The Personalization of Mediated Political Communication: A Review (PDF)

427802 2011 JOUXXX10.1177/1464884911427802Van Aelst et al.Journalism Article Journalism...

427802 2011 JOUXXX10.1177/1464884911427802Van Aelst et al.Journalism Article Journalism The personalization 13(2) 203­–220 © The Author(s) 2011 Reprints and permission: sagepub. of mediated political co.uk/journalsPermissions.nav DOI: 10.1177/1464884911427802 communication: A review jou.sagepub.com of concepts, operationalizations and key findings Peter Van Aelst University of Antwerp, Belgium Tamir Sheafer Hebrew University of Jerusalem, Israel James Stanyer Loughborough University, UK Abstract Personalization has become a central concept in discussions on how political news, and election coverage in particular, has changed over time. The general belief is that the focus of news coverage has shifted from parties and organizations to candidates and leaders. However, the evidence is far from conclusive. This is due in no small part to a lack of conceptual clarity and an absence of common operationalizations which are a major cause of the unclear or conflicting conclusions about the personalization of political news. This article seeks to remedy this shortcoming. It presents a model for comprehending the personalization of political news based on a review of relevant studies. The article makes a series of recommendations for how the concept might be operationalized for an analysis of media content in order to enable cross-nationally comparative research. Keywords content analysis, personalization, political news, politicians, privatization Corresponding author: Peter Van Aelst, University of Antwerp, Sint Jacobstraat 2 (M.281), 2000 Antwerp, Belgium Email: [email protected] 204 Journalism 13(2) The personalization of politics will remain a – and perhaps the – central feature of democratic politics in the twenty-first century. (McAllister, 2007: 585) Introduction In recent discussions about the changing features of politics in advanced industrial democracies personalization is considered one of the key developments (Karvonen, 2010; McAllister, 2007). The rise of ‘candidate-centered politics’ is seen as the result of two interconnected factors: first, the weakening of traditional affective ties between vot- ers and parties (Dalton et al., 2000; Mair, 2005); and, second, the changing media envi- ronment, and especially the growing role of television in political communication (Mazzoleni, 2000; Meyrowitz, 1985; Swanson and Mancini, 1996). The concept of per- sonalization is used in relation to the behavior of voters, political actors and the media (Karvonen, 2010; Rahat and Sheafer, 2007). Voters may increasingly make their elec- toral choices based on leaders or candidates, politicians may behave more as individual actors and less as members of a party, parties may put their leader at the center of their communication, and the media may represent politics as a confrontation more of indi- viduals than of collectivities. In this article we focus only on the role of that last actor, the media, and specifically the news media. Personalization has become a central concept in the discussions on how political news and election coverage have changed over time. The general impression is that the focus of news coverage has shifted from parties and organizations to candidates and leaders and that, in addition, those individual politicians are increasingly portrayed as private persons. However, these impressions are not always supported by empirical studies, leading to confusion and disagreement amongst scholars. For instance, accord- ing to McAllister (2007), there is ‘substantial evidence’ that supports the view that dur- ing campaigns leaders have become increasingly visible in the media at the expense of parties (see also Plasser and Lengauer, 2008). However, the empirical evidence across western democracies seems less supportive of the personalization thesis. Karvonen (2010) concludes on the basis of his overview that the evidence is somewhat mixed, with several studies showing weak or negative support. Kriesi (2011) finds hardly any evidence for a personalization trend, not even in Germany where others have shown that personalization is recently on the rise (see Reinemann and Wilke, 2007). According to Kriesi, only the Netherlands shows a clear personalization trend, but this again does not seem in line with other studies of the Netherlands (Kleinnijenhuis et al., 2009, Vliegenthart et al., 2010). Several reasons may account for this lack of consensus, rang- ing from differences in time periods to differences in institutional settings (Adam and Maier, 2010). In addition, it is the lack of conceptual clarity and the absence of common operationalizations which are an important cause of the unclear or conflicting conclu- sions about the personalization of political news. Solving these problems is the major goal and contribution of this study. On the basis of an extended literature review we distinguish between two forms of per- sonalization in relation to news content. The first form of personalization concerns a focus on individual politicians as central actors in the political arena, including their ideas, capac- ities and policies. This type of personalization is not necessarily in contrast to substantive Van Aelst et al. 205 political news content, but it does imply a shift in media visibility from parties to individual politicians, or from government to individual cabinet members. We therefore label this first form ‘individualization’. The second form of personalization implies a shift in media focus from the politician as occupier of a public role to the politician as a private individual, as a person distinct from their public role. We label this shift in focus ‘privatization’. It is this second dimension of personalization that has often received less attention than it might, but we would argue that it is central to the wider process of personalization. In what follows we explain these two dimensions and create four sub-dimensions that result in a comprehensive model of media personalization. Special attention is given to the different definitions and operationalizations used in previous studies and their main findings. As a next step we suggest common ways of operationalizing the concept to improve the comparability of the different dimensions of personalization for future con- tent analyses. Theoretical and conceptual foundations The personalization of political news has received attention, but so far this has not led to a common understanding of what it means, let alone how it should be operationalized and measured. That said, there seems to be consensus about three things. First, personali- zation refers to a trend, a process of change over time (Brettschneider and Gabriel, 2002; Kaase, 1994; Karvonen, 2010). The central idea is that mediated political communica- tion is more personalized than in the past. Or as Hart (1992: 68) puts it: ‘it is an argument of degree and not of kind – there is simply more of it today’. Hence most studies on personalization are longitudinal studies employing temporal comparative analysis. Second, scholars seem to agree that personalization of political news is a consequence of both media technologies and the strategies of political actors. The values embedded in television and in the privately owned media organizations are often blamed for affecting this process (Blumler and Kavanagh, 1999; Mazzoleni and Schulz, 1999). Meyrowitz, for example, argues that television has blurred the line between front and back regions in politics, creating a new ‘middle region’ which is characterized by a managed informality (1985: 47). An example might be when supposedly off-duty politicians cannot relax and make mistakes away from the prying eye of the camera. Others observe that because of its visual nature, television tends to focus on personalities rather than on abstract entities such as parties and groups (Peri, 2004). Politicians quickly adapt to these trends, creating events that emphasize personalities over parties (Sheafer, 2001; Strömbäck, 2008). In addition, some scholars also point to the influence of changes in institutions on the pres- ence of media personalization (Rahat and Sheafer, 2007). Third, there seems to be a broad consensus in the literature that the personalization of political news is multi-dimensional. However, when it comes to the number of separate dimensions and how they should be distinguished from one another there is less agree- ment. Some scholars have come up with separate labels (Langer, 2010; Rahat and Sheafer, 2007), while others distinguished between different dimensions of, or perspec- tives on, personalization (Adam and Maier, 2010; Reinemann and Wilke, 2007; Van Santen and Van Zoonen, 2009). Most authors make the basic distinction between person- alization at the expense of parties and personalization as the shifting boundaries between the public and the private. 206 Journalism 13(2) In the first form of personalization – individualization – individual politicians have become more central in media coverage, while parties and government institutions have become less relevant. From this perspective, personalization might be seen as a threat to parliamentary systems where traditionally the party, not the candidate, stood at the center of the political process (Shenhav and Sheafer, 2008). More generally, the political system is presented as the domain or battlefield of individual actors and little insight is given to more fundamental power structures (Bennett, 1996: 51). However, this trend does not necessarily imply a shift away from substantive news coverage. For instance, Oegema and Kleinnijenhuis (2000: 54) showed on the basis of their content analysis of Dutch election coverage that there is no contradiction between personalized news and attention to issues. On the contrary, news articles that featured politicians were even more about issues than news that focused on parties (see also Reinemann, 2010, for similar findings). The concern about less substantial or even ‘depoliticized’ news may also apply to the second form of personalization, privatization, which refers to the rising importance of the politician as ‘ordinary’ person. The politician is no longer presented solely as a policy maker or as a spokesperson but rather as a dedicated parent or a passionate music lover. The politician is portrayed as a private individual (Holtz-Bacha, 2004; Van Zoonen, 2000). This trend has been discussed by several academics, most notably Richard Sennett in his 1974 book The Fall of Public Man and by Joshua Meyrowitz (1985), who, as men- tioned, observes that technologies have eroded the barriers between public and private space. Hart (1992) notes that television has moved this trend of growing intimacy between voters and leaders to the next level – a trend he considers not only damaging to the substantial political debate, but potentially also a cause of growing cynicism among citizens. The closer we come to our politicians, the more we will be disappointed when they betray us. According to Thompson (2004) ongoing technological innovation has further decreased the privacy of political actors. For instance, new covert interception technologies allow others to eavesdrop on politicians’ private conversations making them publically available. To emphasize that individualization and privatization are two different forms of per- sonalization we discuss both separately and develop a model that treats them as two distinct dimensions of the same concept. Conceptual definitions and dimensions Rahat and Sheafer (2007: 67) define (media) personalization as a ‘change in the presen- tation of politics in the media, as expressed in a heightened focus on individual politi- cians and a diminished focus on parties, organizations and institutions’. This definition matches accurately our first form of personalization or individualization. Although the definition is pretty straightforward, there is some confusion about what is meant by ‘indi- vidual politicians’. Does this imply all politicians, or rather only political leaders? Does this mean all candidates participating in an election or just the main candidates?1 Both are clearly different and in some ways even contradictory as growing attention on a few leaders can be at the expense of ‘ordinary’ politicians in general (Balmas et al., 2010). Some scholars have related this focus on a select number of politicians to the concept of ‘presidentialization’ (Van Aelst, 2007; Vliegenthart et al., 2010). Presidentialization Van Aelst et al. 207 Individualization Privatization Politicians versus Politician as role occupant versus institutions politician as private individual Focus: shifts from parties to Focus: shifts from the public to the politicians personal General Concentrated Personal Personal life visibility visibility characteristics Focus: shifts to Focus: shifts Focus: shifts to Focus: shifts to private life and to individual leaders non-political personal interests politicians traits Figure 1. Dimensions of personalization in the news refers to the more central role of the prime minister in parliamentary systems, which in election times means a growing focus on the leading candidates for that position2 (Mughan, 2000; Poguntke and Webb, 2005). In countries like Germany3 and the UK, with two dominant parties, this implies a focus on the two (or three) top candidates for the position of prime minister (e.g. Kaase, 1994; Langer, 2010; Reinemann and Wilke, 2007; Schoenbach et al., 2001), while in countries with a more fragmented party land- scape like Sweden and the Netherlands it refers to the attention given to the main candi- date of each party (Johansson, 2008; Oegema and Kleinnijenhuis, 2000). Kriesi (2011) tries to operationalize this ‘concentration of personalization’ (which is different from ‘personalization in general’) using both a focus on the top 10 candidates and a more nar- row focus on the top two. Balmas and colleagues (2010) make a similar distinction between what they call ‘centralized’ and ‘decentralized’ personalization. The first implies a focus on the party’s leaders while the latter implies a more diffuse focus on all indi- vidual politicians except those party leaders. Following these recent studies, we believe it is important to distinguish between the ‘general visibility’ of politicians in the news (as opposed to parties) and ‘concentrated visibility’ on a limited number of political leaders as two separate sub-dimensions of personalization. The second type of personalization, termed privatization, is clearly more complex to define and has more diverse interpretations in previous research. Rahat and Sheafer define privatization as ‘a media focus on the personal characteristics and personal life of individual candidates’ (2007: 68). Although privatization seems the most common label to define this trend, other concepts are also used. For instance, Van Zoonen speaks of ‘intimization’, which she describes as a process whereby ‘values from the private sphere are transferred to the public sphere’ (1991: 223). Langer terms the process ‘politicization of the private persona’ as an increased media focus on personal life (family, upbringing, etc.) and personal qualities, but politicized because personal revelations cannot be divorced from the political (2010: 61, see also Langer, 2007: 379). In our conceptualiza- tion of privatization we consider the attention on ‘personal life’ on the one hand and the attention on more ‘personal characteristics’ on the other, as two distinct sub-dimensions that should be best studied separately (see Figure 1). 208 Journalism 13(2) This distinction between the personal and the political is most difficult to make when studying the characteristics of politicians. According to Adam and Maier (2010: 216) the personalization of characteristics refers to a change from features regarding their profes- sional competence and performance to features concerning personality traits related to their personal life. They admit though it is often hard to define which traits are politically relevant and which are not. Consider, for example, the following list of candidate char- acteristics studied by scholars: competence (Bean, 1993; Kinder et al., 1980; Pancer et al., 1999), leadership (Kinder, 1986; Miller and Miller, 1976), power (Bean, 1993), intelligence (Balmas and Sheafer, 2010; King, 1997), credibility and morality (Benoit and McHale, 2004; Kinder, 1994; Pancer et al., 1999), empathy (Benoit and McHale, 2004) or relations to others (Reinemann and Wilke, 2007), appearance and looks (Reinemann and Wilke, 2007; Tsfati et al., 2010), and charismatic communication skills (Sheafer, 2001). While some characteristics, like competence and leadership, are more clearly related to ‘political’ life, most others, like intelligence, reliability, and morality, can relate to both political and private aspects. Therefore, Adam and Maier (2010: 216) suggest a continuum of political and non-political traits. We argue that a pre-defined grouping may not be the best option because characteristics such as credibility and morality involve both private and ‘political’ life. Hence an article that discusses a lead- er’s credibility may be coded as discussing a political characteristic (e.g. the leader broke a campaign promise), as a personal characteristic (e.g. the leader was caught cheating on his wife), or even both (e.g. the leader cheated on his wife and lied about it to Congress). Figure 1 presents our model of media personalization inspired by a review of relevant studies. The general distinction between the two main dimensions, individualization and privatization, is made by most scholars, although some prefer to label these concepts differently. The two sub-dimensions are less consistently distinguished in the literature, probably because they are seen as less fundamental compared to the first dimension. However, we are convinced they are necessary to operationalize and empirically study the personalization of mediated political communication. The second-level categories enable us to distinguish between different aspects of individualization and privatization and improve the comparability of studies across countries and time. Please note that although it is argued above that most personal characteristics involve both public and private aspects, the personal characteristics category is placed under privatization. The reason for this is that the focus of the personalization model should be on the shift over time from political to non-political characteristics and not so much on the visibility of traits as such. Key findings In general it seems that the personalization of political news has received ample empiri- cal confirmation, but not in all countries and not on all dimensions. Note that a compari- son of these findings is sometimes problematic because of conceptual confusion (many of the studies we discuss use the general term personalization when referring to different aspects of it) and different operationalizations. Rather than stressing the many differ- ences in labeling and measuring personalization we have tried to create conceptual clar- ity by discussing some of the key findings of relevant studies according to our model. We Van Aelst et al. 209 discuss what we term individualization and privatization separately, distinguishing between the two sub-dimensions. Individualization: Increasing visibility of political leaders and other politicians Although our model differentiates between general and concentrated visibility of politi- cians, this distinction is not always made explicit in previous studies. The vast majority of earlier empirical analyses have focused, to the best of our understanding, on political leaders and not on politicians in general. We discuss the main findings of these studies and where possible differentiate between the two sub-dimensions of individualization. McAllister (2007) claims that there is ‘substantial evidence’ in favor of the (media) personalization thesis. This strong belief is mainly based on a collection of longitudinal content analyses of election coverage in five countries (1956–96).4 The visibility of can- didates (at the expense of parties) in the press grew in France, the USA, Austria and to a lesser extent in the UK. Only in Canada was a clear trend absent and the degree of per- sonalization variable. Using a similar method, Sandnes (2004) found a substantial increase of personalization in election coverage between 1953 and 1998 in the Norwegian and, to a lesser extent, the Swedish press. A study of the Belgian press between 1958 and 1999 (Van Aelst and Van Mierlo, 2003) found no signs of a trend in personalization, although a later study in 2003 did find a slightly upward trend (Van Aelst, 2007). In Israel, Rahat and Sheafer (2007), Shenhav and Sheafer (2008) and Balmas et al. (2010) found strong signs of media personalization. Other studies have not used a ratio of per- sons versus parties but rather counted the relative attention given to leaders in news coverage. These have found some evidence of individualization; for example, Ellis and Nyblade (2005) found signs of growing leader visibility in Japan. In general, scholars that have provided an overview of studies on media personaliza- tion of politics agree with McAllister but are more nuanced. Karvonen concludes on the basis of his literature review that the evidence is somewhat mixed, and that at present ‘the issue is genuinely unsettled’ (2010: 14). On the basis of their comprehensive over- view of personalization studies, Adam and Maier find clear evidence for a movement from parties and issues to people in media coverage (2010: 33). The authors, however, do mention several studies that do not fit this overall picture. The most interesting case in this respect is Germany where several studies have found little or no increase in the visibility of leaders in newspaper coverage5 (Kaase, 1994; Wilke and Reinemann, 2001), mainly because the candidates for the position of chancellor have always dominated the news to a very large extent (Schoenbach, 1996). More recently, Reinemann and Wilke (2007) reported an increase in articles on the main candidates but as a consequence of the absolute amount of campaign coverage and not in relative terms. Probably the strongest evidence against the rising visibility of politicians in the news is provided by Kriesi (2011). His comparative study of election coverage in six west European countries shows hardly any evidence of a personalization trend in recent dec- ades (1972–2007). As noted earlier, only in the Netherlands is there a clear trend toward personalization. At first sight this seems to contradict the conclusion of Kleinnijenhuis et al. (2009) that personalization in the Netherlands remained absent between 1990 and 210 Journalism 13(2) 2006. However, a closer look shows that the attention on individual politicians compared to on parties has risen significantly between the early 1970s and the mid-1990s, but has stayed pretty stable since then. As mentioned before, Kriesi is one of the few that distinguish between the visibility for the leading politicians and all politicians. According to his study, the trends for both forms of personalization correspond to a large degree. Vliegenthart et al. (2010) also provide evidence of both personalization and presidentialization in their study of Dutch and British newspapers (1990–2006). This is, however, not the case in a study of political personalization in Israel (Balmas et al., 2010). In their analysis of media coverage of politicians from 1949 to 2003, Balmas et al. show there was a clear trend of what they call ‘centralized personalization’ with the growing visibility of leading politicians, but that there was no corresponding overall trend for ‘ordinary’, or non-leading, politicians. As noted, individualization refers to a trend, a process of change over time. Several related studies, though, are based on cross-sectional analyses, comparing the visibility of (leading) politicians (versus institutions) in various areas at a certain point in time. Although these studies cannot support a temporal personalization hypothesis, they are central to improving our understanding of the phenomenon. One form of cross-sectional analysis is comparing levels of personalized politics across countries. For example, the longitudinal analysis of Rahat and Sheafer (2007) reveals a clear process of media indi- vidualization in Israel, over the period 1949 to 2003, while a similar longitudinal study of the German media by Reinemann and Wilke (2007) found little evidence of a trend, with some other aspects of personalization picking up only recently. However, if we compare these studies, the focus of the media on party leaders during campaigns was almost always higher in Germany than in Israel. Further, in the Portuguese media cover- age of the 1999 legislative elections (Salgado, 2007) this degree of concentrated visibil- ity seems to be even higher than that of Germany. Besides comparing countries at a certain moment in time, several other cross-sec- tional analyses might be relevant, such as comparing between different media. One can expect, for example, to find a greater focus on leaders in television coverage than in newspaper coverage of elections due to the nature of the medium (Salgado, 2007; Van Aelst, 2007). Another comparison might be between campaign and non-campaign peri- ods. In Belgium, contrary to their expectation, Van Aelst and De Swert (2009) found a lower ratio of general visibility (the ratio of politicians to parties) in media coverage of non-campaign periods than in campaign periods. Privatization: Personal characteristics and personal life of politicians We first present studies that focused on the shift from political to non-political or more personal characteristics of politicians and second discuss the research on the personal life of politicians in the media. Numerous studies point to the focus of media coverage of personal characteristics of politicians, yet fewer studies consist of longitudinal empirical analysis that can account for this privatization. Reinemann and Wilke (2007) present probably the most detailed and long-term analysis. Focusing on five main dimensions of candidate evaluations (competence, personality, appearance, relations to others and atti- tudes) between 1949 and 1998, they found no clear shift from purely political towards Van Aelst et al. 211 more personal characteristics but discovered a sharp increase in the appearance category (media performance, looks and rhetorical skills) in the 2002 and the 2005 elections com- pared with previous campaigns. The introduction of televised debates was seen as the main cause for this shift. Several other studies have not found any trend from purely political to more personal characteristics. Langer (2007) examined coverage of British prime ministers between 1945 and 1999 in the Times newspaper on six analogous weeks in November spread across their first three years in office (i.e. two weeks per year). She found that references to leadership qualities were relatively infrequent across the entire period. She also found, however, that there was some degree of fluctuation in the proportions, depending on the prime minister, where the more ‘presidential’ the coverage of a prime minister, the more his or her leadership qualities appeared to have been emphasized (2007: 379). Studies of the media focus on the personal lives of politicians have mainly gathered evidence from longitudinal content analyses of print media. Given the scale of the exer- cise it is not surprising that these studies have been judicious in their sampling. The find- ings that emerge are somewhat mixed. Errera analyzed coverage of French politicians’ private lives in two magazines Paris Match and VSD over a seven-year period between 1990 and 1997 (2006). She found that politicians’ relationships, personal health, their home and family life, personal financial issues and their past life were very much to the fore in the magazines’ coverag, especially of leading French politicians such as Jacques Chirac and Francois Mitterrand. In terms of newspaper articles referring to UK national leaders’ personal lives, Langer found a clear upward trend over time (2007: 383). The coverage of their private lives rose from around 1 percent of the leader’s coverage in 1945 to 8 percent during Tony Blair’s tenure in office (2007). A follow up study (period 2007–8) showed that the coverage of opposition leader David Cameron was even more focused on his private life than that of Tony Blair, while that of Prime Minister Gordon Brown was less, suggesting the importance of specific leaders for the amount of attention their private life receives (Langer, 2010). Finally, Rahat and Sheafer, who looked at elec- tion coverage in two leading Israeli newspapers for 16 campaigns between 1949 and 2003, found no significant trend in media coverage of candidates’ personal lives, with the focus on personal life never exceeding 15 percent of the news items over time (2007: 74). Very few studies, however, examine the less savory aspects of privatization. Stanyer’s study of the publicity afforded politicians’ marital infidelity in the seven democracies (Australia, France, Germany, Italy, Spain, the UK and the USA) found that exposure of politicians’ extramarital sex lives was predominately a feature in US and UK news and had increased dramatically since the 1990s (Stanyer, forthcoming; see also Thompson, 2000). The problems in terms of drawing any comparison arise not just from the sam- pling of material to analyze but also, as the next section shows, in the way the concepts are operationalized. Towards a common operationalization Besides the need for conceptual clarity, research on the personalization of political news could benefit from a harmonization in the way concepts are operationalized for content analysis. Based on a brief discussion of previous studies we give concrete suggestions 212 Journalism 13(2) for the operationalization of the following: the visibility of politicians (and parties); the visibility of the personal characteristics of politicians; and the visibility of information about the private lives of politicians. Note that we take the news article or item as the central unit of analysis, but that our approach can also be used for the coding of smaller units such as paragraphs or sentences. Visibility of political leaders and other politicians The visibility of political actors is mainly studied in two ways. The most simple one is counting the number of articles that mention one or more of the main candidates or lead- ing politicians (Langer, 2007; Wilke and Reinemann, 2001). These articles can then be presented in relation to the total campaign coverage or even the total news coverage. A second, slightly advanced approach is to take into account references to both politicians and parties. In this way one can easily gauge the attention ratio of individual politicians versus parties (or institutions such as government) (Kleinnijenhuis et al., 2009; Kriesi, 2011; Wattenberg, 1996), a ratio of leaders versus parties (Kaase, 1994), or even a ratio of leaders versus all other politicians of the same party (Johansson, 2008; Sani and Legnante, 2001; Van Aelst, 2007). Most scholars prefer to count the number of refer- ences to parties and politicians, but some prefer to code whether the main focus of the article was on a politician, party or both (Rahat and Sheafer, 2007). The study of individualization demands coding of both politicians and parties. Following Kriesi (2011) we believe that our conceptualization (Figure 1) enables researchers to quite easily distinguish between general visibility and a more ‘centralized’ or concentrated visibility that focuses on a limited number of politicians. Additionally, it allows the calculation of the relative attention paid to the party leader as a share of the attention to all other politicians of that party. This measure indicates the extent to which a political party is reduced to its leader. In the Appendix we suggest concrete variables and coding instructions. Generally there are several simple categories, each counting the number of times an actor is mentioned (e.g. ‘regular’ politicians, a leader, a party) within the article. The question of how many politicians should be included when focusing on ‘leading politicians’, or in election periods on the candidates for the ‘highest office’, depends on the political system and the number of relevant parties. Therefore, it seems unrealistic to suggest one approach that would fit all political systems equally. The characteristics of politicians Media presentation of the characteristics of politicians was studied in two main ways: first, by including a single coding category asking for criteria by which candidates are judged in evaluative statements (Reinemann and Wilke, 2007). Coders in this study were presented with a single category with several characteristics (competence, character, appearance, attitudes and relations to others) from which they were instructed to choose one. The second way of content analyzing character is by creating separate categories for each of the characteristics studied. For example, Balmas and Sheafer (2010) included in their coding book separate categories for candidates’ leadership, intelligence, reliability and morality. These characteristics have repeatedly been found as fundamentally Van Aelst et al. 213 important for establishing a political image (Benoit and McHale, 2004; Kinder et al., 1980). For each characteristic category, coders were instructed to code whether the char- acteristic was present or absent in the news item. This was repeated for each of the can- didates studied. Consistent with our categories of personalization, we recommend presenting a special category for each of the characteristics analyzed, as in Balmas and Sheafer. In order to study privatization at the level of characteristics, it is necessary to distin- guish between characteristics that relate to a politician’s political life (e.g. the leader broke a promise to one of his party members) and characteristics that relate to his or her personal life (e.g. the politician’s behavior toward his partner was presented as immoral). This is done by taking into consideration the context in which the characteristics are discussed in a news item. We define personal life in broad terms as the domain outside the political arena where the actions and words of the politician are not explicitly related to his or her public role. Of course, actions in this personal context can have far-reaching political consequences. Consequently, each characteristic analyzed in the content analy- sis has two categories and two coding options: one category codes the unit of analysis as presenting the characteristic (e.g. competence) as political (the characteristic is presented in a political context or not); and a second category codes it as presenting the character- istic as personal (the characteristic is presented in a personal context or not). This allows the identifying of a certain characteristic in one news item as being mentioned in both political and personal contexts. We know of no theoretical justification for a specific set of characteristics that must be included in any personalization study. Competence, leadership, credibility and morality are the most frequently used in public opinion surveys and hence including them in an analysis would enable findings to be compared with a larger body of knowledge. References to rhetorical skills, also referred to as verbal intelligence (Greenstein, 1995: 140), and to candidates’ appearance are very much relevant to the analysis of privatiza- tion, and hence it is recommended that they are included in any analysis. Another option is to use statistical methods to select the characteristics of interest. Reinemann (2010) has used factor analysis to decide which categories of characteristics would be analyzed in the 2009 German elections. A factor analysis of content analysis data from prior elections produced four dimensions: appearance, credibility, leadership qualities and competence. The personal life of politicians Several studies have sought to use an index of factors to track the exposure of private lives. Langer’s study focused on five elements of privatization: family, personal appear- ance, life-style (i.e. hobbies, likes/dislikes and recreational activities), upbringing and religion (2007: 381). Errera’s (2006) examination of the coverage of politicians’ private lives also used an index, but of six areas: romantic life, health, the home, family life, the past and finance. The overlap in the way the elements of the concept have been opera- tionalized by these authors points perhaps to possible directions for developing effec- tive transnational comparability. However, a combined index may be too large to be practicable when conducting multi-country studies. Therefore our proposed index includes: (1) coverage of the family, (2) past life and upbringing, (3) leisure time 214 Journalism 13(2) (hobbies, vacation) and (4) love life. All four categories are clearly personal and easy to capture. We consider this index as the minimal common ground for comparative studies, but of course scholars can easily extend the number of categories in their specific study. The coders determine the presence or absence of manifest references to four or more indicators in each article or unit of analysis. The findings for each indicator could then be combined to form a ‘private life index’. We suggest that such an index would not contain information on the personal characteristics of politicians, as we believe this is a conceptually different aspect of privatization (see Figure 1). The focus on private life might in some news stories be discussed simultaneously with the personal characteristics of politicians, but this is not necessarily always the case. Therefore, the extent to which both sub-dimensions of privatization overlap remains an empirical question. Conclusion Personalization has become a central concept in the discussions on how political news, and election coverage in particular, has changed over time. Although the scholarly inter- est in this phenomenon has grown substantially over the last two decades, the empirical evidence across western democracies is mixed, and often studies within the same country present contradictory conclusions. In this article we have argued that it is the lack of conceptual clarity and the absence of a common operationalization that are major causes of the unclear or conflicting conclusions about the personalization of political news. To address this problem we presented a conceptual model and suggested concrete operation- alizations. Both are based on the wide evaluation of studies of media personalization in many western democracies, and, so we believe, are conducive to allowing cross-national comparisons and research cumulativity. The typology presented in Figure 1 distinguishes between two main forms of person- alization in media content. The first form of personalization, termed individualization, concerns an increased focus over time on either political leaders or on politicians in gen- eral, both in relation to a focus on political parties. This type of personalization is not necessarily in contrast with substantive political news, since a focus on individual politi- cians may still occur alongside a focus on policy issues. The second form of personaliza- tion implies a shift in media focus from the politician as occupier of a public role to the politician as a private individual, as a person distinct from their public role. Following others, we labeled this shift in focus ‘privatization’. Here we distinguish between two sub- dimensional processes: first, a shift in focus from political to personal characteristics; and, second, a shift in focus from the politician’s public life in office to their private life. The operationalizations suggested here closely match the definitions of the various concepts in the model. These are user-friendly content analysis categories that were already used in various earlier studies. Our suggestion is to base any codebook on dichot- omous categories, which ask whether an object (i.e. leader, party, certain characteristic, reference to private life) is present or not within the unit of analysis. This makes the content analysis easy to conduct, increases intercoder reliability and allows for comput- erized content analysis. This article has not dealt with the institutional and behavioral aspects of personaliza- tion, which have been explored fully elsewhere (e.g. Adam and Maier, 2010); rather, it is Van Aelst et al. 215 exclusively focused on the development of a quantitative content analysis. As such it has overlooked qualitative analyses but this is not to argue that they are unimportant. They might be important, for example, for exploring aspects of privatization (e.g. politicians talking about their emotions) which otherwise might be difficult to grasp in a quantitative study. They might also be important for understanding the visualization of aspects of personality, which are not discussed and operationalized. These are often difficult to operationalize, especially in the case of privatization, but perhaps provide a further pos- sible indicator of personalization. We see this area as the next step that requires develop- ment in the examination of the personalization process. Notes 1 For a similar discussion of the personalization of voting behavior see Van Holsteyn and Andeweg (2010). 2 In the US primaries this focus on a limited number of successful candidates is called winnowing. It means that news media deliberately narrow down the number of potential candidates for the nomination (Matthews, 1978). 3 Kaase (1994: 215) admits that this focus on the two or three main candidates is open to debate as about 30 percent of other individuals appear as main actors in the news. 4 It is not a true comparative study, but rather a collection of similar studies by different schol- ars (Wattenberg: USA; Gordon: France, UK, Canada; and Prochart: Austria). The same data were presented earlier in Dalton et al. (2000). 5 Schultz and Zeh (2005) found for television news coverage in campaign periods that person- alization, measured by the attention to the two main candidates for Chancellor, did increase between 1990 and 2002. References Adam S and Maier M (2010) Personalization of politics: A critical review and agenda for research. In: Salmon C Communication Yearbook 34. London: Routledge, 213–257. Balmas M and Sheafer T (2010) Candidate image in election campaigns: Attribute agenda set- ting, affective priming, and voting behavior. International Journal of Public Opinion Research 22(2): 204–229. Balmas M, Rahat G, Sheafer T and Shenhav S (forthcoming) Two Routes to Personalized Politics: Centralized and Decentralized Personalization. Party Politics. Bean C (1993) The election influence of party leader images in Australia and New Zealand. Comparative Political Studies 26: 111–132. Bennett L (1996) News. The Politics of Illusion, 3rd edn. New York: Longman. Benoit WL and McHale JP (2004) Presidential candidates’ personal qualities: Computer content analysis. In: Hacker K Presidential Candidate Images. New York: Rowman & Littlefield, 49–63. Blumler J and Kavanagh D (1999) The third age of political communication: Influences and fea- tures. Political Communication 16: 209–230. Brettschneider F and Gabriel OW (2002) The nonpersonalization of voting behavior in Germany. In: King A Leaders’ Personalities and the Outcomes of Democratic Elections. Oxford: Oxford University Press, 127–157. Dalton RJ, McAllister I and Wattenberg MP (2000) The consequences of partisan dealignment. In: Dalton RJ and Wattenberg MP Parties without Partisans. Political Change in Advanced Industrial Democracies. Oxford: Oxford University Press, 37–63.