![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
The effects of agenda-setting and priming are well established in regard to the news media. Considerably less attention has been paid to these phenomena in entertainment media, in spite of the fact that entertainment media enjoy larger audiences than do news media and often address political topics. This article argues that the psychological mechanism hypothesized to lead to agenda-setting and priming effectsthat is, changes in construct accessibilityapplies as equally to entertainment media as it does to news media. Moreover, we contend that the frequency, consistency, and duration of entertainment media treatments of political issues encourage chronic accessibility of those issues. We test these hypotheses looking at television crime dramas as a source of political information. Using data from two controlled laboratory experiments and the 1995 National Election Study Pilot Study, we demonstrate that viewing crime dramas significantly increases concerns about crime and that these concerns significantly affect viewers' opinions of the president. The NES Pilot Study data suggest that these effects are restricted to frequent viewers of crime dramas, supporting a chronic accessibility model of agenda-setting and priming. These findings extend our growing understanding of how non-news sources of political information contribute to the construction of political attitudes.
A previous version of this article was presented at the annual meeting of the Midwest Political Science Association, April 2003, Chicago, Illinois. We thank Kathleen McGraw, Paul A. Beck, Justin Taylor, Elizabeth Theiss Smith, participants in the Ohio State University's Research in American Politics series, and the anonymous reviewers for their especially helpful comments. Research for this project was funded in part by a grant from the Ohio State University's Program for the Enhancement of Graduate Studies. All data are available by request.
R. Andrew Holbrook is a PhD candidate in the Department of Political Science at Ohio State University. Timothy G. Hill is Assistant Professor in the Department of Political Science at Doane College.
Notes
1. These dramas were ABC's NYPD Blue, CBS's Without a Trace and Robbery Homicide Division, and NBC's Third Watch.
2. Bernard Cohen (1973) deserves credit for this memorable, if grammatically incorrect, phrase. It bears noting that prior to the work of Iyengar and Kinder, Cohen and others (e.g., Lippmann, 1922; McCombs & Shaw, 1972) had done much theoretical and empirical work to flesh out this so-called agenda-setting effect. Previous authors, however, could not make the strong causal claims Iyengar and Kinder made with the use of controlled experiments.
3. More precisely, these authors contend that agenda-setting is a specific example of the more general priming effect. This conceptualization of agenda-setting is consistent with psychological explanations of these media effects, and is one to which we adhere. Our differentiation between agenda-setting and priming is merely a distinction involving the outcome variable of interest, not the psychological processes argued to be at work. We adopt this distinction for reasons of clarity and to remain consistent with much of the recent research in this area (e.g., Gross & Aday, 2003).
4. Two episodes were selected to better rule out the possibility that effects are driven by a single episode of any one program. Only episodes of The West Wing lacking portrayals of violent crime were used, while both episodes of Third Watch provided a number of examples of such activities. This distinction is argued to be a crucial one between crime and noncrime dramas (Raney & Bryant, 2002). Still, some crime-related content did appear in the West Wing episodes. In one episode, the story revolved around convincing five Congressmen to vote for a gun control bill. In the other episode, members of the Secret Service step in to protect the president's daughter from three unruly patrons at a bar. Despite this, only 5.9% and 16.1% of viewers of those episodes, respectively, named crime as the most important problem, compared to 23% and 33.3% of viewers of the two crime drama episodes.
5. This design choice was necessitated by the fact that participants were available for only one session. In such “one-shot” designs, a pretest questionnaire carries with it the possibility of priming considerations relevant to the agenda-setting and priming hypotheses. Previous experimental demonstrations of agenda-setting and priming (e.g., Iyengar et al., 1982) have avoided this possibility by allowing some time to elapse between the pretest and the experiment, usually by bringing participants back for more than one session. Lacking this opportunity, we rely on a binary, posttest-only design to help guard against possible confounding effects associated with a pretest.
6. Intercoder reliability was high (Pearson's r = .81, .90, and .90) for the three response opportunities. The few disagreements which affected the results of this study were resolved by the authors without reference to the experimental condition of the subjects in question. The graduate student coders were not used to examine data from the first experiment. The authors coded those open-ended responses, blind as to the condition to which participants were assigned.
7. Disaggregating the analysis by program did little to alter these findings; 39.4% of Robbery Homicide Division and 36.4% of Without a Trace viewers mentioned crime as the most important problem, while only 14.3% of Everwood and 11.6% of American Dreams viewers did the same.
8. MacKinnon-White heteroscedasticity consistent standard errors were used as opposed to Huber-White standard errors because the latter are inconsistent when sample sizes drop below 250 (MacKinnon & White, 1985; Long & Ervin, 2000). The OLS models estimated here relied on sample sizes of 154 and 133, respectively. Using the MacKinnon-White standard errors minimizes the chances that incorrect inferences will be drawn.
9. An indicator of newspaper readership was excluded from the analysis because it proved to be highly collinear with television news consumption and thus compromised model performance.
10. Unfortunately, only 213 of the 486 pilot study participants reported valid responses for both of these variables. Coupled with missing data for other variables included in the priming model, using this variable decreases the sample size to 133.