3. Politics always come first: Politically-driven approaches to personalisation
5.1. Methodology
Given that the main aim of this analysis is to determine the extent to which media reporting is person- and persona-centred, quantitative content analysis was a logical choice since it is the method most suited to detecting the frequency of references to certain content and identifying recurrent patterns (Berger, 2011;
Deacon, 2007). A more qualitative approach, such as discourse analysis, was considered in order to get a more nuanced understanding of the context in which leaders’ personae were mentioned. However, I decided against this, since discourse analysis is a tool better suited for revealing underlying motives and ideologies (Hesmondalgh, 2006; Smith & Bell, 2007), which was not within the scope of this analysis. Also, given that the aim of this analysis was to determine trends over time in the leaders’ media visibility, it was important to use a method which would allow certain generalizations from the data. Qualitative methods are not suited to making generalizations, while content analysis is considered to be the best quantitative textual method that allows for generalized conclusions (Berger, 2011; Hesmondalgh, 2006). In addition, two pilot studies carried out on a sample of British and Croatian newspapers confirmed that the context in which the references to persona have been made and their nuanced study can be analysed with the content analysis method. Furthermore, studies looking to establish the extent to which media focuses on individual political actors usually employ content analysis (e.g. Bjerling, 2012;
Grbeša, 2008; Langer, 2011; Rahat & Sheafer, 2007; Reinemann & Wilke, 2007;
Wilke & Reinemann, 2001), so it will be easier to compare the findings from this study with that from other countries by gathering the data through content analysis.
It should also be noted that the findings from this study are based on the manifest meaning of media text, since content analysis is not well suited to the analysis of latent, hidden meanings (Hesmondalgh, 2006). In addition, given that content analysis is limited to providing descriptive information about media texts, this study will not be able to reveal why the media reported in a particular way, i.e. reveal their intentions, nor how the audiences received the messages from the media (Berger, 2011)
The analysis covers a time frame from 1974 to 2013, providing data for all Yugoslav and Croatian political leaders who had executive power and stayed in office for more than three years. In total, the analysis includes four Presidents and six Prime Ministers. Croatia had a semi-presidential political system in the 1990s which means that executive power was divided between the President and the Prime Minister (Ilišin, 2001). Since 2000, Croatia has had a parliamentary political system and according to the Constitution the head executive is the Prime Minister.
However, the President kept some of the executive powers, such as the right to co-create foreign policy, appoint ambassadors, command army etc. (ibid.), and also, Presidents are considered the most influential and popular political figures in the society (Baylis, 2007). Therefore, I decided to look at both the semi-presidential and parliamentary system in Croatia as having “dual leaderships” in which there are differing levels of power between Presidents and Prime Ministers. In a semi-presidential system the President can be seen as having more power, but the Prime Minister is also a powerful executive figure. In a parliamentary system more formal power is given to the Prime Minister, but the President is still an important part of the executive. In short, given that both Presidents and Prime Ministers can be seen as having executive power, I decided to include them both in the analysis.
5.1.1. Material
The main material for this analysis is the national daily newspaper Večernji list, which is the only existing mass media outlet that dates back to the communist era with a full archive that is readily available. Editions of Večernji list are taken as a main source and included in analyses of coverage of every President and Prime Minister. Two other dailies are used to supplement the findings from the Večernji list.
The first one is Vjesnik which ceased to exist in April 2012 so it was not used for the
2005, so it was used only for analyses of Kosor, Josipović and Milanović. Therefore, there are at least two sources used in the analysis of every President and Prime Minister’s coverage with an aim to avoid relying on only one source, to increase the validity of findings, and enhance the ability to generalize. These three dailies represent an interesting mix of different types of daily newspapers operating in the Croatian news market. Vjesnik was established as a communist publication and was not privatized in the democratic era (Novak, 2005). Therefore, it is an excellent example of a state-owned media. Also, in the commercial media market of democratic Croatia it was considered to be the daily that was closest to being a quality paper (Jergović, 2004). Večernji list was privatized at the end of 1990s and can be seen as a representative of a daily that was transformed from a state-owned to a commercial daily (Malović, 2004; Tuđen, 2007). Also, according to its content and format, it is usually characterized as a semi-tabloid (Kanižaj, 2006). Therefore, its ownership, content and format are different from Vjesnik’s. Finally, 24sata is considered to be the only real tabloid in the Croatian market (Car & Andrijašević, 2012). It was established and is still owned by a private media conglomerate (ibid.).
Therefore, 24sata represents a third type of daily in the Croatian newspaper market:
a daily established in a democracy, by private owners, with tabloid characteristics.
Unfortunately, only in the analyses of Kosor and Josipović is the coverage of all three dailies included. In total, 392 daily newspapers were analyzed (175 editions of Večernji list, 168 editions of Vjesnik and 49 editions of 24sata).
The unit of analysis is an article, defined as a totality of words, pictures and illustrations that form an independent part of a newspaper and whose elements usually revolve around the same topic.
5.1.2. Sampling and time frame
The time frame was defined by taking into account two main elements. The first one is the availability of archives. Večernji list's archive dates back to 1959 when the daily was established, so in order to have at least one consistent source the start of the analysis had to be after 1959. Secondly, the first year of analysis of each leader’s coverage needed to be the one representative of intense political coverage, since this analysis also aims to discover whether the personalisation of media reporting is a general characteristic of political communication in these systems, or perhaps the focus on the leader and his/her personae is enhanced in intense political periods such as elections or appointments. Given that communist leader Josip Broz Tito was several times appointed President, the decision to take the 1974 appointment as a starting point for analysis was based on the fact that at
that time Tito was declared President for life, so more intense coverage than in a case of regular appointment can be expected. In addition, in 1974 there are two sources available for analysis, so the findings from the main source were able to be
“tested” against the other source what enhances the validity of findings and ability to make more generalized conclusions.
The sample included every news story that referred to each of the Presidents and Prime Ministers (either by name or post) and/or their respective parties in the week preceding their election/appointment to office, and any news story that mentioned the President/Prime Minister during two weeks in March spread across their second and third year in office.
The aim was to obtain a sample of articles that was representative of both the intense political coverage and the ‘normal’ coverage of the Presidents and Prime Ministers, and also to be as comparable as possible. Therefore, the first week of analysis is aimed to capture intense coverage and is usually the one preceding the election in which the President/Prime Minister came to power. However, in three cases (Josip Broz Tito, Nikica Valentić and Jadranka Kosor) the leader came to power by appointment, so the first analyzed week is the one preceding appointment.
I believe that is the period of the most intense coverage of the leaders and in that way comparable with campaign coverage. In order to find articles representative of
‘normal’ periods I decided to focus on March since it is the only month in which there were no elections (presidential, parliamentary, or local) throughout the examined period. Also, the focus is on either the first or the second week in March to avoid the Easter holidays. Exceptions were made in cases of Josip Broz Tito and Franjo Tuđman whose first years in office were not included in the analysis. In the case of Tito, the exception was made not to include the first years in office since the material (newspapers archive) was not available for these years (1945-1948). Instead, the year in which Tito was appointed President for the last time (1974) was chosen as the first year of analysis, for the reasons listed above. The second exception was made in relation to Franjo Tuđman who acted as the President of Croatia after 1990, but Croatia did not formally declare independence until 1991, so the 1992 presidential elections can be seen as the first formal elections in the independent state. Therefore, I decided to include the last week of the 1992 presidential campaign as representative of Tuđman’s first year in office.
Finally, the sample consists of articles mentioning the leader and/or the party in the first examined week, and only the leader in the second and third week. The ratio of leader to party mentions is usually used as an indicator of person-centred
are those mentioning their party, media reporting can be seen as more person- than party-centred. However, since the introduction of a parliamentary system in 2000, Croatian Presidents need to resign from their party membership once they are elected. Therefore, they act as party candidates in the presidential elections which are captured in the first examined week. However, in the second and third examined week they are non-party actors, and hence, there is no party whose media visibility their own media prominence might be compared to. Because of this, the decision was made that in all cases only in the first week does the sample consist of articles mentioning the leader, his/her party, or both of them, and only the leader in the second and third week.
5.1.3. Research design
The code sheet is divided into four sections. The first section was designed to capture structural details, the second section was constructed to examine the presence or absence of references to the leader and his/her party, the third section focused on the dominant topic, and context (political, private, mixed) in which a leader’s persona was mentioned, while the final section was designed to capture the extent to which specific features of a leader’s persona are visible, and the context (political, private, mixed) in which they are reported.
In the first section articles were coded for the newspapers they come from, year of publication, and week of analysis (first, second, third).
In the second section articles were coded for the presence or absence of references to the leader and their party in the first examined week, and consequently only for the presence of the reference to the leader. If an article mentioned both the leader and his/her party in the first week, it was coded for both.
The main aim of the third section was to determine the dominance of certain contents in the article. The first variable in this section coded for the main topic of the article. In order to simplify the comparison across the newspapers which have historically and comparatively different sections, only three coding values were created. One relates to the political content and includes topic such as national, international, and regional political news, economy etc.; the other relates to the non-political content, for example arts, fashion, celebrity, sports etc. The final code value was designed for articles not clearly pertaining to any of these topics. The second variable in this section coded for the pre-dominant context in which a leader’s persona was mentioned, the three main code values being political, private and political/private (mixed) context. These distinctions were based on the differences between private and political persona as suggested by Corner (2000), Van Aelst et
al. (2011) and van Zoonen and Holtz-Bacha (2000). If the features of a leader’s persona were mentioned pre-dominantly in relation to him/her as a person who is performing a political role, the political context value was to be chosen. If they were mentioned dominantly in relation to the leader as a person performing roles that are usually associated with private sphere (father, spouse, sports enthusiast etc.), the private context value was to be chosen. Finally, if the article referred equally to the features of a leader’s persona in both contexts, and/or it made explicit connections between a leader’s political and private persona, the value political/private was offered.
The final section of this coding sheet was created to examine the references to a leader’s persona in a more detailed, nuanced way. The leader’s persona was for the purposes of this study operationalized as consisting of the leader’s personal qualities, i.e. traits and skills, and his/her performances or actions in different areas of life, by drawing on the work of Corner (2000), Van Aelst et al. (2011) and van Zoonen and Holtz-Bacha (2000). In other words, features of a leader’s persona whose media visibility is researched in this study are leader’s personal traits and skills and his/her personal life. Leader’s persona was operationalized as consisting of seven personality traits (openness, extroversion, conscientiousness, niceness, emotional stability and temper, intellect and assertiveness) by drawing on the “big five” model of personality traits which was developed in psychology and lexical research with an aim to group all human personality traits in several broad factors13 (Goldberg, 1992; Hofstee, Kiers, de Raad, Goldberg, & Ostendorf, 1997; John &
Sristava, 1999; Peabody & De Raad, 2002). Also, three skills were included in the analysis - people skills, skills in data gathering and processing, and skills in dealing with things and tools variables, drawing upon Fine’s “things-data-people” skills sets typology which was created to categorize skills needed for performing different jobs (Fine & Cronshaw, 1999). The list of different life areas and activities was made by drawing upon variables used in previous similar research (C. Errera, 2006; Langer, 2006; Stanyer, 2013; Van Aelst et al., 2011) and supplemented with the category to code for references to a leader’s professional life (work), given that work and activities related to it represent a vital part of ones persona, though often neglected in personalisation research. The final personal life index consisted of ten variables:
work, youth, education, family life, love life, appearance, life-style, religion, feelings,
13 The big five model is based on the first five above mentioned variables, but two pilot studies conducted on the British and Croatian newspapers revealed that it is useful to split the original Openness and Conscientiousness variables into two, because of the large number of quite different
personal relationships, health and finances. A detailed account of which traits, skills and activities were coded under which variable is available in the Appendix 1.
Unlike in other similar studies (C. Errera, 2006; Grbeša, 2008; Langer, 2011;
Stanyer, 2013; Wilke & Reinemann, 2001), the affiliation of any of the leader’s qualities or life areas is not predetermined to be either political or private. The pilot study conducted on the sample of 285 articles from British newspapers tried out a technique proposed by Van Aelst et al. (2011) which allows for personal qualities to be coded as presented in either a political or private context. This was applied to both qualities and life areas. The pilot study revealed that not predetermining what is political and what is private provides a more accurate and nuanced picture of the mediated leader’s persona and media’s framing of political and private spheres.
However, it also showed that some cases are not easily categorized as either political or private, but they are rather a combination of both. In other words, the political and private are sometimes connected, most often the private is being politicized (e.g. he lied to his wife, so he will lie to his voters too), so a third code value was introduced in order to account for this mixed context – political/private context. In sum, every personality trait, skill and life aspect had a possibility of being coded as presented in either political, private or mixed, political/private context.
Final sample consists of 3133 articles. Inter-coder reliability test was conducted with another coder who coded 300 randomly chosen articles (9,6% of the sample). Average reliability score calculated using Holsti’s method of agreement14 across all categories was 0.96, with individual variable scores ranging from 0.82 to 1. Detailed results for each category can be found in the Appendix 2.