. The Rising Tide of Conservatism in Turkey , New- York: MacMillan Palgrave.

Çarko ğlu, A and, Yavuz, G., 2010. ‘Press-party Parallelism in Turkey: An Individual Level Interpretation’, Turkish Studies , 11(4): 613-624.

Çarko ğlu, A., Vitiello, T. and Moral M., 2012. ‘Voter Advice Applications in Practice: Some Answers to Some Key Questions from Turkey’, International Journal of Electronic Governance , forthcoming.

Carmines, E. G., Stimson, J. A., 1980. ‘The Two Faces of Issue Voting’. The American Political Science Review 74(1): 78-91.

Cautrès, B., and Muxel, A., 2009. ‘Le processus de la décision électorale’. In Cautrès B. and Muxel A. (Eds.), Comment les électeurs font-ils leur choix ? (pp.27-43). Paris: Les Presses de SciencesPo.

107

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Cedroni, L., and Garzia, D., 2010. Voting Advice Applications in Europe: The State of the Art . Napoli: Scriptaweb.

Coleman, S., 1999. ‘The New Media and Democratic Politics’, New Media and Society , 1(1): 67-74.

Connell, I., 1998. ‘Mistaken Identities: Tabloid and Broadsheet News Discourses’, The Public , 5(3): 11–31.

Converse, P. E. (1964) ‘The Nature of Belief Systems in Mass Publics’, in David E. Apter (ed) Ideology and Discontent , pp. 206-261. London: Free Press.

Costa Lobo, M., Vink, M and Lisi, M. (2010) ‘Mapping the Political Landscape: A Vote Advice Application in Portugal’, in Lorella Cedroni and Diego Garzia (eds.), Voting Advice Applications in Europe. The State of the Art , ScriptaWeb, Napoli, pp.143 - 185.

Couper, M.P., 2000. ‘Web Surveys. A Review of Issues and Approaches’, Public Opinion Quarterly , 64: 464-494.

Dahlberg, L., 2001. ‘Democracy via Cyberspace: Examining the Rhetorics and Practices of Three Prominent Camps’, New Media & Society , 3(2): 187-207.

Dalhgren, P., 2005. ‘The Internet, Public Spheres, and Political Communication: Dispersion and Deliberation’, Political Communication , 22: 147-162.

Dalton, R. J., and Wattenberg, M. P., (Eds.), 2000. Parties without Partisans: Political Change in Advanced Industrial Democracies . Oxford: Oxford University Press.

Dalton, R., 2004. Democratic Challenges, Democratic Choices: The Erosion of Political Support in Advanced Industrial Democracies . New York: Oxford University Press.

Davis, A., 2002. Public Relations Democracy. Public Relations, Politics and the Mass Media in Britain . Manchester: Manchester University Press.

Davis, R., and Owen, D., 1998. New Media and American Politics . New York: Oxford University Press.

Della Vigna, S. and Kaplan, H., 2007. ‘The Fox News Effect: Media Bias and Voting’. The Quarterly Journal of Economics : 122 (3): 1187-1234.

Delli Carpini, M. X., and Keeter, S., 1996. What Americans Know About Politics and Why it Matters . New Haven: Yale University Press.

Downs, A., 1957. An Economic Theory of Democracy . New York: Harper & Row.

Druckman, J.N., Kifer, M.J., and Parkin, M., 2007. ‘The Technological Development of 108

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Congressional Candidate Web Sites: How and Why Candidates Use Web Innovations’, Social Science Computer Review , 25: 425-442.

Eijk van der, C., van der Brug, W., Kroh, M., and Franklin, M., 2006. 'Rethinking the dependent variable in voting behavior: On the measurement and analysis of electoral utilities', Electoral Studies , 25(3): 424–447.

Entman, R. M., 1993. ‘Framing: Toward Clarification of a Fractured Paradigm’. Journal of Communication 43 (4): 51-8.

Erikson, R. S., and Tedin., K., 2007. American Public Opinion: Its Origin, Content, and Impact. New York: Wiley.

Esen, B., and Ciddi S., 2011. ‘Turkey’s June 2011 Elections: An Emerging Dominant Party System?’, MERIA Journal , 15(3), [online]. Available at: http://www.gloria- center.org/2011/10/turkey%E2%80%99s-2011-elections-an-emerging-dominant-party- system/ (Accessed on 5 January 2012).

Evans, G., Heath, A.F. and Lalljee, M.G., 1996. 'Measuring left-right and libertarian- authoritarian values in the British electorate’, British Journal of Sociology , Vol. 47 No.1, pp.93 - 112.

Evans, J., 2004. Voters and Voting – An Introduction , Sage Publications, London.

Farell, D.M., and Webb, P., 2000. ‘Political Parties as Campaign Organizations’, in Dalton, R. J., and Wattenberg, M. P., (Eds.), Parties without Partisans: Political Change in Advanced Industrial Democracies . Oxford: Oxford University Press, pp.102 – 112.

Farrell, D.M., and Schmitt-Beck, R., 2008a. Non-Party Actors in Electoral Politics. The Role of Interest Groups and Independent Citizens in Contemporary Election Campaigns , Nomos, Baden-Baden.

Farrell, D.M., and Schmitt-Beck, R., 2008b. ‘Introduction: The Age of Non-party Actors?’, in Farrell, D.M., and Schmitt-Beck, R. (Eds.), Non-Party Actors in Electoral Politics. The Role of Interest Groups and Independent Citizens in Contemporary Election Campaigns , Nomos, Baden-Baden, pp.13-24.

Fivaz, J. and Schwartz, D., 2007. 'Nailing the Pudding to the Wall – EDemocracy as Catalyst for Transparency and Accountability', presented at the International Conference on Direct Democracy in Latin America , Buenos Aires.

Fivaz, J., and Nadig, G., 2010. "Impact of Voting Advice Applications (VAAs) on Voter Turnout and Their Potential Use for Civic Education," Policy & Internet :, 2(4).

Franklin, M., 2003. Voter Turnout and the dynamics of electoral competition , Cambridge: Cambridge University Press.

109

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Franklin, B., 1994. Packaging Politics. Political Communication in Britain’s Media Democracy . London: Edward Arnold.

Franklin, M. N., Mackie T. T. and Valen, H., 1992. Electoral Change: Responses to Evolving Social and Attitudinal Structures in Western Countries . Cambridge: Cambridge University Press.

Gans, H.J., 2003. Democracy and the News. Oxford: Oxford University Press.

Gemenis, K., 2011. ‘Estimating the positions of political actors through voting advice applications: some methodological considerations’, Acta Politica forthcoming. Available at: http://www.utwente.nl/mb/pa/staff/gemenis/

Gerber, A.S., Karlan, D., and Bergan D., 2009. ‘Does the Media Matter? A Field Experiment Measuring the Effect of Newspapers on Voting Behaviour and Political Opinions’. American Economic Journal: Applied Economics , 1 (2): 35-52.

Gibson, R.K., Margolis, M., Resnick, D., and Ward, S.J., 2003. ‘Election Campaigning on the WWW in the USA and UK’, Party Politics , 9(1): 47-75.

Giddens, A., 1990. The Consequences of Modernity . Cambridge: Polity.

Gilmor, D., 2004. We the media: Grassroots journalism by the people, for the people. Sebastopol, CA: O´Reilly.

Groot, L.F.M., 2003. De kwaliteit van stemprogramma’s en de Stemvork , SISWO, Amsterdam.

Gurevitch, M., and Blumler, J. G., 2004. ‘State of the Art of Comparative Political Communication Research: Poised for Maturity?’, in Esser, F. & Pfetsch, B. (Eds), Comparing Political Communication. Theories, Cases, and Challenges . Cambridge, Cambridge University Press, pp.325 - 343.

Habermas, J., 1962. Strukturwandel der Öffentlichkeit: Untersuchungen zu einer Kategorie der bürgerlichen Gesellschaft [The Structural Transformation of the Public Sphere: An Inquiry into a Category of Bourgeois Society]. Neuwied: Herman Luchterhand Verlag.

Hall, S., Chrichter, C., Jefferson, T., Clarke, J. and Roberts, B., 1978. Policing the Crisis: Mugging, the State, and Law and Order . London: Macmillan.

Hallin, D. C., and Mancini, P., 1984. ‘Political Structure and Representational Form in US and Italian TV news’, Theory and Society , Vol. 13, N. 40, pp.829 – 850.

Hallin, D. C., and Mancini, P., 2004a. ‘Americanization, Globalization, and Secularization: Understanding the Convergence of Media Systems and Political Communication’, in Esser, F. & Pfetsch, B. (Eds), Comparing Political Communication. Theories, Cases, and Challenges . Cambridge: Cambridge University Press, pp.25 – 44. 110

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Hallin, D. C., and Mancini, P., 2004b. Comparing Media Systems. Three Models of Media and Politics . Cambridge: Cambridge University Press.

Hellström, J., 2008. ‘Who Leads, Who Follows? Re-examining the Part-Electorate Linkages on European Integration’. Journal of European Public Policy , 15(8): 1127-1144.

Herman, E., and McChesney, R., 1997. The Global Media. The New Missionaries of Corporate Capitalism . London: Cassel.

Holbrook, T. M., 1996. Do Campaigns Matter? Thousand Oaks, CA: Sage.

Holtz-Bacha, C., 2004. ‘Political Campaign Communication. Conditional Convergence of Modern Media Elections’, in Esser, F. & Pfetsch, B. (Eds), Comparing Political Communication. Theories, Cases, and Challenges . Cambridge: Cambridge University Press, pp. 213-230.

Hooghe, M. and Teepe, W., 2007. 'Party profiles on the web: an analysis of the logfiles of nonpartisan interactive political internet sites in the 2003 and 2004 election campaigns in Belgium', New Media & Society , 9(6): 965-985.

Hopmann, D. N., Elmelund-Praestekaer, C., Albaek, E., Vliegenthart, R., & de Vreese, C. H., 2012. Party media agenda-setting: How parties influence election news coverage. Party Politics , 18(2): 173-191.

Hultén, O., 2007. ‘Between Vanishing Concept and Future Model: Public Service Broadcasting in Europe on the Move’, in Meir, W. E., and Trappel, J. (eds.), Power, Performance and Politics. Media Policy in Europe . Baden-Baden: Nomos, pp. 197-222.

Ihl, O., 2002. ‘Socialisation et événements politiques’, Revue Française de Science Politique , 2-3 : 125-144.

Inglehart, R., 1977. The Silent Revolution : Changing Values and Political Styles Among Western Publics . Princeton: Princeton University Press.

Inglehart, R., 1990. Culture Shift in Advanced Industrial Society. Princeton: Princeton University Press.

Jacobson, G. C., 1983. The Politics of Congressional Elections. Boston: Little, Brown.

Jadot, A., Cautrès, B., Krouwel, A. and Vitiello, T., 2012. ‘Designing a Voting Advice Application about a presidential election: issues or candidates? Insights from the French 2012 “Vote Compass”’, paper presented at Advancing Electoral Research, ELECDEM Final Conference 28-30 June Florence/EUI.

Jenkins, R., 2006. ‘Mass Media and Third-Party Insurgency’ in Brady, H.E. and Johnston, R. (eds) Capturing Campaign Effects , pp. 356-382. Michigan: The University of Michigan 111

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Press.

Jerit, J., 2004. Survival of the Fittest: Rhetoric during the Course of an Election Campaign. Political Psychology , 25(4): 563-576.

Johnston, R., Andre B., Brady, H. E., Crete, J., 1992. Letting the People Decide: Dynamics of a Canadian Election. Kingston: McGill-Queen's Press.

Kaid, L.L., and Holtz-Bacha, C., eds., 1995. Political Advertising in Western Democracies. Parties and Candidates on Television . Thousand Oaks, CA: Sage.

Kaid, L.L., 2009. ‘ Changing and Staying the Same: Communication in Campaign 2008’, Journalism Studies , 10(3): 417-423.

Kalaycıo ğlu, E., 1999. ‘The Shaping of Party Preferences in Turkey: Coping with the Post- Cold War Era’, New Perspectives on Turkey , 20: 47-76.

Kalaycıo ğlu, E., 2005. Turkish Dynamics, Bridge Across Troubled Lands , New York: Palgrave Macmillan.

Karvonen, L., 2010. The Personalisation of Politics. A Study of Parliamentary Democracy . Colchester: ECPR Press.

Kavanagh, D., 1995. Election Campaigning. The New Marketing of Politics. Oxford: Blackwell.

Kaya, R., and Çakmur, B., 2010. ’Politics and the Mass Media in Turkey’, Turkish Studies , 11(4): 521–537. Kaymas, S.M., 2011. ’Media Policy Paradigm Shift in Turkey: Rethinking Neo-Authoritarian Media Systems in the Age of Neo Liberalism’, Acta Universitatis Danubius , 5(1): 40 - 69.

Kersting, N., (Ed.), 2012. Electronic Democracy . Leverkusen: Barbara Budrich Publisher.

Key, V.O., 1966. The Responsible Electorate: Rationality in Presidential Voting, 1936-1960 . Cambridge: Harvard University Press.

Kleinnijenhuis, J. en Krouwel, A.P.M., 2007. ‘The Nature and Influence of Party Profiling Websites’, Gepresenteerd op het politicologenetmaal 2007 te Antwerpen.

Kleinnijenhuis, J. & Krouwel, A.P.M., 2008. Simulation of decision rules for party advice websites. In F. Welsch, J.V. Carrasquero, A. Oropeza & C.B. Chen (Eds.), Proceedings of the 2nd international multi-conference on society, cybernetics and informatics (pp. 138-145). Orlando: IIIS.

Kleinnijenhuis, J., Scholten, O., Atteveldt, W.H. van, Hoof, A.M.J. van, Krouwel, A.P.M., Oegema, D., Ridder, J.A. de, Ruigrok, P.C. & Takens, J., 2007. Nederland vijfstromenland: De rol van de media en stemwijzers bij de verkiezingen van 2006. Amsterdam: Bert Bakker. 112

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Kleinsteuber, H. J., 2004. ‘Comparing Mass Communication Systems. Media Formats, Media Contents, and Media Processes’, in Esser, F. & Pfetsch, B. (Eds), Comparing Political Communication. Theories, Cases, and Challenges . Cambridge: Cambridge University Press, pp. 64-86.

Klingemann, H., Hofferbert, R.I., and Budge, I. (1994) Parties, policies, and democracy, Westview Press, Boulder (CO).

Kriesi, H., 2008. Political Mobilisation, Political Participation and the Power of Vote. West European Politics , 31(1-2), 147-168.

Krouwel, A., and Fiers, S., 2008. 'From Text to Position.Text-extraction and Authorisation of Party Positions along Dimensions of Political Competition'. Paper presented at: Voting Advice Applications between Charlatanism and Political Science, Antwerp May 16.

Krouwel, A., Vitiello, T. and Wall, M., 2012. ‘The Practicalities of Issuing Vote Advice: A new methodology for profiling and matching.’ International Journal of Electronic Governance , forthcoming.

Ladner, A., Fivaz, J., Pianzola, J. P., 2010. 'Impact of Voting Advice Applications on Voters’ Decision-Making', paper presented at Internet, Politics, Policy 2010: An Impact Assessment, Oxford Internet Institute (OII) , University of Oxford.

Ladner, A., and Pianzola, J., 2010. ‘Do Voting Advice Applications Have an Effect on Electoral Participation and Voter Turnout? Evidence from the 2007 Swiss Federal Elections’, in Tambouris, E. et al (Eds.), Electronic Participation , International Federation for Information Processing, Berlin, pp. 211–224.

Langer, J., 2003. ‘Tabloid Television and News Culture’, in Cottle S. (ed.), News, Public Relations and Power , London: Sage, pp. 135–152.

Laver, M., and Garry, J., 2000. ‘Estimating Policy Positions from Political Texts’, American Journal of Political Science, 44(3): 619 – 34.

Lazarsfeld, P.E., Berelson, B., and Gaudet, H., 1948. The People's Choice- How the Voter Makes Up His Mind in a Presidential Campaign. New York: Columbia University Press.

Lewis-Beck, M.S., and Alford, J.R., 1980. ‘Can Government Regulate Safety? The Coal Mine Example’, The American Political Science Review , 74(3): 745-756. van der Linden, C., Dufresne, Y., Hove, J., Eady, G and Loewen, P., 2012. ‘Avoiding the Traps in VEA’s. Pre-campaign Market Research and Micro-level Weighting Adjustments’, paper presented at the panel: Dangerous Toys? Assessing the Electoral Integrity of Voter Engagement Application Websites, IPSA World Congress , Madrid, July 2012

Lijphart, A., 1971. ‘Comparative Politics and the Comparative Method’, American Political 113

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Science Review , 65(3): 682-693.

Lindkvist, K., 1981. Approaches to Textual Analysis. In Rosengren, K. E. (Ed). Advances in Content Analysis (pp. 23-41). London: Sage Publications.

Louwerse, T. and Rosema, M., 2011. ‘The Design Effects of Voting Advice Applications: Comparing Methods of Calculating Results’, Paper presented at the European Consortium for Political Research (ECPR) General Conference , August 24-27 2011. Reykjavik, Iceland.

Luhmann, N., 1971. ‘Offentiche Meinung [Public Opinion]’, in N. Luhmann (ed.) Politische Planung [Political Planning], pp. 9–34. Opladen: Westdeutscher Verlag.

Lupia, A., and Philpot, T., 2005. ‘Views from Inside the Net: How Websites Affect Young Adults’ Political Interests’. The Journal of Politics , 67 (4): 1122-1142.

Luskin, R. C., 1990. ‘Explaining Political Sophistication’. Political Behavior , 12 (4): 331- 361.

Magleby, D.B. (Ed.), 2000. Getting Inside the Outside Campaign: Issue Advocacy in the 2000 Presidential Primaries . Provo/UT: Center for the Study of Elections and Democracy.

Maier, J., and Rittberger, B., 2008. ‘Shifting Europe’s Boundaries: Mass Media, Public Opinion and the Enlargement of the EU’. European Union Politics , 9 (2): 243-267.

Mair, P., 1997. Party System Change. Approaches and Interpretations. Oxford: Clarendon Press.

Manin, B., 1997. The Principles of Representative Government . Cambridge: Cambridge University Press.

Marks, G., Hooghe, L., Steenbergen, M.R., Bakker, R., 2007. ‘Crossvalidating data on party positioning on European integration’, Electoral Studies , 26(1): 23 - 38.

Marschall, S., 2005. 'Idee und Wirkung des Wahl-O-Mat'. Aus Politik und Zeitgeschichte 51- 52: 41-46.

Marschall, S., and Schmidt, C.K., 2010. 'The Impact of Voting Indicators: The Case of the German Wahl-O-Mat', in: Cederoni, L., Garzia, D. (eds.), Voting Advice Applications in Europe. The State of the Art (E-Book), pp. 61-68.

Mazzoleni, G., 1995. ‘Toward a ‘Videocracy’? Italian Political Communication at a Turning Point’, European Journal of Communication , 10(3): 291-319.

McAllister, I., 2007. The Personalisation of Politics, in: Dalton, R. J., Klingemann, H-S, (Eds), Oxford Handbook of Political Behavior . Oxford: Oxford University Press.

McQuail, D., 2005. Mass Communication Theory . London: Sage Publications, 5th Edition. 114

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

McQuail, D., 2007. ‘Revisiting Diversity as a Media Policy Goal’, in Meir, W. E., and Trappel, J. (eds.), Power, Performance and Politics. Media Policy in Europe . Baden-Baden: Nomos, pp. 41-58.

Meir, W. E., 2007. ‘National and Transnational Media Ownership Concentration in Europe: A Burden for Democracy?’, in Meir, W. E., and Trappel, J. (eds.), Power, Performance and Politics. Media Policy in Europe . Baden-Baden: Nomos, pp. 75-104.

Mikhailov, S., Laver. M. and Benoit, K., 2008. Coder reliability and misclassification in Comparative Manifesto Project Codings, Prepared for presentation at the 66th MPSA Annual National Conference , Palmer House Hilton Hotel and Towers, April 3–6, 2008.

Mughan, A., 2000. Media and the Presidentialization of Parliamentary Elections . London: Macmillan.

Negrine, R., 1996. The Communication of Politics . London: Sage.

Negroponte, N., 1995. Being Digital. New York: Knopf.

Neveu, E., 1999. ‘Politics on French Television. Toward a Renewal of Political Journalism and Debate Frames?’, European Journal of Communication , Vol. 14, N. 3, pp.379 – 409.

Norris, P., 2000. A virtuous circle: Political communications in post-industrial societies . New York: Cambridge University Press.

Norris, P., 2001. Digital Divide? Civic Engagement, Information Poverty and the Internet in Democratic Societies . New York: Cambridge University Press.

Norris, P., 2004. ‘Global Political Communication: Good Governance, Human Development, and Mass Communication’, in Esser, F. & Pfetsch, B. (Eds), Comparing Political Communication. Theories, Cases, and Challenges . Cambridge: Cambridge University Press, pp.115 - 150.

Oates, S, Owen, D. and Gibson, R.K., (Eds.), 2006. The Internet and Politics: Citizens, Voters and Activists . London: Routledge.

Özbudun, E., 2000. Contemporary Turkish Politics: Challenges to Democratic Consolidation, London: Lynne Rienner Publishers.

Page, B.I. and R.Y. Shapiro, 1992. The Rational Public: Fifty Years of Trends in Americans’ Policy Preferences . Chicago: Chicago University Press.

Panebianco, A., 1988. Political Parties: Organization and Power . Cambridge: Cambridge University Press.

Patterson, T. C., 1994. Out of Order . New York: Vintage. 115

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Patterson, T. E., and Donsbach, W., 1993. ‘Press-Party Parallelism: A Cross-National Comparison.” Paper presented at the annual meeting of the International Communication Association, Washington, DC.

Peter, J., 2004. ‘Our Long Return to the Concept of Powerful Mass Media – A Cross-national Comparative Investigation of the Effects of Consonant Media Coverage’. International Journal of Public Opinion Research , 16 (2): 144-168.

Pfetsch, B., and Esser, F., 2004. ‘Comparing Political Communication. Reorientations in a Changing World’, in Esser, F. & Pfetsch, B. (Eds), Comparing Political Communication. Theories, Cases, and Challenges . Cambridge, Cambridge University Press, pp.3 - 22.

Picard, R. G., 2004. ‘Commercialism and Newspaper Quality’, Newspaper Research Journal , Vol. 25, N. 1, pp.54 – 65.

Poguntke, T., Webb, P., (Eds.), 2005. The Presidentialization of Politics: a Comparative Study of Modern Democracies . Oxford: Oxford University Press.

Popkin, S.L., 1991. The Reasoning Voter: Communication and Persuasion in Presidential Campaigns. Chicago: Chicago University Press.

Prior, M., 2007. Post-Broadcast Democracy. How Media Choice Increases Inequality in Political Involvement and Polarizes Elections . Cambridge: Cambridge University Press.

Prior, M., 2009. ‘The Immensely Inflated News Audience: Assessing Bias in Self-Reported News’. Public Opinion Quarterly , 73 (1): 130-143.

Przeworski, A., and Teune, H., 1970. The Logic of Comparative Social Inquiry . Malabar, FL: Krieger.

Rabinowitz, G. and Macdonald, S.E., 1989. ‘A Directional Theory of Issue Voting’, The American Political Science Review , 83(1): 93 - 121.

Ramonaite, A., 2009. ‘Bringing ideology back in? The impact of electronic voting assistance tool on democratic process in a post-communist society’. Paper presented at the ECPR Joint Sessions of Workshops , Workshop 20 “Parliaments, Parties, and Politicians in Cyberspace”, 14-19 April 2009, Lisbon.

Reif, K., and Schmitt, H., 1980. ‘Nine Second-Order National Elections: A Conceptual Framework for the Analysis of European Election Results’. European Journal of Political Research , 8: 3-44.

Riffe, D., Lacy, S., and Fico, F.G., 1998. Analyzing Media Messages: Using Quantitative Content Analysis in Research . Mahwah, NJ: Lawrence Erlbaum.

Rohrschneider, R., 2002. ‘Mobilizing versus Chasing: How do Parties Target Voters in 116

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Election Campaigns?’, Electoral Studies , 21: 367-382.

Rose, R., and McAllister, I., 1986. Voters Begin to Choose: From Closed-Class to Open Elections in Britain . London: Sage Publications.

Rubin, B., and Heper, M., (Eds.) 2002. Political Parties In Turkey , London: Routledge.

Ruusuvirta, O., Rosema, M., 2009. 'Do online vote selectors influence electoral participation and the direction of the vote?', presented at: European Consortium for Political Research (ECPR) General Conference , Potsdam.

Sartori, G., 1976. Parties and Party Systems: A Framework for Analysis, Cambridge University Press, New York.

Sayarı, S., 2002. ‘The Changing Party System’, in Sayarı, S. and Esmer, Y. (eds.) Politics, Parties and Elections in Turkey , London: Lynne Riener Publishers, pp. 9-32. Sayarı, S., 2007. ‘Towards a New Turkish Party System’, Turkish Studies , 8(2): 197-210.

Scheufele, D. A., and Nisbet, M. C., 2002. ‘Being a Citizen Online: New Opportunities and Dead Ends’, Harvard Journal of Press/Politics , 7(3): 55-75.

Scheufele, D. A., and Tewksbury, D., 2007. ‘Framing, agenda setting, and priming: The evolution of three media effects models’. Journal of Communication , 57: 9-20.

Schudson, M., 1995. The Power of News . Cambridge, MA: Harvard University Press.

Semetko, H.A., 2007. ‘Political Communication’, in Dalton, R.J., and Klingemann, H-D., (Eds.), The Oxford Handbook of Political Behavior , Oxford: Oxford University Press, pp. 123-143.

Seymour-Ure, C., 1974. The Political Impact of Mass Media . London: Constable.

Siebert, F. S., Peterson, T., and Schramm, W., 1956. Four Theories of the Press . Urbana: University of Illinois Press.

Smith, A., 2009. The Internet’s Role in Campaign 2008 . Pew Internet and American Life Project, April 2009.

Stimson, J.A., 2004. Tides of Consent: How Opinion Movements Shape American Politics. Cambridge: Cambridge University Press.

Stromer-Galley, J., 2002. ‘On-line Interaction and Why Candidates Avoid It’, Journal of Communication , 50(4): 111-132.

Swanson, D.L., 2004. ‘Transnational Trends in Political Communication: Conventional Views and New Realities’, in Esser, F. & Pfetsch, B. (Eds), Comparing Political Communication. Theories, Cases, and Challenges . Cambridge: Cambridge University Press, 117

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607 pp.45 - 63.

Swanson, D.L., and Mancini, P., (Eds.) 1996. Politics, Media, and Modern Democracy. An International Study of Innovations in Electoral Campaigning and Their Consequences . Westport, London: Praeger Publishers.

Tewkesbury, D., 2003. ‘What Do Americans Really Want To Know? Tracking the Behavior of News Readers on the Internet’, Journal of Communication , 53(4): 694-710.

Tezcür, G.M., 2012. ‘Trends and Characteristics of the Turkish Party System in Light of the 2011 Elections’, Turkish Studies , 13(2): 117–134.

Trechsel, A., and Mair, P., 2011. 'When Parties (also) Position Themselves: An Introduction to the EU Profiler', Journal of Information Technology & Politics , 8: 1 - 20.

Trenaman, J., and McQuail, D., 1961. Television and the Political Image . London: Methuen.

Triga, V., Baka, A., and Figgou, L., 2012. 'Neither agree, nor disagree': A critical analysis of the middle answer category in Voting Advice Applications. Paper prepared for the Workshop “VAA Research – State of the Art and Perspectives , Schloss Mickeln, Dü sseldorf, March 2-3, 2012.

Vincent, D., 2000. The Rise of Mass Literacy: Reading and Writing in Modern Europe . Cambridge: Polity Press. de Vreese, H. C., and Semetko, H. A., 2002. ‘Cynical and Engaged: Strategic Campaign Coverage, Public Opinion, and Mobilization in a Referendum’. Communication Research , 29 (6): 615-641. de Vreese, H. C., and Semetko, H. A., 2004. ‘News Matters: Influences on the vote of the Danish 2000 Euro Referendum Campaign’. European Journal of Political Research , 43: 699- 722. de Vreese, H. C., and Boomgarden, H. A., 2006a. ‘Media Effects on Public Opinion about the Enlargement of the European Union’. Journal of Common Market Studies , 44 (2): 419-436. de Vreese, H. C., and Boomgarden, H. A., 2006b. ‘News, Political Knowledge and Participation: The Differentiate Effects of News Media Exposure on Political Knowledge and Participation’. Acta Politica , 41: 317-341.

Wagner, M., and Ruusuvirta, O., 2011. “Matching voters to parties: Voting advice applications and models of party choice.” Acta Politica , forthcoming.

Walgrave, S. and van Aelst, P. 2006. ‘The Contingency of the Mass Media's Political Agenda Setting Power. Towards A Preliminary Theory’, Journal of Communication 56(1): 88-109.

Walgrave, S., van Aelst, P., Nuytemans, M., 2008. ‘Vote Advice Applications as New 118

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607

Campaign Players? The Electoral Effects of the “Do the Vote Test” during the 2004 Regional Elections in Belgium', in Farrell, D.M., and Schmitt-Beck, R. (Eds.), Non-Party Actors in Electoral Politics. The Role of Interest Groups and Independent Citizens in Contemporary Election Campaigns , Nomos, Baden-Baden, pp. 237-257.

Walgrave, S., Nuytemans, M. and Pepermans, K., 2009. 'Voting Aid Applications and the Effect of Statement Selection', West European Politics, Vol. 32 No. 6, pp. 1161 - 1180.

Wall, M., Krouwel, A., and Vitiello, T., 2012. ‘Do Voters Follow the Recommendations of Voter Advice Application Websites? A study of the effects of Kieskompas.nl on its users’ vote choices in the 2010 Dutch legislative elections’, Party Politics , forthcoming.

Wall, M., Krouwel, A., and Kleinnijenhuis, J., 2011. ‘Comparing Voters and Parties Online: The Effects of Multi-Attribute Utility Decision Rule Section in Vote Advice Applications’, paper presented at PoliticologenEtmaal , June 9-10 2011. Amsterdam, the Netherlands.

Wall, M., Sudulich, M.L., Costello, R. and Leon, E., 2009. 'Picking your party online – An investigation of Ireland's first online voting advice application', Information Polity , 14(3): 203-218.

Ware, A., 1996. Political Parties and Party Systems , Oxford: Oxford University Press.

Webb, P., Farrell, D., and Holliday, I. (Eds.), 2002. Political Parties in Advanced Industrial Democracies . Oxford: Oxford University Press.

Wilensky, H. L., 1964. ‘The Professionalization of Everyone’, American Journal of Sociology , 70(2): 137-158.

Wlezien, C., 2005. ‘On the Salience of Political Issues: The Problem with ‘Most Important Problem’, Electoral Studies , 24(4): 555-579.

Zaller, J. R., 1996. "The Myth of Massive Media Impact Revived: New Support for a Discredited Idea." In Political Persuasion and Attitude Change, ed. Diana C. Mutz, Paul M. Sniderman, and Richard A. Brody. Ann Arbor: University of Michigan Press.

Zaller, J., 2002. ‘The Statistical of Election Studies to Detect Media Exposure Effects in Political Campaign’. Electoral Studies , 21: 297-329.

119

The ELECDEM project was funded by the FP7 People Programme ELECDEM 238607