To link to this article:

Transcription

1 This article was downloaded by: [University of Orebro] On: 04 June 2013, At: 23:24 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: Registered office: Mortimer House, Mortimer Street, London W1T 3JH, UK Journalism Studies Publication details, including instructions for authors and subscription information: BIASED INTERROGATIONS? Mats Ekström, Göran Eriksson, Bengt Johansson & Patrik Wikström Published online: 06 Jun To cite this article: Mats Ekström, Göran Eriksson, Bengt Johansson & Patrik Wikström (2013): BIASED INTERROGATIONS?, Journalism Studies, 14:3, To link to this article: PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

2 BIASED INTERROGATIONS? A multi-methodological approach on bias in election campaign interviews Mats Ekström, Göran Eriksson, Bengt Johansson, and Patrik Wikström This study, based on Swedish data from three elections (2002, 2006 and 2010) and on a revised version of Clayman s and Heritage s conceptualization of aggressive questioning, examines bias in election campaign interviews with leading political figures. In the first part of the study, the prevalence of partisan bias is explored, and this analysis confirms that such bias does not exist. Informed by Conversation Analysis, a limited number of interviews from the 2006 election are investigated in the second part. This analysis also involves questions scripted by journalists, and it compares both quantitatively and qualitatively the differences between the manuscripts and live interaction. The results question the assumption that bias is solely related to journalistic values and actions. The level of aggressiveness in the interviews is also dependent on how the politicians manage the interview questions. KEYWORDS accountability; adversarialness; aggressiveness; bias; election campaigns; interviews; journalism; partisan journalism; politics Introduction In this article, we take as our point of departure a longitudinal study of Swedish election campaigns and analyze the phenomena of bias in live broadcast interviews. Interviews with leading politicians are, in Sweden as in many other countries, key moments in election campaigns. Policies and responsibilities, as well as the politicians performance skills and credibility, are seriously tested, and as a rule these interviews receive much attention in news media. As voters have become less loyal to political parties and more changeable in their preferences, the impact of such media events has most likely increased. In political interviews, questions frequently express or imply criticism; they are used to create challenging situations and communicative conflicts for interviewees to deal with (Bull, 2003; Clayman and Heritage, 2002; Ekström, 2009). But are all politicians subjected to an equally tough scrutiny? Or is there a bias in the way different politicians are interviewed? While bias in election campaign news coverage is an extensively researched phenomenon (D Alessio and Allen, 2000; Hopmann et al., 2012), there are very few studies on interviews (Huls and Varwijk, 2011). This article contributes to the research on bias with a particular focus on what has been conceptualized as the dimension of aggressiveness in political interviews (Clayman and Heritage, 2002; Clayman et al., 2006, 2010). Research into bias in the media is most often understood as an evaluation of the norm of impartiality in professional journalism. Journalists are supposed to be balanced Journalism Studies, 2013 Vol. 14, No. 3, , # 2013 Taylor & Francis

3 424 MATS EKSTRÖM ET AL. and not take stances on political issues. There are two main hypotheses explored in previous research. First, the hypothesis of a partisan bias reflecting the journalists political orientation, and second, the idea of a structural bias created by the way journalism operates in the process of news production. In this study these two hypotheses are examined with a focus on aggressiveness in election campaign interviews. Both hypotheses assume that journalists are responsible for differences in the level of aggressiveness, but this cannot be taken for granted. In the second part of this study this assumption is critically examined. Live interviews are governed by scripts, reflecting journalistic norms and values, but are also locally managed in the interaction between the interviewer and interviewee. It is thus possible that bias is essentially scripted, but also that the interviewee s responses might contribute to how critical the interrogation will be. In this study these alternative explanations are explored in a unique design in which we analyze and compare the two different stages in the production of election campaign interviews: the editorial production of the script, and the dynamic interaction in the live interview. Previous Research and Outline of the Study Although the terminology differs*concepts such as media bias, coverage bias, and gatekeeping bias circulate (D Alessio and Allen, 2000)*this is a well-researched area and a number of studies in Western democracies have observed bias in media reporting (D Alessio and Allen, 2000; Fico and Feldman, 2008; Gulati et al., 2004; Haug et al., 2010; Huls and Varwijk, 2011; Kaid and Strömbäck, 2008; Zeldes et al., 2008). A common distinction in the research is between partisan and structural bias (Fico and Feldman, 2008; Kaid and Strömbäck, 2008; Zeldes et al., 2008). The former is, as Zeldes et al. (2008, p. 566) state, produced by journalist political orientations subverting norms for fairness and could be based on a common societal ideology, the media owner s political view, or the dominating political ideology of the newsroom staff. The general view is that a strong partisan bias presupposes that there is a systematic tendency for the media to favor or disfavor certain parties or politicians over several elections (Kaid and Strömbäck, 2008, p. 426). Structural bias can generate the same form of favoritism in news stories as the partisan version (Zeldes et al., 2008), but it will not favor the same party or candidate over time; in other words it varies from one election to another (Kaid and Strömbäck, 2008, p. 426). This form of bias is linked to the news-room level or associated with organizational aspects such as management and economic resources (Fico and Feldman 2008; see also Zeldas et al., 2008, p. 566). Hopmann et al. (2012) single out journalistic work routines and news values as central aspects of structural bias. Editorial decisions, such as paying more attention to potential political conflicts or deeming what is of political relevance, can lead to imbalanced reporting. Related to the relevance issue is the question of incumbency bonus, which means that incumbent politicians are often considered as more newsworthy and supposed to receive more media attention for their campaign efforts (see Hopmann et al., 2011). However, this bonus does not necessarily mean more positive evaluation. Instead, studies show incumbents can suffer from more attacks compared to other candidates/parties (Esser and Hemmer, 2008; Shaefer et al., 2008). Other forms of structural bias can be related to coverage of female politicians, where research shows a harsher treatment of female politicians (Gildengil, 2008; Kroon

4 BIASED INTERROGATIONS? 425 Lundell and Ekström 2008). Research has mainly focused on news coverage but there are also a few studies of election campaign interviews. Haug et al. (2010, p. 82) studied interviews on Norwegian Public Service (NRK) and Commercial TV (TV2) in the 2005 election campaign. The study reports a significant bias in how politicians were treated, but it fails to analyze how questions are designed, which is a central aspect of interviewing. This is more carefully examined in Huls and Varwijk s (2011) study of interviews in Dutch media during the 2006 election campaign. Twelve interviews are analyzed, based on Clayman and Heritage s (2002; see also Clayman et al., 2006, 2010) innovative conceptualization of different dimensions of aggressive questioning. This study claims to uncover a bias since the questions asked of left-wing politicians were less adversarial than questions asked of politicians on the right. The general conclusion drawn by Huls and Varwijk (2011, p. 48) is that interviewers are partial in shaping the way in which politicians can present themselves to the public. The study clearly indicates that detailed analysis of aggressive questioning is a productive way of studying how negatively or positively politicians are treated. Undoubtedly, these studies have helped to develop the research on bias in election campaign interviews. However, they also have shortcomings and pose challenges for further research. First, they are based on data from a single election campaign. This means that we cannot know whether the observed bias is partisan or structural. To test the idea of systematic partisan favoring we need longitudinal data. Second, the Conversation Analytical (CA) approach to interviews applied in Huls and Varwijk (2011, p. 48) raises a critical question of how differences between interviews might be produced. A general idea in CA is that talk in interaction is dynamically produced by the participants. This means that the questions asked in an interview cannot be simply understood as an expression of journalists political values or factors concerning the journalistic process of production. Huls and Varwijk (2011, p. 57) also emphasize that an over-representation of aggressive questions in some interviews might be, at least partly, an effect of how the politicians respond. Evasive answers can motivate the journalists to ask more critical questions. In their statistical analyses Huls and Varwijk (2011) control for the type of answer and conclude that this does not explain the observed bias. However, an election campaign interview consists of basically two stages of production*the production of the script and the live interview*which to varying degrees are affected by journalistic values and strategies. In our study both of these stages are analyzed in order to deepen our knowledge of how differences in aggressiveness are created. This study consists of two interrelated parts. In the first part we test the hypothesis of partisan bias. We analyze interviews on Swedish public service television in three elections (2002, 2006 and 2010). As with Huls and Varwijk s (2011) study, this part of the study is based on a revised version of Clayman s and Heritage s (2002; Clayman et al., 2006) conceptualization of aggressive questioning (presented below), and focuses on how questions are designed. Two key questions are asked: Is there any support for the idea of a partisan (left or right) bias? Can indicators of structural bias, such as incumbent versus opposition status, or gender of the interviewee, explain differences in how politicians are treated in the interviews? In the second part we analyze more carefully a limited number of interviews. Informed by CA, we focus on the relationship between the script and the dynamic interplay between journalists and politicians in the live interaction. As will be shown, the interviews are rigorously planned and anchored in independent journalistic practices. But

5 426 MATS EKSTRÖM ET AL. the script has to be performed in live interaction. The overall research question is to what extent the level of aggressiveness in the interviews reflects editorial planning, or turn-byturn activities of the interviews? This part of the study is delimited to the 2006 election. The analysis involves a quantitative comparison of question design (levels of aggressiveness) in scripts and live interviews, and also an analysis of the interaction on a micro level focusing especially on politicians answers and journalists follow-up questions. Data and Method In Swedish public service television, the tradition of party leader interviews during election campaigns is an institution in itself. It goes back to the 1960s and is associated with high claims of impartiality. The format has changed to some extent, but during the three election campaigns studied in this article (2002, 2006 and 2010) the interviews were organized in essentially the same way. Each program is approximately 50 minutes long. It is a live studio interview with one party leader who sits opposite two interviewers at a desk. Beside the main interview, there are short sequences of questions from the public (in the form of a studio audience, edited segments, or animated questions from the Web). This study examines the questions asked by the journalists, which is the main part of the programs. In order to contextualize the present study, some notions concerning the political context should be mentioned. Supported in Parliament by the Left Party and the Green Party, the Social Democrats formed a minority government in Sweden between 1994 and At the time of the 2002 campaign the Left Party, with their leader Gudrun Schyman, played an active role in negotiations with the Social Democratic government. Even though the Social Democratic party lost its mandate in Parliament, Prime Minister Göran Persson managed to remain in power. In the 2006 election a coalition of four right-wing parties called the Alliance won the election, and the leader of the Conservative Party, Fredrik Reinfeldt, obtained the position of Prime Minister. The period between the elections of 2006 and 2010 is characterized by dramatic changes in public opinion. As a response to the Alliance coalition, the Social Democrats, the Left Party, and the Green Party formed a coalition and were more than 15 percent ahead in the opinion polls two years before the election. But the gap between the challengers and the incumbent parties diminished, and on Election Day the right-wing Alliance won their second election (Table 1). In this study we analyze interviews with four party leaders in three election campaigns. We have chosen to focus on the interviews with the two main challengers for the post of prime minister, and party leaders for two major coalition parties in the government and the opposition. All interviews were taped and transcribed (based on CA conventions). More specifically, the data for the longitudinal part of the study consists of TABLE 1 Party leaders as incumbent and challengers, 2002, 2006 and Social Democrats (Persson/Sahlin) Incumbent Incumbent Challenger Left Party (Schyman/Ohly) Challenger Challenger Challenger Conservatives (Lundgren/Reinfeldt) Challenger Challenger Incumbent Centre Party (Olofsson) Challenger Challenger Incumbent

6 BIASED INTERROGATIONS? 427 approximately 2050 question and answer sequences from 12 interviews. The second part is based on the scripts and transcribed live interviews from the 2006 election, as well as news room studies conducted by a colleague (Kroon Lundell, 2009). Bias is in this study analyzed in terms of the level of aggressive interviewing. The conceptualization and coding scheme were originally developed by Clayman and Heritage (Clayman et al., 2006, 2010), and based on research within the tradition of CA. It aims at capturing the degree of aggressiveness in the journalists question turns, and is organized around five dimensions: adversarialness, accountability, assertiveness, directness and initiative (Clayman et al., 2010, pp. 2323). In this study we exclude the last two dimensions. Initiative is a less relevant dimension (measuring question complexity and the number of follow-up questions) when the journalist does not have to compete for the floor, as in the US presidential press conferences. Directness, which concerns whether questions are deferential (containing e.g. self-referencing frames) or more direct, is also excluded as such frames are almost non-existent in this context. In line with Huls and Varwijk (2011), we added the dimension of interruptions in order to grasp another aspect of journalistic aggressiveness. Below we briefly describe the four dimensions and the indicators used (the first three based on Clayman et al., 2006, 2010). Adversarialness concerns the extent to which a journalist s question suggests an oppositional stance to the interviewee or their party. Oppositional is operationalized as statements that are critical of the politician or their party; that suggest that the politician s or the party s policies are unfair or have failed; that claim splits or disagreements within the party or the government; or that maintain that there are contradictions between the politicians words and/or deeds. Taking such a stance could be confined to the preface to a question only, or it could be at a more global level and concern the question as a whole. In this study we use the global aspect as a measure of adversarialness. Accountability refers to the extent to which the question asks the politician to explicitly justify or defend policies or actions. The first, and less hostile form, are questions that are constructed in the form of, Why did you do X? The second and more hostile forms are the How could you? type of questions (formulations which have direct parallels in Swedish). Such questions are aggressive since they do not accept politicians statements or policies at face value and openly ask for explanations. We have also coded a third form. It is used when the questions are constructed as a hybrid form of the two above, or when the question is holding the politician accountable in other ways, sometimes in an indirect way ( Isn t it strange that a government which claims to work for equality administers inequality? ). These three forms are integrated into one measure. Assertiveness concerns the extent to which questions invite or push for a certain answer. In order to facilitate the coding of this construct, it is delimited to yes/no responses. Yes/no questions can push for specific answers in at least two ways. One aspect is the preface tilt, in other words whether the preface to a question slants the question towards certain answers. Second, questions can be negatively formulated. Such questions start with constructions as isn t it or aren t you or don t you think (the formulations have direct parallels in Swedish). This example includes both aspects: An investigation shows that 122,000 young persons were unemployed in June this year; isn t that an awful lot of youths that haven t got a job? In the present study we have used the latter indicator (negatively formulated questions). Interruption has been related to power and dominance in interaction (Hutchby, 1996; Thornborrow, 2002). It is an action which can be used by interviewers to stop the

7 428 MATS EKSTRÖM ET AL. interviewees from continuing what they are doing, to limit their access to the floor and the possibilities for them to influence the agenda, and also to indicate that the answer in process is unsatisfactory (Bull and Mayer, 1988; Ekström and Bérczes, 2008). This makes interruption a central aspect of aggressiveness in political interviewing. Interruption is a rather complex phenomenon and all overlaps in conversation are not interruptions. The definition used in this article is based on Schegloff s (1987, p. 214) description of interruption as talk by another when a prior speaker is still talking and is not in the vicinity of possible turn completion. For it to be an interruption it also has to be clear that the overlap is oriented towards taking over the turn (and is not only, for example, a continuer). The coding is undertaken by three of the researchers in the group, on the basis of detailed instructions for each variable. After the coding of all interviews, the reliability was tested by the researchers shifting and coding each other s data. Relatively low reliability was identified in the measure of accountability which led to clarifications in the coding instruction and a recoding of all interviews. Partisan Bias in Election Campaign Interviews? The first question to address is whether partisan bias can be found in the Swedish election campaign interviews. In Table 2 we present results showing the extent to which dimensions of question aggressiveness can be found systematically in the elections of 2002, 2006 and By comparing election interviews of party leaders of parties which are ideologically to the left (Social Democrats and the Left Party) and the right (Conservatives and the Centre Party) there is the possibility of testing whether there is partisan bias in interviews. If partisan bias appears in this way there should thus be a clear difference between questioning left-wing and right-wing party leaders, visible in the last columns of Table 2, where aggressive questioning in the four dimensions is summarized. However, the results do not support such an interpretation. In fact, they reveal the opposite. On a general level, party leaders with an ideological left/right affiliation, are treated equally in the election campaign interviews over the period of time as a whole. TABLE 2 Aggressive questioning in election campaign interviews by year and partisanship (%) Total L R L R L R L R Adversarialness 55 (212) 42 (148) 47 (160) 44 (127) 52 (141) 75 (194) 51 (513) 52 (469) Cramer s V 0.12** *** 0.01 Accountability 6 (22) 5 (19) 8 (26) 10 (30) 20 (53) 12 (31) 10 (101) 9 (80) Cramer s V * 0.02 Assertiveness 3 (2) 4 (3) 23 (30) 25 (26) 21 (20) 15 (11) 17 (52) 16 (40) Cramer s V Interruption 40 (184) 37 (138) 28 (105) 29 (89) 17 (47) 22 (58) 30 (336) 30 (285) Cramer s V N is given in parentheses. L, left-wing parties; R, right-wing parties. *pb0.05, **pb0.01, ***pb0.001.

8 BIASED INTERROGATIONS? 429 There are, however, differences in treatment of party leaders when comparing the aggressive questioning of different election campaigns. Adversarialness was more pronounced in interviews with left-wing than right-wing party leaders in 2002 (55 percent versus 42 percent). On the other hand, we find the opposite situation in 2010 when rightwing party leaders receive more criticism in the questions (75 percent versus 52 percent). Other significant differences found were the share of both accountability and assertiveness, where the left-wing party leaders were asked significantly more such questions in But this was not the case in 2002 and On the other hand, the leaders of parties with a right-wing ideology received a larger number of interruptions during their interviews in This is also not consistent over time. The conclusion of this first part of the analysis opposes the suggestion of a partisan bias. Treatment of party leaders in election campaign interviews may vary between elections, but if so the treatment is not systematic and based on the partisanship of the interviewee. Even if partisan bias is not evident, other kinds of factors associated with the interviewee might cause bias in aggressive questioning. In the next stage, we investigate whether factors indicating structural bias can explain differences in how politicians are treated in the interviews. One question is whether female politicians are more aggressively treated in election campaign interviews than their male counterparts. By comparing bias in relation to whether the party leaders are the main candidates for the prime ministerial position and if the party leader is an incumbent or challenger, structural premises of opinion strength and political power can be tested against the chosen indicators of aggressive questioning. To control for the context of the interview, election year is included in the model in Table 3. To make sure there are no suppressed relationships, political affiliation of the interviewee is taken into consideration as well. Table 3 reveals a number of interesting findings. First, as the previous analyses demonstrated and as confirmed by the multivariate model, there is no evidence of partisan bias. Moving on to the question of whether female politicians are treated differently to male politicians, the answer is yes, but not in the assumed direction. Adversarialness is more prevalent against male politicians (b0.42) than female politicians; as is the case with accountability questions, where politicians are asked to explicitly justify or defend policies or actions (b0.50). But other factors help to explain variance in aggressive questioning as well. Party leaders who are not the main candidates for the prime minister post tend to receive more criticism (b0.36), but on the other TABLE 3 Multiple regression predicting bias in election campaign journalism (b, logistic regression) Adversarialness Accountability Assertiveness Interruption Gender (male) 0.42*** 0.50*** Prime minister candidate 0.36*** 0.34* Incumbent 0.61*** Year (2010) 0.44*** 1.24*** 1.96*** 0.98*** Year (2006) * 2.23*** 0.47*** Ideological position (Right) R N *pb0.05, **pb0.01, ***pb0.001.

9 430 MATS EKSTRÖM ET AL. hand, the main candidates face more accountability questions (b0.34). Another factor explaining adversarialness appears to be whether the party leader is incumbent or challenger. Incumbents addressed significantly higher shares of questions which included criticism (b0.61). However, even if these factors can better explain differences in treatment during election campaign interviews than partisanship, the most significant factor seems to be election year. When interviews in the election campaign of 2010 are compared with the previous campaigns, all measures of question aggressiveness are significant, but it is not a linear trend. Adversarialness, accountability questions, and assertiveness are more prevalent in 2010, but on the other hand, we find fewer interruptions in campaign interviews in 2010 compared to previous elections. The explanatory power of the models, using R 2 as a measure, varies between 4 and 12 percent, depending on the dimension analyzed. Which level of explained variance can be tolerated is hard to determine, since it always has to be considered in relation to what is expected and numbers of independent variables included in the model. Nevertheless, the levels of R 2 clearly indicate the significance of other factors explaining levels of aggressive questioning in election campaign interviews. But even if there might be other such factors explaining different treatment of party leaders in election campaign interviews, the situational context can be important as well. This can be illustrated by comparing the level of aggressive questioning of individual party leaders. The results in Table 4 show differences indicating the importance of the situational context of the election campaign interviews. There are a number of clear differences between the interviews. The percentages of adversarialness vary from 38 to 62 percent in 2002, and from 46 to 85 percent in Party leaders who participated in more than one election campaign are treated differently depending on the year of the election campaign. The only party leader who was interviewed in all three election campaigns, the Centre Party leader Maud Olofsson, faced different levels of adversiarialness, accountability questions and assertiveness in 2002, 2006 and Only when comparing levels of interruption is the questioning equal in the TABLE 4 Bias in election campaign interviews in 2002, 2006 and 2010 by party leader (%) Adversarialness Accountability Assertiveness Interruption 2002 Social Democrats (Persson) 46 (83) 8 (15) 0 (0) 33 (69) Left Party (Schyman) 62 (129) 3 (7) 6 (2) 47 (115) Conservatives (Lundgren) 47 (82) 7 (12) 6 (2) 42 (77) Centre Party (Olofsson) 38 (66) 4 (7) 2 (1) 32 (61) 2006 Social Democrats (Persson) 54 (79) 8 (12) 18 (10) 30 (50) Left Party (Ohly) 42 (81) 7 (14) 27 (20) 27 (55) Conservatives (Reinfeldt) 46 (64) 12 (17) 23 (12) 26 (52) Centre Party (Olofsson) 41 (63) 8 (13) 28 (14) 31 (52) 2010 Social Democrats (Sahlin) 46 (53) 22 (24) 23 (8) 22 (26) Left Party (Ohly) 56 (88) 19 (29) 20 (12) 13 (21) Conservatives (Reinfeldt) 63 (75) 10 (12) 8 (3) 14 (17) Centre Party (Olofsson) 85 (119) 14 (19) 24 (8) 28 (41) N is given in parentheses.

10 BIASED INTERROGATIONS? 431 election campaigns analyzed. Of the party leaders participating in two of the three included campaigns, we find the same pattern. Levels of aggressive questioning change from one election campaign to another. The only party leader who was treated approximately the same in two consecutive election campaign interviews was the leader of the Social Democrats, Göran Persson (2002 and 2006). This indicates that situational factors are important in understanding and explaining aggressiveness in election campaign interviews. Bias in Script and Live Interaction: A Comparison To deepen the analysis we will here focus on situational factors related to two stages in the production of the interviews; the production of the script and the live interview. Two questions are explored. First, is the level of aggressiveness and the difference between the interviews manifest in the script, and in that sense planned in advance, or is it primarily an outcome of the interaction in the live interview? Second, are there any specific interviewee actions that seem to prompt the journalists to add critical questions or comments which are not in the script? The answer to these questions will give us a deeper understanding of how the level of aggressiveness is related to editorial work on scripting or to the turn-by-turn dynamic of live interviews. The election campaign interviews are carefully prepared and rigorously scripted (Kroon Lundell, 2009). Scripts are part of the general process of formatting in broadcast production (Ytreberg, 2004). The script is regarded by the editors as crucial to the accomplishment of a critical evaluation of the politicians, and at the same time crucial to the avoidance of partisan journalism. Much work is devoted to the design of questions that can fulfill such journalistic norms and goals. Based on interviews with editors and hosts in the election campaign interviews from 2006, Kroon Lundell (2009) describes the extraordinary amount of planning of the interview questions. In the completed script, questions are formulated in a conversational style, divided into topic-based sections, and marked for one of the two interviewers. The fact that the questions are deeply anchored in journalistic ideas and editorial practices makes them a particular test of biased journalism. Even though questions are planned in detail, the interviews are never a copy of the script. The interview has to be talked into being. From previous research we know that an interview can develop into a more hostile confrontational interaction, partly as a result of how the interviewee answers the questions (Schegloff, 19889; see also Clayman and Heritage, 2002). Questions can of course depart from the manuscript in several ways. We can roughly distinguish two types. First, there are what Kroon Lundell (2009, p. 283) describes as micro-level changes. These are changes in the turn design (wordings are altered, questions are shortened and so on), in order to integrate the questions into the live dialogue. Second, there are changes in the organization of the interview, and what questions are actually asked. In this analysis we will focus on the added interviewer turns. A comparison of the script and the live interview shows that the interviews appear, at first sight, to follow the script fairly closely. Most questions in the script are asked, and with very few exceptions, in the order planned. However, as Table 5 shows, the live interviews contain more questions than the manuscripts. All the politicians answered considerably more questions than planned beforehand. The leader of the Left Party (Ohly) received the most unscripted questions, with 79

11 432 MATS EKSTRÖM ET AL. TABLE 5 Number of questions in manuscript and live interviews, 2006 Manuscript Live interview Change (%) Social Democrats (Persson) Left Party (Ohly) Conservatives (Reinfeldt) Centre Party (Olofsson) questions more in the live interaction than planned, whereas the leader of the Conservative Party (Reinfeldt) received the smallest number of unscripted questions, with 31 more questions in the live interviews than in the manuscript. The interviews are obviously unforeseeable events and progress differently. In Table 6 the differences between the interviewers scripted and performed questions in relationship to the political actors are scrutinized further. Here we delimit ourselves to investigate the dimensions of adversarialness and accountability. Concerning the latter dimension, the number of accountability questions in the manuscripts is equally distributed between the two main candidates for the prime ministerial post, Persson and Reinfeldt (16 each) and between the politicians representing the minor parties (12 for Ohly and 11 for Olofsson). A similar pattern can be observed in live interviews. Rather few accountability questions emerged in them. Actually, Persson received fewer such questions than was planned, while the other three politicians received one (Reinfeldt) or two (Ohly and Olofsson) more questions than were found in the manuscripts. Thus, when it comes to the accountability dimension, both the manuscripts and the live interviews treated the interviewees fairly equally. The other dimension, adversarialness, suggests something else. Looking at the main candidates for the post of prime minister, there are interesting differences between the two. In the manuscripts, the journalist planned 42 adversarial questions for Reinfeldt (39 percent of the total number of scripted questions), and 28 for Persson (26 percent of the total number). A reasonable conclusion is that the manuscripts tended to favor Persson. TABLE 6 Number of adversarialness and accountability questions in manuscript and live interviews, 2006 Manuscript Live interview Adversarialness Accountability Adversarialness Accountability Social Democrats (Persson) 28 (26) 16 (15) 79 (54) 12 (8) Left Party (Ohly) 31 (27) 12 (10) 81 (42) 14 (7) Conservatives 42 (39) 16 (15) 64 (46) 17 (12) (Reinfeldt) Centre Party (Olofsson) 35 (31) 11 (10) 63 (41) 13 (8) Percentages in parentheses.

12 BIASED INTERROGATIONS? 433 In general, the numbers of adversarial questions are clearly higher in the live interviews than in the manuscripts. The most significant shift towards adversarialness takes place in the interview with the Prime Minister Göran Persson. While it was planned that the leader of the opposition, Fredrik Reinfeldt, would receive the most adversarial questions in the script, the situation was the reversed in the live interview (79 for the Prime Minister and 64 for the opposition leader). To conclude, the data clearly indicate a general increase in adversarialness from script to live talk. The change, however, differs between the interviews. Clearly something of importance happens when the script is talked into being. The journalists apparently found reasons to pursue Persson more thoroughly with unscripted adversarial follow-up questions. As this is data from only one election campaign and four interviews we should be careful not to draw conclusions which are too general from these observations. Nevertheless, they clearly indicate the importance of analyzing, in more detail, what is going on within the live interview in order to understand what it is that drives some interviews into a more aggressive interaction. The Live Interview and the Dynamic of Adversarial Follow-up Questions In the four interviews we have analyzed, the added unscripted interviewer turns were in almost all cases produced as follow-up turns. A follow-up is defined here as a question or statement which clearly reacts to an answer, and does not bring up an entirely new topic (cf. Smith, 1990, pp. 1067). Follow-ups can take the form of, for example, a questioning of the answer, or an invitation to clarify or elaborate a topic (Eriksson, 2011; Greatbatch, 1986, p. 94). As Eriksson (2011, p. 3333) notes, a follow-up in one way or another treats the preceding answer as inadequate or insufficient. In order to illustrate how unscripted actions in the interaction can generate an extension of adversarialness, we will look into a specific form of interviewee response, what we call a challenging of the question. In previous research two types of challenges have been distinguished: the announced refusal to answer, and the challenging of the question (Ekström, 2009; Harris, 1991). In the latter, which will be analyzed here, the interviewee questions propositions or assumptions in the question. It is a form of response which challenges the power of the question, and indirectly also the interviewer and his project. Of course, the specific form of the responses does not explain all changes in the level of adversarialness from script to live interview. It is not our intention here to analyze the various reasons why certain interviews developed into a more adversarial direction, but to explore a common pattern in the data. In view of its challenging aspect, this response might have a particular potentiality to enhance the adversarialness of the interview. The analysis shows that the interview with the highest increase in adversarialness (that with the Prime Minister, Persson) also contained the most frequent use of challenging responses. Not all challenging interviewee responses are followed by an unscripted and adversarial follow up. However, in the interview with the Prime Minister there were a number of such situations which clearly contributed to the increase in adversarialness. In this interview, the adversarial unscripted follow-up questions were in most cases preceded by a response turn including a challenging of the question. This will here be illustrated in a more detailed analysis of three examples.

13 434 MATS EKSTRÖM ET AL. Example 1: From a scripted non-adversarial question to an adversarial follow-up 1 P ((...)) we can t put ourselves in that 2 situation [that electricity prices wipe out basic industry 3 J [.h but is it- could it be time Göran 4Persson to re-regulate the market? 5 P.h I ve actually said that a number of 6 times (0.3) a:h and a:h 7 J how many times are you going to say it 8 before it comes true then? The first question in this example (line 3) derives from the script. The wording is slightly changed and it is produced in overlap with the politician s answer. The second question (line 7) is, however, not in the script. The question is designed as a follow-up and clearly related to the politician s answer (line 5). More specifically, the journalist deals with a situation where the politician (at least implicitly) has questioned the question. The politician treats the question as if it is, to some extent, inappropriate. The actually (Swedish ju) in line 5 appeals to shared knowledge, and in this context it might indicates that this is something that the journalist should know (cf. Heinemann et al., 2011). In the follow-up, the interviewer regains the initiative. The interviewer uses the politician s answer as an opportunity to formulate a rather critical follow-up which indicates that the politician should act instead of talking. This alludes to a general popular understanding of politicians as a category of people who talk too much and do too little. The sequence develops from a non-adversarial question (line 3) to an adversarial follow-up (line 7), and what makes this follow-up relevant is the politician s answer. In our second example, the opening question (line 1) asks the interviewee to explain something that can reasonably be understood as a failure of the policies that the interviewee is responsible for. It is a typical adversarial and accountability question. The question was set out in the script with exactly the same wording. The interview, however, develops beyond the script, and into an argumentation which extends the adversarialness. Example 2: From a scripted adversarial question to off-script adversarial argumentation 1 J but you ve been Prime Minister for- for ten years 2 The Social Democrats have ruled for twelve years 3 why have the politics that have been conducted not 4been able to reduce the differences 5 P oh yes we ve reduced the differences 6 you ll see it too if you look 7 at those statistics 8.h beginning in two thousand the differences in Sweden 9 start declining 10 J m:: but this development is at 11 quite a standstill now [when it comes to poor people] 12 P [no:: it s also head-] 13 it s also headed in the right direction 14[and it 15 J [but slowly [slowly 16 P [yeah but the fact is that it 17 doesn t happen so very quickly ((...)) Doing what the question asks him to do (deliver an explanation) would mean that the politician in this case also accepts the failure implied in the question turn. In his response turn (line 5) the politician instead questions the assertion built into the question,

14 BIASED INTERROGATIONS? 435 and claims that the journalist is wrong. The response can be described as a counterassertion. In this situation the journalist goes off-script and delivers another counterassertion (line 10). The interview develops into argumentation, which continues for a few turns after what is shown in the example 2 extract. As Montgomery (2011) shows, sequences of assertions and counter-assertions are rather common in adversarial political interviews. This also applies to the interviews in this study, but in the script all planned actions are designed as questions. There are no arguments. Although the journalists could probably predict that the interviewee would not accept all the statements which they have built into the questions, there are no attempts to anticipate this in the script. The sequences of argumentation develop dynamically in the situation. With regard to the degree of adversarialness, these situations, on the one hand, expand the number of turns oriented towards criticism. On the other hand, the criticism is countered by the politicians. In the third example as well, the question at the end of the journalist s turn (line 7) is almost identical to that posed in the script, but the preface is slightly changed. The journalist implies a conflict between the politician s way of life and what he claims to represent. It is a yes/no question and the interviewee answers it, at least implicitly, with a yes. In her next turn (line 15), which is not in the script, the interviewer questions the answer. The first question was designed with the presumption of a no answer. By referring to how it has actually always been and describing the politician s home as a multimillion-crown house, the preface of the question clearly indicates that it is not possible to live that way and be rooted in the popular (line 7). However, the politician does not admit that his way of life threatens his relations with ordinary people. At the same time he questions the criticism implied in the question. In this situation the interviewer (line 15) delivers an assertion in which the criticism is more clearly stated. Example 3: Implied criticism in script is expressed more clearly in a follow-up argument 1 J but it s actually always been 2 some kind of social democratic tradition, 3 that one should live (.).h as close 4to the people as possible sort of I mean 5 now you ve built a a 6 multimillion-crown house.h ah wh- (.) 7 Is there- is it possible to be rooted in the popular 8 and still live that way? 9 P.hh I don t think the popular 10 association is in any way 11 rooted in (.)the house(.).h but is rather a:h 12 rooted in the way one approaches his 13 task and the contacts one ha:s 14and (.) [in the company one keeps 15 J [but at any rate they re conditions that 16 greatly deviate from those of normal people The examples above illustrate how the politician contributes to the aggressiveness of the interview by challenging the questions and opening up the scope for a critical follow-up. The journalists go off-script and develop their projects of critical interviewing when they are threatened. But what makes the challenge relevant is the scripted adversarial question in the first place. Actions in the interaction do not determine the next action, but they do make a particular response relevant and even expected. A challenging

15 436 MATS EKSTRÖM ET AL. response is not the only possible way to deal with the question, as the adversarial followup is not the only way to deal with a challenging response. In other situations the journalist, for example, does not respond to the challenge of the question but instead asks the next question in the script. The level of adversarialness is partly locally managed and interactionally produced. Conclusion Live interviews in various formats constitute an important part of mediated politics (Bull, 2003; Clayman and Heritage, 2002; Ekström and Patrona, 2011). Although we know that politicians can be treated in different ways, with varying degrees of critical and tough questioning, the research on bias has devoted comparatively little attention to interviews (Huls and Varwijk, 2011). In this study we have suggested a multi-methodological approach on bias in political interviews, based on a conceptualization of aggressive questioning developed by Clayman and Heritage (Clayman et al., 2006). In agreement with the research on news coverage (e.g. Kaid and Strömbäck, 2008), we emphasize that hypotheses of partisan bias can only be tested in a longitudinal design. However, in order to understand how bias is created in interviews, we must also take into account that interviews are both scripted, as the result of journalistic norms and practices, and formed in interaction between the participants. We have illustrated the significance of an approach covering these practices. Through a longitudinal approach, the study shows that there is no partisan bias* measured in terms of aggressive questioning*in the Swedish election campaign interviews. There is no indication that the journalists over time systematically favor a political party or ideological orientation. The results are in line with the majority of the previous research on news coverage. In the ongoing discussions about journalism and its future, critics claim an increased politicization, that is, that the media, and particularly the commercial side, take a clear stance and give voice to certain political interests. In the light of this discussion, the election campaign interviews broadcast by the Swedish public service company still appear to be a vital, and well-functioning, form of scrutiny of political candidates. The journalists perform their watchdog role independently. This is not to say that bias does not exist. Our analyses also show that different levels of aggressive questioning occur in the interviews; some party leaders are treated more favorably than others, but this differs from election to election. Some of these differences can be explained by other factors such as being a main candidate for the prime minister post or being an incumbent. But altogether, variance in aggressive questioning can only be explained by such factors to a limited extent. Previous research links bias to structural and organizational aspects of journalistic work. In this study, interview scripts are analyzed as an indicator of journalistic norms and objectives. The scripts are rigorously prepared and each question has been considered carefully. The empirical comparison of the scripts and the interaction in the live interview clearly indicates that both the structured aspects of journalism and the situational aspects of the interview are important to consider in analyses of bias. Significant changes in the degree of aggressiveness from script to live interviews are shown. More important, there are differences in how the live interviews with various politicians developed. For example, the leader of the opposition who was supposed to receive the highest share of adversarial

16 BIASED INTERROGATIONS? 437 questions*according to the script*ended up facing lower shares of such questions than the Prime Minister. As interviews are locally managed and produced in the dynamic interplay between interviewer and interviewee, it is not surprising that the politicians way of answering can influence the degree of aggressiveness in the interviewing. Responses challenging the question are in this study identified as a common promoter of extended adversarialness. These results question the general assumption that journalism is in all cases responsible for the prevailing bias. However, it is an important topic for future research to develop the analysis of how adversarialness can be heightened or declined in political interviews. More generally, this study clearly shows that there are no simple explanations to the fact that some politicians are treated harder in interviews during an election campaign. Both structural and more situational factors are important to consider. This is also an argument for research combining quantitative analyses of longitudinal trends, ethnographic research on journalistic editorial practices, and detailed analyses of live interaction. REFERENCES BULL, PETER (2003) The Microanalysis of Political Communication, London: Routledge. BULL, PETER and MAYER, KATE (1988) Interruptions in Political Interviews: a study of Margaret Thatcher and Neil Kinnock, Journal of Language and Social Psychology 7, pp CLAYMAN, STEVEN, ELLIOT, MARK, HERITAGE, JOHN and BECKETT, MEGAN (2010) A Watershed in White House Journalism: explaining the post-1968 rise of aggressive presidential news, Political Communication 27, pp CLAYMAN, STEVEN, ELLIOT, MARK, HERITAGE, JOHN and MCDONALD, LAURIE (2006) Historical Trends in Questioning Presidents , Presidential Studies Quarterly 36, pp CLAYMAN, STEVEN and HERITAGE, JOHN (2002) The News Interview. Journalists and public figures on the air, Cambridge: Cambridge University Press. D ALESSIO, DAVE and ALLEN, MIKE (2000) Media Bias in Presidential Elections: a meta analysis, Journal of Communication 50, pp EKSTRÖM, MATS (2009) Announced Refusal to Answer: a study of norms and accountability in broadcast political interviews, Discourse Studies 11, pp EKSTRÖM, MATS and BÉRCZES, JUDIT (2008) Avbrott i politiska mediesamtal [ Interruptions in political media talk ], Nordicom 30, pp EKSTRÖM, MATS and PATRONA, MARIANNA (Eds) (2011) Talking Politics in Broadcast Media, Amsterdam: John Benjamins. ERIKSSON, GÖRAN (2011) Follow-up Questions in Political Press Conferences, Journal of Pragmatics 43, pp ESSER, FRANK and HEMMER, KATHARINA (2008) Characteristics and Dynamics of Elections News Coverage in Germany, in: Jesper Strömbäck and Lynda Lee Kaid (Eds), The Handbook of Election News Coverage Around the World, New York: Routledge, pp FICO, FREDERICK and FELDMAN, ERIC (2008) Biasing Influence on Balance in Election News Coverage: an assessment of newspaper coverage of the 2006 U.S senate elections, Journalism & Mass Communication Quarterly 85, pp GILDENGIL, ELISABETH (2008) Media Matter: election coverage in Canada, in: Jesper Strömbäck and Lynda Lee Kaid (Eds), The Handbook of Election News Coverage Around the World, New York: Routledge, pp