uncorrected proofs

3 downloads 0 Views 932KB Size Report
Oct 31, 2013 - ment, during the 1969 US presidential election, of the relationship between the patterns of news ..... gubernatorial and US senatorial elections and the 2010 Texas gubernatorial election ..... St. Paul, MN: West. Stromback, J.
14

PR O

O

Maxwell E. McCombs and Lei Guo

FS

Agenda-setting Influence of the Media in the Public Sphere

U

N

C

O R

R

EC TE

D

Nearly a century ago, Walter Lippmann (1922) succinctly identified the setting for the news media’s role in the formation of public opinion with the phrase “the world outside and the pictures in our heads.” This phrase was the title of his opening chapter in Public Opinion, where, through historical examples and ­anecdotes, Lippmann made the argument that the news media are the principal bridge between the broad arena of public affairs and our perceptions of this arena. He elaborated the role of the news media in transmitting a truncated ­version of that world outside to the public, whose mental pictures are the pseudoenvironment that is the basis of public opinion and behavior. A quarter-century later, Maxwell McCombs and Donald Shaw (1972) ­empirically tested Lippmann’s thesis that the truncated versions of the world ­outside presented by the news media are a primary source of citizen’s perceptions of public affairs. Their seminal Chapel Hill study undertook the precise measurement, d ­ uring the 1969 US presidential election, of the relationship between the  patterns of news coverage for public issues and the voters’ perceptions of what were the most important issues of the day. Specifically, McCombs and Shaw ­examined the degree of correspondence between the news coverage for the top five issues in the 1968 presidential campaign and the level of importance that voters accorded these issues. Their hypothesis was that the salience of these issues on the  media’s agenda influenced the salience of the same issues on the public’s agenda.  The agenda-setting role of the news media is the transmission of issue salience from the media’s agenda to the public’s agenda. Or, in metaphoric terms, McCombs and Shaw hypothesized that the media’s agenda set the public’s agenda. The term “agenda,” as used here, is a neutral descriptive term, unlike in “to have an agenda.” The media’s agenda is the pattern of news coverage for the major The Handbook of Media and Mass Communication Theory, First Edition. Edited by Robert S. Fortner and P. Mark Fackler. © 2014 John Wiley & Sons, Inc. Published 2014 by John Wiley & Sons, Inc.

0002065169.INDD 251

10/31/2013 10:27:32 PM

252

Maxwell E. McCombs and Lei Guo

EC TE

D

PR O

O

FS

issues of the day. In the Chapel Hill study, this agenda was determined by a ­content analysis encompassing 25 days during the fall campaign covered by nine major news sources used by voters – local and national newspapers, network ­television news, and news magazines. The public’s agenda was determined by a survey that asked voters which issues they considered to be most important in the election. The substantial correlations found in Chapel Hill between the rank order of these five issues in the news coverage and the rank order of these same issues in voters’ perceptions of the most important issues launched the agendasetting tradition in communication research. Subsequently the agenda-setting influence of the news media on the public has been documented in hundreds of studies worldwide, both during election campaigns and in non-election periods, for a vast array of issues and topics (McCombs, 2004). Here we will examine three distinct aspects of this agenda-setting influence of the news media on the public as well as the psychological principles that explain when this influence is strong and when it is weak. The continuing evolution of the  theory of agenda setting has now extended over nearly a half century, from the seminal 1968 Chapel Hill study to the present. Two aspects of the influence of the media on the public sphere date from the earliest years of agenda-setting research. Explication of a third aspect of this influence dates from recent years. And the explication of the psychology of the agenda-setting process goes back to the  early years and extends to the present. We now turn to an overview of this continuing evolution of agenda-setting theory, which describes and explains the role of the news media in the formation of public opinion.

R

Basic Agenda-Setting Effects

U

N

C

O R

The basic agenda-setting role of the news media is to focus public attention on a small number of key issues and topics. Although there are dozens of issues and other aspects of the world outside competing for attention, the news media can cover only that handful deemed most newsworthy. In turn, the accumulated empirical evidence indicates that, to a considerable degree, the public accepts this agenda of the media as its own agenda of the most important issues of the day. As previously noted, this is not a deliberate and premeditated effort by the news media to influence public opinion, but rather an inadvertent result of the media’s need to focus on a few key topics in their presentation of the news. This was documented in the Chapel Hill study, with its tight focus on voters in a single community. Shortly after, Ray Funkhouser’s (1973) benchmark study, with its exceedingly broad focus, documented this relationship between the media’s agenda and the public’s agenda at the national level across the entire ­decade of the 1960s. In his study the correlation was + 0.78 between the pattern of coverage for public issues in the major news magazines and the public’s pattern of responses to the Gallup Poll question “What is the most important problem ­facing this c­ ountry today?”

0002065169.INDD 252

10/31/2013 10:27:32 PM



253

Agenda-setting Influence of the Media in the Public Sphere

U

N

C

O R

R

EC TE

D

PR O

O

FS

Another longitudinal study focused on a single issue, civil rights, and examined the match, over a 23-year period, between the percentage of the public’s ­naming of civil rights in response to the Gallup Poll’s “most important problem” (MIP) question and the pattern of coverage for civil rights on the front pages of the New York Times (Winter & Eyal, 1981). The correlation between the media’s agenda reflected by the news coverage and the public’s agenda reflected in responses to the MIP question was + 0.71 across a diverse period of time – 1954 to 1976. Fast forwarding to the present with its vastly expanded media landscape and its patterns of news media use varying widely among different generations, Coleman and McCombs (2007) found little difference in agenda-setting effects among the younger, middle, and older generations in state-wide surveys from North Carolina and Louisiana. Correlations between the prominence of five major issues in the newspapers of each state and the importance of these issues in each of the three age groups ranged from + 0.80 to 1.00. Particularly compelling is the comparison, in Louisiana, of the issue agendas of low and high Internet users to the issue agenda of the state’s major newspapers. For low Internet users, the correlation with the newspaper agenda is + 0.90. For high Internet users, the correlation is + 0.70, a lower but still robust correspondence. These finding from two state-wide surveys suggest that these strong agenda-setting effects result from the pervasive diffusion of news by many media rather than from the particular impact of individual media, be they traditional media such as newspapers or the array of new electronic media that are favored by members of the younger generation. This macro view is further buttressed by an investigation of the 2006 Swedish national election that measured the impact of daily news use across nine major news media – both newspapers and television – on issue salience (Stromback & Kiousis, 2010). To enhance the measurement of perceived issue salience among members of the public, each respondent’s reply to the “most important problem” question was followed by this question: “How important is that issue for how you will vote in the parliamentary election?” Stromback and Kiousis (2010, p. 288) found that “attention to political news exerts a significant and rather strong ­influence on perceived issue salience and that attention to political news matters more than attention to various specific news shows on television and in radio, or to  different newspapers.” In other words, agenda-setting effects result from the ­transmission of issue salience to the public from the collective voice of the news media. This outcome is buttressed by widespread exposure, ranging from incidental e­ xposure to studied attention paid by much of the public to a range of news media over the course of a week (McCombs & Lee, in press). These basic agenda-setting effects, sometimes called first-level effects, which are reflected in the substantial match between the media’s agenda and the public’s agenda in these examples, have been replicated in a broad variety of cultural, ­political, and media settings worldwide and with a mix of research methodologies that test the causal relationship. These replications range from an early series of election studies in the United States (Shaw & McCombs, 1977; Weaver, Graber,

0002065169.INDD 253

10/31/2013 10:27:32 PM

254

Maxwell E. McCombs and Lei Guo

Attribute Agenda-Setting Effects

FS

McCombs, & Eyal, 1981) and public concern across a year or more for a variety of national issues in the USA and Germany (Brosius & Kepplinger, 1992; Eaton, 1989) to local issues in Japan, Spain, and Argentina (Canel, Llamas, & Lennon, 1996; Lennon, 1998; Takeshita, 1993).

U

N

C

O R

R

EC TE

D

PR O

O

Beyond these basic, first-level agenda-setting effects on the public’s focus of ­attention and its perception of what the most important issues of the day are, there is a second level of agenda-setting effects: attribute agenda setting. If we think of the items on the agenda at the first level as a set of objects (analogous to attitude objects), then it is clear that these objects have certain attributes that are e­ mphasized to varying degrees on the media’s and the public’s agendas. When news ­stories report on an object, an issue, a public figure, or whatever, some aspects of the object are emphasized, some are mentioned less frequently, and other aspects not at all. The same is true when individuals think about an object or talk about that object with others. In other words, for each object on the first-level agenda, there is an agenda of attributes that can be rank-ordered in terms of their appearance in the news and in terms of their appearance in people’s descriptions of p ­ublic issues and other objects. At the second level of agenda setting, the theoretical proposition is that the media’s attribute agenda influences the salience of these attributes on the public’s agenda. Again, there is considerable empirical evidence in support of this proposition. In the 1996 Spanish general election, the ranking orders of the attributes describing the major candidates presented in seven news media were substantially correlated with the ranking order of the attributes in Navarra voters’ descriptions of these same men (McCombs, Lopez-Escobar, & Llamas, 2000). For example, the correlations between the voters’ attribute agenda and the attribute agenda of El Pais, a national newspaper, was + 0.83 for Aznar, who won the election; + 0.84 for Gonzalez, the incumbent prime minister; and + 0.86 for Anguita, the candidate of a far-left coalition. For all six comparisons with two national newspapers (3 ­candidates × 2 newspapers), the median correlation was + 0.81. For two national TV channels, the median correlation was + 0.52. For the six comparisons between the voters and two local newspapers, the median correlation was + 0.70. Evidence of similar attribute agenda-setting effects was found in a longitudinal study of the 2006 Israeli national election, which examined the attribute agenda  of  the three major candidates in two leading newspapers across three months and  among voters in three surveys during the final two months of the ­election (Balmas & Sheafer, 2010). Nine comparisons of these attribute agendas (3 candidates × 3 time periods) yielded a media correlation of + 0.62. The range of the attribute agenda-setting effects was + 0.52 to + 0.89. Additional analysis of the changes in these agendas over the three time periods found that changes in the media’s attribute agenda were paralleled by changes in the public attribute agenda

0002065169.INDD 254

10/31/2013 10:27:32 PM



255

Agenda-setting Influence of the Media in the Public Sphere

U

N

C

O R

R

EC TE

D

PR O

O

FS

(+ 0.81). Finally, a comparison of light and heavy newspaper readers found weak  correlations among the light readers for the attribute agendas of all three ­candidates  (a median correlation of + 0.26) and strong significant correlations among heavy readers for all three candidates (a median correlation of + 0.61). Similar results have been found for the attributes of issues. A national longi­ tudinal study of a Swiss referendum on changing the country’s political asylum law also found strong attribute agenda-setting effects evolving across the summer (Wirth et al., 2010). Interviews with a representative sample of citizens found ­little  impact of the media’s agenda in the initial two waves of interviewing, but strong cumulative effects in a third wave, among persons reporting heavy reliance on newspapers and TV for information about politics and the asylum law. The ­correlation between the wave 1 media’s agenda and the wave 3 public’s agenda for  persons reporting heavy media reliance was + 0.79. For the wave 2 media’s agenda and the wave 3 public’s agenda, it was + 0.82. For the wave 3 media’s agenda and the wave 3 public’s agenda, it was + 0.92. All the other correlations were nonsignificant. At the local level in the United States there was strong correspondence (+ 0.60) between the attribute agenda of a local newspaper and the pictures in citizens’ minds for six facets of an environmental issue – the development of a large ­man-made lake in Indiana (Cohen, 1975). These second-level, attribute agenda-setting effects have been replicated in a broad variety of settings and with a mix of methodologies testing the causal ­relationship. For candidate images, these replications have included presidential elections in the United State and local elections in Spain and Taiwan (Becker & McCombs, 1978; King, 1997; McCombs, Llamas, Lopez-Escobar, & Rey, 1997; Weaver et al., 1981). For the attributes of public issues, attribute agenda-setting effects have been found for the issues of the economy in the United States and of the environment and political reform in Japan (Benton & Frazier, 1976; Mikami, Takeshita, Nakada, & Kawabata, 1994; Takeshita & Mikami, 1995). In sum, there is strong evidence of media’s agenda-setting effects at two levels: the public’s views about what the most important topics of the day are, and its perceptions of the specific aspects of these topics. Returning to Lippmann’s phrase “the pictures in our heads,” the media influence what these pictures are about and what details of these pictures are prominent in our minds. Contemporary research continues to build on this foundation of media ­influence and is characterized by two trends: ●●

●●

a centrifugal trend of research, expanding to domains beyond the original focus on public affairs; a centripetal trend of research, further explicating agenda-setting theory’s core concepts.

Although the expansion to domains as disparate as professional sports, corporate reputations, religion, and cultural organizations is intriguing, our focus here is on

0002065169.INDD 255

10/31/2013 10:27:33 PM

256

Maxwell E. McCombs and Lei Guo

Exploring a Third Level of Effects

FS

two major aspects of the centripetal trend: a further explication of the nature of the  media’s influence on “the pictures in our heads”; and, most importantly, ­further exploration of the psychology of this process, which determines whether these agenda-setting effects are strong, moderate, or weak.

U

N

C

O R

R

EC TE

D

PR O

O

The elements of “the pictures in our heads” that have been investigated to date at the first and second level of agenda setting are discrete objects and attributes. That is, they are separate and disconnected elements that have been disaggregated from the whole in which they are embedded. On the media’s agenda, these elements are various objects and attributes that have been isolated in the content analysis of news stories. On the public’s agenda, these elements usually are the various objects and attributes that have been isolated in the content analysis of individuals’ responses to open-ended questions. Are the news media able to transfer the salience of a more integrated image – the salience of the larger, more comprehensive picture? Recent research on the network agenda-setting model is the opening gambit on the exploration of a third level of effects, where it is theorized that the salience of the networked relationships among objects and attributes is transferred from the news media to the public. Theoretically central to the network agenda-­ setting  model is an associative network model of memory. Scholars in various disciplines – cognitive psychology, philosophy, geography, and communication – have theorized this associative memory model in similar ways, yet under different descriptions. Examples of such descriptions are “the associative network model”  (Anderson, 1983; Anderson & Bower, 1973), “cognitive mapping” (Kaplan, 1973), “the cognitive network model” (Santanen, Briggs, & de Vreede, 2000), “the connectionist model” (Monroe & Read, 2008), and “the spreading activation model” (Collins & Loftus, 1975). Rather than conceptualizing our mental representations as a hierarchical or ­linear structure – as implied in the traditional understanding of agenda-setting theory – this associative network model holds that the representation operates pictorially, diagrammatically, or cartographically (Armstrong, 1973; Barsalou, ­ 1998; Braddon-Mitchell & Jackson, 2007; Cummins, 1996). In this network model, individuals’ cognitive representation is presented as a network-like structure where any particular construct or node will in general be connected to numerous other constructs or nodes. Here a construct or node in the network can refer to any unit of information: objects and their attributes; goals, values, and motivation; affective or emotional state; and even macro units like schema or frame (Lindsay & Norman, 1977; Price & Tewksbury, 1997; Rumelhart & Norman, 1978). Take political communication as an example. When an individual considers a political candidate and uses some attributes to describe that candidate, it is not  necessary for the individual to articulate a hierarchy of attributes ranked by

0002065169.INDD 256

10/31/2013 10:27:33 PM



257

Agenda-setting Influence of the Media in the Public Sphere

U

N

C

O R

R

EC TE

D

PR O

O

FS

their importance. Instead an assortment of attributes as well as other related ­constructs  can constitute a network-shaped picture describing that candidate in the individual’s mind. News media are a crucial factor impacting some of our cognitive networks, ­especially in relation to public affairs. In a restatement of the basic proposition of agenda-setting theory, McCombs (2004) suggested that news media have the capability to influence the audience’s network-like mental structure. And he ­paraphrased Lippmann’s (1922) thesis: “the news media, our windows to the vast world beyond direct experience, determine our cognitive maps of that world” (McCombs, 2004, p. 3). Along this line, one less studied concept in agenda-setting theory – a “compelling argument” – also implicitly suggests that news media have the potential to affect the audience’s cognitive map by transferring the relationships, or the connections, among various agenda elements to the public’s mind. Specifically, the “compelling argument” hypothesis suggests that the news media emphasis on a certain attribute of an object, namely an attribute that has a particular resonance with the public, provides people with cues to modify their perceived salience of the object that ­possess that attribute (McCombs & Ghanem, 2001; Yioutas & Segvic, 2003). According to this hypothesis, news media not only may shape the perceived ­importance of attributes and objects separately, but can actually bundle an object with an attribute and make them salient in the public’s mind simultaneously. In consequence, the audience may not only treat a certain attribute as a part of the object but also regard the two as connected – as elements integrated in audience members’ mental “pictures.” For the issue of crime in Texas, Ghanem (1996, 1997) found that a low psychological distance between a crime event in the news and an individual was a compelling argument for the salience of the issue. Although there is also additional empirical support for the “compelling argument” hypothesis (Schoenbach & Semetko, 1992; Williams, Shapiro, & Cutbirth, 1983), the “compelling argument” effect remains greatly understudied (Kiousis, 2005). Expanding beyond the “compelling argument” and its specific linking of an attribute and an object, the network agenda setting model asserts that the media can bundle a variety of elements and make them salient in the public’s mind simultaneously. These bundles could be sets of objects, sets of attributes, or some combination of objects and attributes. To explore these possibilities, Guo and McCombs (2011a) initially examined sets of attributes, specifically the attributes of political candidates, and re-analyzed data from a previous study, which had found strong attribute agenda-setting effects on the basis of an analysis grounded in the traditional discrete sets of attributes. Using network analysis for the new round of analysis, the strategy of Guo and McCombs for this opening gambit in the exploration of a third level of agenda setting was to compare traditional attribute agenda-setting effects with networked attribute agenda-setting effects. How does the strength of the association between networked media’s and ­public’s agendas compare to the strength of traditional measures of attribute agenda setting?

0002065169.INDD 257

10/31/2013 10:27:33 PM

258

Maxwell E. McCombs and Lei Guo

Specifically, two interrelated hypotheses were tested with the help of a network analysis of data originally collected for Kim and McCombs’ (2007) study of the 2002 gubernatorial and US senatorial elections in Texas.

O

●●

The salience of the relationship network of political candidate attributes on the media’s agenda will be positively associated with the public salience of this attribute network. The centrality of political candidate attributes in the media attribute network will be positively associated with the centrality of the attributes on the public network agenda.

FS

●●

U

N

C

O R

R

EC TE

D

PR O

Two sets of content analysis data from the Austin American-Statesman were ­available from 2002 for reanalysis: (1) news coverage of Republican gubernatorial candidate Rick Perry and of Democratic gubernatorial candidate Tony Sanchez, as each competed in his party’s primary gubernatorial election in the spring (2002 Spring Dataset); and (2) news coverage of the same two candidates as well as of two US senatorial candidates, Republican John Cornyn and Democrat Ron Kirk, in the general elections in November (2002 Fall Dataset). The unit of measurement in the content analysis was an assertion in an article that a candidate possessed a particular attribute that described his qualifications or character. Only attributes concerning candidate qualifications and character were considered, because they were by far the most prominent attributes highlighted both by the Austin media and by voters. In other words, what this new study analyzed are specific sub-­ networks of attributes portraying the political candidates. Nine attributes were found to define personal qualifications and character in the analyzed news stories: (1) leadership; (2) experience; (3) competence; (4) credibility; (5) morality; (6) caring about ­people; (7) communication skills; (8) pride in family/backgrounds, roots, and race/ethnicity; (9) non-politician. During the 2002 election period, a telephone survey was also conducted among randomly selected Austin residents. To measure the public’s attribute agenda, respondents were asked: “Suppose that one of your friends has been away a long time and knows nothing about the political candidates. What would you tell your friend about (Cornyn, Kirk, Perry, and Sanchez)”? Responses were recorded ­verbatim and later coded using the same nine attributes’ categories. From the network analysis of these two data sets, graphs were created to ­visualize  these media and the public network agendas. Figure  14.1a and 14.1b ­presents graphs that represent the media and the public attribute agenda networks based on the 2002 Combined Dataset. Visually, these two networks are highly similar. For example, three attributes – “experience,” “competence” and “credibility” – are central in both the media and the public networks and demonstrate the strongest mutual connections, as shown in both graphs. Table  14.1 details the correlations measuring the degree of correspondence between the media network agendas and the public network agendas. Specifically, quadratic assignment procedure (QAP) correlation and regression coefficients

0002065169.INDD 258

10/31/2013 10:27:33 PM



259

Agenda-setting Influence of the Media in the Public Sphere Communication

Family

Competence

Other

Caring

O

Leadership

FS

Non-Politican

PR O

Experience

Credibility

Morality

D

Figure 14.1a

EC TE

Communication

Caring

O R

R

Non-Politican

Experience Competence

C

Leadership

Family Other

U

N

Credibility

Morality

Figure 14.1b

for all three 2002 comparisons are positive and statistically significant. The QAP correlations (Pearson’s r) between the agendas range from + 0.67 to + 0.84. All three support the proposition that the salience of the candidate attribute relationships in the media is positively associated with the public salience of those same attribute relationships.

0002065169.INDD 259

10/31/2013 10:27:34 PM

260

Maxwell E. McCombs and Lei Guo

Table 14.1  QAP correlation/regression and centrality correlation/regression between the media’s and the public’s attribute agenda networks

2002 Spring Dataset 2002 Fall Dataset 2002 Combined Dataset 2010 Fall Dataset

QAP Regression (R2)

Centrality Correlation (Pearson’s r)

Centrality Regression (R2)

0.75**1 0.67* 0.84** 0.71**

0.56** 0.45* 0.71** 0.51**

0.84*2 0.76* 0.91** 0.81**

0.70* 0.57* 0.84** 0.66**3

FS

Dataset

QAP Correlation (Pearson’s r)

** means p