The theory of agenda setting is still at a formative level. In spite of the problems in method and time span mentioned earlier, the findings in agenda setting are consistent enough to permit some first steps toward theory building. To begin, longitudinal studies of agenda setting have permitted some tentative causal statements. Most of this research has supported the interpretation that the media’s agenda causes the public agenda; the rival causal hypothesis—that the public agenda establishes the media agenda—has not received much support (Behr & Iyengar, 1985; Roberts & Bachen, 1981). Thus, much of the recent research has attempted to specify the audience-related and media-related events that condition the agenda-setting effect.
It is apparent that constructing an agenda-setting theory will be a complicated task. Williams (1986), for example, posited eight antecedent variables that should have an impact on audience agendas during a political campaign. Four of these variables (voter interest, voter activity, political involvement, and civic activity) have been linked to agenda setting (Williams & Semlak, 1978). In addition, several studies have suggested that a person’s “need for orientation” should be a predictor of agenda holding. (Note that such an approach incorporates uses and gratifications thinking.) For example, Weaver (1977) found a positive correlation between the need for orientation and a greater acceptance of media agendas.
These antecedent variables define the media-scanning behavior of the individual (McCombs, 1981). Important variables at this stage of the process are the use of media and the use of interpersonal communication (Winter, 1981). Other influences on the individual’s agenda-setting behavior are the duration and obtrusiveness of the issues themselves and the specifics of media coverage (Winter, 1981). Three other audience attributes that are influential are the credibility given to the news media, the degree to which the audience member relies on the media for information, and the level of exposure to the media (Wanta & Hu, 1994b).
Despite the tentative nature of the theory, many researchers continue to develop models of the agenda-setting process. Manheim (1987), for example, developed a model of agenda setting that distinguished between content and salience of issues. Brosius and Kepplinger (1990) used time series analysis in their study of German news programs to test both a linear model and a nonlinear model of agenda setting. The linear model assumes a direct correlation between coverage and issue importance; an increase or decrease in coverage results in a corresponding change in issue salience. Four nonlinear models were also examined: (1) the threshold model—some minimum level of coverage is required before the agenda-setting effect is seen; (2) the acceleration model—issue salience increases or decreases to a greater degree than coverage; (3) the inertia model—issue importance increases or decreases to a lesser degree than coverage; and (4) the echo model—extremely heavy media coverage prompts the agenda-setting effect long after coverage recedes. Their data showed that the nature of the issue under study was related to the model that best described the results. The acceleration model worked better for issues that were considered subjectively important by the audience (taxes) and for new issues. The linear model seemed to work better with enduring issues (the environment). Some support was also found for the threshold model. There was, however, little support for the inertia model, and not enough data were available for a convincing test of the echo model. In sum, these data suggest an agenda-setting process more complicated than that envisioned by the simple linear model.
Recent developments have been focused on integrating agenda setting with other theories from communication and psychology. Jeffres, Neuendorf, Bracken and Atkin (2008), for example, attempt to use the third-person effect to link agenda setting and cultivation. Jorg (2008) conducted a panel study to show that a person’s need for orientation was a predictor of the agenda-setting effect and Liu (2008) demonstrated the usefulness of the elaboration likelihood model in explaining agenda setting.