Content-Type: text/html
Identifying Dimensions of Media Salience: A Factor Analysis of Media Coverage during the 2000 Presidential Election
Identifying Dimensions of Media Salience: A Factor Analysis of Media Coverage during the 2000 Presidential Election
Identifying Dimensions of Media Salience: A Factor Analysis of Media Coverage during the 2000 Presidential Election
By
Spiro Kiousis
Assistant Professor
Greenlee School of Journalism & Communication
124C Hamilton Hall
Iowa State University
Ames, IA 50011-1180
Office: 515-294-9967
Fax: 515-294-5108
[log in to unmask]
In August 2002, Assistant Professor of Public Relations, University of Florida
Submitted to the Communication Theory & Methodology Division for the annual convention of the Association for Education in Journalism and Mass Communication in Miami, Florida, held on August 7-10, 2002.
Identifying Dimensions of Media Salience: A Factor Analysis of Media Coverage during the 2000 Presidential Election
Identifying Dimensions of Media Salience: A Factor Analysis of Media Coverage during the 2000 Presidential Election
Identifying Dimensions of Media Salience: A Factor Analysis of Media Coverage during the 2000 Presidential Election
Abstract
Media salience - the key independent variable in agenda-setting research - has traditionally been explicated as a singular construct. Nevertheless, scholars have defined it and measured it using a number of different conceptualizations and empirical indicators. To address this limitation in research, this study builds a conceptual model of media salience suggesting it is a multidimensional construct consisting of three core elements: attention, prominence, and valence. Furthermore, the model is tested through an exploratory factor analysis of New York Times news coverage of eight major political issues during the 2000 presidential election. The data show that two dimensions of media salience emerge: visibility and valence. Based on the factor analysis, two indices are created to measure the construct that could be used as a research tool in future investigations. Finally, the implications of these findings are also considered.
Identifying Dimensions of Media Salience: A Factor Analysis of Media Coverage during the 2000 Presidential Election
Introduction
For nearly 30 years, the core concept in agenda-setting theory that scholars have sought to gauge is the transfer of issue salience. In most instances, researchers have examined how media emphasis on political issues raises their salience in public opinion (McCombs & Shaw, 1972; Funkhouser, 1973; McLeod, Becker, & Byrnes, 1974). Of course, the scope of agenda-setting research has been expanded to investigate the salience of political candidates (Weaver, Graber, McCombs, & Eyal, 1981), political advertising (Roberts, 1997), policy agendas (Brewer & McCombs, 1996), public relations materials (Turk, 1986), and more. While the concentration on salience resonates throughout the literature, few inquiries have formally explicated the concept. For those that have, the tendency has been to touch on aspects of public salience (e.g., Evatt, 1997; Evatt & Ghanem, 2001) or the broader concept of agendas themselves (e.g., Manheim, 1986). Few studies have exclusively converged on agenda se
tting's key independent variable, media salience. Thus, the chief purpose of this paper is to fill this void in research by engendering a conceptual model of media salience and empirically testing it by performing a factor analysis on media coverage of eight major political issues during the 2000 presidential election. Subsequently, we generate media salience indices for future scholarship in this area.
General Views of Salience
Salience is a concept that draws on several disciplines, including communication, political science, and psychology. Over time, a wide range of research has emerged utilizing the concept in empirical domains, but few parsimonious theoretical explications have been carried out (Salwen & Matera, 1996; Evatt & Ghanem, 2001). The term has been used interchangeably with concepts such as awareness, attention, concern, popularity, and importance (Edelstein, 1993; Burns, 1998). It is also closely connected to the constructs of interest, conspicuousness, relevance, awareness, and involvement (e.g., Schuman, Ludwig & Krosnick, 1986; Carter, 1965; Hill, 1991; see Evatt, 1997 for extensive discussion).
A critical obstacle that many researchers face when sifting through the concept's various theoretical explications is that they are frequently fragmented and consequently difficult to categorize. One strategy for distinguishing among the various descriptions is to isolate an element along which nearly all the definitions vary - namely, whether salience is internally or externally evident in an object.[1] For example, Augoustinos and Walker (1995), from the purview of cognitive psychology, understand salience as "the extent to which a stimulus, or referent object in the surrounding situation, stands out from other stimuli, or from other aspects of the situation" (p.86). Further, Sullins (1989) contends that salience encompasses how well an object stands out from its surroundings. Finally, Dearing and Rogers (1996), discussing the concept's meaning in agenda setting, articulate that salience refers to "the degree to which an issue is perceived as relatively important" (p.8, em
phasis added) - i.e., relative to other objects on an agenda. Therefore, a principal characteristic shared by such definitions is that salience consists of how an object is depicted among groups of other objects, stimuli, etc. - making the external environment central to these conceptions.
In comparison, other scholars define the salience of an object more in terms of its internal properties than in its relation to other objects. Carter (1965), for instance, discerns between salience and pertinence, with the former stressing one's experience with an object, while the latter highlights how one thinks about an object. A similar view is provided by Sears and Whitney (1973), who conclude that salience is associated with personal relevance. Edelstein (1993) submits that salience has typically been employed by researchers simply to discriminate among topics, leaving other facets of the concept largely ignored. In particular, borrowing from Weaver (1982) and Carter (1965), he argues that salience not only involves thinking of issues, but thinking about them as well. As a result, he speculates that news should be viewed in terms of "problematic situations." According to this framework, Edelstein, Ito, and Kepplinger (1989) propose five types of "problematics" (conditi
ons of need, deprivation, blocking, conflict, and uncertainty) and five possible responses to problematics (procedural, defining, conditional, completed, and evaluation of steps). Thus, the salience of objects, in this light, rests less on their external position, but more on their internal characteristics.
Of course, hybrid perspectives of salience are also prevalent, incorporating both internal and external elements into their definitions. For example, Evatt (1997), in an exhaustive analysis of how emotional news content shapes community salience of political issues, conducted a factor analysis that identified four types of public salience: social salience, excitement and personal salience, need, and worth. Likewise, Manheim (1986) identifies three dimensions of the media agenda: visibility, audience salience, and valence. By adopting the internal/external dichotomy introduced above, we see both external (e.g., social and visibility) and internal (e.g., need and valence) aspects in these definitions.
A somewhat novel description of salience can be extrapolated from a historical thesaurus, which places words into an elaborate, yet elegant classification system to organize their meanings (e.g., Roget, 1882).[2] McCombs and Ghanem (2001) recommend such a system for studying second-level agenda setting, but the same approach might be used for object salience - as the current project suggests. Specifically, the thesaurus affords a depiction of salience as both internally and externally derived. When looking at associated words, such as "projection," "sharpness," "manifestation," and "importance," this distinction is made apparent. For example, on the one hand, salience is linked to "projection" and "sharpness," which are indexed under the broad category of "space" - showing that location is integral to an object's salience (i.e., externally determined). On the other hand, salience is connected to "importance" and "manifestation," which are grouped under the general categories of "intellect" and "volition" - both of which typically describe an object's internal qualities. Thus, salience, in this context, is internally determined. Ultimately then, the salience of an object simultaneously hinges on its internal attributes, as well as how it compares to other objects, stimuli, and environments.
In summary, the salience of an object appears to be governed by both its internal and external characteristics. Though some general discussions of salience have been spawned in the literature, little agenda-setting scholarship has rigorously attempted to conceptualize media salience to the extent that the current project does. Nevertheless, it is clear that agenda-setting researchers have basically adopted an external definition of the concept, where an object is evaluated based on its relationship to other objects in the media (e.g., issues are salient relative to how other topics are situated on the media, public, & policy agendas). Even though direct scholarly consideration of media salience is minimal, myriad indirect references can be found in the agenda-setting literature. By synthesizing the direct discussions of salience with the indirect facets evident in operational measures, we can build a conceptual and methodological framework that future scholarship can utilize.
Accordingly, we now examine some of the key dimensions of media salience underscored in prior research.
Media Salience in Agenda Setting
When perusing through agenda-setting scholarship, three dimensions of media salience seem to emerge: attention, prominence, and valence. Although some of these facets are explicitly discussed in investigations, most emanate from operational measures of the "media agenda." In order to connect the portrayal of media salience evident in operational measures with the more theoretical explanations of salience specified above, this section shows that the first two dimensions are externally grounded (attention & prominence), while the third is internally grounded (valence). Before continuing, it should be recognized that this conceptualization is not intended to be an exhaustive account of every aspect of media salience, nor are the indicators detailed below distinctly separate, for there certainly is overlap. But instead, this should be viewed as a starting outline of the concept's primary components.
Attention
Perhaps the most common approach for measuring media salience is by attention. From this perspective, salience is akin to media awareness of an object, usually gauged by the sheer volume of stories or space about topics in newspapers, television news, etc. (e.g., Benton & Frazier, 1976; Gordon & Heath, 1981; Golan & Wanta, 2001; Pritchard, 1986). For instance, Dearing and Rogers (1996) argue that "the number of news stories measures the relative salience of an issue on the media agenda" (p.18, emphasis in original). Several scholars have implemented this approach for estimating media salience (e.g., Palmgreen & Clarke, 1977; Atwood, Sohn, & Sohn, 1978; Burns, 1998). Funkhouser (1973), for one, implicitly connoted his application of this scheme for collecting news magazine stories when writing that:
no attention was paid to the actual amount of space, nor to the content of the material, nor to any bias or slant that might be involved. The tallies for these magazines were added together, thus providing a rough measure of what the news media were emphasizing, when (p.37).
Further, Benton, and Frazier (1976) incorporated a similar procedure for tracking media salience of economic issues in television and print news. In particular, they scrutinized the total amount of newspaper space and television time that concentrated on fiscal issues. Palmgreen and Clark (1977), in an examination of national and local issues, analyzed newspapers and television news transcripts in this same manner.
The attention approach has also been valuable for measuring salience in intermedia investigations, such as Gilberg, Eyal, McCombs, and Nicholas (1980), where total news content from the New York Times, Washington Post, and national television news broadcasts constituted the media agenda. Finally, in recent efforts, researchers have extended this approach to include news content from editorial/op-ed pages, cartoons, and letters to the editor (e.g., Dalton, Beck, and Huckfeldt., 1998). In fact, Burns (1998) employed this strategy for ascertaining the influence of radio station playlists on people's perceived popularity of music. Hence, attention is a significant dimension of media salience in agenda-setting inquiries. In relation to the general definitions of salience mentioned above, attention is externally based because issues are salient with respect to the total length of space allotted to them in the entire media or in an entire media text.
Prominence
Another instrumental part of media salience is prominence. As applied here, prominence refers to the position of a story within a media text that communicates its importance. Ghanem's (1997) explanation of "framing mechanisms" may best capture this dimension whereby a story's salience is influenced by its placement, size, pictures, pull quotes, and other aesthetic devices (see Williams, 1985 for similar examples). This is different from the attention dimension because placement and position are the most paramount characteristics instead of volume. Manheim (1986) collapses attention and prominence into the category of visibility, but some literature seems to suggest differences between them. For example, Watts, Mazz, and Snyder (1993) argue that "coverage prominence measures are superior to story counts because they acknowledge structural and presentational elements of the news story. Stories in the media indicate their importance (and thus the agenda of the media) to the aud
ience by virtue of their placement, length, or treatment" (pp. 414).
The widespread application of the prominence approach can be found in studies assessing lead coverage (e.g., Behr & Iyengar, 1985; Winter & Eyal, 1981). McCombs and Shaw's (1972) inaugural agenda-setting study, for instance, employs such a procedure by only counting content that appeared within the first three stories of television news, on the front-page of the newspaper, in the beginning news section of magazines, and in the lead editorial column of editorial pages. Any investigation that has monitored front-page newspaper coverage is essentially gauging salience as prominence (e.g., McLeod et al., 1974; Gordon & Heath, 1981; Wanta, 1997; Smith, 1987).
Beyond placement and position, the presence of a story in a prestigious news source is another signal of prominence that relays salience cues to audiences. Thus, inquiries probing elite, national media such as the New York Times and Washington Post are also rooting their measures of media salience in prominence (e.g., Rogers & Chang, 1991; Reese & Danielian, 1989; Singer & Ludwig, 1987). As Winter and Eyal (1981) succinctly state when explaining their rationale for using the New York Times, "it is the elite U.S. newspaper" (pp. 379, emphasis in original). In addition to print channels, inquiries that trace elite, national television coverage are also measuring prominence (e.g., Wanta & Hu, 1994; Zucker, 1978; Zhu & Boroson, 1997; West, 1994). The essential assumption of these researchers is that elite media sources influence the agendas of non-elite media sources and consequently enhance the generalizibility of using data from these news outlets. As a result, prominence plays a crucial role in shaping media salience. Similar to attention, this area of media salience is also externally driven because a story is salient with respect to its location within media texts (i.e., within a cluster of other stories, articles, advertisements, etc.)
Valence
Despite the traditional focus of agenda setting on what and how people think about news, affective elements of news are also tied to the concept of media salience, especially when pondering the recent conceptualization of second-level agenda setting. Specifically, second-level agenda setting suggests that media not only affect the salience of "objects" (e.g., issues, candidates, etc.) in the news, but the salience of object "attributes" (e.g., the properties & qualities that describe those objects) as well (e.g., McCombs & Estrada, 1997; McCombs, 1995; Lopez-Escobar, Lennon & Llamas, 1998; Lopez-Escobar, Llamas, & McCombs, 1998; Huckins, 1998; Takeshita, 1997). Substantive (cognitive) and affective (emotional) attributes represent two of the major classes of attributes in the news (McCombs & Ghanem, 2001; Ghanem, 1997; King, 1997). The latter class, of course, deals with the valence dimension of media salience.
The valence of news has not just been germane to second-level investigations, but to traditional first-level examinations too. For example, Manheim (1986) explicitly declares that valence is a central dimension of the media agenda. Furthermore, Mutz (1998) articulates that the conventional question scrutinizing public salience ("what is the most important problem facing the nation today?") has affective overtones in it. After all, the term "problem" will not usually secure purely cognitive responses from audiences. In addition, Mutz (1998) also shows that some studies have operationalized media salience in terms of negative news content (e.g., Mackuen & Coombs, 1981). In particular, she writes that:
for many issues, salience and direction coincide. For example, valence issues such as employment, inflation, crime, education, or poverty have only one side in the sense that typically no one is for crime or against education. Thus, in practice, much evidence of agenda setting turns out to be evidence that more negative coverage leads people to think certain issues are more problematic" (pp. 71).
In second-level examinations, research has underscored how affective attributes of news shape audience perceptions of those attributes (e.g., McCombs, Llamas, Lopez-Escobar, & Rey, 1997; Ghanem, 1996). Though further analyses of affective attributes at the second level are needed, attribute coverage also conveys cues that shape the overall affective salience of issues, candidates, and other objects (e.g., how interesting or appealing they are). Therefore, affective elements in news can also enhance or reduce overall salience of objects (Kiousis, Bantimaroudis, & Ban, 1999).
In methodological terms, one effective technique for calibrating valence might be to locate the amount of conflict in a story - the higher the conflict, the higher the valence score. Another approach could be to code the number of stories in print or broadcast media that had a positive or negative tone towards the object of a story - the higher number, the higher the valence score. In such situations, more positive stories would indicate higher salience in some instances while more negative stories would indicate higher salience in other cases. A final approach might be to have stories that have either a positive or negative tone measure salience. In short, the distinction here would be between valenced and neutral content. This third alternative appears more in line with the classical trajectory of agenda-setting research, which makes few distinctions with regard to positive or negative news coverage. Consequently, the distinction between valenced or neutral news seems most
appropriate. Compared to the first two dimensions, affective measures of salience are internally driven because content within stories fuels this dimension of media salience.
A Comprehensive Model
Based on the aforementioned theoretical explication, the salience of an object is simultaneously determined by its position within an environment (external) and by its implicit qualities or properties (internal). Thus, we assert that media salience is a multidimensional construct. For agenda setting, three aspects of media salience operate within this dichotomy. In particular, they are: attention (external), prominence (external), and valence (internal). Visually, this model might look like this.
- - - Figure 1 Here - - -
Research Question
At this juncture, it is vital that we empirically test the model submitted, for too often researchers fail to combine conceptual models with empirical evidence. Therefore, we will monitor media coverage of a set of issues and ascertain how well the model functions. Eventually, more complex factor analyses will be needed to pinpoint other areas of media salience that may exist.
Methods
Sample
To ascertain the utility of the theoretical model advanced above, an exploratory content analysis was performed on media coverage of eight political issues during the 2000 presidential campaign in order to create indices for measuring media salience. Several agenda-setting studies have been completed during election years (e.g., McCombs & Shaw, 1972; Gilberg et al., 1980; Wanta, 1997), making this choice ideal for building on existing research about the theory. The news outlet used for the analysis was the New York Times, which has been identified as a key gatekeeper in national news coverage by many scholars (Dearing & Rogers, 1996; Rogers & Chang, 1991; Reese & Danielian, 1989). In some sense, we are limiting the analysis to a "prominent" source, but such a step was required to make the analysis manageable and generalizible.
To secure a comprehensive look at media coverage, content was collected over the entire 2000 election year. Monthly tallies were kept from January through December to assess trends in coverage over time. Thus, the unit of analysis was the number of stories in the New York Times per month. Stone and McCombs (1981) maintain that this is a good indicator for measuring salience because of its high correlation with other measures of space such as column inches. For most issues, daily and weekly tallies did not generate enough variance for statistical comparisons, prompting the decision to use the longer time frame as the unit of analysis.
The eight political issues chosen were the economy, education, crime/violence, health care, taxes, morality/values, social security, and defense (military). The choice of eight issues seemed appropriate given that past communication research has reasoned that the public pays attention to between five and seven different issues at one time (e.g., Shaw & McCombs, 1977; McCombs & Ghanem, 2001), and that past psychological research has also noted that people can remember between five and nine topics at once (Klatzsky, 1975). The issues were selected through a preliminary examination of various public opinion polls rating the most important issues of the campaign and nation in 2000. The specific polls inspected were the Harris, CBS/New York Times, ABC News/Washington Post, Newsweek, and Gallup. Polls were probed throughout the year to isolate issues that were consistently salient in public opinion over time.
An independent, trained human coder conducted keyword searches of New York Times stories in Lexis/Nexis to gather the sample. The keywords used were "economy," "education," "crime" or "violence," "health care," "taxes," "morality" or "values," social security," and "defense" (with a military focus). Keywords had to appear within the headline or lead of the story. Given the inverted pyramid style of print news coverage, this pinpointed the most relevant stories.
Operational Measures
To review, the three dimensions of media salience that needed to be operationalized were attention, prominence, and valence. The attention dimension of media salience was measured by two variables. For the first indicator, the total number of stories per month (anywhere in the newspaper) containing the keyword in the headline or lead was used. This gave us a measure of total mentions of issues in news coverage. For the second indicator of attention, the coder judged which stories from the initial list were "relevant." All subsequent measures were based on the list of "relevant" stories. In total, these two indicators measured the cumulative volume of attention. A coder reliability check, utilizing a subsample of stories, was conducted by two other coders to validate the judgments made by the principal coder. The coder reliability figures from the check exceeded 80 percent, acceptable by most standards of content analysis (Riffe, Lacy, & Fico, 1998).
Similar to attention, prominence was operationalized by two variables as well. The first indicator was the total number of stories per month containing the keywords on the front-page of the New York Times. The second indicator was the total number of stories containing the keywords per month appearing in the front-section of the newspaper. Thus, these indicators captured the importance of where stories are placed relative to other news content.
Valence was operationalized by three variables. In addition to the appearance of keywords, each story was classified as positive, neutral, or negative by the principal coder. For example, the first set of indicators gauged the total number of "relevant" stories per month that were positive, neutral, or negative. The second set of indicators gauged the number of front-page stories per month that were positive, neutral, or negative. Finally, the third set of indicators gauged the number of front-section stories per month that were positive, neutral, or negative. To simplify the analysis, the number of positive and negative stories was collapsed into a single measure to essentially create an indicator measuring the number of "valenced" stories (as discussed in the literature review). The amount of neutral stories was not used in this analysis. After the transformation, we were left with three indicators that assessed valence: the (1) total number of valenced stories, (2) the n
umber of front-page valenced stories, and (3) the number of front-section valenced stories. A second coder reliability check was performed on the valence measures collected. Similarly, the reliability figures again surpassed 80 percent, increasing our confidence in the data.
Data Analysis
The data-analysis strategy for this project was loosely based on the approach used by Evatt and Ghanem (2001) in developing a scale for measuring public salience. To analyze the data, a principal component factor analysis with varimax rotation was executed to determine whether the conceptual model advanced above corresponded with the empirical evidence provided by the data. A total of 96 cases (8 issues x 12 months) were used for the analysis. The total number of stories underlying those cases were as follows: economy=1,978, taxes=2,886, education=3,536, health care=1,461, social security=537, crime/violence=5,951, morality/value=2,919, and defense=1,449. For purposes of interpretation, only factor loadings exceeding the conservative threshold of ((.50 were considered meaningful, an approach recommended by many statistical scholars (e.g., Hair, Anderson, & Latham, 1987). The final step in the data analysis involved calculating the alpha-reliability scores of the variable sub
sets that emerged from the factor analysis to assist in generating indices for measurement of media salience in future research.
Results
In the literature review, we posited that three primary dimensions of media salience existed: attention, prominence, and valence. Our broad research question asked whether these dimensions would appear in an empirical analysis of political issue coverage during a national election year. Prior to examining the main research question, it is useful to look at the aggregate data to ascertain general trends in New York Times coverage. Figure 2 shows the proportion of coverage for the eight issues analyzed, and Figure 3 shows the distribution of coverage for the eight issues over the 12-month time period. These data were based on the total mentions in the newspaper.
- - - Figure 2 Here - - -
- - - Figure 3 Here - - -
With respect to overall mentions, the media agenda was led by crime/violence and education, reflecting a large focus on social issues during the election year. In particular, crime/violence appeared to be the most volatile issue, fluctuating from approximately 400 to 600 stories at its lowest and highest monthly points. Coverage for both issues peaked in October, understandable given the national-election date in early November. The decline in November and December can probably be heavily attributed to the rise of the Florida vote-count controversy following the presidential election. At the other end of the spectrum, social security and defense received the least coverage and remained at fairly consistent levels throughout the year. The top fiscal issue was taxes, not surprising given the extensive debates among the major political parties and presidential candidates concerning how to handle the nation's budget surplus at that time. Overall, the sample seemed to adequately
represent what might be expected of national coverage during the 2000 election year. After examining the general coverage trends, we can now turn to the more pertinent findings of how many factors of media salience surfaced from the data analysis.
In general, the data suggested that two factors of media salience exist. Table 1 shows the factor loadings. The two factors have been labeled visibility and valence.
- - - Table 1 Here - - -
The first factor, consisting primarily of the attention and prominence indicators, accounts for 63 percent of the data's total variance. There is some overlap between factors, though, because the front-section valence variable loaded highly on both factors. In general, the data show that few distinctions are apparent in terms of salience between content located in prominent areas of the newspaper and content located throughout the entire newspaper. Thus, Manheim's (1986) dimension of visibility (attention and prominence) is supported by the evidence here, but more research is needed to replicate the present study's findings.
In comparison, the second factor consists exclusively of the valence indicators and accounts for 16 percent of the total variance. As a result, regardless of whether valenced issue content appeared on the front page, front section, or anywhere in the newspaper, this factor loading indicates that it substantially differs from content that is not valenced. Of course, we must recognize that the factor loadings are not clearly distinguishable given the overlap of the front-section valenced-content measure, yet the collective data do hint at differences.
In addition to the factor analysis, the next step in our investigation was to test the reliability of the variable sets that scored highly on the factor loadings. Accordingly, the following items were included in the valence factor: front-page stories that are valenced, front-section stories that are valenced, and the number of relevant stories that are valenced. The alpha reliability score for this variable set was (=+.66. When removing the front-page indicator, reliability increased to (=+.79. As a result, the revised group consists of the remaining two variables.
The following items comprised the visibility factor: front-section stories that were valenced, relevant stories, total stories, front-page stories, and front-section stories. The alpha reliability score for this variable set was (=+.60. When the item measuring valenced stories in the front-section was removed, the reliability increased to (=+.62. Because this item fits better conceptually with the valence factor and because the reliability of the set slightly improves, the revised group is comprised of the remaining four variables. Nunnally (1978) suggests that alpha reliability estimates above +.60 are acceptable. Table 2 shows the final variable sets after the factor analysis and reliability check.
- - - Table 2 Here - - -
Implications & Conclusions
To reiterate, the current project asserted that media salience is a multifaceted concept composed of three dominant dimensions: attention, prominence, and valence. The first two dimensions measure an object's external salience, while the latter dimension measures an object's internal salience. An exploratory empirical test of the model exposed that it is promising, but further work is needed to judge its potential. Since its inception, agenda-setting scholars have operated under the impression that media salience is a singular construct. The current project demonstrates the problems with such a view and hopes to yield future work that will refine definitions of media salience to the point where theoretical and methodological discussions can more accurately capture the construct's primary components.
Though the initial model speculated that media salience consisted of three core dimensions, the empirical data uncovered two. In particular, the first dimension deals with the visibility of coverage given to issues in media content, and the second seems to deal with the degree to which media coverage is valenced. At this point, little evidence was documented to make a distinction between prominence and the other two dimensions. Of course, the exploratory nature of the project means that it should not yet be dismissed as a key dimension of media salience, but this initial evidence - and the support of Manheim's (1986) assertions - does point in that direction. In addition to the indicators developed here, other variables should be introduced to locate new factors and test out the robustness of those identified in the present study.
Based on the findings, it seems that future agenda-setting work should make a distinction between the visibility of coverage paid to public issues and the extent to which that coverage contains valenced content. A natural progression in this line of research would be to assess the relative influence of these different dimensions of media salience on perceived salience of public issues. That is, is the visibility of media coverage a more powerful force in impacting public opinion or is the amount of valenced content surrounding an issue a stronger force? The answers to such questions have significant implications on our understanding of the relationship between the media and public. For example, while agenda setting usually converges on questions involving how people think about public issues, a stronger impact for valenced content on perceived salience might suggest that more agenda-setting research be executed in the domain of public attitudes instead of only cognitions. In
deed, priming theory (e.g., Iyengar & Simon, 1993; Krosnick & Kinder, 1990; Iyengar & Kinder, 1987) and some recent studies of second-level agenda setting (e.g., McCombs, & Shaw, 1993; McCombs et al., 1997; Kiousis, Bantimaroudis, & Ban, 1999) point toward such attitudinal consequences as a result of shifts in media salience.
While this study looked at the traditional "object" of issues, more contemporary literature proposes that we test the current paper's model on political candidates and other topics in the news as well. In addition, the model should be tested on both "first-level" and "second-level" media salience. For example, the valence dimension may be more influential at one level, while the visibility dimension may be more paramount at another. These distinctions can then help us better establish the role of agenda setting in society
Despite its potential, the current study had its shortcomings. For example, the reliability figures produced from the factor analysis were somewhat low. Of course, the exploratory nature of the study decreases the chances of gaining higher figures at this point. The small number of variables used to develop each factor was also a limitation. Future scholars would be wise to generate more variables to represent the identified factors. Finally, the use of only a single newspaper is problematic. Still, the dominance of the New York Times in the national media should help allay such concerns (Dearings & Rogers, 1996). Future research should include broadcast and online news as well.
In summary, the central independent variable in agenda-setting theory for nearly three decades has been media salience. While historically deemed a singular construct, the current project proposes that it may be multidimensional. As a result, future research would benefit by further explicating this crucial variable in media effects.
Notes
Identifying Dimensions of Media Salience: A Factor Analysis of Media Coverage during the 2000 Presidential Election
Figures
Figure 1: Conceptualization of Media Salience
Media Salience
Internal
External
Prominence
Attention
Valence
Identifying Dimensions of Media Salience: A Factor Analysis of Media Coverage during the 2000 Presidential Election
Figure 2: Proportion of Total New York Times Coverage of 8 Political Issues in 2000
Figure 3: Monthly Proportion of Total New York Times Coverage of 8 Political Issues in 2000
Identifying Dimensions of Media Salience: A Factor Analysis of Media Coverage during the 2000 Presidential Election
Tables
Table 1: Factor Loadings of Media Salience during the 2000 Presidential Election
Factor 1: Visibility
Factor 2: Valence
Media Salience Dimensions
Front-Page Valenced Stories
0.89
Front-Section Valenced Stories
0.62
0.54
Total Valenced Stories
0.85
Total Story Frequency
0.85
Relevant Stories
0.89
Front-Section Stories
0.94
Front-Page Stories
0.88
Explained Variance
63%
18%
Table 2: Indices Created for Dimensions of Media Salience
Visibility (alpha=+.62)
Valence (alpha=+.79)
Relevant Stories
Total Stories
Front-Page Stories
Front-Section Stories
Front-Section Valenced Stories
Total Valenced Stories
References
Atwood, L.E., Sohn, A.B., & Sohn, H. (1978). Daily newspaper contributions to community discussions. Journalism Quarterly 55, 570-576.
Augoustinos, M., & Walker, I. (1995). Social Cognition: An Integrated Introduction.
Thousand Oaks, CA: Sage.
Behr, R., & Iyengar, S. (1985). Television news, real-world cues and changes in the public agenda. Public Opinion Quarterly 58, 479-508.
Benton, M., & Frazier, P. (1976). The agenda setting function of the mass media at three levels of "information holding." Communication Research 3, 261-274.
Brewer, M., & McCombs, M. (1996). Setting the community agenda. Journalism &
Mass Communication Quarterly 73, 1, 7-16.
Burns, J.E. (April, 1998) The Agenda Setting Process in Music Radio: A Purposive Sample of Listeners. Paper presented at Broadcast Education Association annual convention, Las Vegas, NV.
Carter, R.F. (1965). Communication and affective relations. Journalism Quarterly 42,
203-212.
Dalton, R.J., Beck, P.A., Huckfeldt, Koetzle, W. (1998). The test of media-centered
agenda setting: Newspaper content and public interests in a presidential election.
Political Communication 15, 463-481.
Dearing, J.W., & Rogers, E.M. (1996). Communication Concepts 6 Agenda-Setting. London: Sage.
Edelstein, A.S. (1993). Thinking about the criterion variable in agenda-setting research.
Journal of Communication 43, 2, 85-99.
Edelstein, A.S., Ito, Y., & Kepplinger, H.M. (1989). Agenda-setting in the multicultural context. In Communication and Culture: A comparative approach (pp. 224-255). New York: Longman.
Evatt, D. (1997). The Influence of Emotion-Evoking Content of News on Issue
Salience. Ph.D. Dissertation, University of Texas at Austin.
Evatt, D.S., & Ghanem, S.I. (2001). Building a scale to measure salience. Paper presented to the World Association of Public Opinion Research annual conference in Rome, Italy, from September 20-22, 2001.
Funkhouser, G.R. (1973). The issues of the sixties: An explicatory study of the
dynamics of public opinion. Public Opinion Quarterly 37, 62-75.
Ghanem, S. (1997). Filling in the tapestry: The second level of agenda-setting. In M. McCombs, D.L. Shaw, & D. Weaver (Eds.), Communication and Democracy(pp.3-14). Mahwah, New Jersey: Lawrence Erlbaum Associates.
Ghanem, S. (1996). Media Coverage of Crime and Public Opinion: An Exploration of
the Second Level of Agenda setting. Ph.D. dissertation, University of Texas at
Austin.
Gilberg, S, Eyal, C., McCombs, M., & Nicholas, D. (1980). The state of the union
address and the press agenda. Journalism Quarterly 57, 584-588.
Golan, G. & Wanta, W. (2001). "Second-level agenda setting in the New Hampshire primary: A comparison of coverage in three newspapers and public perceptions of candidates. Journalism & Mass Communication Quarterly 78, 247-259.
Gordon, M.T., & Heath, L. (1981). The news business, crime, and fear. In D.A. Lewis
(Ed.), Reactions to Crime (pp.227-251). Beverly Hills, CA: Sage.
Hill, D.B. (1991). Viewer characteristics and agenda-setting by television news. In D.L. Protess & M. McCombs (Eds.), Agenda Setting: Readings on Media, Public Opinion, and Policymaking (pp.141-147). Hillsdale, NJ: Lawrence Erlbaum Associates.
Iyengar, S., & Kinder, D.R. (1987). News that Matters. Chicago, Illinois: 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. In S. Iyengar & R. Reeves (Eds.) Do the Media Govern? Politicians, Voters, and Reporters in America (pp. 248-257). Thousand Oaks, CA: Sage.
King, P. (1997). The press, candidate images, and voter perceptions. In M. McCombs, D.L. Shaw, & D. Weaver (Eds.), Communication and Democracy (pp.29-40). Mahwah, New Jersey: Lawrence Erlbaum Associates.
Kiousis, S., Bantimaroudis, P., & Ban, H. (1999). Candidate image attributes: Experiments on the substantive dimension of second level agenda setting. Communication Research 26, 4, 414-428.
Klatzky, R. L. (1975). Human memory: Structures and processes. New York: Freeman and Company.
Krosnick, J.A., & Kinder, D.R. (1990). Altering the foundations of support for the president through priming. American Political Science Review 84, 2, 497-512.
Lopez-Escobar, E., Llamas, J.P., & McCombs, M. (1998). Agenda Setting and Community Consensus: First and Second Level Effects. International Journal of Public Opinion Research10 (4) 335-348 (1998).
Lopez-Escobar, E., Llamas, J.P., McCombs, M., & Lennon, F. (1998). Two levels of agenda setting among advertising and news in the 1995 Spanish elections. Political Communication 15, 225-238.
MacKuen, M.B. & Coombs, S.L. (1981). More Than News: Media power in public affairs. Beverly Hills: Sage.
Manheim, J.B. (1986). A model of agenda dynamics. In M.L. McLaughlin (Ed.), Communication Yearbook 10 (pp. 499-516). Newbury Park, CA: Sage.
McCombs, M. (1995, September). The media outside and the pictures in our heads:
Surveying the second-dimension of agenda-setting. Paper presented at "New trends in communication," A conference of the Universita Degli di Roma La Sapienza, Rome, Italy.
McCombs, M., & Ghanem, S.I. (2001). The convergence of agenda setting and framing. In S.D. Reese, O.H. Gandy. Jr., & A.E. Grant (Eds.), Framing Public Life: Perspectives on media and our understanding of the social world (pp. 67-83). Mahwah, NJ: Lawrence Erlbaum Associates, Publishers.
McCombs, M., & Estrada, G. (1997). The news media and the pictures in our heads. In S. Iyengar & R. Reeves (Eds.), Do the Media Govern? (pp. 237-247). London: Sage
McCombs, M., Llamas, J.P., Lopez-Escobar, E., & Rey, F. (1997). Candidate images in Spanish elections: second-level agenda-setting effects. Journalism & Mass Communication Quarterly 74,4, 703-717.
McCombs, M.E., & Shaw, D.L. (1993). The evolution of agenda-setting research: Twenty-five years in the marketplace of ideas. Journal of Communication 43, 58-67.
McCombs, M.E., & Shaw, D.L. (1972). The agenda-setting function of mass media. Public Opinion Quarterly 36, 176-187.
McLeod, J., Becker, L.B., & Byrnes, J.E. (1974). Another look at the agenda-setting
function of the press. Communication Research 1, 131-166.
Mutz, D. (1998). Impersonal Influence: How perceptions of mass collectives affect political attitudes. New York: Cambridge University Press.
Nunnally, J.C. (1978). Psychometric Theory, 2nd edition. New York: McGraw Hill.
Palmgreen, P., & Clarke, P. (1977). Agenda-setting with local and national issues.
Communication Research 14, 435-452.
Pritchard, D. (1986). Homicide and bargained justice: The agenda-setting effect of crime news on prosecutors. Public Opinion Quarterly 50, 143-159.
Reese, S., & Danielian, L. (1989). Intermedia influence and the drug issue: Converging
on cocaine. In P. Shoemaker (Ed.), Communication Campaigns about Drugs (pp.
29-46). Hillsdale, NJ: Lawrence Erlbaum Associates.
Riffe, D., Lacy, S., & Fico, F.G. (1998). Analyzing Media Messages: Using quantitative content analysis in research. Mahwah, New Jersey: Lawrence Erlbaum Associates.
Roberts, M. (1997). Political Advertising's Influence on News, the Public and Their
Behavior. In M. McCombs, D.L. Shaw, & D. Weaver (Eds.), Communication
and Democracy (pp.85-98). Mahwah, New Jersey: Lawrence Erlbaum Associates.
Rogers, E.M., & Chang, S. (1991). Media coverage of technology issues: Ethiopian
drought of 1984, AIDS, Challenger, and Chernobyl. In L. Wilkens & P. Patterson (Eds.), Risky Business: Communicating Issues of Science, Risk, and Public Policy (pp. 75-96). New York: Greenwood.
Roget, P.M., M.D., F.R.S. (1882). Thesaurus of English Words and Phrases. New
York: John R. Anderson & Co.
Salwen, M.B., & Matera, F.R. (1996). Agenda-setting with an Ethnic-Relevant Topic:
Ethnicity and Public Salience of Illegal Immigration. Paper presented to the Association for Education in Journalism and Mass Communication. Anaheim, CA: August 1996.
Schuman, H., Ludwig, J., & Krosnick, J. A. (1986). The perceived threat of nuclear war, salience, and open questions. Public Opinion Quarterly 50, 519-536.
Sears, D.O., & Whitney, R.E. (1973) Political persuasion. In I. De Sola Pool (Ed.), Handbook of Communication. Chicago, Illinois: Randy McNally College Publishing Company.
Shaw, D.L. & McCombs, M.E. (Eds.). (1977). The Emergence of American Political Issues: The agenda setting function of the press. St. Paul, MN: West.
Singer, E., & Ludwig J. South Africa's Press Restrictions: Effects on Press Coverage and Public Opinion Toward South Africa. Public Opinion Quarterly 51, 3, 315-334.
Smith, K. (1987). Newspaper coverage and public concern about community issues. Journalism Monographs 101.
Stone, G.C., & McCombs, M.E. (1981). Tracing the time lag in agenda-setting. Journalism Quarterly 58, 151-155.
Sullins, E.S. (1989). Perceptual salience as a function of nonverbal expressiveness.
Personality and Social Psychology Bulletin 15, 4, 584-595.
Turk, V. Information subsidies and media content: A study of public relations influence on the news. Journalism Monographs 100.
Wanta, W. (1997). The Public and The National Agenda. Mahwah, New Jersey: Lawrence Erlbaum Associates.
Wanta, W., & Hu, Y. (1994). Time-lag differences in the agenda-setting process: An
examination of five news media. International Journal of Public Opinion Research 6, 225-240.
Watts, J.H., Mazza, M., & Snyder, L. Agenda setting effects of television news coverage and the effects decay curve. Communication Research 20, 3, 408-435.
Weaver, D.H. (1982). Media agenda-setting and media manipulation. In D.C. Whitney,
E. Wartella, & S. Windahl (Eds.), Mass Communication Review Yearbook 3
(pp. 537-554). Newbury Park, CA: Sage.
Weaver, D.H., Graber, D.A., McCombs, M.E., Eyal, C.H. (1981). Media agenda- setting in a presidential election. New York: Praeger.
West, D.M. (1994). Television advertising in election campaigns. Political Science Quarterly 109, 5, 789-809.
Williams, W., Jr. (1985). Agenda-setting research. In J.R. Dominick & J.E. Fletcher (Eds.), Broadcasting Research Methods (pp. 189-201). Boston, MA: Allyn and Bacon, Inc.
Winter, J., & Eyal, C. (1981). Agenda setting for the civil rights issue. Public Opinion Quarterly 45, 376-383
Zucker, H.G. (1978). The variable nature of news media influence. In B. D. Ruben
(Ed.), Communication Yearbook 2 (pp. 225-245). New Brunswick, NJ:
Transaction Books.
[1] Recent theoretical work has stressed that "object" and "attribute" salience are the key areas of focus in traditional and contemporary agenda-setting research (e.g., McCombs & Ghanem, 2001). Objects, from this perspective, are similar to the notion of attitude objects in psychology.
[2] For example, the six basic categories are abstract relations, space, matter, intellect, volition, and affections. However, these general categories can be broken down into 1000 subclasses (see Roget, 1882 for further discussion).