Mass Communication & Society
Q Routledge
Taylor &. Francis Groi
ISSN: 1520-5436 (Print) 1532-7825 (Online) Journal homepage: https://www.tandfonline.com/loi/hmcs20
Agenda-Setting, Priming, and Framing Revisited: Another Look at Cognitive Effects of Political Communication
Dietram A. Scheufele
To cite this article: Dietram A. Scheufele (2000) Agenda-Setting, Priming, and Framing Revisited: Another Look at Cognitive Effects of Political Communication, Mass Communication & Society, 3:2-3, 297-316, DOI: 10.1207/S15327825MCS0323 07
To link to this article: https://doi.org/10.1207/S15327825MCS0323 07
Published online: 17 Nov 2009.
Submit your article to this journal C7
Article views: 14621
5l  View related articles ß1
Citing articles: 105 View citing articles C?
Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/joumallnformation?joumalCode=hmcs20
MASS COMMUNICATION & SOCIETY, 2000, 3(2&3), 297-316
Agenda-Setting, Priming, and Framing Revisited: Another Look at Cognitive Effects of Political Communication
Dietram A. Scheufele
Department of Communication Cornell University
Agenda-setting, priming, and framing research generally has been examined under the broad category of cognitive media effects. As a result, studies often either examine all 3 approaches in a single study or employ very similar research designs, paying little attention to conceptual differences or differences in the levels of analysis under which each approach is operating. In this article, I revisit agenda-setting, priming, and framing as distinctively different approaches to effects of political communication. Specifically, I argue against more recent attempts to subsume all 3 approaches under the broad concept of agenda-setting and for a more careful explication of the concepts and of their theoretical premises and roots in social psychology and political psychology. Consequently, it calls for a reformulation of relevant research questions and a systematic categorization of research on agenda-setting, priming, and framing. An analytic model is developed that should serve as a guideline for future research in these areas.
The original formulation of the agenda-setting hypothesis (Cohen, 1963; McCombs & Shaw, 1972) phrased it more like a speculative idea or heuristic (Kosicki, 1993) than as a theory in its most conventional sense. In an attempt to construct a more comprehensive theoretical model, Weaver, McCombs, and Shaw (1998) recently suggested that priming and framing should be viewed as natural extensions of agenda-setting. Priming is the impact that agenda-setting can have on the way individuals evaluate public officials by influencing the thematic areas or issues that individuals use to form these evaluations. Framing can be considered an extension of agenda-setting as it "is the selection of a restricted number of themati-
Requests for reprints should be sent to Dietram A. Scheufele, Department of Communication, Cornell University, 315 Kennedy Hall, Ithaca, NY 14853-4203. E-mail: dietram.scheufele@cornell.edu
298 SCHEUFELE
cally related attributes for inclusion on the media agenda when a particular object is discussed" (McCombs, 1997, p. 6). In other words, whereas agenda-setting is concerned with the salience of issues, frame-setting, or second-level agenda-setting (McCombs, Llamas, Lopez-Escobar, & Rey, 1997) is concerned with the salience of issue attributes.
Empirical work, however, seems to contradict such theorizing, suggesting that although both are important, perceived importance of specific frames rather than salience is the key variable (Nelson, Clawson, & Oxley, 1997; Nelson & Kinder, 1996). In other words, "frames influence opinions by stressing specific values, facts, and other considerations, endowing them with greater apparent relevance to the issue than they might appear to have under an alternative frame" (Nelson et al., 1997, p. 569). In contrast to McCombs and his associates (McCombs, Llamas, Lopez-Escobar, & Rey, 1997), Nelson et al. directly operationalized salience of frames by measuring response latency (for an overview, see Bassili, 1995), and found support for their theorizing. Causal modeling revealed that various dimensions of perceived importance accounted for major proportions of the variance in framing effects, with salience or accessibility of frames playing only a minor role.
PARSIMONY VERSUS PRECISION: THE BENEFITS OF DIFFERENTIATING MODELS OF MEDIA EFFECTS
McCombs, Llamas et al. (1997) based their attempts to combine agenda-setting, priming, and framing into a single theoretical framework on the assumption that integrating theory is always desirable. However, if theories are based on distinctively different premises, and if they differ with respect to the empirical patterns observed, this strategy might in fact be counterproductive. As I argue in this article, agenda-setting and priming differ from framing with respect to their assumptions and premises. At the same time, they derive distinctively different theoretical statements and hypotheses from these premises. Consequently, these three approaches to media effects should be taken for what they are: related, yet different approaches to media effects that cannot be combined into a simple theory just for the sake of parsimony.
Thus, using consistent terminology for essentially incompatible theoretical models has done little to remedy the lack of conceptual clarity surrounding agenda-setting, priming, and framing in the area of media effects research (Entman, 1993; McNamara, 1992; Scheufele, 1999). Possibly as a result of the vagueness of the constructs and theoretical statements involved, the original authors admit, "evidence from ... agenda-setting studies is mixed" (Weaver et al., 1998, p. 1).
To provide a systematic overview of the different possible approaches to agenda-setting, priming, and framing in media effects research, it is necessary to develop clear conceptualizations of agenda-setting, priming, and framing based on
COGNITIVE EFFECTS REVISITED 299
their theoretical premises. This will allow researchers to define the role that the three concepts can play in media effects research and to determine if there are, in fact, theoretical overlaps. Based on these conceptual definitions, this article develops an analytic model incorporating all three theoretical approaches to develop strategies and recommendations for future research.
Theoretical Premises of Agenda-Setting, Priming, and Framing
As argued earlier, attempts to integrate agenda-setting, priming, and framing into a single model have largely ignored the differences between the premises underlying these three theoretical models. Agenda-setting, on the one hand, and priming as a direct extension or outcome of agenda-setting (e.g., Iyengar & Kinder, 1987), on the other hand, are based on the same assumptions or premises. Contrary to McCombs's (1997) theorizing, however, framing is based on premises that differ from those of agenda-setting or priming.
Salience: The theoretical premises of agenda-setting and priming. The
theoretical foundation of agenda-setting and priming can be traced back to psychological concepts of priming in work on cognitive processing of semantic information (Collins & Loftus, 1975; Tulving & Watkins, 1975). By receiving and processing information, individuals develop memory traces (Tulving & Watkins, 1975) or activation tags (Collins & Loftus, 1975); that is, concepts or issues are primed and made more accessible in an individual's memory. Activation tags or memory traces, therefore, influence subsequent information processing (Salancik, 1974). "When a concept is primed, activation tags are spread.... When another concept is subsequently presented, it has to make contact with one of the tags left earlier and find an intersection" (Collins & Loftus, 1975, p. 409).
The metaphor of activation tags or memory traces was later replaced by the construct of accessibility. The idea of accessibility is the foundation of a memory-based model of information processing, which assumes that individuals make judgments about other people or issues based on information easily available and retrievable from memory at the time the question is asked (Hastie & Park, 1986; Iyengar, 1990). According to a memory-based model, judgments and attitude formation are directly correlated with "the ease in which instances or associations could be brought to mind" (Tversky & Kahneman, 1973, p. 208).
Empirical work on accessibility generally falls into one of two categories. First, studies often test a memory-based model of information processing against a competing model of online processing, which sees attitudes as being formed when incoming information is initially processed (e.g., Hastie & Park, 1986; Moy, Scheufele, Eveland, & McLeod, in press). According to an online model, attitudes are stored in memory as what have been called summary or judgment tallies
300 SCHEUFELE
(Lodge, McGraw, & Stroh, 1989), judgment operators (Hastie & Park, 1986), or information integrators (Wyer & Srull, 1986). New impressions are processed and integrated into a running tally, shaping one's current attitude (Tourangeau, 1984, 1987; Tourangeau & Rasinski, 1988). Although the attitude (or judgment tally) changes, the original cognition or argument that changed the attitude is eventually forgotten (Lodge et al., 1989). It is beyond the scope of this article to resolve this ongoing conflict. More recent research, however, has provided strong evidence for cognitive processes like rationalization (Rahn, Krosnick, & Breuning, 1994), priming (Petty & Jarvis, 1996; Sudman, Bradburn, & Schwarz, 1996), and affective priming (Moy et al., in press) that are all based on a memory-based model of information processing; that is, on the assumption of attitude accessibility. These operationalizations infer attitude accessibility as a construct from the behavioral or cognitive patterns observed.
Second, attempts have been made to measure attitude accessibility more directly. Bassili (1995) suggested an operationalization of attitude accessibility as a continuous variable that involves measuring the amount of time that it takes a respondent to answer a given question. The underlying rationale is that the longer it takes a respondent to answer a given question, the less accessible the evoked cognition. "Accessibility is conceived in terms of associational strength in memory and measured in the metric of response time" (Huckfeldt & Sprague, 1997, p. 6).
Both agenda-setting and priming are based on this assumption of attitude accessibility and, in particular, a memory-based model of information processing. Mass media can influence the salience of certain issues as perceived by the audience; that is, the ease with which these issues can be retrieved from memory. As hypothesized in the priming model, perceived issue salience becomes the independent variable and influences the role that these issues or considerations play when an individual makes a judgment about a political actor. Mass media, Iyengar and Kinder (1987) argued, affect "the standards by which governments, policies and candidates for public office are judged" (p. 63). Political issues that are most salient or accessible in a person's memory will most strongly influence perceptions of political actors and figures.
Attribution: The theoretical premises of framing. A macroscopic approach to framing that examines media frames as outcomes of journalistic norms or organizational constraints is based on what Pan and Kosicki (1993) called the sociological approach to framing research. This approach, although drawing on theoretical approaches in various fields and disciplines, is commonly linked to attribution theory (Heider, 1930,1959,1978; Heider & Simmel, 1944) and frame analysis (Goffman, 1974).
In his work on attribution theory, Heider (1930) assumed that human beings cannot understand the world in all its complexity. Therefore the individual tries to infer underlying causal relations from sensory information. These assumptions were sup-
COGNITIVE EFFECTS REVISITED      301
ported by experimental evidence. A vast majority of individuals who were shown movies with abstract movements of geometrical shapes interpreted these movements as actions of human beings with a certain underlying motivation (Heider & Simmel, 1944). Based on these studies, Heider (1978) defined attribution as the link between an observed behavior and a person who is considered responsible for this action. In his later work, Heider (1959) expanded this definition of attribution to environmental factors; that is, an observed behavior can be attributed to both personal and societal or environmental factors. Drawing on Heider's distinction between societal and individual attributions of responsibility, Iyengar (1991) argued that people try to make sense of political issues by reducing them to questions of responsibility. In other words, responsibility for social issues or problems can be framed as individual responsibility or the responsibility of society at large.
Although he did not explicitly refer to Heider's (1959) findings, Goffman (1974) too assumed that individuals cannot understand the world fully and therefore actively classify and interpret their life experiences to make sense of the world around them. The individual's reaction to sensory information therefore depends on schemes of interpretation called "primary frameworks" (Goffman, 1974, p. 24). These frameworks or frames can be classified into natural and societal frames: Natural frames help to interpret events originating from natural and nonintentional causes, whereas societal frames help "to locate, perceive, identify, and label" (Goffman, 1974, p. 21) actions and events that stem from intentional human action. The most important implication for the field of mass communication research, then, is that there are various ways of looking at and depicting events in news media that depend on the framework employed by the journalists. Or as Goffman put it, "The type of framework we employ provides a way of describing the event to which it is applied" (p. 24).
Unlike the sociological approach, a microscopic or psychological approach (Fischer & Johnson, 1986) examines frames as individual means of processing and structuring incoming information. This psychological approach is summarized in work on frames of reference (Sherif, 1967) and prospect theory (Kahneman & Tversky, 1972, 1979, 1984; Quattrone & Tversky, 1988).
In his work on frames of reference, Sherif (1967) assumed that individual judgments and perceptions not only are influenced by cognitive or psychological factors but also occur within an appropriate frame of reference. Therefore, it is possible "to set up situations in which appraisal or evaluation of a social situation will be reflected in the perceptions and judgments of the individual" (p. 382).
Although this work does not suggest how mass media can influence individual judgments and perceptions, research on prospect theory (e.g., Quattrone & Tversky, 1988) points to a possible link between mass media coverage and the framework individuals employ to interpret events. Specifically, experimental research has shown that how a decision-making situation is framed will affect what people believe will be the outcome of selecting one option over the other
302 SCHEUFELE
(Kahneman & Tversky, 1984). Edelman (1993) applied their findings to social settings: "The social world is ... a kaleidoscope of potential realities, any of which can be readily evoked by altering the way in which observations are framed and categorized" (p. 232).
EXPLICATING AGENDA-SETTING AND PRIMING
The interactive construction of reality by mass media and audiences as hypothesized by McQuail (1994) has important implications for the conceptualization of agenda-setting and priming as theories of media effects. An analysis of the roles of both audience members and mass media in this constructivist approach requires research on various levels of analysis.
Agenda-setting and priming, therefore, need to be conceptualized at two separate levels. On the macroscopic level, agenda-setting has to be examined based on the media agenda; that is, "the importance assigned to issues and personalities in the media" (Winter & Eyal, 1981, p. 376). Priming, an inherently individual psychological outcome of agenda-setting, does not operate at this level of analysis. Therefore, on the microscopic level, agenda-setting has to be examined based on the audience agenda or the salience or accessibility of certain issues in a person's memory, and, closely related, the priming of "criteria that citizens use to evaluate their leaders" (Behr & Iyengar, 1985, p. 38).
Research on agenda-setting and priming as media effects, therefore, can be classified along two dimensions. First, agenda-setting needs to be examined across levels of analysis; that is, both as media agendas and as audience agendas. Second, agenda-setting needs to be examined as both independent and dependent variables. Based on this distinction, three distinct processes can be differentiated: agenda-building, agenda-setting, and priming (see Figure 1).
Agenda-Building
Agenda-building as a term was introduced by Cobb and Elder (1971) who "are concerned with how issues are created and why some controversies or incipient
Elites, Interest Groups, etc.
Media Agenda
Audience Agenda
Evaluations
Agenda-Building
IL
Agenda-Setting
Priming
FIGURE 1   An overview of agenda-setting research.
COGNITIVE EFFECTS REVISITED 303
issues come to command the attention and concern of decision makers, while others fail" (p. 905). In other words, in the process of agenda-building, the media agenda is considered the dependent variable. Rogers, Dearing, and Bregman (1993)—who used the term media agenda-setting instead of agenda-building—noted that agenda-setting research has widely accepted the media agenda as a given and only a few studies have considered the process by which it is constructed.
Without explicitly referring to agenda-building, Funkhouser (1973a) was one of the first researchers to examine the relation between real-world events and the amount of media coverage on these events. Assuming that "the contents of weekly news magazines would be a reasonable indicator of the contents of the news media nationwide" (p. 533), he compared the amount of media coverage—measured as the amount of coverage of a number of issues in news magazines between 1960 and 1970—to official statistics regarding these issues during the same time period. Overall, there was little support for the hypothesis that real-world events drive media coverage (Funkhouser, 1973a).
Explicitly referring to agenda-building, Lang and Lang (1981) conducted a case study of the Watergate scandal. They argued that previous research has largely ignored the question of how the public agenda is formed; that is, there is little or no "recognition of the process through which agendas are built or through which an object that has caught public attention, by being big news, gives rise to a political issue" (p. 448). Lang and Lang proposed a four-step model of agenda-building. In a first step, mass media highlight some events, activities, groups, or personalities. In a second step, these elements of a conflict are combined into a common frame or the description of some problem or concern. In a third step, the issue is linked "to secondary symbols, so that it becomes a part of the recognized political landscape" (p. 465). In a fourth and final step, spokesmen and spin masters play an important role in promoting issues and symbols and establishing a feedback loop to media coverage to increase issue coverage or at least to keep the issue alive. This formulation of the process of agenda-building, then, attributes a key role both to mass media for initially picking up an issue and to political actors for keeping an issue prominent in the media agenda or even increasing its prominence.
Without referring explicitly to the term agenda-building, Behr and Iyengar (1985) provided one of the few empirical examinations of the agenda-building process. To examine the impact of the media agenda on the audience agenda, they combined trend data from three national surveys with a content analysis of the CBS Evening News. They also collected a number of indicators of current conditions on all three issue areas under study. In addition, they recorded instances in which the president delivered speeches to the nation concerning any of these issues. Their analyses revealed that "television news coverage is at least partially determined by real-world conditions and events" (p. 47). Similarly, presidential addresses increased levels of news coverage for the respective issue
304 SCHEUFELE
for all issues under study. Finally, Behr and Iyengar (1985) found that levels of public concern were driven by television news coverage for two of the three issues under study. The reverse, however, was not true: "News coverage is for the most part unaffected by public opinion and the assumption that agenda-setting is a recursive process is on solid ground" (p. 47).
Agenda-Setting
A hypothesized positive relation between the media agenda as the independent variable and the audience agenda as the dependent variable was first examined empirically by McCombs and Shaw (1972). Their reasoning was based on earlier work by Cohen (1963), who argued that mass media "may not be successful much of the time in telling people what to think, but is stunningly successful in telling its readers what to think about" (p. 13). McCombs and Shaw examined the agenda-setting hypothesis during the 1968 presidential campaign. Employing a cross-sectional survey design, they compared 100 undecided voters' perceptions of issue salience with the amount of coverage of these issues, measured by a content analysis of television, newspaper, and magazine coverage. Based on zero-order correlations, they concluded "that media appear to have exerted a considerable impact on voters' judgments of what they considered the major issue of the campaign" (p. 180).
In 1996, Dearing and Rogers (1996) listed as many as 350 publications about the agenda-setting effect. It is therefore impossible to examine all or even a reasonable sample of studies. It is necessary, however, to summarize problematic areas common to a large number of studies in the area. All in all, criticisms of previous studies fall into one of two major categories.
First, agenda-setting is an inherently causal theory. Studies generally have found some form of positive association between the amount of mass media content devoted to an issue and the development of a place on the public agenda for that issue. The research designs and statistical methods employed, however, are in few cases suited to make causal inferences. As a result, "causal direction must remain an open question for now, at least in terms of most survey studies" (Kosicki, 1993, p. 106). Exceptions are experimental studies like Iyengar and Kinder's (1987) experimental research on agenda-setting and priming.
At least three conditions need to be fulfilled to infer some form of causal relation. First, some form of covariation needs to be identified between the two variables under study. Second, potential alternative explanations need to be ruled out. Third, a temporal order between independent and dependent variable has to be established, with the hypothesized cause preceding the effect (Lazarsfeld, 1957). Whereas the first condition (i.e., a covariation between media and audience agenda) is tested in most studies, third variable explanations and temporal order are controlled for in considerably fewer studies.
COGNITIVE EFFECTS REVISITED 305
As far as third variable explanations are concerned, previous research has identified a number of them as well as contingent conditions under which agenda-setting effects will occur. Demers, Craff, Choi, and Pessin (1989), for example, identified issue obtrusiveness as a key contingency in the agenda-setting process. Findings on spurious explanations are, at best, mixed. Most important, the notion that real-world events rather than the portrayals of these events in mass media drives audience agendas is supported by few studies (Dearing & Rogers, 1996).
As far as the temporal order of variables is concerned, agenda-setting assumes that there is a process by which the media agenda influences the audience agenda over time. Across studies, however, the issue of time lag between media agenda-setting and audience effects "is insufficiently theorized and underspecified" (Kosicki, 1993, p. 107). Various researchers employed a combination of survey designs and content analytic designs with longer time lags (e.g., Allen, O'Loughlin, Jasperson, & Sullivan, 1994; Behr & Iyengar, 1985; Funkhouser, 1973b; Iyengar & Simon, 1993). Others used experimental designs with shorter periods of time between administration of the stimulus and measurement of agenda-setting effects (e.g., Iyengar & Kinder, 1987). Theoretical arguments typically are not the driving force behind the choice of the research design; rather, "time lags are tested in numerous ways until the optimal one is found" (Kosicki, 1993, p. 107).
The second area of criticism of previous agenda-setting research refers to the measurement of the criterion variable; that is, the perceptions of issue salience by the audience. In their seminal study, McCombs and Shaw (1972) operationalized issue salience among audience members as judgments about the perceived importance of issues. Later studies replaced perceptions of importance with terms such as salience, awareness, attention, or concern (Edelstein, 1993).
The conceptual difference between these concepts also has important operational consequences. If individual-level salience of issues is the key criterion variable, measures of perceptions of issue importance are inadequate indicators. Iyengar (1990), for example, described the power of television in the context of agenda-setting as the ability "to make information 'accessible' or more retrievable from memory" (p. 2). Measures of perceived importance, however, do not capture the ease with which considerations can be retrieved from memory. Rather, salience should be measured indirectly through variables like response latency (e.g., Bassili, 1995) or should be demonstrated in question order experiments (e.g., Zaller & Feldman, 1992).
Priming
The priming hypothesis states that mass media, by making some issues more salient than others, influence "the standards by which governments, presidents, policies, and candidates for public office are judged" (Iyengar & Kinder, 1987, p. 63). In this sense, studies testing priming effects examine agenda-setting as
306 SCHEUFELE
the independent variable and priming effects as outcomes of agenda-setting or as the dependent variable.
In a series of experiments, Iyengar and Kinder (1987) examined the impact of network newscasts on viewers' perceptions of issue salience on the criteria they use to make judgments about political candidates. They hypothesized that the salience of certain issues as portrayed in mass media influences individuals' perceptions of the president because respondents will use issues that they perceive as more salient as standards for evaluating the president. Even though they considered priming "a robust effect" (p. 72), they found mixed statistical support for their hypothesis. Strong priming effects emerged for only some issues and weak effects for others, such as unemployment. Iyengar and Kinder attributed this weak priming effect to the generally high salience of unemployment in American politics.
Finally, priming effects were studied during the Gulf War. Iyengar and Simon's (1993) combination of survey data and content analyses of network television news supported the basic priming hypothesis: Foreign policy issues that had been made more salient for individuals by intensive Gulf War coverage tended to override other issues as influences on assessments of presidential performance.
EXPLICATING FRAMING
Similar to agenda-setting and priming, framing has to be examined not only across levels of analysis but also as a dependent and independent variable. Two concepts of framing need to be specified: media frames and audience frames (Scheufele, 1999).
Media frames have been defined as "a central organizing idea or story line that provides meaning to an unfolding strip of events. ... The frame suggests what the controversy is about, the essence of the issue" (Gamson & Modigliani, 1987, p. 143). Media or news frames serve as working routines for journalists, allowing them to quickly identify and classify information and "to package it for efficient relay to their audiences" (Gitlin, 1980, p. 7). Audience frames are defined as "mentally stored clusters of ideas that guide individuals' processing of information" (Entman, 1993, p. 53).
A Typology of Framing
Similar to priming research, different approaches to framing can be classified along two dimensions. A between-level dimension conceptually defines media frames on a macroscopic level and audience frames on a microscopic level and hypothesizes potential relations between them. A within-level dimension conceptualizes media frames and audience frames separately as both independent and dependent variable. Similar to the research on agenda-building, agenda-setting, and priming, studies of framing commonly examine one of three distinct processes: frame-setting, frame-building, and individual-level outcomes of framing (see Figure 2).
COGNITIVE EFFECTS REVISITED 307
EllteS' Media A.,HJm^ Attributions of
Interest Groups, — ™£ —   Causa, / T—
eLt" Responsibility
I_II____IL
Frame- Frame- Individual-level
Building Setting "Consequences"
of Framing
FIGURE 2   An overview of framing research.
Frame-building. Although studies have examined both extrinsic and intrinsic factors influencing the production and selection of news (e.g., Gans, 1979; Shoemaker & Reese, 1996; Tuchman, 1978), little evidence has yet been systematically collected on how various factors influence the structural qualities of news in terms of framing. Based on previous research (e.g., Shoemaker & Reese, 1996; Tuchman, 1978), at least five factors may potentially influence how journalists frame a given issue: social norms and values, organizational pressures and constraints, pressures of interest groups, journalistic routines, and ideological or political orientations of journalists.
Frame-setting. In many cases, studies examining media frames as the independent variable also examine audience frames as the dependent variable. Whereas some of this research experimentally manipulates media frames to examine their impact on audience frames, other studies include measures of both media and audience frames.
The most promising approach conceptualizing both media frames as the independent variable and audience frames as the dependent variable is reflected in the combination of content analytic data and survey data by Huang (1995,1996). Using the Anita Hill-Clarence Thomas controversy as an exemplar, she analyzed "to what extent media frames are operative in audience frames" (Huang, 1996, p. 1). She measured frames on both a macro- and microlevel, comparing news frames, based on content analyses of how the network evening news and two local newspapers framed the controversy, and audience frames—based on coding answers to open-ended questions on the controversy. Her results (Huang, 1995) showed that media frames can find their way into audience frames. When media and audience frames overlap, however, the media and the audience accord different weights to those frames.
Other studies measure only the dependent variable and experimentally manipulate media frames as the independent variable. Iyengar's (1987) content analysis of network television newscasts showed that network newscasts are framed in episodic or thematic terms. Episodic newscasts depict public issues as concrete instances or specific events, whereas thematic newscasts report on a more abstract level in the form of general outcomes. Iyengar's (1991) analyses also indicated that networks rely extensively on episodic framing. He hypothe-
308 SCHEUFELE
sized that the type of media framing has an impact on the attribution of responsibility by audiences and differentiated two dimensions of attribution of responsibility: causal and treatment responsibility (Iyengar, 1987, 1991). Iyengar's results showed that a relation between media frames and audience frames is strongly contingent on the issue under study. An experimental manipulation of highly salient issues like unemployment, for instance, had little or no impact on individual attribution of responsibility.
Price, Tewksbury, and Powers (1995, 1997) offered the most elaborate approach to audience frames as dependent variable. Undergraduate students were asked to read news articles about possible cuts in state funding to the university that were experimentally prepared to manipulate various news frames. In a posttest questionnaire, the participants were asked to write down all thoughts and feelings they had while reading the news article, including those thoughts that were not necessarily directly relevant to the article. Coding of the open-ended question showed that issue frames of news stories had a significant influence on the respondents' cognitive responses. The most interesting finding was a phenomenon that Price et al. (1995) called "a kind of 'hydraulic' pattern, with thoughts of one kind, stimulated by the frame, driving out other possible responses" (p. 23).
Individual-level consequences of framing. To find an explicit and direct link between audience frames as independent variable and individual information processing or political action, one can turn to the social movements literature. Generally, the frames individuals use to interpret conflicts may have an impact on the "mobilization for collective action aimed at social change" (Gamson, 1985, p. 620). More specifically, other work has focused on how master frames invented by social movements can potentially influence the motivation for individuals to support these movements and to form consensus (Entman & Rojecki, 1993; Klandermans, 1988, 1992; Klandermans & Oegema, 1987). Gerhards and Rucht (1992) tried to synthesize the previous findings into a single model, differentiating three types of framing: diagnostic framing, or identifying a problem and attributing blame; prognostic framing, which specifies what needs to be done; and motivational framing, or the "call to arms for engaging in ameliorative or corrective action" (Snow & Benford, 1988, p. 199).
AGENDA-SETTING, PRIMING, AND FRAMING: MODELS FOR FUTURE RESEARCH
This article is concerned with agenda-setting, priming, and framing as distinctively different approaches to effects of political communication and argues for a more careful explication of the concepts and of their theoretical premises and roots in social and political psychology. I have shown that agenda-setting and priming, on the
COGNITIVE EFFECTS REVISITED 309
one hand, and framing, on the other hand, are, in fact, based on distinctively different assumptions and therefore translate into equally different theoretical statements.
Theoretical Differences
Agenda-setting and priming rely on the notion of attitude accessibility. Mass media have the power to increase levels of importance assigned to issues by audience members. They increase the salience of issues or the ease with which these considerations can be retrieved from memory if individuals have to make political judgments about political actors. In other words, media influence the standards by which audience members evaluate political figures. Framing, in contrast, is based on the concept of prospect theory; that is, on the assumption that subtle changes in the wording of the description of a situation might affect how audience members interpret this situation. In other words, framing influences how audiences think about issues, not by making aspects of the issue more salient, but by invoking interpretive schemas that influence the interpretation of incoming information. Although the process of issue selection or agenda-setting by mass media necessarily needs to be a conscious one, framing is based on subtle nuances in wording and syntax that have most likely unintentional effects or at least effects that are hard to predict and control by journalists.
An Analytic Model of Agenda-Setting, Priming, and Framing
This article has addressed McCombs, Shaw, and Weaver's (1997) proposition that priming and framing are extensions of agenda-setting from a largely theoretical perspective. The question that arises from this conceptual work, of course, concerns implications for research in these areas. More generally, testing these models of media effects requires linking macro- and microlevels of analysis. In other words, models of agenda-setting and framing link media content as the unit of observation to audience characteristics. This creates analytic and statistical problems when trying to assess the relation between media content and audience characteristics. Previous research has commonly addressed this issue in one of two ways.
First, researchers have compared aggregate measures of media content to aggregate measures of public opinion, either in cross-sectional designs (comparing the rank-orders of media and audience agendas) or in longitudinal designs (using time points as their unit of observation). In one of the earlier studies of agenda-setting, for example, Funkhouser (1973b) compared the average amount of issue coverage in mass media to the average importance attributed to these issues by the public. He found a rank-order correlation between media and audience agenda of .78. Studies
310 SCHEUFELE
like Funkhouser's (1973a, 1973b) have at least two inherent problems. First, relatively high rank-order correlations can be observed even if the actual ranking of issues differs substantially between media and audience agenda or if some issues covered in the media do not find their way onto the audience's agenda at all.
Second, studies examining audience and media agendas at the aggregate level potentially encounter ecological fallacy problems. It is possible that audience members who do not follow a given medium at all show the highest levels of congruency with the media agenda, whereas audience members who are highly exposed to the medium show only weak agenda-setting effects. Regardless of these differences, however, the rank-orders of media and audience agendas might be very similar.
These problems are addressed by the second approach to testing agenda-setting effects. McLeod, Becker, and Byrnes (1974), for example, suggested a micro level approach to agenda-setting, comparing individual-level measures of perceived issue importance to a measure of media reliance weighted by the issue agenda predominant in that medium. Specifically, they examined the agenda-setting function of two competing community newspapers. Based on respondents' media reliance and the predominant agenda in the two local newspapers, they calculated a weighted index of received issue saliences that measured the weighted exposure to the agenda of one or both newspapers, depending on the media reliance patterns of each respondent. This analytic technique has the advantage that it allows for an individual-level estimation of agenda-setting effects.
The model outlined in Figure 3 is based on McLeod et al.'s (1974) core analytic model. It goes beyond their analyses, however, and applies their analytic strategies to the models of priming and framing. The goal of this model is to allow researchers to compare the distinctively different models of agenda-setting and priming, on the one hand, and framing, on the other hand. Specifically, it identifies the key differences between the respective models as far as content characteristics, audience variables, and media effects are concerned. To allow researchers to control for contingencies or third variable explanations, it also includes key controls identified in previous research (e.g., demographic controls, strength of partisanship, issue ob-trusiveness, information processing strategies, etc.). Specifically, the models for agenda-setting and priming, on the one hand, and framing, on the other hand, include six blocks of key variables.
Demographic controls and preexposure orientations. In addition to standard demographic controls, previous research on agenda-setting, priming, and framing has identified a number of preexposure orientations that influence how audiences receive and process information and that might therefore influence agenda-setting, priming, or framing effects.
For example, the obtrusiveness of issues or other qualities might influence subsequent information seeking and processing. Wanta and Wu (1992) intro-
COGNITIVE EFFECTS REVISITED      31 1
Agenda-Setting/Priming Framing
FIGURE 3 Analytic model of agenda-setting, priming, and framing.
Evaluations of Political Actors			Attributions of Causal/Treatment Responsibility
t			t
Audience Agenda (Salience, Response Latency)			Audience Frames (Frequency/Hierarchy of Frames)
t			t
Information Processing			
t t			
Weighted Media Use (Reliance weighted by Media Agenda)			Weighted Media Use (Reliance weighted by Frequency/Hierarchy of Media Frames)
i	f t		
Strength of Partisanship, Issue Obtrusiveness Gratifications Sought from Mass Media, etc.			
t			
Demographic Controls			
duced the notion of interpersonal issues; that is, issues that are not covered in mass media. Interpersonal discussion with others, for these issues, is the main predictor of a person's individual agenda and "may interfere with media agenda-setting effects" (p. 850).
In addition to the quality of the issue, McLeod et al. (1974) identified four contingent or contributory audience orientations "under which the [agenda-setting] hypothesis holds in greater or lesser strength" (p. 55). The first orientation is the strength of partisan orientation (independent of its direction). Weaker partisans, McLeod et al. found, show significant agenda-setting effects, even after controlling for the direction general partisanship. The second orientation is the dependence on newspapers as information source. As can be expected, newspaper agenda-setting effects were more likely to occur for people who reported relying on newspapers than those who reported not relying on newspapers. The third orientation is political interest. Respondents who reported lower levels of political interest were more susceptible to agenda-setting effects. The fourth orientation is gratifications sought from newspapers. Specifically, older respondents who reported using newspapers to keep up with latest events showed strong agenda-setting effects.
312 SCHEUFELE
Weighted media use. As argued earlier, indexes of weighted media use are a prerequisite for testing influences of media content on audience variables using individual-level data. As Figure 3 shows, however, different weighted indexes need to be calculated separately for agenda-setting and priming, on the one hand, and framing, on the other hand. Both indexes are based on respondents' reliance on a given medium for their political information. In most cases, the scope of the content analysis will make it unrealistic to include all print and broadcast media in a given market in the analyses. Rather, the predominant print and broadcast media should be selected. The content of each medium is then coded with respect to the frequency and hierarchy of issue frames or with respect to the salience attributed to different issues. Following McLeod et al. (1974), weighted indexes of exposure to media agendas or frames then are created based on a person's predominant media reliance.
Postexposure orientations: Information processing strategies. The individual-level effects of media agendas and news frames are very likely mediated by the way individuals process information they have gained from news media. Specifically, Kosicki and McLeod (1990) identified three dimensions of news processing. The first dimension refers to a respondent's tendency to scan media content for news of particular interest to him or her and not pay particular interest to or think about other stories. Framing and agenda-setting effects should be highest for this group. The second and third dimensions refer to a respondent's tendency to actively process and reflect on news content to get the real story behind the news. Reflecting on news content and pondering alternative frames and related issues can be expected to diminish the influences of media coverage on individual-level agendas and frames.
Audience agendas and audience frames. The final block of antecedent variables in the analytic model contains individual-level perceptions of salience and individual-level frames. These variables are the outcomes of agenda-setting or frame-setting processes, respectively. The theoretical differences between agenda-setting and priming, on the one hand, and framing, on the other hand, make it necessary to measure the two groups of variables in distinctively different ways. Individual-level salience of issues is ideally tapped indirectly through measures of response latency, whereas individual-level frames have mostly been measured by post hoc coding answers to open-ended survey questions or posttest questionnaires in experimental designs.
Outcomes: Evaluations versus attributions. As outlined earlier, previous research on priming has commonly examined evaluations of political actors as the final dependent variable. Framing research, in contrast, has focused on attributions of causal and treatment responsibility for social problems as the final depend-
COGNITIVE EFFECTS REVISITED      31 3
ent variable. To distinguish agenda-setting and priming from framing requires an assessment of discriminant validity. If agenda-setting, priming, and framing are related constructs, antecedents and outcomes of all three models should be similar. In other words, the blocks of antecedents outlined in Figure 3 should have a similar impact on individual-level perceptions of salience (i.e., on what people think about) and on the frequency and hierarchy of frames respondents invoke when they describe issues (i.e., on how they think about an issue). If priming and framing indeed differ from each other and from the concept of agenda-setting, they also should have different antecedents and outcomes. More important, each of the two models of media effects should have unique effects that remain significant, even if the other model is controlled for. For presidential evaluations as the dependent variable, for example, Iyengar and Kinder (1987) demonstrated significant influences of individual perceptions of issue salience. What remains unanswered is the question of whether the framing of an issue—regardless of its perceived salience—might have a significant effect on evaluations of political actors that goes above and beyond priming. This article has made a theory-driven argument in favor of precision rather than parsimony; that is, in favor of carefully distinguishing among agenda-setting, priming, and framing as different models of media effects. Following the analytic model outlined, future research will have to address this issue empirically.
ACKNOWLEDGMENTS
A previous version of this article was presented at the annual convention of the International Communication Association, San Francisco. I thank Patricia Moy, David Demers, Virginia Sapiro, and Zhongdang Pan for valuable comments on earlier versions of this article.
REFERENCES
Allen, B., O'Loughlin, P., Jasperson, A., & Sullivan, J. L. (1994). The media and the Gulf War: Framing, priming, and the spiral of silence. Polity, 27, 256-284.
Bassili, J. N. (1995). Response latency and the accessibility of voting intentions: What contributes to accessibility and how it affects vote choice. Personal and Social Psychology Bulletin, 21, 686-695.
Behr, R. L., & Iyengar, S. (1985). Television news, real-world cues, and changes in the public agenda. Public Opinion Quarterly, 49, 38-57.
Cobb, R. W., & Elder, C. (1971). The politics of agenda-building: An alternative perspective for modern democratic theory. Journal of Politics, 33, 892-915.
Cohen, B. C. (1963). The press and foreign policy. Princeton, NJ: Princeton University Press.
Collins, A. M., & Loftus, E. F. (1975). A spreading-activation theory of semantic processing. Psychological Review, 82, 407-428.
Dearing, J. W., & Rogers, E. M. (1996). Agenda-setting. Thousand Oaks, CA: Sage.
Demers, D. P., Craff, D., Choi, Y., & Pessin, B. M. (1989). Issue obtrusiveness and the agenda-setting effects of national network news. Communication Research, 16, 793-812.
314 SCHEUFELE
Edelman, M. J. (1993). Contestable categories and public opinion. Political Communication, 10,231-242.
Edelstein, A. S. (1993). Thinking about the criterion variable in agenda-setting research. Journal of Communication, 43, 85-99.
Entman, R. M. (1993). Framing: Towards clarification of a fractured paradigm. Journal of Communication, 43, 51-58.
Entman, R. M., & Rojecki, A. (1993). Freezing out the public: Elite and media framing of the U.S.
anti-nuclear movement. Political Communication, 10, 155-173. Fischer, G. W., & Johnson, E. J. (1986). Behavioral decision theory and political decision making. In R.
R. Lau & D. O. Sears (Eds.), Political cognition: The 19th annual Carnegie symposium on cognition
(pp. 55-65). Hillsdale, NT: Lawrence Erlbaum Associates, Inc. Funkhouser, G. R. (1973a). The issues of the sixties: An exploratory study in the dynamics of public
opinion. Public Opinion Quarterly, 37, 62-75. Funkhouser, G. R. (1973b). Trends in media coverage of the issues of the sixties. Journalism Quarterly,
50, 533-538.
Gamson, W. A. (1985). Goffman's legacy to political sociology. Theory and Society, 14, 605-621. Gamson, W. A., & Modigliani, A. (1987). The changing culture of affirmative action. In R. G. Braungart &
M. M. Braungart (Eds.), Research in political sociology (Vol. 3, pp. 137-177). Greenwich, CT: JAI. Gans, H. (1979). Deciding what's news. New York: Pantheon.
Gerhards, J., & Rucht, D. (1992). Mesomobilization: Organizing and framing in two protest campaigns in West Germany. American Journal of Sociology, 98, 555-595.
Gitlin, T. (1980). The whole world is watching: Mass media in the making & unmaking of the new left. Berkeley: University of California Press.
Goffman, E. (1974). Frame analysis: An essay on the organization of experience. Cambridge, MA: Harvard University Press.
Hastie, R., & Park, B. (1986). The relationship between memory and judgment depends on whether the
task is memory-based or on-line. Psychological Review, 93, 258-268. Heider, F. (1930). Die Leistung des Wahnehmungssystems [The system of human perceptions].
Zeitschrift für Psychologie, 114, 371-394. Heider, F. (1959). The psychology of interpersonal relations (2nd ed.). New York: Wiley. Heider, F. (1978). Über balance und attribution [About balance and attribution]. In D. Görlitz, W.-U. Meyer,
& B. Weiner (Eds.), Bielefelder symposium über attribution (pp. 19-28). Stuttgart, Germany: Klett. Heider, F., & Simmel, M. (1944). An experimental study of apparent behavior. American Journal of
Psychology, 57, 243-259. Huckfeldt, R., & Sprague, J. (1997). Accessibility and the political utility of partisan and ideological
orientations (Working paper). Bloomington: Indiana University. Huang, K. S. (1995). A comparison between media frames and audience frames: The case of the
Hill-Thomas controversy. Unpublished doctoral dissertation, University of Wisconsin, Madison. Huang, K. S. (1996, May). A comparison between media frames and audience frames: The case of the
Hill-Thomas controversy. Paper presented at the annual convention of the International Communication Association, Chicago. Iyengar, S. (1987). Television news and citizens' explanations of national affairs. American Political
Science Review, 81, 815-831. Iyengar, S. (1990). The accessibility bias in politics: Television news and public opinion. International
Journal of Public Opinion Research, 2, 1-15. Iyengar, S. (1991). Is anyone responsible? How television frames political issues. Chicago: University
of Chicago Press.
Iyengar, S., & Kinder, D. R. (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.
Kahneman, D., & Tversky, A. (1972). Subjective probability: Ajudgment of representativeness. Cognitive Psychology, 3, 430^54.
COGNITIVE EFFECTS REVISITED      31 5
Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263-291.
Kahneman, D., & Tversky, A. (1984). Choices, values and frames. American Psychologist, 39,341-350.
Klandermans, B. (1988). The formation and mobilization of consensus. In B. Klandermans, H. Kriesi, & S. Tarrow (Eds.), International social movement research: Vol. 1. From structure to action: Comparing social movement research across cultures (pp. 173-196). Greenwich, CT: JAI.
Klandermans, B. (1992). The social construction of protest and multiorganizational fields. In A. D. Morris & C. McClurg Mueller (Eds.), Frontiers in social movement theory (pp. 77-103). New Haven, CT: Yale University Press.
Klandermans, B., & Oegema, D. (1987). Potentials, networks, motivations, and barriers: Steps towards participation in social movements. American Sociological Review, 52, 519-531.
Kosicki, G. M. (1993). Problems and opportunities in agenda-setting research. Journal of Communication, 43, 100-127.
Kosicki, G. M., & McLeod, J. M. (1990). Learning from political news: Effects of media images and information-processing strategies. In S. Kraus (Ed.), Mass communication and political information processing (pp. 69-83). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Lang, G. E., & Lang, K. (1981). Watergate: An exploration of the agenda-building process. In G. C. Wilhoit & H. de Bock (Eds.), Mass communication review yearbook (Vol. 2, pp. 447^68). Beverly Hills, CA: Sage.
Lazarsfeld, P. F. (1957). Interpretation of statistical relations as a research operation. In P. F. Lazarsfeld & M. Rosenberg (Eds.), The language of social research (pp. 115-125). Glencoe, IL: Free Press.
Lodge, M., McGraw, K. M., & Stroh, P. (1989). An impression-driven model of candidate evaluation. American Political Science Review, 83, 399-419.
McCombs, M. F. (1997, August). New frontiers in agenda-setting: Agendas of attributes and frames. Paper presented at the annual convention of the Association for Education in Journalism and Mass Communication, Chicago.
McCombs, M. F., Llamas, J. P., Lopez-Escobar, E., & Rey, F. (1997). Candidate images in Spanish elections: Second-level agenda-setting effects. Journalism & Mass Communication Quarterly, 74, 703-717.
McCombs, M. F., & Shaw, D. L. (1972). The agenda-setting function of mass media. Public Opinion Quarterly, 36, 176-187.
McCombs, M. F., Shaw, D. L., & Weaver, D. (1997). Communication and democracy: Exploring the intellectual frontiers in agenda-setting theory. Mahwah, NJ: Lawrence Erlbaum Associates, Inc.
McLeod, J. M., Becker, L. B., & Byrnes, J. E. (1974). Another look at the agenda-setting function of the press. Communication Research, 1, 131-166.
McNamara, T. P. (1992). Theories of priming. Journal of Experimental Psychology, 18, 1173-1190.
McQuail, D. (1994). Mass communication theory: An introduction (3rd ed.). Thousand Oaks, CA: Sage.
Moy, P., Scheufeie, D. A., Eveland, W. P., Jr., & McLeod, J. M. (in press). Support for the death penalty and rehabilitation: Question artifact or communication effect? Journal of Applied Social Psychology.
Nelson, T. E., Clawson, R. A., & Oxley, Z. M. (1997). Media framing of a civil liberties conflict and its effect on tolerance. American Political Science Review, 91, 567-583.
Nelson, T. E., & Kinder, D. R. (1996). Issue framing and group-centrism in American public opinion. Journal of Politics, 58, 1055-1078.
Pan, Z., & Kosicki, G. M. (1993). Framing analysis: An approach to news discourse. Political Communication, 10, 55-75.
Petty, R. E., & Jarvis, W. B. G. (1996). An individual differences perspective on assessing cognitive processes. In N. Schwarz & S. Sudman (Eds.), Answering questions: Methodology for determining cognitive and communicative processes in survey research (pp. 221-257). San Francisco: Jossey-Bass.
Price, V., Tewksbury, D., & Powers, E. (1995, November). Switching trains of thought: The impact of news frames on readers' cognitive responses. Paper presented at the annual conference of the Midwest Association for Public Opinion Research, Chicago.
316 SCHEUFELE
Price, V., Tewksbury, D., & Powers, E. (1997). Switching trains of thought: The impact of news frames on readers' cognitive responses. Communication Research, 24, 481-506.
Quattrone, G. A., & Tversky, A. (1988). Contrasting rational and psychological analyses of political choice. American Political Science Review, 82, 719-736.
Rahn, W. M., Krosnick, J. A., & Breuning, M. (1994). Rationalization and derivation processes in survey studies of political candidate evaluations. American Journal of Political Science, 38, 582-600.
Rogers, E. M., Dearing, J. W., & Bregman, D. (1993). The anatomy of agenda-setting research. Journal of Communication, 43, 68-84.
Salancik, J. R. (1974). Inference of one's attitude from behavior recalled under linguistically manipulated cognitive sets. Journal of Experimental and Social Psychology, 10, 415-427.
Scheufele, D. A. (1999). Framing as a theory of media effects. Journal of Communication, 49, 101-120.
Sherif, M. (1967). Social interaction: Process and products. Chicago: Aldine.
Shoemaker, P. J., & Reese, S. D. (1996). Mediating the message (2nd ed.). White Plains, NY: Longman.
Snow, D. A., & Benford, R. D. (1988). Ideology, frame resonance, and participant mobilization. In B. Klandermans, H. Kriesi, & S. Tarrow (Eds.), International social movement research: Vol. 1. From structure to action: Comparing social movement research across cultures (pp. 197-217). Greenwich, CT: JAI.
Sudman, S., Bradburn, N. N., & Schwarz, N. (1996). Thinking about answers: The application of cognitive processes to survey methodology. San Francisco: Jossey-Bass.
Tourangeau, R. (1984). Cognitive sciences and survey method. In T. B. Jabine, M. L. Straf, J. M. Tanur, & R. Tourangeau (Eds.), Cognitive aspects of survey methodology: Building a bridge between disciplines (pp. 73-100). Washington, DC: National Academy Press.
Tourangeau, R. (1987). Attitude measurement: A cognitive perspective. InH.-J. Hippler, N. Schwarz, & S. Sudman (Eds.), Social information processing and survey methodology (pp. 149-162). New York: Springer-Verlag.
Tourangeau, R., & Rasinski, K. A. (1988). Cognitive processes underlying context effects in attitude
measurement. Psychological Bulletin, 103, 299-314. Tuchman, G. (1978). Making news: A study in the construction of reality. New York: Free Press. Tulving, E., & Watkins, M. J. (1975). Structure of memory traces. Psychological Review, 82,261-275. Tversky, A., & Kahneman, D. (1973). Availability: A heuristic forjudging frequency and probability.
Cognitive Psychology, 5, 207-222. Wanta, W., & Wu, Y. (1992). Interpersonal communication and the agenda-setting process. Journalism
Quarterly, 69, 847-855.
Weaver, D., McCombs, M., & Shaw, D. L. (1998, Ma>'). Agenda-setting research: Recent developments in place and focus. Paper presented at the annual convention of the World Association for Public Opinion Research, St. Louis, MO.
Winter, J. P., & Eyal, C. H. (1981), Agenda-setting for the civil rights issue. Public Opinion Quarterly, 45, 376-383.
Wyer, R. S., & Srull, T. K. (1986). Human cognition in its social context. Psychological Review, 93, 322-359.
Zaller, J. R., & Feldman, S. (1992). A simple theory of the survey response: Answering questions versus revealing preferences. American Journal of Political Science, 36, 579-616.