Candidate Image in Election Campaigns: Attribute Agenda Setting ...

6 downloads 23009 Views 332KB Size Report
Official Full-Text Paper (PDF): Candidate Image in Election Campaigns: Attribute ... second-level agenda setting can be best seen in terms of political candidate.
International Journal of Public Opinion Research Advance Access published May 11, 2010 International Journal of Public Opinion Research ß The Author 2010. Published by Oxford University Press on behalf of The World Association for Public Opinion Research. All rights reserved. doi:10.1093/ijpor/edq009

CANDIDATE IMAGE IN ELECTION CAMPAIGNS: ATTRIBUTE AGENDA SETTING, AFFECTIVE PRIMING, AND VOTING INTENTIONS Meital Balmas and Tamir Sheafery *Hebrew University of Jerusalem, Department of Communication, Mount Scopus, Jerusalem 91905, Israel y Hebrew University of Jerusalem, Department of Communication and Political Science, Mount Scopus, Jerusalem 91905, Israel

ABSTRACT Combining three telephone surveys during the 2006 Israeli elections and a content analysis of leading Israeli newspapers, this study extended the function of attribute agenda setting and priming. Analysis documented that throughout the course of a campaign, public opinion fluctuates in tandem with the saliency of candidate attributes emphasized in the news. We also found an important consequence of attribute agenda setting, affective priming of candidate attributes, by which the prominent tone of the media’s candidate attributes functions as a criterion for evaluating a candidate’s suitability. Finally, this public evaluative tone was linked with voting intentions for the candidate’s political party.

Political communication has been described as undergoing a process of personalization, in which media coverage of politics increasingly focuses on individual politicians rather than on political parties, even in parliamentary systems, in which the vote goes to parties and not to individual candidates (Wattenberg, 1995; Blumler & Kavanagh, 1999; Mazzoleni & Schulz, 1999; Wilke & Reinemann, 2001; Peri, 2004; Rahat & Sheafer, 2007). It is not surprising, therefore, that the political image of candidates in election campaigns has become a central theme in studies of political communication (Pfiffner, 1994; Ohr & Oscarsson, 2003; Hacker, 2004; Stephen, Harrison, Husson, & Albert, 2004; Sheafer, 2008).

Corresponding author: Meital Balmas, e-mail: [email protected] This article was first submitted to IJPOR January 23, 2009. The final version was received January 20, 2010. This research was supported by The Israel Science Foundation.

2

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH

The main goals of this study are related to the implications of this rise of candidate-centered politics in the media. One of the most common theories that has been used to examine the relationship between the way the news media emphasize subject attributes (i.e. politicians’ traits) and the eventual saliency of those attributes in the public’s mind, is commonly referred to in the literature as ‘‘second-level agenda setting’’ (Becker & McCombs, 1978; Weaver, Graber, McCombs, & Eyal, 1981; Robinson & Sheehan, 1983; Kiousis, Mitrook, Wu & Seltzer, 2006). It is noteworthy that most secondlevel agenda-setting studies are based on a single observation (a single crosssectional analysis). It has been suggested, however, that in order to establish the causal effect of attribute agenda setting on the attitudes and opinions of the public, the research design should include time-series analyses (Kleinnijenhuis, Van Hoof, Oegema, & De Ridder, 2007). Accordingly, the first goal of this study is to investigate how changes in the news media attribute agenda in the course of the campaign affect the public attribute agenda. The second goal of this study is to address the possible association between individual candidate attributes and candidate evaluations. Specifically, the focus of this investigation is on the association between the tone (either positive or negative) of the most accessible candidate attributes in individual memory and the general judgment regarding the candidate’s suitability for the position of prime minister. This effect, which is similar to priming but has a clear direction (positive or negative), is termed here affective priming of candidate attributes. The third goal of this study moves on to the next stage of affective priming. It links the tone of the most accessible candidate attribute in the public’s mind and the public’s voting intentions for the leader’s party. Therefore, it extends the original priming hypothesis in another way: instead of focusing only on the general evaluation of the candidate as the dependent variable, this study also focuses on voting intentions for the candidate. The theoretical discussion below starts with the importance of political image in election campaigns. This is followed by a discussion of attribute agenda setting and affective priming of candidate attributes. Next, we suggest a general introduction to the 2006 Israeli elections as a case study. We argue that the Israeli case provides an ideal setting for examining attribute agenda setting and affective priming.

THE I MPORTANCE OF P OLITICAL IMAGE IN MODERN ELECTION CAMPAIGNS Candidate images are an integral part of any modern election campaigns (Hacker, 1995). Studies have consistently confirmed that voters are far more

CANDIDATE IMAGE IN ELECTIONS

3

interested in candidate characteristics than in specific issues and are more likely to vote on the basis of the candidate’s image (Miller, Wattenberg, & Malanchuk, 1985; Hellweg, Dionisopoulos, & Kugler, 1989; Pfiffner, 1994; Hacker, 2004; Stephen et al., 2004; Sheafer, 2008). Moreover, research indicates that perceptions of candidate personal characteristics can even affect party popularity (Shanks & Miller, 1990; Brown, 1992; Stokes, 1996). (Other scholars, though, did not find evidence of significant electoral effect, see the edited volume by King, 2002.) Why is political image so important? First, it has been found that voters react better to personal perceptions of the candidates than to objective reality (Sears, 1983). According to Sears (1969), ‘‘persons represent unusually simple stimuli, easily cognized and retained’’ (p. 364). Second, perceptions of candidate traits provide individuals with a good way to organize all the daily information that becomes available about political issues (Kinder, 1986). Kinder also suggested that personality traits are seen as stable over time, and by ascribing traits to their political leaders, individuals have some basis for gauging the reaction of their political leadership to future demands of their office. There are also systemic aspects: the political weight of the individual actor in the political process increases over time, while the centrality of the political group (i.e. political party) declines (Miller & Miller, 1976; Wattenberg, 1995; Wilke & Reinemann, 2001; Rahat & Sheafer, 2007). Many researchers have documented that mass media coverage of politics tends to focus attention on leadership personalities, and as a result, the public defines the substance of politics in similar terms (Mendelsohn, 1994). For this reason candidates consider news to be one of their most important vehicles of influence (Davis, 1992). Review of the literature reveals that there are a number fundamental attributes. The most important appear to be competence (Kinder, Peters, Abelson, & Fisk, 1980; Bean, 1993; Pancer, Brown, & Barr, 1999), leadership (Miller & Miller, 1976; Kinder, 1986), power (Bean, 1993), intelligence (King, 1997) and credibility and morality (Kinder, 1994; Pancer et al., 1999; Benoit & McHale, 2004).

SECOND-LEVEL AG E NDA SETTING The second-level of agenda setting (sometimes called attribute agenda setting) deals with the specific attributes of a topic and how this agenda of attributes influences public opinion (McCombs & Evatt, 1995). 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’’ subjects. The most basic application of second-level agenda setting can be best seen in terms of political candidate

4

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH

image (Becker & McCombs, 1978; Weaver et al., 1981; Robinson & Sheehan, 1983; Kiousis et al., 2006). For example, if media coverage of a political candidate includes a certain attribute, such as credibility, this attribute is expected to become most accessible in the public’s mind (King, 1997; Kiousis, Bantimaroudis, & Ban, 1999; Golan & Wanta, 2001; Kim, Scheufele, & Shanahan, 2002; Wanta, Golan, & Lee, 2004). McCombs, Lopez-Escobar, and Llamas’ (2000) study of the 1996 Spanish general election, for instance, found a high degree of correspondence between the attribute agenda of seven different mass media and the voters’ attribute agenda for each of the candidates. Kim and McCombs’ (2007) analysis of the 2002 Texas gubernatorial and U.S. senatorial elections revealed that attributes positively or negatively covered in the news were perceived in a similar way by the public. In addition, Kiousis et al. (2006) discovered that the salience of affective candidate attributes in media coverage was related to the perceived salience of affective candidate attributes in public opinion. As noted, previous research has documented significant links between public agenda and the general affective tone on news coverage about political objects. Therefore, replicating prior research on attribute agenda setting, we expect to find support for: H1a (attribute agenda setting): The saliency of candidate attributes in the public agenda changes in tandem with the saliency of candidate attributes in the news during the course of the campaign.

We also expect that attributes emphasized in the news are more likely to have an impact on heavy readers than on light readers (Miller & Krosnick, 1996; Kim & McCombs, 2007), thus our next hypothesis is: H1b (attribute agenda setting): Heavy newspaper readers will echo the salience of candidate attributes as they were presented in the newspaper to a greater extent than light newspaper readers.

These two hypotheses are tested at the aggregate level because they examine the correspondence between the rank-order attributes of newspapers and the rank-order attributes of the public, as is done in other agenda-setting studies (McCombs et al., 2000; Kim & McCombs, 2007). AFFECTIVE P RIMING OF CANDIDATE IMAGE As noted, according to the attribute agenda-setting hypothesis, the news coverage influence public opinion by emphasizing certain candidate attributes over others. The amount of media attention devoted to certain attributes increases their accessibility and consequently influences the degree of the public’s concern with these attributes. Similarly, the tone or direction of candidate attributes emphasized in the media becomes more salient in the

CANDIDATE IMAGE IN ELECTIONS

5

public’s mind (e.g. Sheafer, 2007). This process affects the priming effect, which is the next stage of the extended agenda-setting process (Kim et al., 2002; Kim & McCombs, 2007). According to priming, issues that are most salient in the public’s mind become the criteria for candidate evaluation (e.g. Iyengar & Kinker, 1987; Krosnick & Kinder, 1990; Iyengar & Simon, 1993; Pan & Kosicki, 1997). Likewise, candidate attributes (e.g. leadership, morality) that are most salient in the public’s mind become the criteria for candidate evaluation (Kim & McCombs, 2007). Please note that while the saliency of candidate attributes in the public’s mind is the dependent variable in the attributes agenda-setting hypothesis, it becomes the independent variable in the priming hypothesis, while the evaluation of the candidates’ general fit for the job is the dependent variable in priming. Because priming can alter the criteria citizens use to evaluate political leaders, priming effects can have important consequences for the outcome of elections and approval ratings of political actors (Sheafer & Weimann, 2005; Althaus & Kim, 2006). However, Sheafer (2007) argued that there is a missing link in the ‘‘regular’’ process of priming. How can we predict whether the priming effect on evaluation of a candidate will be positive or negative? Iyengar and Kinder (1987) explain that if the news media primed 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 a 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 regarding this question. On the one hand, scholars argue that voters are miserly in expending cognitive efforts when processing political information (Fisk & Taylor, 1991; Popkin, 1994). Therefore, voters evaluate the president based on the most accessible issue in their memories as an information shortcut. But on the other hand, voters are expected to invest great cognitive effort in assessing the success of the president in handling this problem. This rationale does not make sense. A better explanation, suggested by Sheafer (2007), can be found to a large extent in the affective attributes or objects. Testing this proposition, Sheafer found that the more individuals are exposed to a more negative media presentation of the economy, the lower their evaluation is of the incumbent party. Therefore, the role of the evaluative tone in the media is central in the process of priming effect. Sheafer (2007) has termed this process affective priming. Although the current study focused on priming of political candidate attributes and not of political issues, the role of the evaluative tone is fundamental. The traditional priming hypothesis can only explain the weight of a certain candidate’s attribute (like morality) in the general evaluation of the

6

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH

candidate. But many times it cannot explain whether the general evaluation of the candidate will be positive or negative. A recent study by Kim and McCombs (2007) is a solid example of this missing link in the area of candidate attribute. Their work focused on the relationship between the public’s attribute agenda for political candidates and the public’s general attitudes toward these candidates, taking into account the affective tone of each candidate’s attribute. They found that the affective tone (i.e. positive vs. negative) of the candidate attributes in the public’s mind predicts general attitudes (favorable vs. unfavorable) toward the candidate. Hence, positive attributes predicted a positive general attitude toward a candidate, and vice versa. One concern in using general attitude (favorable vs. unfavorable) toward the candidate as the dependent variable in affective priming of candidate attribute is causing a tautology. This threat to validity may result from the close similarity between the independent variable (candidate attribute, measured as the positive or negative candidate attribute provided as a response to an open question) and the dependent variable (general attitudes toward the candidate, which is a feeling thermometer). To avoid this, we have used a different dependent variable: general evaluation of the candidate’s suitability for the position of Prime Minister. Given this, our second hypothesis is: H2 (affective priming of candidate attribute): The tone of the most accessible candidate attribute (e.g. leadership vs. lack of leadership) in an individual’s memory will positively predict the evaluation of the candidate’s suitability for the position of Prime Minister.

Affective priming of candidate image is a process, in which the saliency of certain candidate attributes in the news affects the weight that individuals place upon those attributes when evaluating candidate. Therefore, we expect to find stronger affective priming among heavy newspaper readers than among light newspaper readers. Most media priming research has focused on political candidates (or presidents’) general evaluation. However, there is no reason to strictly limit our research on affective priming of candidate image only to evaluation of candidates and not extend it also to voters’ electoral behavior. Actually, there is good reason for extending affective priming effect: according to Iyengar and Kinder (1987, pp. 102–103; see also Brosius & Kepplinger, 1992 and Sheafer & Weimann, 2005), there is a high positive correlation between evaluation of presidents’ performances (the dependent variable in most priming effect studies) and voting for or against them. By extension, this concept also can be applied to affective priming of candidate attributes. According to this extension of the priming hypothesis, changes in the tone of the candidate’s most salient attribute in the public mind (and in the news) during the course of the campaign should ‘‘move’’ in tandem with voting intentions for the party led

CANDIDATE IMAGE IN ELECTIONS

7

by that candidate. Political parties are therefore expected to enjoy an electoral gain when the news, and respectively, the public, emphasize their leader’s most salient attribute in a positive way, and vice versa. Thus, our third hypothesis is: H3 (affective priming of candidate attribute): The tone of the candidate’s most salient attribute in the public mind (and in the news) during the course of the campaign will be positively associated to the public’s intention with respect to voting for the leader’s party.

T HE 2 0 06 ISRAELI E LECTIONS AS A CASE STUDY The 2006 Israeli elections marked a significant change in the political alignment. Until then there were only two main parties in Israel, the Likud (on the right) and Labor (on the left), and voters could define themselves as rightwingers or left-wingers. In late November 2005, Prime Minister Ariel Sharon of the Likud and a number of other Likud ministers and members of the Knesset announced that they were leaving the party to form a new centrist party, which was eventually named Kadima (’forward’ in Hebrew). The formation of Kadima turned the elections into a three-party race. Kadima was initially led by Sharon, Labor by Amir Peretz and Likud by Benjamin Netanyahu. On January 4, 2006, Sharon suffered a hemorrhagic stroke, leaving him in a coma. Subsequently, Ehud Olmert officially became Kadima’s new candidate for Prime Minister. Israel provides an ideal setting for examining attribute agenda setting and attributes priming. Unlike the USA, Israel witnessed the heyday of mass political party. Parties have been the dominant actors in the Israeli polity since the country’s independence (Galnoor, 1982; Horowitz & Lissak, 1989). Thus, Israel supplies a clearly non-personalized and party-dominated starting point for measuring the possible implications of the rise of candidate-centered politics in the media (Mendelsohn, 1994). Except for a short experiment with direct elections for prime minister from 1996 to 2001, the structure of the Israeli regime is clearly non-personalized (Rahat, 2001; Shugart, 2001). It is a parliamentary system, in which voters cast one ballot for a political party, and not directly for candidates. This makes it less likely that Israeli voters will rely on the candidates’ personal attributes and suitability for the position of prime minister when making their voting preferences for a political party.

METHODS The analyses are based on data from the last Israeli election for the Knesset (2006). This method combines three separates telephone surveys of Israeli

8

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH

voters at three points in time during the campaign, and a content analysis of two main Israeli newspapers. THE MEDIA AGENDA The agenda-setting literature suggests that, in general, newspapers more frequently dominate agenda setting (McCombs, 2004) and that newspapers set the agenda for television news (Roberts & McCombs, 1994). In this study, a content analysis examined only printed news stories (excluding opinion editorials and letters to the editor) about the three main party leaders in two leading Israeli daily newspapers, Yediot Aharonot and Ha’aretz during a 2month period, beginning January 27 and ending March 28, 2006, which was the Election Day. In order to make a comparison between different media environments during the course of the campaign, we divided the content analysis into three periods (1 ¼ January 27 to February 6; 2 ¼ February 7 to March 12; 3 ¼ March 13 to March 22). These periods correspond with the three telephone surveys, as is described below. Following extensive training, five student coders independently identified every item about the three candidates. Overall, 467 articles were analyzed (in period 1: 56 from Ha’aretz and 27 from Yediot Aharonot; period 2: 140 from Ha’aretz and 115 from Yediot Aharonot; period 3: 72 from Ha’aretz and 57 from Yediot Aharonot), but because the unit of analysis (Wimmer & Dominick, 1991) was the media coverage of a single candidate in a single article, there was a total of 738 coding items (usually more than one candidate is covered in a single news item). Overall, Olmert was mentioned in 332 items, Netanyahu in 222 and Peretz in 184. Inter-coder reliability, tested on 20 percent of the coding items, resulted in a mean reliability (using Alpha Krippendorf) of .87, and no lower than 0.74 (for the lowest coding category). Candidates’ images in the media were measured by five variables: (1) leadership (2) intelligence, (3) reliability, (4) morality, and (5) ‘‘other’’ comments about the candidate’s personal qualifications and character. For example, morality was measured by the category ‘‘Is there a clear reference to the morality of Olmert/Peretz/Netanyahu?" (1 ¼ no attitude; 2 ¼ yes, moral; 3 ¼ yes, immoral; 4 ¼ unclear/not relevant). As discussed above, these attributes have repeatedly been found as fundamentally important for establishing a political image (Kinder et al., 1980; Benoit & McHale, 2004). Note that this coding includes the tone of coverage (i.e. positive-moral vs. negative-immoral) to measure the affective aspects of each candidate’s image. Therefore, each of the five attributes was reported separately for positive and negative tone, resulting in a total of 10 categories. For descriptive statistics see Tables A1 and A2.

CANDIDATE IMAGE IN ELECTIONS

9

THE PUBLIC AGENDA Three separate telephone surveys were conducted during the course of the campaign among a representative sample of Israeli Jewish voters. Israeli Arabs (comprising about 18 percent of the Israeli population) were not included in the current study. The first survey was conducted on February 6 and included 493 respondents, 52 percent of which were female and the average age was 48 (SD ¼ 16.25). The second poll was conducted on March 12 with 601 respondents, 51 percent of the sample were female and the average age was 46 (SD ¼ 16.01). The third was done on March 22 with 423 responders, 57 percent of the sample were female and the average age was 44 (SD ¼ 15.53). The three random samples had a margin of error of about plus or minus 4 percent at the 95 percent confidence level. Within households, adults over 18 were selected. Response rate, calculated according to AAPOR guidelines was RR1 ¼ 0.22 (AAPOR, 2009). This is a rather low response rate, but it is similar to rates obtained in other contemporary public opinion surveys in Israel. The main variables were (1) the attribute for each of the three candidates, and (2) evaluation of each candidate’s suitability for the position of Prime Minister. To measure the public’s attribute agenda, respondents were asked, ‘‘Assuming you have a friend who doesn’t know anything about Ehud Olmert/Amir Peretz/Benjamin Netanyahu, and you are required to describe him in a single word, what is the attribute or the characteristic that best describes him?’’ This survey’s question is commonly used in the first- and the second-level agenda-setting studies (Kim & McCombs, 2007; Sheafer, 2007). The answer to this open question represents the most accessible candidate attribute in a respondent’s memory. Responses were coded using the 10 attributes defined in the content analysis above. For example, if the single word that a respondent chose to describe a candidate was ‘‘liar’’, we recoded this statement as a ‘‘lack of reliability’’, while if the respondent’s statement was ‘‘truthful’’, than we recoded it as ‘‘reliability’’ (Other neutral statements were no more than 10percent and excluded from the analyses.) Inter-coder reliability (using Alpha Krippendorf), based on 20 percent of the coding items, resulted in a reliability of 0.88. To measure the public’s evaluation of candidate suitability for the position of a Prime Minister, respondents were asked, ‘‘Is the candidate fit for the public office of Prime Minister?’’ The variable was measured on a scale of 1 (not fit at all) to 7 (very much fit) (Olmert: M ¼ 3.91, SD ¼ 2.07; Peretz: M ¼ 3.04, SD ¼ 2.02; Netanyahu: M ¼ 3.20, SD ¼ 2.22). This question was asked only on the second survey.

10

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH

MEDIA EXPOSURE Individuals’ newspaper exposure was measured by the survey question, ‘‘How frequently do you read a daily newspaper?’’ (1 ¼ rarely through; 7 ¼ every day; M ¼ 3.52, SD ¼ 2.10). This question was asked only on the second survey as well. COVARIATES Political identity was measured through the survey question, ‘‘Which party would you vote for if the elections were held today?’’ Respondents could choose from a list that included all the nominated parties. The results were then recoded as 1 for Left-wing identification, 2 for Center identification, and 3 for Right-wing identification (first survey: left-wing ¼ 30.9 percent; center ¼ 42.1 percent; right-wing ¼ 27 percent; second survey: left-wing ¼ 31.7 percent; center ¼ 30.9 percent; right-wing ¼ 36.7 percent; third survey: left-wing ¼ 33.2 percent; center ¼ 38.3 percent; right-wing ¼ 28.5 percent). Candidate knowledge was constructed via a survey question, in which respondents were asked to recognize the economic position of each of the three candidates. Economic position was based on the candidates’ political platform, according to which Peretz was identified as a socialist, Netanyahu as a capitalist, and Olmert in a central position between them. The measure was done on a seven-point scale, where 1 was a capitalist position and 7 was a socialist position (Correct answers positioned Netanyahu at the capitalist end, Olmert at the center and Peretz at the socialist end). Results were then recoded as 0 for a wrong answer (when the respondent gave a wrong answer as to the candidate’s position), and 1 for a correct answer (correct answer: Olmert ¼ 45.8 percent; Peretz ¼ 85.7 percent; Netanyahu ¼ 67.9 percent). Demographic variables included in the statistical analyses were, age, religion (1 ¼ orthodox through 5 ¼ secular) and years of education. RESULTS The current study had three goals that built upon each other. The first was to investigate attribute agenda setting during the course of a campaign. The second goal addressed the possible association between the tone of the most accessible candidate attribute in individual memory and the general evaluation of the candidate suitability for the position of Prime Minister. This investigation represented the outcome of the second-level agenda setting. The third and last goal led forward to an exploration of the impact of affective priming of candidate attribute on voter behavior, which is one of the most challenging applications in political communication.

11

CANDIDATE IMAGE IN ELECTIONS

SECOND-LEVEL AGENDA SETTING Figures 1, 2, and 3 (each figure for every candidate: Olmert, Peretz, and Netanyahu, respectively) compare cross correlations between news agenda and public agenda across three time points of measurement. These correlations are based on the five image attributes described above, which coded for positive and negative attributes, totaling 10 attribute categories (for attributes rank-list in each point in time refer to Table A1). Three distinct patterns are apparent from Figures 1, 2, and 3. The first pattern comes from examinations of the correlations across the vertical lines, between media agenda (i.e. the media presentation of the candidate’s image) and public agenda (i.e. the attribute mentioned by the respondent in response to an open question) at each point in time. It is important to mention here that the media agenda measured at each point of time represents all media coverage from the last point of time (or from the beginning of the campaign) and up to a

FIGURE 1 Cross correlations between attribute agenda of Olmert’s news coverage and the public attribute agenda over the course of the campaign .18

News

News

-.18

News

.09

News .28

.62**

.78***

.60*

.62**

.07 Public

.20 Public

Public

Public -.12

-.04 Time 1 Time 2 Rozelle-Campbell baseline=.12

News

Time 1 Time 3 Rozelle-Campbell baseline=.11 .66*

News

.97*** .78**

.60* .68**

Public

Public .73**

Time 2

Time 3

Rozelle-Campbell baseline=.39

Note: The media agenda measured at each point of time represents all media coverage from the last point of time (or from the beginning of the campaign) and up to a day before the survey was conducted

12

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH

FIGURE 2 Cross correlations between attribute agenda of Peretz’s news coverage and the public attribute agenda over the course of the campaign .06

News

News

News

.81**

News

.61*

.46 .59*

.59*

.60*

.70* .24

.09 Public

Public

Public

.30

Public .74 *

Time 1 Time 2 Rozelle-Campbell baseline=.33

News

Time 1 Time 3 Rozelle-Campbell baseline=.41 .26 News -.53

.60*

.62* .16 Public

Public -.10

Time 2 Time 3 Rozelle-Campbell baseline=.09

Note: The media agenda measured at each point of time represents all media coverage from the last point of time (or from the beginning of the campaign) and up to a day before the survey was conducted

day before the survey was conducted. Such a time-sequence design specifically fits agenda-setting analyses, in which the researcher is interested in assessing the impact of media agenda on the public agenda. In other words, although such a design cannot completely rule out a reverse causation (i.e. public agenda affects media agenda), it greatly minimizes this threat. In such a temporal design the first expectation is to find causal correlations between the media and the public agendas within each period of time (i.e. along the vertical lines). In this regard, we found that almost all the correlations along the vertical lines were positive and statistically significant at the p < 0.05 level. Only one correlation (Figure 3, point time 2) was significant at the p < 0.10 level. These findings mean that there was a strong relationship between the saliency of candidate attributes as presented in the news, regarding each of the three candidates, and the saliency of candidate attributes in the public’s mind, as demonstrated in the responses to the open survey question. For all nine correlations (between media agenda and public agenda, at all three

13

CANDIDATE IMAGE IN ELECTIONS

FIGURE 3 Cross correlations between attribute agenda of Netanyahu’s news coverage and the public attribute agenda over the course of the campaign .60*

.29

News

News

News .89**

News .11

.89*

.89**

.52

.73** .37

.44 Public

Public

Public

Public .27

.96** Time 1 Time 2 Rozelle-Campbell baseline=.46

Time 1 Time 3 Rozelle-Campbell baseline=.19 .44

News

News

.17 .52

.73 ** .40 Public

Public .22

Time 2

Time 3

Rozelle-Campbell baseline=.18

Note: The media agenda measured at each point of time represents all media coverage from the last point of time (or from the beginning of the campaign) and up to a day before the survey was conducted

points in time), the median value was 0.62, which constitutes strong support for attribute agenda-setting effect. These findings, however, would be similar to a one-shot analysis if the rank order of candidate image attributes during the course of the campaign did not change. But as can be seen in Figures 1, 2 and 3, this was not the case. Changes in the rank order of candidate image attributes during the course of the campaign can be seen when looking at the horizontal correlations between attribute agendas (media/public) of the same type during different points of time (e.g. between media agenda at time 1 and media agenda at time 2). High and significant correlations mean that the agenda had not changed much, while low correlations were evidence of a change in the agenda. As can be seen in the figures, agendas did change sometimes during the campaign. As expected according to the agenda-setting hypothesis, when there was no change between two points in time in the news agenda (i.e. news agendas

14

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH

at these two points of time were significantly correlated), there was also no change between the same points in time in the public agenda. But when the news agenda did change, so did the public agenda. These findings indicate that the saliency of candidate attributes in the public agenda changed in tandem with the saliency of candidate attributes in the news during the course of the campaign. This finding also strengthens our confidence in the causal relationship found along the vertical lines. The third examination deals with the correlations along the diagonals, that is, the correlations between the news agenda at one point in time (i.e. time 1) to the public agenda at the next point in time (i.e. time 2) and the other way around, between public agenda at one point in time to the news agenda at the next point in time. The results demonstrated, first, that only when there was a significant correlation between two points in time in the news agenda, as tested above in the correlations along the horizontal lines, there was also a significant correlation along the diagonals (i.e. between news agenda at one point in time to the public agenda at the next point in time). Second, for all three candidates, almost only the diagonals from the news media to the public media are positive and statistically significant (i.e. Olmert: from time 2 in the news to time 3 in the public; Peretz: from time 1 in the news to time 3 in the public; Netanyahu: from time 1 in the news to time 2 in the public). In the case of Olmert, the direction from the public agenda in time 2 to the news agenda in time 3 is also significant; however, this correlation is significantly lower than the other way around (Fisher’s Z test indicates that the difference between these two correlations is significant at the p < .01 level). Third, in all cases, the correlations between the news agenda at one point in time to the public agenda at the next point in time exceeded the Rozelle–Campbell baseline, and there was no evidence of reciprocal influence. In short, the pattern of attribute agenda setting in Figures 1, 2, and 3 identifies the media agenda as the prime mover. The public agenda, on the other hand, has a modest influence on the media agenda of substantive attributes. Therefore, the first research hypothesis (H1a) is supported. According to hypothesis 1b, heavy newspaper readers will echo the saliency of candidate attributes in the newspaper to a greater extent than light newspaper readers. To test this hypothesis we conducted a series of correlations (Spearman’s rho) between candidate attribute saliency in the news agenda and the saliency of candidate attributes among (1) light newspaper readers and (2) heavy newspaper readers. Supporting the hypothesis, all three correlations among heavy readers were positive, strong and statistically significant (see the last column in Table 1). On the other hand, the correlations among light readers were lower and statistically insignificant. Heavy readers, therefore, reflect the salience of the

15

CANDIDATE IMAGE IN ELECTIONS

TABLE 1 Rank-order correlations between media candidate attributes and public candidate attributes for light and heavy newspaper readers

Olmert Peretz Netanyahu

Light readers n ¼ 10

Heavy readers n ¼ 10

0.29 0.26 0.15

0.75 0.61 0.60

Note: Fisher’s Z tests for significance of the difference between two correlations of each candidate (light readers and heavy readers) indicate that the differences for Olmert is significant at the p < .08 level, for Peretz at the p < .20 level, and for Netanyahu at the p < .05 level. Due to small N, relatively low significance was expected. This table is based only on time point 2, at which the main survey was conducted, as well as the largest content analysis (N ¼ 255).  p < .05;  p < .01;  p < .001.

three candidates attributes in the news agenda to a greater extent than light readers.

AFFECTIVE PRIMING OF CANDIDATE ATTRIBUTE According to the second hypothesis, the tone of the most accessible candidate attribute in a persons’ memory positively predicts the evaluation of the candidate’s suitability for the position of Prime Minister. To test this hypothesis we first divided respondents into two groups, heavy and light newspaper readers. Six separate regression analyses were then conducted, one for each candidate for light readers and for heavy readers. In each regression model, respondents’ evaluation of candidate suitability for the position of Prime Minister (the dependent variable) was regressed on all attributes. Each attribute included in the regression was dummy-coded so that 1 represents the single attribute mentioned by the respondent, and 0 represents the attributes not mentioned by the respondent. In addition, political identity, level of political knowledge, education and age served as control variables (see Table 2). All these control variables have been shown repeatedly to be central predictors of evaluation toward and voting for political candidates. Among Olmert’s attributes, leadership, lack of morality, lack of credibility, lack of leadership and credibility were the top five attributes that received the largest amount of news attention. Consistent with our second hypothesis, for light readers only two attributes of the top five (lack of morality and leadership) were significantly predicted evaluations of Olmert, while among heavy readers, four of the top five attributes (leadership, lack of morality, lack of credibility, and credibility) significantly predicted his evaluation. Similar patterns were observed for the other two candidates. For Peretz, no single attribute (positive or negative) predicted light newspaper readers’ evaluations, whereas two negative attributes (lack of intelligence and lack of leadership),

16

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH

TABLE 2 Affective priming of candidate attributes Attributes

Light readers

Heavy readers

Olmert

Credibility Morality Intelligence Leadership Lacks credibility Lacks morality Lacks intelligence Lacks leadership Adjusted R2

1.81 (1.34) 0.28 (1.13) 0.22 (0.97) 1.76 (0.48) 0.34 (0.50) 1.18 (0.48) 1.20 (1.11) 1.64 (.52) 0.21

1.92 (0.51) 0.23 (1.68) 1.29 (0.77) 1.78 (0.44) 1.30 (0.42) 1.34 (0.47)  0.38 (.45) 0.36

Peretz

Credibility Morality Intelligence Leadership Lacks credibility Lacks morality Lacks intelligence Lacks leadership Adjusted R2

0.59 (0.74) 0.98 (1.20)  0.13 (0.58) 0.48 (0.54) 0.01 (0.63) 0.75 (1.74) 0.38 (0.35) 0.26

0.56 (0.42)  0.95 (0.93) 0.86 (0.56) 0.50 (0.67)  1.28 (0.58) 0.61y (0.32) 0.47

Netanyahu

Credibility Morality Intelligence Leadership Lacks credibility Lacks morality Lacks intelligence Lacks leadership Adjusted R2

2.56 (1.07)  1.00 (0.65) 0.62 (.47) 0.53 (0.36) 0.49 (0.48) 1.39 (0.93) 0.76 (0.84) 0.41

2.19 (0.75) 2.22 (1.60) 2.20 (0.56) 1.77 (0.52) 0.57 (0.32) 0.51 (0.45)  0.52 (0.28) 0.49

Note: Entries are unstandardized OLS regression coefficients with standard errors in parentheses, after controlling for political identity (measured by party vote intention), candidate knowledge (measured by recognition of the candidates’ economic positions), year of education, and age. The dependent variable is evaluation of candidates’ suitability for the position of Prime Minister. This regression is based only on time point 2, in which the main survey was conducted.  p < .05;  p < .01;  p < .001; yp < .10.

which were the top two attributes in terms of the amount of media attention received, predicted heavy newspaper readers’ evaluations of his suitability for the position of Prime Minister. For Netanyahu, no attribute significantly predicted light newspaper readers’ evaluations, and four attributes (lack of credibility, intelligence, leadership, and credibility) were significant predictors of heavy newspaper readers’ evaluations. It is important to mention here that Olmert’s attributes received the most news coverage (71 percent) compared to Netanyahu’s attributes (47 percent) and Peretz’s attributes (39 percent). The percent are not summarized in 100

CANDIDATE IMAGE IN ELECTIONS

17

percent given that there were three separate questions in the survey, one for each candidate. For example: ‘‘Is Olmert appearing in the item?’’ (1 ¼ No; 2 ¼ Yes, in the headline; 3 ¼ Yes, in the first paragraph; 4 ¼ Yes, after the first paragraph). The results were then recoded (for each candidate) as 1 ¼ No, and 2 ¼ Yes (the candidate appearing in the item). For this reason, light readers were more exposed to Olmert’s attributes than to those of Peretz and Netanyahu. This may explain why there were more attributes that predicted light readers evaluations of Olmert compared with the two other candidates. Taken together, Table 2 demonstrates the general pattern that the attributes most emphasized in the news, and their evaluative tone, are likely to predict general evaluation of the candidate’s suitability for the position of Prime Minister to a larger degree for heavy newspaper readers than for light newspaper readers. Generally, the second hypothesis regarding affective priming is mostly supported. AGGREGATION OF VOTING INTENTION FOR PARTIES As discussed above, based on findings in earlier studies (e.g. Sheafer & Weimann, 2005) we have assumed that the process of affective priming can be extended beyond evaluations of the candidate’s fitness for the position of Prime Minister. Accordingly, we expected to find an association between the tone of the candidate’s attributes and voting intentions for the party led by the candidate (H3). This assumption has been tested based on the relationship between the changes in the tone of the candidate’s most salient attribute in the public mind and the changes in aggregate voting intentions for the party led by the candidate. The most salient attribute for each candidate was the attribute (either in its positive or negative version) that, at the aggregate level, was most frequently mentioned in all three polls. For Olmert and Peretz, the most salient attribute was leadership/lack of leadership, and for Netanyahu it was credibility/lack of credibility (please refer to Table A2, for a more detailed ranklist of the attributes in the surveys). As can be seen in Figure 4a, b, and c, the graphs of the public and the news attribute agendas move together. This is expected based on the previous analyses presented here. For Olmert and Peretz, the graph of voting intentions and the graph of the most salient attribute (public and news) parallel each other. This pattern of change suggests that changes in the tone of the most salient attribute in the public’s mind may have electoral-behavioral consequences. When the public perceived the candidate’s attribute in a negative manner, the voting intentions for the leader’s party declined, and vice versa. A different pattern was observed for Netanyahu. It seems that throughout the campaign there were no changes in the news and public in terms of

18

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH

FIGURE 4 (a–c) Changes in aggregate vote intentions for a party compared with changes in the aggregated salient attribute of the leader

Note: The X-axis represents the three points in time during the campaign. The left Y-axis represents the changes in the most salient candidate attribute (1 ¼ positive, 1 ¼ negative), and the right Y-axis the changes in the proportion (by percent) of the voting intentions for the leader’s party

CANDIDATE IMAGE IN ELECTIONS

19

attribute saliency. Netanyahu was consistently represented in the news as lacking credibility and the public agreed. The changes in the voting intentions were also rather minimal. However, aggregate-level figures are not sufficient as evidence of a relationship between variables that move together. In this case we do not have better data to support and substantiate these findings.

DISCUSSION The basic second-level agenda-setting hypothesis states that the attributes of the subject that are emphasized by the news media affect the saliency of those attributes in the public’s mind. The evidence presented here not only provides additional support for this kind of relationship, but also extends the body of knowledge in this area by exploring second-level agenda-setting effects in more than a single observation. Based on three surveys that were conducted in different points in time during the 2006 Israeli elections, we documented the dynamics of the changes in candidate attributes. It was demonstrated that the saliency of each candidate attribute in the public agenda changed in tandem with the saliency of candidate attributes in the news throughout the course of the campaign. Remarkably, when the saliency of a candidate’s attributes changed in the news agenda, similar changes also appeared in the public’s agenda; when there were no changes in the news agenda, there were no changes in the public agenda as well. It is important to note that this similar matching appears not to be coincidental. The saliency of attributes did not change simultaneously for the three candidates. In other words, the news had different agendas for different candidates at different points in time during the course of the campaign. And yet, despite this complication, the public’s agenda changed respectively. Moreover, additional analyses that took into account the extent of exposure to news agenda revealed, as expected, a stronger relationship between news agenda and public agenda among heavy readers compared with light newspaper readers. These analyses revealed the impact of exposure to the newspaper. An important consequence of attribute agenda setting, which we call affective priming of candidate attributes, deals with the impact of news coverage on the public’s evaluation of candidates. More specifically, according to the affective priming hypothesis, the most salient candidate attributes in the public’s mind will become the criterion for evaluating the candidate. And, the evaluative tone of that attribute in the memory of the voter will be a decisive factor that will generate the direction (positive vs. negative) of candidate evaluation. The ‘‘regular’’ priming hypothesis ignores this affective aspect and therefore does not fully explain how people use the primed attribute when evaluating political actors and making electoral decisions. Therefore,

20

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH

the ‘‘regular’’ priming hypothesis only explains the weight of a certain attribute (like leadership ability) in the general evaluation of the candidate. But it says nothing about whether the effect of priming on the general evaluation of the candidate will be positive or negative. In the electoral competition, the tone of the attribute is the critical factor. The affective priming hypothesis carries with it a hidden assumption that if a certain attribute, leadership for example, is primed, people will invest resources to assess the candidate’s leadership ability and consequently will decide how to evaluate the candidate. This does not usually happen and therefore many priming studies have a missing link, for which the priming hypothesis does not fully account. In the process of affective priming of candidate attributes, people use the candidate’s most salient attributes (positive or negative) as an information shortcut that assists them in making general political evaluations about him/her. The analyses in Table 2 show important findings in this regard. The attributes and their tone were found to be positively associated with the general evaluation of the candidates’ suitability for the position of Prime Minister. In addition, these findings were found to be somewhat stronger among heavy newspaper readers than among light newspaper readers. This only makes sense, since it was found that the public’s candidate attributes agenda is influenced by media coverage, and that this attribute agenda-setting process is stronger among heavy news consumers. Therefore, the more individuals were exposed to news coverage, the more their evaluation of the candidate’s suitability was built upon those attributes and their tone. Finally, this study linked the tone of the candidate’s most salient attribute in the public’s mind to voting intentions for the leader’s party. This represents an important but natural theoretical and empirical extension of the priming and affective priming hypotheses. The evidence presented in this study indicated that, for two of the three candidates (Olmert and Peretz), when the public perceived the candidate’s most salient attribute in a negative manner, the voting intention declined, and vice versa. These findings are especially significant for parliamentary systems in general and in particular for the Israeli political system as a case study. Unlike the presidential election campaigns in the U.S. and other countries, where voters cast the vote for an individual candidate, Israeli voters can only cast their vote for a political party. Therefore, theoretically it seems less likely that Israeli voters will change voting intentions for a political party based on the party leader’s personal attributes. Yet these findings can be explained according to Rahat and Sheafer’s (2007) study on the personalization of politics in Israel. They found a process of political personalization in Israel in three realms: institutional (political changes), media (increased focus on politicians at the expense of covering political parties) and political behavioral (an

21

CANDIDATE IMAGE IN ELECTIONS

increase in personal activity of politicians). Thus, it seems that the wide process of personalization in Israel has entered the realm of voter behavior as well.

APPENDIX

TABLE A1 Frequency of attributes on the media and the public attribute agendas about candidate’s personal qualifications and character Attributes

Credibility Morality Intelligence Leadership Lacks credibility Lacks morality Lacks intelligence Lacks of leadership

Media agenda

Public agenda

Period 1

Period 2

Period 3

Period 1

Period 2

Period 3

7 8 12 23 24 18 15 22

21 11 13 21 20 21 12 21

10 8 10 33 44 30 15 30

53 2 49 157 210 91 62 150

65 8 50 150 273 164 48 210

37 0 79 122 220 93 36 124

Note: The category ‘‘other positive/negative comments’’ was excluded from the table.

Note: The category ‘other positive/negative comments’ was excluded from the tables.

Panel C: Netanyahu’s rank-order attributes in the news and the public Credibility 6 8 7 6 Morality 8 6 5 8 Intelligence 5 7 3 3 Leadership 2 3 4 1 Lacks credibility 1 2 2 2 Lacks morality 3 4 1 5 Lacks intelligence 5 5 8 7 Lacks leadership 4 1 6 8 6 8 5 4 1 2 7 3

5 6 7 4 3 5 2 1

Panel B: Peretz’s rank-order attributes Credibility 5 Morality 7 Intelligence 8 Leadership 6 Lacks credibility 3 Lacks morality 4 Lacks intelligence 2 Lacks leadership 1

in the news and public 2 5 3 4 7 6 6 8 7 1 3 1 7 4 2 3 6 5 8 2 8 4 1 4

5 8 6 3 2 4 7 1

5 8 3 1 2 4 7 6

1 8 7 2 7 4 3 5

3 8 4 3 5 6 7 2

Supporters

Supporters

Non-supporters

Time 2

Time 1

Time 3

Time 1

Time 2

Public agenda (rank)

News agenda (rank)

Panel A: Olmert’s rank-order attributes in the news and the public Credibility 6 5 6 2 Morality 8 7 7 8 Intelligence 5 6 5 3 Leadership 1 3 2 1 Lacks credibility 2 4 1 5 Lacks morality 4 2 3 6 Lacks intelligence 7 8 8 7 Lacks leadership 3 1 4 4

Attributes

7 8 5 4 1 2 5 3

6 7 8 5 6 4 2 1

6 7 5 4 3 2 8 1

Non-supporters

5 8 3 1 2 6 7 4

2 8 6 3 4 7 5 1

3 8 2 1 5 6 7 4

Supporters

Time 3

7 8 5 4 1 3 6 2

6 8 7 4 3 5 2 1

6 8 5 3 1 2 7 4

Non-supporters

TABLE A2 Rank-order attributes on the media and among the public about the candidates’ personal qualifications and character, during the course of the campaign

22 INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH

CANDIDATE IMAGE IN ELECTIONS

23

REFERENCES AAPOR, (2009). Standard definitions: Final dispositions of case codes and outcome rates for surveys, (6th Ed.), Lenexa, KS: AAPOR. Althaus, S. L., & Kim, Y. M. (2006). Priming effects in complex information environments: Reassessing the impact of news discourse on presidential approval. The Journal of Politics, 68, 960–976. Bean, C. (1993). The election influence of party leader images in Australia and New Zealand. Comparative Political Studies, 26, 111–132. Becker, L., & McCombs, M. (1978). The role of the press in determining voter reactions to presidential primaries. Human Communication Research, 4, 301–307. Benoit, W. L., & McHale, J. P. (2004). Presidential candidates’ personal qualities: Computer content analysis. In K. Hacker (Ed.), Presidential candidate images (pp. 49–63). New York: Rowman & Littlefield Publishers, Inc. Becker, L., & McCombs, M. (1978). The role of press in determining voter reaction to presidential primaries. Human Communication Research, 4, 301–307. Blumler, J. G., & Kavanagh, D. (1999). The third age of political communication: Influences and features. Political Communication, 16, 209–230. Brosius, H. B., & Kepplinger, H. M. (1992). Beyond agenda-setting: The influence of partisanship and television reporting on the electorate’s voting intentions. Journalism Quarterly, 69, 894–901. Brown, J. A. (1992). The Major effect: Changes in party leadership and party popularity. Parliamentary Affairs, 45, 545–564. Davis, R. (1992). The press and American political. New York: Longman. Fisk, S., & Taylor, S. (1991). Social Cognition. New York: McGraw-Hill. Galnoor, I. (1982). Steering the polity: Communication and politics in Israel. Beverly Hills, CA: Sage. Golan, G., & Wanta, W. (2001). Second-level agenda setting in New Hampshire primary: A comparison of coverage in three newspapers and public perceptions of candidates. Journalism and Mass Communication Quarterly, 78, 247–259. Hacker, K. L. (Ed.), (1995). Candidate images in presidential elections. Westport, CT: Praeger. Hacker, K. L. (Ed.), (2004). Presidential candidate images. Rowman, New York: Littlefield Publishers, Inc. Hellweg, S. A., Dionisopoulos, G. N., & Kugler, D. B. (1989). Political candidate image: A state-of-the-art review in progress. In B. Dervin & M. J. Voigt (Eds.), Progress in communication sciences IX (pp. 43–78). Norwood, NJ: Ablex. Horowitz, D., & Lissak, M. (1989). Trouble in utopia: The overburdened polity of Israel. Albany: State University of New York Press. Iyengar, S., & Kinder, D. (1987). News that matters: Television and American opinion. Chicago: University of Chicago Press. Iyengar, S., & Simon, A. (1993). News coverage of the Gulf crisis and public opinion. A study of agenda-setting, priming, and framing. Communication Research, 20, 365–383.

24

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH

Kim, S. H., Scheufele, D. A., & Shanahan, J. (2002). Think about it this way: Attribute agenda-setting function of the press and public’s evaluation of local issue. Journalism and Mass Communication Quarterly, 79, 7–25. Kim, K., & McCombs, M. (2007). News story descriptions and the public’s opinions of political candidates. Journalism & Mass Communication Quarterly, 84, 299–314. King, P. (1997). The press, candidate image, and voter perceptions. In M. McCombs, D. L. Shaw & D. Weaver (Eds.), Communication and democracy: Exploring the intellectual frontiers in agenda-setting theory (pp. 29–40). Mahwah, NJ: Erlbaum. King, A. (2002). Do leaders’ personalities really matter? In A. King (Ed.), Leaders’ personalities and the outcomes of democratic elections (pp. 1–43). Oxford: University Press. Kinder, D. R. (1994). Reason and emotion in American political life. In R. Schank & E. Kanger (Eds.), Beliefs, Reasoning and Decision Making (pp. 277–314). NJ: Erlbaum Publishers. Kinder, D. R. (1986). Presidential character revisited. In R. R. Lau & D. O. Sears (Eds.), Political Cognition (pp. 233–255). Hillsdale, NJ: Erlbaum. Kinder, R. D., Peters, M. D., Abelson, R. P., & Fisk, S. T. (1980). Presidential Prototypes. Political Behavior, 2, 315–337. Kiousis, S., Bantimaroudis, P., & Ban, H. (1999). Candidate image attributes: Experiments on the substantive dimension second-level agenda setting. Communication Research, 26, 414–428. Kiousis, S., Mitrook, M., Wu, X., & Seltzer, T. (2006). First- and second-level agenda-building and agenda-setting effects: Exploring the linkages among candidate news releases, media coverage, and public opinion during the 2002 Florida gubernatorial election. Journal of Public Relations Research, 18, 265–285. Kleinnijenhuis, J., Van Hoof, A. M. J., Oegema, D., & De Ridder, J. A. (2007). A test of rivaling approaches to explain news effects: News on issue positions of parties, real-world developments, support and criticism, and success and failure. Journal of Communication, 57, 366–384. Krosnick, J. A., & Kinder, D. A. (1990). Altering the foundations of support for the president through priming. American Political Science Review, 84, 497–512. Mazzolini, G., & Schulz, W. (1999). ‘‘Mediatization’’ of politics: A challenge for democracy?. Political Communication, 16, 247–262. McCombs, M. (2004). Setting the agenda: The mass media and public opinion. Cambridge: Polity Press. McCombs, M., & Evatt, D. (1995). Los temas y los aspectos: Explorando una nueva dimension de la agenda setting [Objects and attributes: Exploring a new dimension of agenda setting]. Comunicacion y Sociedad, 8, 7–32. McCombs, M., Lopes-Escobar, E., & Llamas, J. P. (2000). Setting the agenda of attributes in 1996 Spanish general election. Journal of Communication, 50, 77–91. Mendelsohn, M. (1994). The media’s persuasive effects: The priming of leadership in 1988 Canadian election. Canadian Journal of Political Science, 27, 81–97. Miller, J. M., & Krosnick, J. A. (1996). News media impact on the ingredients of presidential evaluations: A program of research on the priming hypothesis.

CANDIDATE IMAGE IN ELECTIONS

25

In D. Mutz & P. Sniderman (Eds.), Political persuasion and attitude change (pp. 79–99). Ann Arbor, MI: University of Michigan Press. Miller, A. H., & Miller, W. E. (1976). Ideology and the 1972 election: Myth or reality—A rejoinder. American Political Science Review, 70, 832–849. Miller, A. H., Wattenberg, M., & Malanchuk, O. (1985). Schematic assessments of presidential candidates. American Political Science Review, 80, 521–540. Ohr, D., & Oscarsson, H. (September 18–21, 2003). Leader traits, leader image and voting choice. Meeting of European consortium for political research. Marburg, Germany. Pan, Z., & Kosicki, G. M. (1997). Priming and media impact on the evaluation of the president’s performance. Communication Research, 24, 3–30. Pancer, S. M., Brown, S. D., & Barr, C. W. (1999). Forming impression of political leaders: A cross-national comparison. Political Psychology, 20, 345–368. Peri, Y. (2004). Telepopulism: Media and politics in Israel. Stanford, CA: Stanford University Press. Pfiffner, J. P. (1994). The modern presidency. New York: St. Martin’s Press. Popkin, S. (1994). The Reasoning Voter: Communication and Persuasion in Presidential Campaigns, (2nd Ed.), Chicago: University. Chicago Press. Rahat, G. (2001). The politics of reform in Israel: How the Israeli mixed system came to be. In M. S. Shugart & M. P. Wattenberg (Eds.), Mixed-member electoral systems: The best of both worlds? (pp. 123–151). Oxford, England: Oxford University Press. Rahat, G., & Sheafer, T. (2007). The personalization(s) of politics: Israel, 1949–2003. Political Communication, 24, 65–80. Robinson, M. J., & Sheehan, M. A. (1983). Over the wire and on TV: CBS and UPI in campaign ’80. New York: Russell Sage Foundation. Roberts, M., & McCombs, M. (1994). Agenda-setting and political advertising: Origins of news agenda. Political Communication, 11, 249–262. Sears, D. O. (1969). Political behavior. In G. Lindzey & E. Aronson (Eds.), The handbook of social psychology, (Vol. 5, 2nd Ed., pp. 315–458). Reading, MA: Addison-Wesley. Sears, D. O. (1983). The person-positivity bias. Journal of Personality and Social Psychology, 44(2), 233–250. Shanks, J. M., & Miller, W. E. (1990). Policy direction and performance evaluation: Complementary explanations of the Reagan election. British Journal of Political Science, 20, 143–235. Sheafar, T. (2007). How to evaluate it: The role of story-evaluation tone in agenda setting and priming. Journal of Communication, 57, 21–39. Sheafer, T. (2008). Charismatic communication skill, media legitimacy and electoral success. Journal of Political Marketing, 7, 1–24. Sheafer, T., & Weimann, G. (2005). Agenda-building, agenda-setting, priming, individual voting intentions and the aggregate results: An analysis of four Israeli election. Journal of Communication, 55, 347–365. Shugart, M. S. (2001). ‘‘Extreme’’ electoral systems and the appeal of mixed-member alternative. In M. S. Shugart & M. P. Wattenberg (Eds.), Mixed-member electoral systems: The best of both worlds? (pp. 25–51). Oxford, England: Oxford University Press.

26

INTERNATIONAL JOURNAL OF PUBLIC OPINION RESEARCH

Stephen, T., Harrison, T., Husson, W., & Albert, D. (2004). Interpersonal communication styles of political candidate: Predicting winning and losing candidates in three U.S. Presidential Election. In K. Hacker (Ed.), Presidential Candidate Images (pp. 177–196). New York: Rowman and Littlefield Publishers. Stokes, D. E. (1996). Some dynamic elements of contests for the presidency. American Political Science Review, 60, 19–28. Wanta, W., Golan, G., & Lee, C. (2004). Agenda setting and international news: Media influence on public perceptions of foreign nations. Journalism and Mass Communication Quarterly, 81, 364–377. Wattenberg, M. (1995). The rise of candidate-centered politics. Cambridge: Harvard University Press. Weaver, D. H., Graber, D. A., McCombs, M. E., & Eyal, C. H. (1981). Media agenda-setting in a presidential election: Issues, image, and interest. New York: Praeger. Wilke, J., & Reinemann, C. (2001). Do the candidate matter? Long-term trends of campaign coverage—A study of the German press since 1949. European Journal of Communication, 16, 291–314. Wimmer, R., & Dominick, J. (1991). Mass Media Research. Belmont, CA: Wadsworth Publishing Co. Wolfsfeld, G. (2004). Media and the Path to Peace. Cambridge: Cambridge University Press.

BIOGRAPHICAL NOTES Meital Balmas is a Ph.D. candidate in the Department of Communication at the Hebrew University of Jerusalem. Her research interests include various facets of public opinion, political personalization, and media effects. Tamir Sheafer is a senior lecturer in the Departments of Communication and Political Science at the Hebrew University of Jerusalem. His research interests include issues such as media effects, political personalization and mediated public diplomacy.