Wildfire imagery reduces risk information-seeking among homeowners as property wildfire risk increases

Climate change is increasing the frequency and intensity of natural hazards, producing powerful negative imagery that practitioners may use in risk communications. Theory and mixed empirical evidence suggest negative imagery and fear appeals can either motivate protective action by increasing negative affect, perceived severity, and attention, or backfire by inducing avoidance, reactance, hopelessness, or compassion fatigue—especially when efficacy is low. Practitioners lack consensus on whether worst-case imagery (e.g., a burning house) engages or repels homeowners. The study investigates whether negative wildfire imagery affects homeowners’ intentions and actual information-seeking, and whether effects vary by property-level wildfire risk. The core hypothesis is that negative imagery will not uniformly increase engagement and may reduce information-seeking among higher-risk homeowners absent strong efficacy cues.

The paper synthesizes literature on affect, framing, and fear appeals. Negative emotions can motivate behavior by increasing attention and reducing psychological distance, and loss framing often increases preparedness attitudes. Availability and affect heuristics can raise perceived risk when consequences are made salient. However, Extended Parallel Process Model and Protection Motivation Theory predict that fear appeals succeed when perceived efficacy is high; otherwise they can trigger maladaptive responses (avoidance, denial, reactance). Climate communication studies note potential backfire from dire messages, compassion fatigue, and credibility loss with strong terminology. Prior work also documents gaps between expert and public risk perceptions and mixed causal evidence linking fear appeals to climate-related actions. This study positions wildfire imagery in this theoretical context and addresses the evidence-advice gap with domain-specific experiments.

The research comprises three studies using two focal images: a negative imagery "flames" photo of a burning house (from National Geographic) and a "status quo" aerial landscape photo commonly used in wildfire communications. Study 1: Practitioner survey. Convenience sample of wildfire practitioners recruited via emails and the Fire Adapted Communities Learning Network message board. Online survey asked which image would be more effective at engaging homeowners and which they would prefer to use; also elicited perceptions of field-wide agreement and open-text rationales (coded for fear-related terms). Proportions compared using Pearson’s chi-squared tests. Targeted at least 100 responses to detect a 10 percentage point difference. Study 2: Online experiment. N=440 U.S. homeowners from Prolific residing in 15 wildfire-prone states. Participants imagined owning a high-risk home and receiving a postcard from the local fire department. Random assignment to one of four photos: two flames images (burning houses, used in media) and two status quo images (used by fire departments). Primary comparisons focus on the two images also used in the field study. Measures: emotional reactions (Likert 0–10), risk assessments (probability and riskiness), behavioral intentions to visit a hypothetical property-specific risk website (5-point Likert), and personal preferences (liking, relevance). Analyses: t-tests with Benjamini–Hochberg correction for multiple comparisons; ANOVA with Tukey HSD for sensitivity/generalizability across similar photos. Power analysis targeted d=0.35 with α=0.05, β=0.80. IRB approved (Protocol 20-0372). Study 3: Field experiment. Setting: Ashland, Oregon (designated Wildfire Hazard Zone). Sample frame: owners of all residential parcels within city limits (N≈6400); after exclusions and de-duplication, 5,785 unique postcard recipients. July 2020 mailing from the local fire department’s wildfire division provided each parcel’s professionally assessed wildfire risk and directed recipients to a personalized webpage; unique access codes enabled household-level tracking. Randomized block design: homeowners evenly assigned to flames vs status quo postcard images within expert risk categories (Low to Extreme). Outcomes: webpage visitation within two months (binary). Analyses: Pearson’s chi-squared test for overall difference; linear probability models with robust SEs, with and without covariates (parcel risk score/100, acreage, property value; ownership characteristics including part-time residency, multiple properties). Moderator analysis: treatment × risk score interaction; robustness checks excluding risk score outliers (IQR criterion; 2.5th/97.5th percentiles). Pre-registered hypotheses and analysis plan (https://osf.io/fa2ms). IRB approved with consent waiver (Protocol #20-0177).

• Practitioner expectations: 83% (100/120) believed flames imagery would be more effective at engaging homeowners than status quo (χ²=135.4, p<0.001). 62% preferred to use the flames photo in communications (χ²=47.5, p<0.001). However, 29% of those selecting flames as most effective would not choose to use it; 29% used fear-related terms in justifications. Perceived field agreement: 48% said most practitioners think flames engage; 43% said there is no agreement.
• Online experiment (homeowners, N=440): Flames photo increased negative emotions relative to status quo—anxious (+2.3 points, 95% CI 1.7–3.0), fearful (+2.0, 1.3–2.6), worried (+2.3, 1.7–2.9)—and reduced positive emotions—calm (−1.8, −2.4 to −1.1), peaceful (−2.1, −2.8 to −1.5), safe (−2.2, −2.8 to −1.6). Slightly higher perceived home risk (+0.8 on 0–10, 95% CI 0.2–1.4), but no differences in assessed wildfire probability or damages and no difference in behavioral intentions to visit a property risk website (mean “somewhat likely”; Mdiff≈0.0). Flames image was less liked (−3.0, −3.7 to −2.3) and less personally relevant (−1.7, −2.5 to −0.9). Patterns were consistent with alternative flames/status quo photos.
• Field experiment (N=5,785): Overall visitation rate 19.3%. No significant overall difference between treatments: Flames 18.8% vs Status Quo 19.7% (χ²(1)=0.7, p=0.4). Regression-adjusted average treatment effect similar (β=−0.7 percentage points; p=0.5; 95% CI [−2.7, 1.3]).
• Moderation by risk: Higher parcel wildfire risk predicted more visits overall (e.g., Risk Score/100 β≈0.048, p<0.01). However, flames imagery reduced visits as risk increased: treatment × risk interaction −0.021 to −0.023 per 100 risk points (p<0.01, Models 2–4). At risk score=0, flames increased visits by about 9.1 percentage points (p<0.01; 95% CI [2.7, 15.5]), but for every 100-point increase in risk score, flames reduced visits by ≈2.2 percentage points relative to status quo. For a homeowner one SD above average risk, flames reduced visitation by ≈3.5 percentage points. Results robust to covariate adjustments and outlier exclusions.

Practitioners largely expected worst-case-scenario imagery to increase homeowner engagement, but empirical results did not support this overall. Negative imagery reliably heightened negative affect yet did not increase stated intentions or actual information-seeking on average. Critically, effects varied by property risk: flames imagery increased visits among very low-risk owners but decreased visits as risk rose, indicating a threat-by-efficacy interaction consistent with the Extended Parallel Process Model and Protection Motivation Theory. When high threat is paired with low perceived efficacy or high response costs (common in wildfire mitigation due to structural, environmental, and financial constraints), maladaptive responses such as avoidance, willful ignorance, reactance, or hopelessness may suppress engagement. Conversely, low-risk owners receiving salient threat cues may feel more able to cope, facilitating engagement. These findings highlight the tradeoff between salience and efficacy in fear appeals and underscore the need for actionable, feasible mitigation options when employing negative imagery. They also suggest that discrepancies between professional risk ratings and homeowner perceptions may exacerbate reactance or denial among high-risk owners. Tailoring communication to risk level and bolstering efficacy may prevent backfire among those at greatest risk.

This work demonstrates an evidence–practice gap in wildfire risk communication: although practitioners expect negative imagery to engage homeowners, it does not increase information-seeking on average and can reduce engagement among higher-risk homeowners. The study contributes causal, domain-specific evidence across lab and field settings, showing that fear-based visuals interact with risk to shape behavior. For practice, negative imagery should be paired with clear, feasible, and affordable mitigation actions to enhance efficacy, and messages should be tailored to risk levels to avoid backfire among high-risk households. Future research should test imagery combined with efficacy-enhancing interventions (e.g., subsidies, step-by-step guidance, social support), examine longer-term behavioral outcomes beyond initial information-seeking, explore heterogeneity across communities and demographics, and investigate mechanisms linking perceived risk, efficacy, and reactance.

• External validity: Emotional responses measured online may not mirror field responses; laboratory demand characteristics and social desirability may influence self-reports.
• Context specificity: Field site (Ashland, OR) is older, wealthier, and more educated than U.S. averages; results may differ in other communities and hazard contexts.
• Practitioner sample: Convenience sample from an engaged professional network; may not represent all wildfire practitioners; response rate unknown.
• Risk perception gap: Discrepancies between professional risk scores and homeowner beliefs could induce cognitive dissonance, denial, or reactance, interacting with negative imagery.
• Efficacy cues: The postcard referenced resources but may not have sufficiently addressed response costs or boosted perceived efficacy, particularly for high-risk owners with limited control or resources.
• Outcome scope and timing: Behavioral outcome limited to short-term webpage visitation; mitigation actions and longer-term effects were not measured.
• Image specificity: Results pertain to selected flames and status quo photos; although sensitivity analyses used similar images, other imagery types or message framings may yield different effects.