Agenda setting, localisation and the third-person effect: an experimental study of when news content will directly influence public demands for policy change

ABSTRACT

Building from the third-person effect model of DRR policy adoption and mediated policy learning, this study provides an experimental examination of how specific elements of news media’s localisation of distant events directly influence public opinion. Controlling for salience effects, the construction of affinities between the distant, stricken community and the newspaper’s audience is argued to create a sense of shared vulnerability to the reported disasters. This is correlated within an increase in the respondent’s intention to act directly and an increase in their willingness to punish elected officials who do not act accordingly. The construction of difference between the communities, even though it is not related to risks related to the disaster, is argued to create implicit reassurances that the observing community does not need to act. This leads to an increased intention to act directly in opposition to efforts to reduce risk, but a neutral response towards political actors who pursue risk reduction policy actions.

KEYWORDS:

• Disaster risk reduction
• agenda setting
• experiment
• news
• necessary conditions model

Acknowledgements

Previous versions of the paper were presented in the Media and Communication Department Seminar Series at the University of Canterbury, and in the Comparative Politics group and the Networked Democracy Lab at the University of Southern California. The authors would like to thank Babak Bahador, Pablo Barberá, Pat James, Gerry Munck, Phil Seib, Nick Weller, participants in the seminars, and the anonymous reviewers for excellent comments and suggestions. Any errors that remain are our own responsibility. We are also grateful to the Center for International Studies and the School of International Relations at USC for funding for the experiment. This study was approved by the USC Institutional Review Board, and received approval number UP-16-00306 on 16 May 2016. The study was pre-registered at the Evidence in Governance and Politics (EGAP) registry on May 25, 2016, and is accessible at http://egap.org/registration/1896.

Notes

1. This is a debatable point. A more complex conceptualisation of framing would include elements, such as the emotive, that would directly impact opinions in ways beyond what logic is used for making sense of incoming information.

2. Unfortunately, a point of disambiguation is required when extending the discussion of this model to the study of public opinion. In the study of public opinion, the term ‘third-person effect’ has been used to refer to the dynamic where even though an individual’s opinion is not changed by an influence such as media coverage, that person believes that the influence will change the opinions of others (Davison 1983; Perloff 1993). It is not clear why a literary term that refers to a non-participating observer was adopted in this way in the study of public opinion, but the third-person effect model discussed here (Van Belle 2015) was initially constructed in the context of the study of media, and closely reflects the original meaning and use of the term. In that model, third person refers specifically to a non-participating observer, and the effect is the influence that the mediated construction of a distant event has upon that observer.

3. It is worth noting that in this way, prospect theory drives an opposite reaction in the observing community than it did in the affected community. In the observing community, framing in terms of threats to possessed goods should bias the response towards reducing the risk to those goods.

4. Soft-storey buildings are structures with a bottom floor that is weaker than the floors above it. Typically, these building are supported by pillars that can collapse under the weight of the building during earthquakes as lateral movement imposes too much stress on them.

5. Like previous studies using this platform, participants in the sample were more educated, more liberal and less religious than Californian residents, more generally. A comparison of the sample with the population of the California is included in Tables A5–A7 in Appendix C.

6. This analysis uses OLS regression for ease of interpretation, but it is important to acknowledge some of the limitations of this method of analysis in regard to this data. Among those, non-stationarity could be a concern where important statistical factors vary over time, threatening the internal validity of the study if there is systematic variation over time. This could especially be the case if the study was a within-subjects design that measured changes in individuals’ attitudes over time, or if the study used repeated measures. However, we do not believe this to be a critical concern in this posttest-only survey experiment. Random assignment of treatment conditions to participants means that any systematic changes over time that could affect the validity of the results are equally distributed between the groups. As a result, any differences that persist between the treatment groups should be attributable to the average treatment effect of the intervention, and not to any other factors that could confound the analysis in the absence of random assignment.

7. In the regression models presented here, the R-squared values are noticeably low. When working with observational data this would be a particularly important concern, given that this would indicate the models do not explain much of the variance. However, the objective of an experiment is not to explain variation in the dependent variables. Instead, OLS regression of experimental data is used to determine how the intervention of the treatment conditions influences the average of the dependent variable, relative to the absence of the intervention. As such, we do not consider the low R-squared values to be a concern in this study.

Additional information

Funding

We are grateful to the Center for International Studies and the School of International Relations at the University of Southern California for funding the experiment.

Notes on contributors

Thomas Jamieson

Thomas Jamieson is a Ph.D. Candidate at the University of Southern California. His research draws from political psychology, political communication, and behavioral economics to better understand how leaders and people in general think, feel, and behave during periods of crisis.

Douglas A. Van Belle

Douglas A. Van Belle is a Senior Lecturer in Media Studies at Victoria University of Wellington. His most recent research focuses on the role of press freedom in democratic politics and studying how variations in the content of news media coverage of disasters might be used to address questions of race, image and governmental practices.