What the evidence shows

Fear appeals — messages that emphasise the danger that will befall people if they fail to act — are among the most contested tools in persuasion and public-health communication. Advocates argue they motivate protective behaviour; critics argue they produce defensive avoidance or backfire. Tannenbaum and colleagues built the largest meta-analytic database on the question so far: 127 articles yielding 248 independent samples (N = 27,372), and tested 16 theory-derived hypotheses, nine of them for the first time meta-analytically. They organised the analysis around a Message–Behaviour–Audience framework: what the message says, what behaviour it recommends, and who receives it.

Most of the evidence concerns individual health and safety behaviours (about 200 of the 248 samples). Only 13 samples concerned environment or society.

Key Findings

• Fear appeals work on average. Across the 248 samples, the average effect on attitudes, intentions and behaviours combined was d = 0.29 (95% CI 0.22 to 0.35). Results varied a lot from study to study (I² ≈ 85%).

• Efficacy statements make them work better, but are not required for them to work. Fear appeals that included efficacy statements had d = 0.43 (92 samples), against d = 0.21 without them (154 samples). The authors also tested whether fear appeals without efficacy statements backfire, and found no support for that. An efficacy statement tells the audience that they can perform the recommended action (self-efficacy) and/or that the action will work (response-efficacy).

• No tested condition produced a backfire effect. Across all the moderators examined, none showed a level at which fear appeals produced worse outcomes than the comparison.

• More fear is not better beyond a point. In 21 samples comparing moderate with high depicted fear, outcomes did not differ significantly. The authors caution that this rests on few samples.

• Personal risk plus serious consequences. Messages that stressed both susceptibility (personal risk) and severity were the only combination with significant positive effects on attitudes, intentions and behaviours separately (combined d = 0.39, 78 samples). Either one alone missed significance on at least one outcome.

• One action beats an ongoing one. Fear appeals recommending a single action (getting vaccinated, signing up for a test) had d = 0.43 (82 samples), against d = 0.21 for repeated or ongoing behaviours (166 samples). They were effective for both types.

Stand out data points

• d = 0.29: average effect of fear appeals across all samples.

• d = 0.43 vs. d = 0.21 (efficacy): with vs. without efficacy statements.

• d = 0.43 vs. d = 0.21 (behaviour type): one-time vs. repeated behaviours. These are the same numbers as the efficacy comparison, by coincidence: two separate comparisons in the paper's Table 4.

Examples of how Understories uses this research

• Actionability. Asking for one clear, discrete action was associated with larger effects than asking for ongoing behaviour change. This is a correlational moderator finding, not an experiment.

• Hope Doom Balance. Fear appeals work, and add more when a credible way to act comes with them (d = 0.43 with efficacy statements vs. 0.21 without). This supports judging a message on whether a threat comes with a specific, believable action, not on tone alone. Without efficacy statements, fear appeals still had a positive effect, so the finding is "stronger with a path forward", not "backfires without one". "Efficacy" here means statements that people can act and that acting works; it is not the same as a hopeful tone. Beyond moderate levels of alarm, showing more added no measurable benefit.

• Personal Relevance. Personalising the risk ("this could affect you") together with serious consequences was the combination that worked across attitudes, intentions and behaviours.

• Justice. Only 29 samples came from collectivist cultures (219 individualist), and the paper found no significant difference in effectiveness between them. The paper's own examples of collectivist samples are East Asian. Participants were mostly university students (145 of 248 samples) with an average age of about 23, and on average 71% White. Understories should flag when applying this research to communities that are not represented in the sample.

  — non-exhaustive.

Key quote

"Fear appeals are particularly effective when the communication depicts relatively high amounts of fear, includes an efficacy message, stresses severity and susceptibility, recommends one-time only behaviors, and targets audiences that include a larger percentage of female message recipients."

© American Psychological Association, 2015 — All rights reserved.

Last update: 20 September 2026

—-

Want to stay ahead of the curve? Join the Reading List.

Fortnightly, I’ll share short explainers, key findings, and data points from selected reports as they go live, plus exclusive content.