How to Evaluate It The Role of Story-Evaluative Tone in Agenda Setting and Priming

ORIGINAL ARTICLE How to Evaluate It: The Role of Story-Evaluative Tone in Agenda Setting and Priming Tamir Sheafer Department of Communication and Department of Political Science, The Hebrew University of Jerusalem, Mount Scopus, Jerusalem 91905, Israel The main contribution of this study to the understanding of agenda-setting and prim- ing effects is its focus on the role of evaluative tone in all stages of the agenda setting/ priming process. First, the public’s evaluation of issue importance, which is the depen- dent variable in most agenda-setting studies, is influenced by the issue saliency in the news and by the evaluative tone of media coverage (positive, negative, or neutral). This evaluative tone or affective attribute attached to the issue is part of the second-level agenda setting. Second, these affective attributes that people attach to issues further play an important role in the process of priming, on which they have both indirect and direct impacts. Priming, therefore, carries with it an affective component: It is a combi- nation of message strength and direction. Third, the political judgments of individuals are also directly influenced by media-affective attributes. All of the arguments are supported by the empirical analyses. doi:10.1111/j.1460-2466.2006.00327.x How to evaluate it: The role of story-evaluative tone in agenda setting and priming This research aims to show that the evaluative tone of media coverage, the affective attributes attached to objects (candidates, issues, events), plays an important role in the agenda-setting and priming processes, and consequently affects the political judgment of voters. While current agenda-setting and priming models are based on message salience, or strength, the argument developed here is that both models include an affective component: They are a combination of message strength and direction. Corresponding author: Tamir Sheafer; e-mail: msstamir@mscc.huji.ac.il Journal of Communication ISSN 0021-9916 Journal of Communication 57 (2007) 21–39 ª 2007 International Communication Association 21 Agenda setting According to the agenda-setting hypothesis (first-level agenda setting), the media influence public opinion by emphasizing certain issues over others. The amount of media attention, or the media salience, devoted to certain issues increases their accessibility and consequently influences the degree of public concern for these issues (Dearing & Rogers, 1996; McCombs, 2004; McCombs & Shaw, 1972). The empirical analysis in this study focuses on the issue of the economy in Israel, and the first-level agenda-setting hypothesis is: Hypothesis 1 (first-level agenda setting): An increase in the level of media coverage of the economy will be associated with an increase in the proportion of survey respondents naming this issue as the country’s most important problem. Indeed, because the economy is such an obtrusive issue, because people may have many channels of information about this issue, the influence of real-world indicators and direct experience may provide an alternate explanation for the agenda-setting hypothesis. Therefore, any test of the hypothesis must control for this alternative explanation. Evaluative tone in first- and second-level agenda setting: The role of affective compelling arguments In the last few years, the focus of many agenda-setting analyses has shifted from first to second level, sometimes called attribute agenda setting. It is a shift from a focus on the media’s role in telling us ‘‘what to think about’’ to their function of telling us ‘‘how to think about’’ objects. According to McCombs (2004), each of the objects on the agenda ‘‘has numerous attributes, those characteristics and properties that fill out the picture of each object. Just as objects vary in salience, so do the attributes of each object’’ (p. 70). The major hypothesis that is associated with second-level agenda- setting (Ghanem, 1997; McCombs, 2004; McCombs & Ghanem, 2003) states that the attributes of the object emphasized by the news media affect the saliency of those attributes in the public’s mind. For example, the media coverage of a political candi- date may include attributes such as the candidate’s issue positions and qualifications. Media emphasis on such attributes is expected to affect the saliency of the attributes in the public’s mind and leads to certain evaluations. This hypothesis is generally sup- ported by several empirical studies (see, e.g., Golan & Wanta, 2001; Kim, Scheufele, & Shanahan, 2002; King, 1997; Kiousis, 2005; Kiousis, Bantimaroudis, & Ban, 1999; Wanta, Golan, & Lee, 2004). The focus in this study, though, is on a rather neglected part of second-level agenda setting, the compelling arguments hypothesis (see Figure 1). According to the compelling-arguments hypothesis (Ghanem, 1997; McCombs, 2004; McCombs & Ghanem, 2003), some object attributes emphasized by the news media affect the accessibility of that object (and not of the attribute) to the public, regardless of the frequency of their appearance in the media message. This is the impact of the second-level on first-level agenda setting. According to McCombs How to Evaluate It T. Sheafer 22 Journal of Communication 57 (2007) 21–39 ª 2007 International Communication Association (2004), ‘‘compelling arguments are frames . that enjoy high success among the public’’ (p. 92). McCombs (McCombs & Ghanem, 2003; McCombs, Llamas, Lopez-Escobar, & Rey, 1997; McCombs, Lopez-Escobar, & Llamas, 2000) presents two general groups of attributes at the second level: cognitive (or substantive) attributes and affective attrib- utes. Cognitive attributes deal with the definition of issues (or objects in general) in the media, whereas affective attributes deal with the tone of media presentation, with eval- uation of issues (i.e., positive, negative, or neutral). The empirical distinction between the two types is not always clear. Consider, for example, the empirical variable tested in this study. Each front-page newspaper item that discusses the economy was coded as pre- senting the economy in a positive, negative, or neutral way. An item was coded as positive (or negative) if it presents the economy as improving (or declining). Such an operation- alization no doubt captures the item tone or evaluation. But other cognitive attributes that define an improving (or declining) economy may be presented as well. Nevertheless, the focus in this study is on the affective component of presentation. The effect of affective compelling arguments (i.e., positive, negative, or neutral media presentation of objects) is hardly studied in empirical analyses. An exception is a study by Schoenbach and Semetko (1992), who found that the positive tone with which a certain issue was covered in the news (i.e., positive attribute) reduced the salience of that issue on the public agenda. I would argue that there are at least two sound explanations for the influence of affective attributes, and specifically of negative affective attributes, on perceived issue importance. The first is a theoretical explanation: Information about negative devel- opments captures our attention far more than information about positive develop- ments. This hypothesis is quite extensively discussed and supported in other fields of research (Cacioppo & Berntson, 1994; Kahaneman, Slovic, & Tversky, 1982; Lau, 1985; Marcus, Neuman, & Makuen, 2000; Mutz, 1998; Schul & Schiff, 1993). A negative object attribute is, therefore, expected to increase object importance and accessibility on the public agenda. A positive tone, on the other hand, is not expected to have such an effect. In fact, as noted, Schoenbach and Semetko (1992) found that positive coverage decreased the perceived issue importance. Media Agenda Public Agenda Object Salience of Object Attribute Salience of attribute Compelling arguments 1 2 3 Figure 1 Compelling arguments. Taken from McCombs (2004, p. 92). Note: (1) Traditional agenda setting (first-level effects); (2) attribute agenda setting (second- level effects); (3) compelling arguments (attribute effects on object salience). T. Sheafer How to Evaluate It Journal of Communication 57 (2007) 21–39 ª 2007 International Communication Association 23 The second explanation is based on the common operational definition of public agenda. This variable is usually measured by the survey question, ‘‘what is the most important problem facing this country today?’’ (Dearing & Rogers, 1996, p. 17). Note that this question has two important ingredients. The first is a question of issue importance. But in the second part of the question people are asked to evaluate the most severe problem, or negative development in their environment. This is an affec- tive-evaluative question that has a clear, one-sided valence (i.e., a negative valence or tone). Consequently, this wording may direct the attention of respondents to a com- bination of importance and negative valence, just as in the affective compelling arguments explanation. This leads to the second hypothesis. It is presented as unidirectional because the theoretical discussion deals only with the impact of negative information, not of positive information (recall that the empirical analysis focuses on the state of the economy): Hypothesis 2 (affective compelling arguments): (a) The higher the salience of media coverage of the economy and the more negative the media presentation of the economy, the greater will be the increase in the proportion of survey respondents naming this issue as the country’s most important problem; (b) this effect is expected to be stronger than the effect of media salience alone. Affective prepriming, priming, and political evaluations Affective attributes may also play a central role in the next stage of the extended agenda- setting process, media priming. The priming hypothesis states that the media agenda affects the criteria people use to evaluate the performance of political actors. Individuals use those issues that are most salient and accessible in their memory to evaluate the performance of political actors (Iyengar & Kinder, 1987). If, for example, the issue of the economy was primed, it would become the basis for evaluating the president’s perfor- mance (Iyengar & Simon, 1993; Krosnick & Kinder, 1990; Pan & Kosicki, 1997). Using the example of the economy, the evaluation of the president’s economic performance is the independent variable in priming analyses, and the evaluation of the president’s overall performance is the dependent variable. Media-affective attributes, or evaluative tone, may affect the priming process in direct and indirect ways. The indirect effect is discussed here and the direct effect in the next section. Voters’ evaluations of the incumbent’s economic performance may be affected by the way the media present the state of the economy. Negative media presentation of the state of the economy is likely to lead to a very different evaluation of the economic performance of the incumbent compared with a positive media presen- tation of the economy. This is an affective prepriming effect that has an indirect effect on priming because it affects a fundamental component of priming. In other words, it affects the independent variable in priming (i.e., evaluation of the incumbent’s economic performance), which is the dependent variable in the affective prepriming effect. Accordingly, the third hypothesis is How to Evaluate It T. Sheafer 24 Journal of Communication 57 (2007) 21–39 ª 2007 International Communication Association Hypothesis 3 (affective prepriming): The more positive is the media presentation of the economy, the more positive are the evaluations individuals will assign to the economic performance of the incumbent party. The fourth hypothesis presents the original priming effect. Until now, most media-priming research has focused on evaluations of the president’s performance (Price & Tewksbury, 1997). However, there is no reason to limit the research on priming effects in natural settings mostly to leaders and not to political parties as well: Hypothesis 4 (priming): The more individuals are exposed to media coverage of the economy, the more weight they assign to the economic domain when they evaluate the overall performance of the incumbent party. It is important to note that in a field study such as this one, which covers a set of election campaigns, testing such a hypothesis requires controlling for two variables: first, the general salience of the economy in the information environment in each period, and second, the media exposure of individuals. Two individuals with a similar media exposure but in different periods, one in which the media accorded a lot of salience to the economy and another in which a much lesser salience was accorded, are not exposed to the same levels of economic information and therefore are expected to assign different weights to the economy in their evaluation of the incumbent party. The same can be said of two individuals in the same period but with different levels of media exposure. Previous priming field studies usually con- trolled for a single variable. For example, Krosnick and Kinder (1990) and Iyengar and Simon (1993) controlled for the information environment (the salience of an issue in the media) but not for the media exposure of individuals. The current study aims to control for personal media exposure as well. Affective attributes, affective media priming, and political evaluations However, how can we predict whether the priming effect on the evaluation of the president or the incumbent party would be positive or negative? This question is not fully answered by the theory. Iyengar and Kinder (1987) explain that if the news media prime the prospects of nuclear annihilation, ‘‘then citizens would judge the president primarily by his success, as they see it, in reducing the risk of war’’ (p. 63). But how can people assess the success of the president in this task? How can we predict whether, following a media priming of this issue, the electoral fortunes of the president will improve or decline? There is a paradox in media research regard- ing this question. On the one hand, scholars argue that voters are miserly in expending cognitive efforts when processing political information (Fiske & Taylor, 1991; Popkin, 1994). Therefore, voters are evaluating the president based on the most accessible issue in their memory as an information shortcut. But on the other hand, voters are expected to invest a lot of cognitive effort to assess the success of the president in handling this problem. This rationale doesn’t make sense. A better T. Sheafer How to Evaluate It Journal of Communication 57 (2007) 21–39 ª 2007 International Communication Association 25 explanation, I would argue, can be found to a large extent in the affective attributes of objects. Indeed, some ideological and highly controversial issues are likely to have rather expected electoral effects simply by being on the agenda. For example, if the social welfare policy is emphasized by the media and becomes the basis for evaluation, it is expected that conservatives will evaluate a conservative president in a positive way (Zaller, 1992). Other issues are ‘‘owned’’ by certain parties, and these parties enjoy electoral gains when they are primed (Kleinnijenhuis, Maurer, Kepplinger, & Oegema, 2001; Petrocik, 1996; Petrocik, Benoit, & Hansen, 2003; Sheafer and Weimann, 2005; Simon, 2002). Yet, many other issues do not naturally lend themselves to such clear evaluations. Economic growth is one example. My argument is that the electoral consequences of these issues being primed is a result of the fact that people attach affective attributes to these issues (see Just, Crigler & Neuman, 1996, and Willnat, 1997, for quite a similar argument). Actually, according to the ‘‘hot cognition’’ psychological hypothesis, all sociopolitical concepts a person has evaluated become affectively charged, positively or negatively, strongly or weakly (Morris, Squires, Taber, & Lodge, 2003). Therefore, when economic growth is primed, people will evaluate the president or the incumbent party based on the affective evaluation (positive or negative) they attach to economic growth (for instance, is it growing and positive or declining and negative). I, therefore, refer to this kind of priming as affective priming.1 Affective priming is the affective- evaluative component that is inseparable from priming. This concept, therefore, does not replace priming; it only adds to it an affective dimension. The media influence affective priming through the affective compelling argu- ments effect, in which they attach an evaluative tone (i.e., positive, negative, or neutral) to objects or issues. In other words, the media help people in assigning affective attributes to these issues. For example, the media may present the economy as growing (positive attribute) or declining (negative attribute), thus affecting indi- viduals’ evaluations of this issue. It appears, therefore, that priming has a built-in affective component, and media attributes activate or prime a specific political judgment (for related arguments see Entman, 2004; Ju, 2006; Shah, Domke, & Wackman, 2003; Snyderman, Brody & Tetlock, 1991). In effect, we can find that in most empirical analyses of priming, subjects were exposed to affectively charged media messages. For example, in Iyengar and Kinder’s (1987) original priming experiments, subjects viewed newscasts that either emphasized a problem in a certain area (i.e., a negative evaluative tone) or not. Following that, they were asked to evaluate the president. In Kim et al.’s (2002) field study, subjects were exposed to pro and con attributes of a certain commercial devel- opment project (a positive or negative evaluative tone). Indeed, Willnat (1997) argues that ‘‘especially missing from the current literature are analyses of how positive and negative news coverage of political issues influences the. priming effect’’ (p. 62). Such affective attributes have political consequences, probably mainly through the process of voters’ attribution of responsibility to the incumbent party. Two How to Evaluate It T. Sheafer 26 Journal of Communication 57 (2007) 21–39 ª 2007 International Communication Association political-science hypotheses that discuss the effects of such attribution of responsi- bility are retrospective voting (Fiorina, 1981) and economic voting (Lewis-Beck & Stegmaier, 2000). For example, according to the economic-voting hypothesis, people reward or punish the incumbent based on the state of the economy, be it the actual state (Lewis-Beck & Stegmaier, 2000) or the state of the economy presented by the media (Hetherington, 1996; Shah, Domke, & Wackman, 1999). Therefore, when the affective priming is negatively valenced, the incumbent is generally expected to lose support. As noted, this process may work both indirectly and directly. It works indirectly, as discussed in the previous section, by affecting evaluations of the incumbent’s performance in the primed area (the affective prepriming effect). This process may also work more directly alongside the ‘‘regular’’ priming effect. It is possible to hypothesize that the level of the evaluative tone—the strength of the direction— matters, as well. This is the direct effect of affective attributes on the general perfor- mance evaluations of the incumbent party: Hypothesis 5 (affective attributes): The more positive is the media presentation of the economy, the more positive are the evaluations that individuals will assign to the general performance of the incumbent party (and not only to its economic performance). Methods The analyses are based on d