Framing terror: an experimental framing effects study of the perceived threat of terrorism

Abstract

An exploratory framing effects experiment was conducted to test whether three controversial news portrayals (or frames) of the terrorist threat increase subjects' perceptions of the danger. A total of 176 subjects were exposed to one of eight different article treatments. The subjects reported higher levels of perceived threat when the danger was associated with ‘radical Islamic groups’ (as compared to homegrown terrorists) and ‘nuclear’ technology (as compared to conventional weapons). Contrary to conventional wisdom, however, this study also showed that the term ‘terrorism’ itself did not affect the perceived threat. These results provide support for a theory of framing that explains the precise ways in which a particular set of communicating texts (terrorism news frames) influence human consciousness. Further theoretical details, as well as the social and political implications of this study, are discussed.

Keywords:

• framing
• terrorism
• media
• Islam
• social attitudes
• stereotypes

Acknowledgements

I convey my gratitude to Stan Kaplowitz, Fred Fico and Toby Ten Eyck for their helpful suggestions on this article. I also thank Piers Robinson and the three reviewers for their helpful comments.

Notes

1. See data from the Gallup Organization in a continually updated report (Terrorism in the United States) available online at http://www.gallup.com/; see also Gillespie (2001), Lewis (2005), Davis (2007).

2. Among the many changes in social attitudes and behaviour after 9/11 were increased levels of presidential approval (Gaines 2002), trust in government (Ford et al. 2003), patriotic feelings, rituals and acts (Collins 2004, Moskalenko et al. 2006), helping behaviour (Putnam 2002, Skocpol 2002), social trust (Smith et al. 2001), support for military actions against terrorism (Huddy et al. 2002), willingness to give up some civil liberties for safety (Greenberg et al. 2004, Huddy et al. 2005) and negative attitudes towards Arabs, Muslims and immigration (Moore 2002, Yum and Schenck-Hamlin 2005, Panagopoulos 2006, Pew Research Center 2006, Jones 2007, Newport 2007). For a review of ‘9/11 effects’, see Woods (2011a, 2011b).

3. This is a common conceptual weakness in the literature. Scholars regularly define news frames by explaining their origins or effects, rather than explaining what they consist of or how they can be identified in communication content. To consider a few more examples, Norris et al. (2003, p. 10) suggest that ‘the idea of “news frames” refers to interpretive structures that journalists use to set particular events within their broader context’. In a study by Terkildsen and Schnell (1997, p. 881), news frames are defined as ‘the “maps” or internal story patterns reporters and editors draw for their readers’. For other examples of concept–theory overlaps, see Bronstein (2005, p. 785) and Lepre et al. (2003).

4. While many studies have shown that hazards produce both cognitive and emotional responses, it is not clear which occurs first (Peters and Slovic, 1996, Peters et al. 2004). Previous research suggests that when people find themselves in a risky situation, their emotions may in fact precede their cognitive processing of the danger (Slovic et al., 2002). Any attempt to establish the time ordering of the three factors of perceived threat – dread, risk judgement and worry – goes beyond the scope of this study. The aim here is simply to test whether certain experimentally manipulated terrorism news frames affect any of these dependent variables.

5. For instance, news portrayals of fatal car crashes may produce lower level risk judgement, dread and worry than terrorism coverage, in part because the dangers of driving are perceived as more familiar and conventional, even though they are far more deadly (42,116 people died in motor vehicle crashes in 2001, while less than 3000 were killed in the attacks on 9/11). See US Department of Transportation, Traffic Safety Facts 2001, available online at: http://www-nrd.nhtsa.dot.gov/Pubs/809476.PDF

6. Race research in sociology and social psychology relies on a similar line of theorising (see e.g. Ford 1997).

7. A note should be made about whether it is appropriate, in the context of a framing study, to compare the radical Islamic and homegrown extremist frames. Some scholars may reasonably argue that the framing literature has focused primarily on comparing the effects of slightly different presentations of essentially the same thing (i.e. alternate labels for the same person, event, type of threat or policy). Some may further suggest that the two manipulations in question (homegrown vs. Islamic) are not the two versions of the same thing – that radical Islam is qualitatively different than homegrown extremism – and therefore the study should utilise a different theory such as priming. My contention, in contrast, is that the two types of threat (and the manner in which they were presented to the subjects) are indeed quite similar. As illustrated in the experimental treatments in Appendix 1, all the subjects in the study were asked to think about ‘small groups’ committing violent acts against people living in cities. All the subjects learned that cities are ‘particularly vulnerable’, that the threat is ‘most serious’, that there is a high risk of people dying in an ‘explosion’ and that increased security efforts are needed to make people safe. The two threats, in short, were described precisely in the same way, except for the ethnic identity of the supposed killers. I argue that the two groups’ mutual interest in killing thousands of people make them, in conceptual terms, far more similar than different, even if they come from diverse regions of the world or pray to different gods. Following this line of reasoning, this study fits within the traditional boundaries of the framing literature.

8. See a report by Peter Bergan and Bruce Hoffman, ‘Assessing the Terrorist Threat: A Report of the Bipartisan Policy Center's National Security Preparedness Group’ (10 September 2010), available online at http://bipartisanpolicy.org/sites/default/files/NSPG%20Final%20Threat%20Assessment.pdf

9. Sunstein uses the term ‘probability neglect’ to explain why people often overestimate the chance of low-probability risks of catastrophes, such as terrorism. According to the theory, when a danger evokes high levels of emotion in people, they tend to focus on the ‘badness of the outcome, rather than on the probability that the outcome will occur’ (Sunstein 2003, p. 121), which explains the common misjudgement of some very frightening yet extremely low-risk hazards.

10. Issues involving representativeness inevitably arise when student samples are used in research. In such cases, a strong argument for external validity can never be made. At best, these findings can be generalised to the wider student body at the university where the study was conducted. Students pursuing degrees in social sciences may be more sensitised to topics such as terrorism than the general public. Future research is needed to consider the potentially ‘confounding variables’ of age, race/ethnicity and gender, as well as the differences that would likely be revealed by comparative, cross-national research. At the same time, if not for its enumerative generalisability, this study can be justified by its potential theoretical generalisability.

11. Pre-test bias was not anticipated to be a problem, given the fact that all participants received both the pre-test and post-test questions and were exposed to roughly the same stimulus (aside from the subtle framing manipulations).

12. In the case of risk judgement, two very small, but significant interactions were found (nuclear × radical Islamic and radical Islamic × terrorism). These findings were difficult to interpret and not seen as meaningful. Identical analyses of the other two indicators of perceived threat (dread and worry) did not uncover any interactions. Future research aimed at interpreting interactions would benefit from a larger sample size.

13. These findings also have a methodological implication that is worth noting. When pollsters ask about the threat of terrorism, some respondents may be thinking of radical Islamic groups from foreign lands, while others contemplate the threats posed by homegrown terrorists. If true, the variance between such respondents could be misconstrued as a difference in general levels of perceived threat, when in fact it should be attributed to a systematic tendency to think either about radical Islamic or homegrown terrorism. Measures of perceived threat should offer greater detail on the specific type of terrorism under study.

14. The idea that multi-victim murders are more likely to be committed by people from the Middle East with radical Islamic beliefs than by people with other radical beliefs seems particularly problematic in the light of the many heinous attacks carried out since 9/11 by non-Muslim American citizens. For instance, the Catholic microbiologist Bruce Ivins allegedly committed the 2001 anthrax attacks, while the computer software consultant Joe Stack carried out a suicide plane attack against the IRS in Texas in February 2010.

15. For FBI hate crime statistics, see the online source available at http://www.fbi.gov/ucr/ucr.htm#hate; for a study of hate crime in the wake of 9/11, see Byers and Jones (2007).