Notes on Perpetrator Identity and Media Framing of Attacks

Perpetrator Identity Influences on Media Framing of Terrorist Attacks

• Research Question: How does perpetrator identity influence media framing of attacks as "terrorism" or "mental illness"?

• Gap Addressed: Systematic analysis of potential biases in how media frame terrorism, specifically concerning perpetrator identity (Muslim vs. non-Muslim, White vs. non-White).

• Methodology: Examined text of print news coverage of all terrorist attacks in the United States between 2006 and 2015, as coded by the Global Terrorism Database (GTD).

Key Findings

• Terrorism Framing:

  • The odds that an article references terrorism are approximately five times greater for a Muslim versus a non-Muslim perpetrator, even when controlling for fatalities, group affiliation, and existing mental illness.

• Mental Illness Framing:

  • The odds that an article references mental illness do not significantly differ between White and non-White perpetrators.

Implications

• Differences in media framing can influence public (mis)perceptions of violence and threats.

• This can ultimately harm counterterrorism policy.

Background and Speculation

• Public speculation suggests media frame attacks by non-White (often Muslim) perpetrators as terrorism, while White perpetrators are portrayed as mentally ill.

• Individual cases provide anecdotal support, like Dylann Roof's 2015 attack (Charleston) and the 2017 Las Vegas shooting.

Prior Research

• Systematic work shows Muslim-perpetrated attacks receive more media attention (Kearns et al., 2019b; Mitnik et al., 2020).

• Qualitative differences research (Powell, 2011; Mitnik et al., 2020; Arva et al., 2017) suggests bias in media coverage, but may be a function of the cases selected for comparison.

Definition of Terrorism (Global Terrorism Database - GTD)

• "The threatened or actual use of illegal force and violence by a non-state actor to attain a political, economic, religious, or social goal through fear, coercion, or intimidation."

• The GTD is the most comprehensive publicly available terrorism dataset.

Specific Inquiries

• Compared references to terrorism in coverage of attacks with Muslim versus non-Muslim perpetrators.

• Compared references to mental illness in coverage of attacks with White versus non-White perpetrators.

Media Framing

• Media establish what the public needs to know and highlight the most relevant and salient dimensions of how information should be understood and discussed (Gerbner, 1998; McCombs, 2014; Scheuefele & Tewksbury, 2007).

• Framing is the process of defining an object by heightening the salience of selected aspects of it.

  • Influences how people perceive and structure the object and its place in the world (McCombs, 2014; Powell, 2011).

  • Framed aspects become more salient, unframed aspects become less salient (Entman, 1993; McCombs, 2014).

• Media framing is powerful in influencing public discussions of issues, events, and people (Bekkers et al., 2011).

• Frames help construct mental models of the world and encourage understanding in certain ways (Price et al., 1997).

Types of Frames

• Emphasis Frames: Draw attention to information meant to be viewed as deserving special consideration (Druckman, 2001).

  • Example: Referring to an attack as a "terror attack" vs. "shooting" vs. "crazed rampage." alter audience interpretations.

• Associative Frames: Highlight the connections between things, encouraging the audience to judge an issue by those connections (Van Atteveldt et al., 2005).

  • Example: News saying a terrorism task force is investigating a perpetrator.

• Media select topics to frame with a bias towards attention-grabbing events (Price & Tewksbury, 1997).

• Audiences are likely to activate a frame when presented with similar objects in the future (Price & Tewksbury, 1997).

• Exposure to frames encourages searching for more information that fits within those frames and makes people less likely to accept competing frames (Druckman et al., 2012).

Impact of Media Framing

• Differential framing based on identity could lead audiences to inaccurately associate terrorism or mental illness with particular demographics.

• Global events are translated into local cultures through media framing, colored by the dominant values and power structures of the cultural context (Altheide, 1987; McCombs, 2014; Nagar, 2010).

• News framing is an accumulative process, reinforcing ideologies (Bryant & Mirion, 2004).

• Repeated frames have stronger effects than individual exposure (Lecheler et al., 2015).

• Dominant ideologies and perspectives come to be viewed as "common sense" (Gitlin, 1978).

Media Depictions of Terrorism and Its Perpetrators

• Media reporting tends to represent minority members more negatively than White individuals (Dixon, 2017; Dixon & Linz, 2000).

• Muslim-perpetrated attacks receive disproportionately more coverage (Dixon & Williams, 2015; Kearns et al., 2019b; Mitnik et al., 2020).

• Studies of terror incidents in the U.S. found Muslim-perpetrated attacks were more likely to be framed as terrorism (Arva et al., 2017; Mitnik et al., 2020; Powell, 2011).

Framing of Muslims in Media

• Muslims are described in more negative tones and linked to extremism (Alsultany, 2012; Dixon & Williams, 2015; Prince, 2009; Shaheen, 2014).

• Violence perpetrated by Muslims is disproportionately framed as terrorism (Nagar, 2010).

• Both entertainment and news media have a history of framing Muslims as terrorists (Alsultany, 2012; Shaheen, 2014).

• Drawing attention to group differences and linking threatening characteristics to an out-group bolsters in-group identity (Greenaway & Cruwys, 2019).

• People may come to primarily associate terrorism with Islam.

• Media have the potential to both shape and reflect public sentiment about Muslims, as well as public perceptions of terrorism (Slater’s reinforcing spirals model, 2007, 2015).

Public Perceptions of Violence

• Experimental research shows people are more likely to classify a hypothetical attack as terrorism if the perpetrator was Muslim (Huff and Kertzer, 2018).

• People are more likely to call an attack terrorism if the perpetrator was Arab-American (D’Orazio & Salehyan, 2018).

Hypotheses

• H1a: Media coverage of a terrorist attack will be more likely to reference terrorism when the perpetrator is Muslim versus non-Muslim.

• H1b: Media coverage of a terrorist attack will be more likely to reference mental illness when the perpetrator is White.

Mental Illness and Terrorism

• Mental health issues are not a driver of violence (e.g., Appelbaum & Swanson, 2010).

• Stress and trauma accrued through terrorist activity may contribute to developing mental illness (Weatherston & Moran, 2003).

• Lone actor terrorists have higher rates of mental illness while group actors have lower rates (Gruenewald et al., 2013).

Partisan Reasoning and Attribution

• People are more likely to rate violent in-group actors as being mentally ill and violent out-group actors as motivated by identity (Noor et al., 2018).

• Americans are far less likely to refer to White supremacist terrorists as terrorists or assume an ideological motivation, and tend to assume that they are mentally ill (D’Orazio & Salehyan, 2018).

• People are less likely to call an attack terrorism if the perpetrator was mentally ill (Huff & Kertzer, 2018).

Additional Hypotheses

• H2a: Media coverage of a terrorist attack will be more likely to reference terrorism when it is associated with a known terrorist group.

• H2b: Media coverage of a terrorist attack will be less likely to reference mental illness when it is associated with a known terrorist group.

• H3a: Media coverage of a terrorist attack will be less likely to reference terrorism when the perpetrator has a mental illness.

• H3b: Media coverage of a terrorist attack will be more likely to reference mental illness when the perpetrator has a mental illness.

• H4: Media coverage of a terrorist attack will be more likely to reference terrorism as the number of fatalities increases.

Alternate Explanations for Referencing Terrorism

1. People are more likely to view a bombing as terrorism (Huff & Kertzer, 2018).

2. People are more likely to call an attack terrorism when there are multiple perpetrators (Huff & Kertzer, 2018).

3. Media coverage of crime is framed less negatively when the victims are minorities (Gilliam & Iyengar, 2000).

4. Attacks against the government receive more media attention (Kearns et al., 2019b; Zhang et al., 2013), there is conflicting evidence on whether targeting the state influences how the public labels an attack (Huff & Kertzer, 2018).

5. Some violent incidents clearly meet the GTD definition of terrorism while others do not meet the threshold for a definitive classification.

Alternate Explanations for Referencing Mental Illness

1. The number of people killed in an attack may have a positive relationship with references to mental illness.

2. The existence of multiple perpetrators may indicate more coordination, thus making the presence of mental illness seem less likely.

3. Individual perpetrator(s) of terrorism to be unidentified (Kearns, 2019). When the individual perpetrator(s) are unknown, media are less able to defensibly speculate about potential mental illness.

4. Violent incidents that do not clearly meet the GTD definition of terrorism may be more open to speculation about the causes, which creates more room to discuss mental illness as a potentially relevant factor.

Methods

• Data Source: Global Terrorism Database (GTD) for terrorist attacks in the United States from 2006 to 2015.

  • 170 terrorist attacks, collapsed into 136 unique attacks.

• Media Coverage: LexisNexis Academic and CNN.com.

• Procedure:

  • Limited coverage to print and online newspaper articles from US-based sources between the date of each attack and the end of 2016.

  • Searched for perpetrator(s), location, and key words about the attack.

  • Two authors separately reviewed each article.

  • Dataset comprised of 3,541 news articles, reduced to 100 attacks after excluding those with no coverage.

• Variables:

  • Dependent: (1) Whether an article references terrorism, (2) Whether an article references mental illness.

  • Predictor: Perpetrator(s) and number of casualties.
    * Membership in a known group (binary).
    * Fatalities (total number of people killed).
    * Perpetrator White (binary).
    * Perpetrator Muslim (binary).
    * Perpetrator mentally ill (binary).

  • Alternative Explanations: Bomb, multiple perpetrators, unknown perpetrator, target minority, target law enforcement/government, does not meet all criteria for terrorism.

Results

• Logistic regression models with standard errors clustered on the attack.

• Odds ratios > 1 indicate a positive relationship, < 1 a negative relationship.

Terrorism References

• Supporting H1a, articles are significantly more likely to reference terrorism when the perpetrator is Muslim.

  • The odds an article references terrorism are approximately five times greater if the perpetrator is Muslim (Model 2).

• Supporting H2a, a two and a quarter times increase in odds that an article references terrorism when the attack is perpetrated by a group.

• When we coded diagnosed mental illness strictly (Model 2), we did not see support for H3a. However, when we coded mental illness more loosely (Model 3), the odds that an article references terrorism decrease by approximately half.

  • H3a: Media coverage of a terrorist attack will be less likely to reference terrorism when the perpetrator has a mental illness.

• Contrary to H4, fatalities were not positively related to the odds of terrorism being referenced. We suggested five alternative explanations for when media refer to an attack as terrorism.

• None of these factors impacted the likelihood that an article references terrorism.

Mental Illness References

• Contrary to expectation in H1b, articles about attacks perpetrated by White people were not more likely to reference mental illness across all of our models.

• Supporting H2b, the odds that an article references mental illness decreased between 69% and 82% if the perpetrator(s) had a group affiliation (Models 7–14).

• Supporting H3b, the odds that an article references mental illness are between approximately one to two and a half times greater when the perpetrator actually had a diagnosed mental health issue at the time of reporting (Models 7–14).

• Fatalities, multiple perpetrators, and unknown perpetrators have impact on media referenced mental illness.

  • When mental illness was coded strictly, the odds that an article references mental illness increased 7% per fatality.

  • Across models, the odds that an article referenced mental illness decreased 62%–74% if the attack had multiple perpetrators.

  • When the perpetrator(s) were unknown, the odds that an article references mental illness decrease 96%–98%.

Discussion

• This study systematically examined when terrorism is referenced and when mental illness is referenced in media coverage of terror attacks, and whether there is variation on the basis of perpetrator identity.

Key Findings Revisited

• Less than 40% of articles referenced terrorism, despite all attacks being included in the GTD.

• The percentage of stories that referenced mental illness is roughly proportional to the percentage of perpetrators with a known history of mental illness.

• Coverage of attacks perpetrated by Muslims were dramatically more likely to reference terrorism.

• Articles covering attacks by individuals affiliated with or inspired by a known group that uses terrorism were also more likely to reference terrorism.

• We found partial support for the expectation that media would be less likely to reference terrorism if the perpetrator had a known mental health issue.

Mental Illness References Revisited

• We found no evidence that media are more likely to reference mental illness when the perpetrator is White. This finding runs counter to public speculation and experimental findings (Noor et al., 2018), but was robust across all of our models.

• Media coverage was more likely to reference mental illness if the perpetrator had a known or suspected mental health issue.

• Attacks with multiple perpetrators or in association with a known group signal greater planning and preparation, which can explain why coverage of these attacks was less likely to reference mental illness.

• Attacks with unknown perpetrators were far less likely to reference mental illness, probably due to a lack of information.

Conclusion

• Results show that incident- and perpetrator-level factors of a terrorist attack influence whether media coverage references terrorism or mental illness.

• Media framing and coverage influences how the public understands violence and its perpetrators, as well as which policies the public will support to address the problem.

• Our results clearly show that incident- and perpetrator-level factors of a terrorist attack influence whether media coverage references terrorism or mental illness. Media cov- erage of terrorism influences how the public understands violence and its perpetrators, as well as which policies the public will support to address the problem.

Limitations

• Limited to textual analysis of U.S.-based print media coverage of U.S. terrorist attacks in a 10-year period.

• It is unclear if findings apply across other contexts or whether the inclusion of image and photo analysis might yield different or more nuanced results.

• Most work on media coverage of terrorism focuses on the U.S. so results may not translate cross-nationally.

• Lack of a public archive for websites like FoxNews.com and HuffingtonPost.com precluded from examining differences in how media frame violence more comprehensively across the political spectrum.

Policy Implications

• Attacks classified as terrorism by experts should be discussed as terrorism by media, but this is often not the case.

• Misperceptions promote prejudice and reduce the likelihood that security threats will be appropriately addressed.