Nuclear weapons risk communication: evaluating the impact of message exposure and format

Abstract

Many experts believe the risk of nuclear war today is the highest it’s been in generations. Yet much of the public has little awareness of nuclear threats and is ill-prepared to take the actions needed to save lives, including their own. Using a survey experiment on a diverse national sample of U.S. citizens fielded across two time periods, this study evaluates the impact of risk communication regarding nuclear weapons threat. We measure the effectiveness of nuclear preparedness messages across formats, finding infographics to be the most effective. Importantly, we also find that any message exposure improves recall of the recommended action, confidence, and message acceptance, with these shifts positively impacting subsequent behavioral intentions.

Keywords:

• Nuclear weapons
• emergency preparedness
• message format
• infographic
• risk communication

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The Department of Homeland Security's (DHS) all-hazards 15 National Planning Scenarios highlight potential scenarios See Bell & Dallas (2007) for a longer discussion of government models of mass casualty estimations by geographic region and results of research into urban health care systems in four American cities.

2 Parikh and colleagues (2016) use “do nothing” as a baseline behavior scenario and add human behavior complexity over five comparative models that include six behaviors: sheltering, evacuation, healthcare-seeking, worry behaviors, household reconstitution, and providing assistance. They find that sheltering and evacuation lessen casualties over doing nothing and outcomes improve as the model includes healthcare-seeking and worry behaviors but gets worse as individuals provide assistance or seek to reunite with family.

3 Nuclear war risk assessments are underway (Baum 2018) but disagreement among experts remains (Saradzyhan, 2019) and nuclear war survivability differs dramatically from single detonation incidents.

4 Across both samples, participants were removed for reporting technical difficulty, for being less than 18 years old, for not being U.S. citizens, and for completing the survey in less than 3 minutes. Given the length of the survey, we chose to remove participants that were clearly “satisficing” or exerting minimal effort; the 3-minute cutpoint marked the bottom 2% of the distribution.

5 The only exception to the between-subjects analyses is when we evaluate lagged effects using only recontacted participants from Time 2 (i.e. within-subjects), as indicated in text and in the respective results table.

6 Members of the control group were still told they would be asked about “Nuclear Weapons Threat” and answered all of the dependent measures except those regarding message recall, confidence, and acceptance for the media example.

7 Images of all four treatments are available in the appendix. Links to the full video files (tv news report and PSA) are also provided.

8 We thank Dr. Robert Levin, public health officer in Ventura County, California for speaking with us prior to this research and encouraging study of the publicly-available materials.

9 The university’s institutional review board (IRB) approved the current studies prior to any data collection.

10 We rely mostly on multivariate regression analyses to examine the impact of key variables and demographic characteristics on our dependent variables. When the dependent measure is dichotomous, we rely on logistic regression analyses. All of the analyses use standardized coefficients, thus allowing for direct comparisons. Where appropriate, the sample time period being analyzed is clearly indicated in the text and in the title of the table.

11 We thank Daniel Basta, an undergraduate research assistant, for assistance coding this data. Responses were categorized as mutually exclusive across four categories: correct, incorrect, dated, and unrelated/nonsense.

12 The two actions that have the most potential to lead to preventable casualties are “evacuation” and attempts to “reunite”, as they both involve potentially exposing individuals to dangerous radioactive fallout. The intent to take either of these actions significantly decreases (compared to the control group) with exposure to any of the treatment conditions.

13 The appendix includes an additional plot that examines the degree to which Time 1 message acceptance accrues at Time 2. In short, the rate of increase for the intent to get inside according to message acceptance is significantly greater (i.e. steeper) among recontacted participants than new participants.

14 U.S. Census Bureau 2010–2016.

15 Pew Research Center, 2016.