The Media Priming Effect: News Media and Considerations Affecting Political Judgments

Gerald M Kosicki. The Persuasion Handbook: Developments in Theory and Practice. Editor: James Price Dillard & Michael Pfau. Sage Publication. 2002.

Public officials, lobbyists, professional communicators, public interest group leaders, social activists, and corporate officials work tirelessly to bring their side of political and social controversies to the public. Through a variety of venues including television interviews, press releases, news conferences, photo opportunities, and paid advertising, potential news sources with a point of view to explain or promote hope to make themselves visible to the public (Pan & Kosicki, 1993; Price & Tewksbury, 1997). Sources and items selected for inclusion in the news are chosen for their news values (Gans, 1979; Manning, 2001; Pan & Kosicki, 2001), a practice that gives the news a certain set of themes or perspectives (Turow, 1992, p. 108). Other values, such as quality, accuracy, completeness, and diversity, can promote effective public discourse, understanding, and judgment or can hinder it.

When considering the impact of news reports on public opinion about political issues, agenda-setting has been a dominant model guiding research (McCombs, Shaw, & Weaver, 1997). In fact, the agenda-setting perspective has become so common that many use the phrase agenda-setting to refer to nearly any type of research question involving public issues and media. Scholars working in this framework have become adept at expanding it in many directions, incorporating new dimensions as part of the expanding area of agenda-setting. However, agenda-setting is more properly viewed as a single type of media effects hypothesis that concerns the connection between the news media and public issues (Kosicki, 1993).

Agenda-setting was arguably a leading edge of the new look in mass communication research of the late 1960s that rejected simple models of persuasion as they were understood at that time and embraced “cognitive effects” that concentrated more on what people learned, how they thought, or what they thought about (Becker, McCombs, & McLeod, 1975; McLeod & Reeves, 1980;

Reeves, Chaffee, & Tims, 1982). This cognitive perspective has tended to focus on indirect effects (Becker & Kosicki, 1995). Work in persuasion was redirected by information processing models, which led to an expansion of media effects studies in both persuasion and other aspects of public affairs media effects.

Agenda-setting research concerns itself with establishing links between what broad topical areas the media consider important by the amount and prominence of their coverage and public acceptance of this list of concerns. Agenda-setting research typically defines public issues as rather broad, topical, and abstract areas such as “trust in government,” “the economy” “environment,” and “crime.” This innovation makes it possible to track the media agenda in a general way over long periods of time and to use a wide variety of survey data to assess the public agenda. Correlations between the rank orders of aggregated media and public agendas are taken as evidence of an agenda-setting effect (McCombs & Shaw, 1972). Survey-based studies continue in this line to the present day, albeit with many refinements, distinctions, and caveats (Kosicki, 1993). Research that was explicitly cognitive began appearing during the 1980s. Most such research used experimental methods to examine media influence on public agendas (Iyengar & Kinder, 1987; Iyengar, Peters, & Kinder, 1982). Agenda-setting remains a model with tremendous heuristic value to many scholars of political communication, and the literature continues to build in the area despite certain theoretical and methodological flaws (Kosicki, 1993).

Iyengar et al. (1982) developed the media priming hypothesis as an outgrowth of some experiments on agenda-setting as a way of showing how people’s views of public issues might have real-world political consequences. One way to do this was to attempt to show what processes guide the formation and expression of views about public policies and political figures. In other words, what are the ingredients of public affairs judgments?

Media priming is a media effects hypothesis that suggests that the news media influence the standards by which political figures or public policies are judged by calling attention to some matters and ignoring others (Iyengar & Kinder, 1987, p. 63). As noted by Price and Tewksbury (1997), media priming refers to the tendency of audience members to evaluate their political leaders according to the particular events and issues that have been highlighted in news reports. This approach to persuasive communication, by focusing on the standards of political judgment and how they are formed, represents a particularly widespread and important phenomenon in political life because it provides the basis for matters such as voting for candidates and issues. It is also a principal mechanism for explaining mass media effects in many political contexts. Media priming is a general type of effect and thus can be applied across many types of persuasive communication and in many types of contexts, from election campaigns to routine news coverage between election periods.

Price and Tewksbury (1997, p. 198) argued that agenda-setting is, in theoretical terms, a special case of media priming. This is so, they argued, because media priming is a temporary accessibility effect. Media messages make a particular construct applicable and activate it. After activation, the construct remains temporarily accessible. When an individual needs to evaluate the performance of a public official, this concept is likely to be summoned to mind. Unless deemed inapplicable to the evaluation, it will be used for making a judgment about that official’s performance. Price and Tewksbury argued that one such judgment could be issue importance, in other words, agenda-setting. This argument makes sense logically but does not account for the historical development of either perspective given that agenda-setting grew out of a very different intellectual tradition from the media priming literature. This theoretical convergence highlights one powerful advantage of the cognitive approach: a comprehensive, flexible set of theory-building tools that can be applied in many settings (Price & Tewksbury, 1997). One of the many strengths of the cognitive approach is that it begins with a series of realistic assessments of human thinking powers. Basic to this approach is the notion that people are faced with many complex mental tasks and that these tasks, if approached with an equal degree of diligence, would be overwhelming. Accordingly, for many matters, individuals do not attempt to make the best possible decisions. Instead, especially for routine matters or relatively unimportant concerns, they employ series of mental shortcuts that are expected to yield decisions that are good enough. Simon (1957) has called this decision-making strategy “satisficing.” It assumes that people do not have the ability, time, or interest to make the best possible decision in every case, so they satisfice rather than optimize.

Cognitive theory suggests that satisficing strategies are typical for most matters most of the time unless the issue is particularly important or salient to the individual. The more important a problem, the more likely individuals are to expend cognitive effort. This also suggests that for most issues, most of the time, people engage in satisficing. This is particularly so when people do not see their stake in an issue as being very great. Lacking reasons for taking more time or energy with a particular decision, people are thought to rely not on an exhaustive search of all their stored information but instead on a subset of information that comes to mind (e.g., Taylor & Fiske, 1978). For many people, media may be the principal source of this information.

The priming hypothesis provides a theoretically driven perspective from which to consider the question of media influence in the world of public affairs and politics. It has been most heavily used to date in studying the criteria that people use to evaluate the president, and that emphasis is reflected in this chapter. Satisficing is important to understanding political judgments because for most people, most of the time, political matters are not of intense interest. In fact, politics often gets slight attention. Media cues are often the main ones that people turn to when asked to make political judgments. Furthermore, few people would feel the need to optimize on formulating their answers to political questions.

This chapter discusses media and their connection to an effect of media called media priming. Media are seen as industrialized processes for the creation and distribution of messages that have certain properties that are influential in the priming process. These properties include systematic problems that may make priming more powerful in certain areas of political life, depending on media performance. Media treatments of public policy issues and candidates suggest that the media will present certain considerations such as facts, emotional appeals, exemplars, and framing. Media priming is thus seen to be a “subtle mechanism by which the media alter the conduct of public policy-making” (Miller & Krosnick, 1996, p. 80). The chapter continues with a discussion of media production factors and content. It moves from content to audience factors in priming and then to public policy considerations.

Defining Mass Communication

While there are many ways to define mass communication, a particularly useful one is provided by Turow (1992): “Mass communication is the industrialized [mass] production, reproduction, and multiple distribution of messages through technological devices” (p. 107). In this definition, “messages” can be any linguistic or pictorial representations that appear purposeful. The word “industrialized” indicates that the process is carried out by mass media firms, that is, by “conglomerations of organizations that interact regularly in the process of producing and distributing messages” (p. 107).

The definition means that industrial application of technology is a key defining feature of mass communication, and this is crucial in separating mass communication from other forms of communication. The industrial process suggests that large numbers of messages and audience members are involved. The definition calls special attention to the industrial nature of message creation, particularly the relationships between media and other sources of social, political, and economic power.

A key element of this perspective is that in stressing attention to the production and distribution processes, along with the considerations that shape news, information, and entertainment, there is a recognition that these cultural patterns lead to similar cultural models across individual stories. As Turow (1992) noted,

Even when the specific subjects that media carry are not the same, the similarity with which creators approach the world can still yield similar perspectives. When local TV [television] stations across the country daily fill their 7:25 a.m. newscasts with tales of overnight murders, robberies, and accidents, the general knowledge and worldview gleaned from those patterns might be the same, even though individual stories might not be. (p. 108)

Although journalists like to say that they simply cover the news and reflect reality, it is commonly recognized in the literature of the field that things are more complex than that. If news simply reflected the world of official announcements and the activities of business and political elites, there would be few reasons to make news treatment of topics a subject of study.

Framing News and Public Affairs

Public deliberation is the process of collective and open reasoning and discussion about the merits of public policy (Page, 1996), and it is fundamental to the health of democratic societies (Kinder & Herzog, 1993). Framing is one perspective from which to approach the study of public deliberation, which includes “the discursive process of strategic actors utilizing symbolic resources to participate in collective sense-making about public issues” (Pan & Kosicki, 2001, p. 36). Framing analysis is an analytical approach connecting normative propositions from “deliberative democracy” and the empirical questions of collective decision making to understand political participation and experiences in democratic society (Baumgartner & Jones, 1993; Kingdon, 1984).

Goffman (1974) defined frames as “schemata of interpretation” that enable individuals to locate, perceive, identify, and label occurrences in everyday life. Framing analysis deals with how various social actors act and interact to produce organized ways of making sense of the world. Frames are “central organizing ideas to understand and organize political reality” (Gamson & Modigliani, 1987, p. 143).

Framing owes much to its roots in cognitive science. Price and Tewksbury (1997) reviewed this literature well, and it is not repeated here. Others such as Iyengar (1991), Cappella and Jamieson (1997), and Scheufele (1999) made this case well in the area of media effects. But the effects tradition does not exhaust what framing brings to the study of public opinion and communication.

Frames define the boundaries of discourse surrounding an issue and categorize each political actor involved in the issue in a particular way. Heated political contestation surrounds these boundaries as well as what is an acceptable definition of a given issue. Pan and Kosicki (2001) argued that framing analysis needs to place political and social actors in the center of the process. The framing potency of any actor comes from the resources one can bring to bear on a framing situation, one’s strategic alliances, and one’s stock of skills and knowledge in the arts of frame sponsorship and management. Through such resources, they argued, political actors weave “webs of subsidies” to privilege the dissemination and packaging of information in the most advantageous directions.

The combination of a frame and its symbolic devices or “packages” functions as a narrative that may resonate in the minds of other actors under certain conditions, which include credibility and other shared experiences (which form the basis for “frame alignment” [Snow & Benford, 1988]). This alignment is enhanced to the extent that an actor can link his or her frame to some enduring values in the society and thus subsidize other actors in processing and packaging information concerning a given issue. Framing potency is also determined by certain sociological factors such as the size and depth of an actor’s web of subsidies and the framing actor’s ability to mobilize such subsidies with strategic targeting (Pan & Kosicki, 2001). In general, the broader a discursive community becomes and the more clear its identity becomes, the greater the likelihood of achieving its goals.

The type of framing analysis discussed here requires the use of tools and strategies for analysis quite different from those in public opinion analysis, although public opinion analysis might be one part of the total picture. Developing an understanding and empirical verification of such framing efforts requires data about the various framing actors and groups, their communication and other organizing strategies, and their messages to other groups and the public. Although many types of data in various combinations may prove useful in understanding these processes, the richness and interconnected nature of the activities and actors implies that case history approaches may be favored for this part of the story (e.g., Gamson & Modigliani, 1987; Gitlin, 1980).

Pan and Kosicki (1997) have also introduced the notion of “issue regime” to denote “Periods in which a news story is so big that it dominates the amount of total media attention available” (p. 4). Examples include the Clinton-Lewinsky scandal during the late 1990s and the saturation coverage of the Gulf War. These giant stories occupy so much time and space, and consume so many media resources, that they tend to crowd out other news items and thus have a profound impact on public discourse and public attention. Following the work of Hilgartner and Bosk (1988), they rely on the notion of public arena, that is, the totality of channels, forums, and means of public deliberation about public policy and the relevant venues in which such matters are discussed. This is based on the insight, among others, that there is only so much channel capacity in the public arena, and so issues compete for resources. The issue regime concept has proven useful in the analysis of public opinion during and after the Gulf War, in which president Bush went from the most popular American president to one of the least popular, in part because of declining media attention to the war and the increasing attention to the economy, which was in recession. Issue regime as a concept allowed the researchers to document the public opinion effects of media switching between the Gulf War and the economy and the consequences of this for Bush’s approval rating.

Similar systematic media coverage or biases in media coverage have been documented over far longer periods of time. For example, Gilens (1996, 1999) used a variety of data in innovative ways to carefully describe media emphasis on social welfare as an issue, particularly the shift in media focus over the years from whites to blacks and the growing use of black examples in news coverage. This leads to a picture of the world in news that is, in certain ways, the opposite of real-world indicators; for example, only about 29% of poor Americans are black, but 62% of poor people portrayed in major newsmagazines between 1988 and 1992 were black (Gilens, 1996). This kind of systematic misrepresentation can be expected to influence people’s judgments about social welfare policies.

Of course, framing of public issues is an inevitable process because all news is necessarily told from some point of view, even as journalists struggle to be objective or fair. Consensus values of the larger society are always present, a phenomenon that Gans (1979) called “enduring values in the news.” Beyond this, however, media cover public policy issues and events in ways that may provide clear signals to audience members about the ways in which the issue is to be understood and processed.

Fundamentally, news reporters tell stories about things that happen in the world that they think are important for people to know or contain elements that people will find interesting. In so doing, journalists provide a steady flow of information to the public about current affairs and politics. While telling their stories, journalists make active choices about what information to select, what issues to stress, and how to present this information. In the political sphere, media might be said to present at least four types of information that are highly relevant to people’s understanding of public issues.

First, the news media provide factual information about conditions in the world. Such reports provide individuals, communities, and nations with a shared information base that people use in forming their basic understanding of the world. Much of this information closely tracks the pronouncements of officials and reports of conditions as provided by official government agencies, corporations, academics, and other types of qualified experts (e.g., Ericson, Baranek, & Chan, 1989; Gans, 1979). If we accept this characterization of journalism as a type of activity relying primarily on human sources, it becomes a major concern as to how these sources are selected and used (e.g., Gans, 1979). Mainstream media make attempts to screen sources carefully, often to the point where they are vulnerable to criticism for sampling experts too narrowly to provide truly diverse discourse and perspectives about issues of the day.

Second, the news media also provide images, examples, and episodes that function as exemplars or explications of more abstract ideas and principles of public policy matters. Stories of the “deserving poor,” or of individuals struggling against racial animosity and inequality, may help to explicate the principle of equal opportunity and what it means or does not mean in contemporary American life. News filled with images of single mothers on social welfare, crack addicts, prostitution, and alcoholic homeless people, without discussion of the social conditions that tend to foster or exacerbate such problems, will encourage the view that affirmative action, social welfare, and related policies tend to give unearned advantages (Pan & Kosicki, 1996; see also Gilens, 1999; Iyengar, 1991; Sniderman, Brody, & Tetlock, 1991).

Third, the vividness and immediacy of news media reports help to arouse, sustain, and renew various affective and emotional experiences about an issue. Television coverage of the Gulf War fostered patriotic emotions and pride that contributed to the rally effect, in which the public supported the country’s leader during difficult times. In this case, that meant contributing increased support to the president’s policies once the war started (Pan & Kosicki, 1994, 1996). In terms of domestic public policy issues such as affirmative action, the media may renew aversive feelings toward blacks with stories that associate blacks with crime, militancy, welfare, and so on (Entman, 1992; Pan & Kosicki, 1996; van Dijk, 1991). Television may be more effective than print media in helping people to activate their emotional responses due to the unique combination of visual and sound stimuli.

Fourth, the news media, through the use of exemplars, emotional appeals, metaphors, visual images, sound bites, and other symbolic devices, frame public policy issues in unique ways (Gamson, 1988; Gamson & Modigliani, 1987). Framing suggests an interpretation of issues or events in the news (Pan & Kosicki, 1993). Framing effects tend to limit the scope of discussion among members of the public, but they also supply the vocabulary for public discussion (Gamson, 1992; Iyengar, 1991). For example, framing policies about race as either “affirmative action” or “reverse discrimination” brings different considerations to the public’s mind and may consequently lead to different opinion responses (Kinder & Sanders, 1990). Chong (1996) discussed the need to create common frames of reference or considerations between elites and public when thinking of public policy.

Question wording experiments that vary these types of word choices in the context of surveys, growing out of a concern about the robustness of the method, are now motivated by concerns arising from cognitive psychology (e.g., Feldman, 1995; Schwarz & Sudman, 1996; Sirken et al., 1999). Although primarily methodological in character, they may be seen as examples of such framing effects and even given substantive interpretations under certain circumstances (Kinder & Sanders, 1996).

This discussion of framing has been intended to describe in a theoretical way variations in news media content as it relates to public issues. In the next section, the priming hypothesis helps to connect these issues illuminated in media content to the study of variations in public opinion in the evaluations of presidents and public policies.

Priming

As has been discussed, the media priming hypothesis suggests that the news media stress certain considerations, and not others, in their news materials about public policies or candidates for office. In so doing, media alter the standards by which these policies or candidates for office are evaluated. Of course, it is well-known that media content alone does not produce effects (McLeod & Reeves, 1980). In other words, learning and many other effects of media messages are subject to the usual limits of physical exposure, attention, and other information processing strategies, in which the audience member may reject or accept the basic premise of the information (e.g., Cappella & Jamieson, 1997; see also Austin & Pinkleton, 1999; Kosicki & McLeod, 1989; Zaller, 1992).

Priming versus Agenda-Setting and Framing

The agenda-setting model has been so Pervasive in the field that virtually any reference to the topic of media and public issues will be said by some to be a case of agenda-setting. Yet we have seen that framing and priming have their places as well; therefore, it is useful to offer a comparison among the various concepts so that each will be understood with minimal ambiguity in the discussion that follows.

Agenda-setting is commonly understood as a model that links the salience or priority of issues in the media with the priorities of the public (Kosicki, 1993). A key to understanding agenda-setting is the typical conceptualization of public issues as relatively broad, abstract categories such as environment, trust in government, and poverty. Mutz (1998, p.70) called agenda-setting a “nondirectional” effect, by which she meant the effect is supposed to result from the amount of salience of the coverage of a particular issue and does not matter whether the coverage is favorable or unfavorable. There are both strengths and weaknesses of this approach. On the one hand, the use of abstract issues makes it possible to study the rise and fall of salience of a slate of issues over many decades. On the other hand, the abstract nature of what is studied is quite removed from the details of the issue. This suggests that often we are left with the feeling that agenda-setting studies do not tell very much about the nature of news discourse or the audience effects that they purport to study. As Kosicki (1993) has described this, issues are, or ought to be, a type of contested discourse, yet “the agenda-setting framework strips away almost everything worth knowing about how the media cover an issue and leaves only the shell of the topic” (p. 112). Cappella and Jamieson (1997) made a similar point concerning media priming:

Media priming, like agenda setting, is not concerned with how issues are treated in news coverage, only with their relative frequency. Priming would treat a story about president Clinton’s decision to send troops to Haiti described as a ploy to drum up support prior to fall elections as equivalent to a story about sending troops to Haiti to ensure the return of democracy. These two stories are framed very differently, but they are equivalent from the viewpoint of media priming. (p. 52)

Despite these reservations about priming research, in contrast to agenda-setting, it is often more complex and detailed. Priming relies on a series of evaluative criteria that vary in detail and relates these to both media and their effects on overall evaluations of issues or political figures. The dependent variables are often evaluations of policies or political figures, not the mere salience of an issue.

Framing, by contrast, is a conceptual framework for examining the details of how issues are conceptualized in public discourse as highly contested matters over which there is often considerable disagreement. Framing takes as a starting point the idea that language matters in the study of media and issues and makes the analysis of metaphor and other lexical choices very prominent (Pan & Kosicki, 1993, 2001). Framing as explained by Iyengar (1991) is a particular type of context effect. News stories are either told as event-driven stories, which he called “episodic,” or reported with some larger background material, which he referred to as “thematic.” In Iyengar’s framing effects work, it does not matter whether the actual context is positive or negative. For this notion of framing effects, Iyengar drew on the work of Kahneman and Tversky (1987), who showed in experiments that equivalent choices could be presented to participants in such a way that their choices would be systematically altered depending on the mode of presentation.

Unpacking the Expectations of Priming

The priming perspective suggests that when people are asked to give an evaluation of a public policy or public figure, they do not evaluate all of the information they have or can find about that topic, weight it according to some priorities, and then calculate a logical response. Instead, people use only a sample of readily available information. This sample of information that is stimulated by priming has nothing to do with motivation. The sample is small and biased because of the limited nature of the cognitive system. Furthermore, this sample is not a random sample of all information but instead a specific subset of what is cognitively accessible (e.g., Feldman, 1995; Higgins, 1996; Higgins & King, 1981; Zaller, 1992). Accessible information for most people is what is presented by news media. Issues that have been portrayed in the news repeatedly are most likely to be evoked in people’s mental calculations at a given time by media. This is the basis of the priming hypothesis.

Media coverage of an issue is an indication of the salience of the issue in media content. This may affect the cognitive processes that audience members use to make sense of a given issue. Pan and Kosicki (1997, p. 10) noted that the process may happen in at least three ways: (a) by increasing the ease with which the related thought elements or considerations are activated, (b) by increasing the breadth of the accessible thought elements related to a given topic or issue, or (c) by tightening the links of the various thought elements. This means that individuals will be more likely to use the aspects of issues activated by media in their calculations for evaluating a given policy or policy actor such as the president.

Note that priming is thus very different from traditional persuasion research, as explained by Miller and Krosnick (1996):

Whereas persuasion focuses on media messages advocating particular positions, priming can be invoked simply by a news story devoting attention to an issue without advocating a position. And whereas persuasion is thought to result from effortful decision making about a message’s likely veracity, priming presumably occurs as a result of automatic and effortless processes of spreading activation in people’s minds. (p. 81)

A key point in any type of determination of media effects is to specify precisely what is the proposed effect (McLeod & Reeves, 1980). Building on these insights, Miller and Krosnick (1996) posed several subhypotheses, which they claimed form the core of the concept of priming. The first of these is known as the target gradient hypothesis. This suggests that priming effects will decrease in strength as attitudes become more remote from those being directly activated by a news story. Miller and Krosnick (1996, p. 83) asserted that this gradient hypothesis actually has two parts. The first part relates to the target and addresses which attitudes have their accessibility altered by priming and by how much. In other words, priming induced by a news story should occur mostly for attitudes directly related to the content of the story. News stories about the president’s economic policies should presumably mainly affect attitudes about the president’s economic policy performance, and these effects should be strongest. The effects might spread to judgments about the president’s health policies because of their possible economic implications (p. 82).

The second part, known as the consequence gradient hypothesis, concerns which overall presidential evaluations are altered and to what extent. The argument here is that the relevance of a given attitude will, in part, determine the extent to which the accessible attitudes make a difference to the judgment being made. Scandal information might not be seen as relevant to economic or international relations performance (Miller & Krosnick, 1996, p. 82).

These hypotheses are best explained by noting that political attitudes are structured in memory as part of an associative network (e.g., Collins & Loftus, 1975; Higgins, 1996). This network is made up of nodes (i.e., political attitudes) and strong links to other sets of related attitudes. Nodes that are less logically related or not related at all will have weaker links or perhaps no links at all. When circumstances are such that a particular node is activated, related nodes are also pulled into short-term memory. This spreading activation is what gives priming its power to connect various thought elements. The extent to which concepts are activated in this type of network depends on many things that are beyond the scope of this chapter.

In the case of candidates, it has become customary to conceptualize evaluations in terms of competence, leadership, integrity, and empathy (Kinder, 1986). Because individuals do not have the motivation or ability to simultaneously think about all considerations, we are likely to see patterns in which those closer to the topic at hand are considered more, whereas others that are more remote are relied on less. Media, as noted earlier, are also limited in terms of the comprehensiveness of their reports and in any case tend to stress certain aspects of stories that are thought to be especially newsworthy. For example, a personal scandal can be expected to have greater effects on integrity than on competence or leadership. Similarly, information made accessible by news coverage that is highly relevant for one of these dimensions may be irrelevant for another. For example, information from the news that activates integrity might not also activate competence.

The dosage hypothesis refers to the amount of media coverage on an issue that enters a person’s consciousness (Miller & Krosnick, 1996, p. 83). In general, the expectation is that the greater one’s use of media content and the more attention that is paid to it, the more one should be influenced by it.

Miller and Krosnick (1996) also discussed the resistance hypothesis, by which they meant that “holding dosage constant, the more knowledge of politics one has, the more resistant one should be” to priming effects. This suggests that more knowledgeable individuals have greater cognitive experiences, which helps them to resist fluctuations in the accessibility of information. On the other hand, those with little knowledge are more likely to be responsive to media primes.

Empirical Tests of the Priming Hypothesis

The basic priming hypothesis has been tested using survey and experimental methods. The results are generally confirming for studies of both types. A typical experiment might manipulate aspects of stories relating to a single theme in the context of a newscast or series of newscasts viewed over time. For example, Iyengar and Kinder (1987) developed early studies in which experimental participants viewed newscasts that emphasized inadequacies in defense preparedness or not. They were testing participants to see whether the treatment would influence the standards that viewers applied to evaluating the president’s overall performance. A day after the final newscast was shown to participants, they were asked to rate the president’s performance on matters such as defense and overall performance. The expectation is that people who saw stories about defense inadequacies should give more weight to the president’s performance on that problem when evaluating his overall performance. In the experiment, for people who saw multiple stories about defense (i.e., those who were primed on that theme), the impact of ratings on the president’s performance on defense was more than twice as great as that for people who were not so primed.

Priming studies using survey research typically attempt to measure various aspects of people’s constructions of the issues and relate these to exposure to aspects of news content.

Iyengar and Kinder (1987) reported some priming results using data from the National Election Studies (NES, 1980 major panel, N = 1,008), with data gathered at four points during the election year. A series of standard items were asked to measure concepts of “competence” and “integrity” of president Carter. The raw items were factor-analyzed separately for Carter, and then overall evaluation was estimated by regression analysis with predictors including inflation performance, energy policy performance, unemployment performance, hostage crisis performance, and Afghanistan performance. The expectations were that the evaluations on specific topics that had been in the news— inflation, energy crisis, and Iranian hostage crisis—would predict Carter’s general evaluation but not judgments of his competence and integrity. Results confirmed these expectations, with the strongest relations shown between inflation and Carter’s general performance, less with judgments of his competence, and only small effects with judgments of his integrity.

Experimental data are generally confirming of the priming gradient and general priming hypothesis but not the dosage hypothesis. It was found that even moderate exposure triggered priming responses (Iyengar, Kinder, Peters, & Krosnick, 1984). Participants viewed one of three 40-minute sets of network news coverage with varied levels of exposure to energy stories. Priming effects were measured by unstandardized regression coefficients on evaluations of general performance, competence, and integrity under experimental conditions of “some coverage.” Effects of energy performance on overall performance ratings were larger in the coverage conditions than in the no coverage conditions. This confirms the general priming hypothesis but not the dosage hypothesis given that there were no differences between the high and moderate exposure conditions. The consequence gradient hypothesis was supported given that priming of energy performance ratings showed greater influence on evaluations of overall job performance than on judgments of competence and integrity (Iyengar et al., 1984). These results were replicated in a second study, but with added qualification. The second study of general attitudes toward Carter’s performance also confirmed that knowledge does help to resist priming, as more knowledgeable individuals are less likely to be primed, and the less knowledgeable consistently are more influenced by their exposure to the experimental media primes. In this case, regression coefficients predicting overall performance ratings from ratings of energy, inflation, and defense performance were considerably lower for the high knowledge group than for the low knowledge group.

Krosnick and Brannon (1993), in their Gulf War study, found that knowledge, exposure, and interest all had the same effect on priming; that is, higher involvement was associated with reduced priming. This finding is quite common in the literature (e.g., Iyengar et al., 1984; Krosnick & Kinder, 1990). But when knowledge, exposure, and interest are studied simultaneously, their effect turned around; that is, high levels of knowledge increased priming, and high levels of exposure and interest reduced priming.

In explaining these results, Krosnick and Brannon (1993) noted that while a traditional view of political involvement suggests that greater exposure and interest would indicate a stronger dose of media content, and that greater knowledge suggests resistance, to new information, such results might not be found in practice. They interpreted knowledge as providing individuals with a greater ability to interpret, store, and retrieve new information, while media exposure and interest are associated with a higher probability of online processing and decision making. If this is true, then one should expect weakened priming effects due to exposure because it opens the individual to a wider range of stimuli that can be activated by media. Krosnick and Brannon concluded that the assumptions of the dosage and resistance models are likely correct due to available evidence from various areas of psychology. However, they noted that the real question relates to their application in priming research (p. 972). More work is clearly needed in this area to understand the roles of these concepts in priming and the possible effect of issue or setting on the results.

Survey data reinforce the basic priming, gradient, and target gradient hypotheses. In a study of the Iran-Contra affair, attitudes toward president Reagan’s performance were better predicted by attitudes toward aid to the Contras and aid to Central America than were items such as general isolationism, U.S. military strength in general, health of the national economy, and federal aid to blacks (Krosnick & Kinder, 1990). Using data from the NES, the authors took advantage of a natural experiment by dividing the sample into “prerevela-tion” and “postrevelation” groups depending on their date of interview in relation to the national television announcement by Attorney General Meese that money from the secret sale of weapons to Iran had been diverted to the Nicaraguan Contras. Regressions predicting Reagan’s overall performance rating were predicted by variables such as aid to Contras, isolationism, U.S. strength, economic assessments, and aid to blacks. As examined in the postrevelation sample, the basic priming hypothesis was supported in that aid to Contras was the most important predictor of overall evaluations. The target gradient hypothesis was supported because a slightly smaller priming effect occurred for views about isolationism, which is related to Contra-Central America support. Priming was not seen for the more distant predictors of economic assessments and aid to blacks.

In the same study, the consequence gradient hypothesis was also supported. However, the effect of Contra-Central America support extended not just to judgments of overall competence but to integrity as well. Miller and Krosnick (1996, p. 88) noted that this is appropriate given that the Iran-Contra affair raised questions that were widely debated about Reagan’s competence and integrity.

In a survey study of priming during the Gulf War, Krosnick and Brannon (1993) found additional support for priming in general and the target gradient hypothesis in that the impact of foreign policy performance and domestic economic policy performance did not change much. However, Pan and Kosicki (1997), using additional data from the NES, were able to show how a change in media emphasis on the war after the gulf crisis to the economy shifted public opinion focus from one issue to the other. These results were shown through a more complete data set that encompassed a longer time frame than that employed by Krosnick and Brannon (1993). In addition, Pan and Kosicki augmented the NES data with a detailed month-by-month content analysis. This was effective in showing the shift in media emphasis from the Gulf War issue regime to the presidential campaign regime in which then Democratic challenger, Bill Clinton, effectively kept the focus on the economy, to the severe disadvantage of president Bush. This study effectively documents the fall of Bush from the most favorable presidential ratings in history to one of the lowest in the space of several months, and it clearly shows the shift in media and public opinion emphasis from Gulf War to the economy.

Conceptualizing the media’s role in the priming process, Pan and Kosicki (1997) employed the notion of “issue regime,” which is a period in which one issue dominates the public arena or “an overall consonance of all the channels and apparatuses of public deliberation concerning public issues during the period” (p. 10). A greater reliance on this dominant issue in evaluating a political figure such as the president results from and indicates the dominance of the issue in the public arena, which includes media coverage, real-world cues, and interpersonal communication. As successful as this strategy is, it still leaves us with the need to demonstrate the direct priming effect of media.

Krosnick and Brannon (1993) linked their analysis to media exposure and argued that people with lower levels of exposure are more affected by priming because these people will absorb the “big message” from media, but without details. This study suggests that people form their judgments by retrieving information only when needed by a task such as retrieving a memory-based judgment. They are, in effect, arguing that media exposure is only a contingent condition for the priming effect, in other words, necessary but not sufficient.

Pan and Kosicki (1997) considered media a priming agent and estimated priming effects directly from media exposure variables. In this formulation, heavy exposure to media coverage of a dominant issue leads to stronger or more frequent activation of thought elements related to that issue. Thus, increased media exposure is related to increased activation. They added the notion of affective valence, that is, the negative or positive tone associated with each issue. In their example, media coverage of the Gulf War carried a positive “big message” for president Bush because he was getting credit for the military victory. Coverage of the economic recession in the United States was negative for Bush.

Pan and Kosicki (1997), using panel data from the NES covering the period from late 1990 to 1992, showed that priming is temporary and situational. Media changed their dominant issue focus, and president Bush’s approval ratings changed together with shifting issue dominance. Pan and Kosicki noted that the timing of data gathering in the context of such fluid political reality is an issue that deserves more consideration by scholars, in terms of the pragmatic issues of when to gather data but also in theoretical terms. If priming activation and decay are to be studied as dynamic concepts, then the current commonly used methods of pre- and posttest studies might not be optimal. More theoretical work is needed in the temporal aspects of priming.

A few other variables have also been investigated in terms of their mediation role in priming with mixed results. Krosnick and Brannon (1993) found greater knowledge related to priming in the context of the Gulf War, although in other settings, Krosnick and Kinder (1990) found the opposite—that priming was limited to those who know the least. Although the analysis in the former study is more detailed, the relationship of priming to knowledge is clearly an area that needs additional attention due to its importance.

In a study of president Clinton’s performance evaluation, Miller and Krosnick (1997) found significant positive results in their examination of the relationship between trust in media and priming. In this case, using a mix of media trust measures that included credibility and respondents’ sense that the media were appropriately focusing on important news, they found that people who were high on trust of media were more likely than others to be primed by issues that the media raised. In a subsequent study, Miller and Krosnick (2000) produced additional support for the effects of media credibility on priming.

Cappella and Jamieson (1997) noted that “framing effects are more subtle than media priming and agenda setting” (p. 83). By this, they meant to draw attention not just to the presence of certain topics in the news but also to how topics are treated, similar to the valence issues discussed previously but extending beyond this into framing effects. Their study, of media and their role in fostering cynicism, employed an explicitly cognitive model incorporating elements of spreading activation with particular journalistic details to suggest that people can be primed in ways that priming theory suggests. However, Cappella and Jamieson studied far more subtle journalistic details than have been proposed to date in the mainstream priming studies. This suggests a major avenue for future studies in the area.

Price, Tewksbury, and Powers (1997, p. 403) reported a pair of experimental studies developed as a clear extension of priming research into more subtle terrain akin to some of the experiments on value framing of Ball-Rokeach, Power, Guthrie, and Waring (1990) and Shah, Domke, and Wackman (1996). Price et al. (1997) studied a process where “by activating some ideas, feelings, and values rather than others, the news can encourage particular trains of thought about political phenomena and lead audience members to arrive at more or less predictable conclusions” (p. 404).

Priming research over the next few years should develop in several directions. First, we can expect more detailed studies to understand the mechanisms of priming. In one of the early reported studies of priming, Iyengar and Kinder (1986, pp. 148–150) discussed the need to demonstrate that the effects they were calling priming were not, in fact, effects of projection. Projection, defined as the opposite of priming, would be present if news coverage of a particular problem caused people to adjust their rating of the political figure on that problem to match their overall rating of the political figure. Iyengar and Kinder (1987) pointed out that this could occur in a situation where supporters of the president might interpret unfavorable economic news as not caused by the president’s policies. At the same time, the president’s opponents see the bad news as a good opportunity to hold him accountable for what they already see are his failed policies (p. 72). It is important to note that this occurs not because the economic news dominates their overall impression but because they interpret the economic news according to their view of the president. Controlling this requires an “over time” design and an analysis strategy in which reciprocal causal effects can be eliminated. Iyengar and Kinder examined the impact of problem performance ratings on the overall performance ratings after adjusting for the impact of overall performance ratings on the problem performance ratings. They conducted their experiment in two waves 6 weeks apart and asked for evaluations of president Reagan at both time points: before and after television news containing priming stimuli. Their results indicate that after projection was eliminated, priming effects were clarified. In fact, they concluded that “if anything, our earlier estimates of priming actually understated the impact of television news” (p. 150). The various priming hypotheses such as gradient, target gradient, resistance, and projection will likely be the focus of considerable additional work during the years ahead.

Another question about priming has been raised by Miller and Krosnick (1996). They noted that an important issue to consider in future research relates to the role of accessibility, which is normally considered the key mechanism in priming research (e.g., Wyer & Srull, 1981). Miller and Krosnick (1996) argued that the mediational role of accessibility has never been fully documented, and they provided a few suggestions for attempting to do so. Although they cast doubt on both accessibility and agenda-setting as mediational processes for priming, they did not provide a clear alternative explanation. Additional research is warranted to identify the mechanisms that account for the effects.

Mutz (1998, p. 72) raised an additional concern about accessibility and its possible confound with negativity. She noted that although the priming effect is often explained in terms of simple accessibility and thus is supposed to be, like agenda-setting, a non-directional effect, this nondirectional aspect has not been adequately demonstrated. She correctly noted that most of the stimuli used in experiments have been of bad news, not good news, and that positive news has not resulted in significant effects. This suggests that while there may be nondirectional effects (as priming theorists argue), there may also (or instead) be directional effects of negative information. This also suggests avenues for additional studies. Because much news is indeed negative, this is a significant issue for scholars to consider carefully in the design of future work. More work is certainly indicated in each of these areas, particularly in terms of making sure that priming and negativity effects are not confounded.

Another area of importance involves judgments of information quality as important mediating factors in priming. Miller and Krosnick (2000) examined cognitive mediators of news media priming, namely trust in media. Because this has proven to be a useful concept, as it has to some extent in agenda-setting, further exploration is certainly warranted here. In addition to trust in the media in general, it is natural to explore the various individual media. This might also be done by issue regime or story given that people may very well have reactions to a whole stream of news such as the media’s coverage of the Clinton-Lewinsky scandal or the Gulf War. It is also important to note that trust and credibility do not exhaust the possible responses that media engender in their viewers. Indeed, people bring a wide range of views about various aspects of media performance to their encounters with media, and many of these have been shown to be influential in shaping the effects of media (Kosicki & McLeod, 1989).

Priming effects research has mainly focused on evaluations of the president, although in principle the theoretical apparatus is such that it can readily be extended to evaluations of other political figures, office holders, candidates for office, and issues. Accessibility effects of the type discussed in priming can certainly be studied in many different venues. It is likely that in future years, priming research will spread out quite far from its current base in evaluations of presidents. As research spreads to other contexts such as other political leaders (e.g., governors, senators), one can expect increased importance of media treatment of the issues and thus greater attention to the total context of the studies. Once such studies appear in large numbers, it will be possible to employ meta-analysis techniques to provide empirical summaries that will highlight the importance of such contexts in public opinion. As priming studies spread to public evaluations of issues such as abortion and trade policy, still other contingencies, such as the way the issues are described by media, are likely to become very important.

Conclusion

Media priming studies have been carried on in many contexts, using many types of research designs and measures of priming effects as well as a wide range of participants. We have also seen some studies incorporate both experimental and survey evidence, whereas others rely on surveys alone. This is important not so much for theoretical reasons as for matters of external validity, although the survey studies have often dealt with an interesting mix of real-world issues and media content and with unique periods in our political history such as the Iran-Contra affair, the Gulf War, and the 1992 presidential election. There would seem to be considerable scalability in priming studies, however, to the extent that despite certain ambiguities in the results, the perspective has, for the most part, been a useful addition to the wide-ranging literature on persuasion. The future is very promising in that political figures and issues besides the president can be examined in detail. In other words, priming seems to represent a good, dependable, theoretically motivated type of media effect that can be detected in many types of political and social contexts. The future research agendas of scholars will determine whether this assessment is correct.

In addition, other recent work, such as that of Price and Tewksbury (1997), Price et al. (1997), Cappella and Jamieson (1997), Nelson and Kinder (1996), Nelson, Clawson, and Oxley (1997), and Nelson, Oxley, and Clawson (1997), suggests that exciting new work will emerge, perhaps blending priming with more subtle cues propagated by the emerging framing research. This emerging literature will likely go well beyond simple dichotomies of episodic versus thematic in describing and studying frames and framing effects. Furthermore, the empirical literature will be developed using both laboratory and real-world contexts. The future of this area of research seems bright.