COVID-19 Risk Perception Measure

1. Abstract

The COVID-19 Risk Perception Measure is a multidimensional psychometric instrument formulated by Lucia Savadori, Oksana Tokarchuk, Massimo Pizzato, and Stefania Pighin (2023) to assess the multifaceted nature of disease risk appraisals and their downstream effects on behavioral decision-making, specifically tourist mobility and travel intentions during the COVID-19 pandemic. Grounded in dual-process cognitive theories and health-protective behavioral paradigms, the instrument conceptualizes risk assessment through three core psychological domains: affective risk perception (visceral emotional states, dread, fear, and worry), analytical risk perception (deliberative cognitive calculations of likelihood, statistical probability, and health consequence severity), and experiential risk perception (heuristic, vivid mental simulations, availability of prior infection events, and intuitive vulnerability). The original inventory comprised 13 items adapted from foundational public health and decision-making literature, presented alongside behavioral travel intentions.

Empirical evaluation of the instrument was executed across two empirical experimental investigations involving adult Italian residents. Psychometric testing utilized confirmatory factor analysis (CFA) to scrutinize the structural validity of the proposed dimensional framework. While an initial thirteen-item model demonstrated adequate global fit (χ² = 375, df = 62, χ²/df = 6.05, Comparative Fit Index [CFI] = 0.951, Tucker-Lewis Index [TLI] = 0.938, Root Mean Square Error of Approximation [RMSEA] = 0.090), specific model diagnostic indices exceeded conservative methodological thresholds. Respecification guided by empirical modification indices yielded an optimized 11-item structural model (χ² = 126, df = 41, χ²/df = 3.07, CFI = 0.984, TLI = 0.978, RMSEA = 0.058), demonstrating exemplary structural fidelity. Internal consistency reliability analysis across all retained dimensions revealed robust metrics, yielding Cronbach's alpha coefficients exceeding the standard psychometric threshold of α > .70. Items are administered using a 7-point Likert response scale calibrated from 1 (e.g., Strongly disagree or Extremely low / Not at all) to 7 (e.g., Strongly agree or Extremely high / Extremely). This measure provides public health researchers, behavioral economists, and tourism psychologists with a psychometrically validated assessment tool for diagnosing health risk cognition and modeling protective consumer choices during biological crises.

2. Keywords

Affective Risk Perception, Analytical Risk Perception, Experiential Risk Perception, COVID-19, Risk Perception, Travel Intentions, Pandemic Decision-Making, Health Belief Model, Dual-Process Theory, Psychometrics

3. Authors

The COVID-19 Risk Perception Measure was developed and psychometrically validated by an interdisciplinary team of behavioral scientists, experimental psychologists, economists, and molecular virologists from the University of Trento, Italy:

• Lucia Savadori, Ph.D. (Corresponding Author)
  Affiliation: Department of Economics and Management, University of Trento, Via Inama 5, 38122 Trento, Italy.
  ORCID: 0000-0003-3957-3132
  Email: [email protected]
• Oksana Tokarchuk, Ph.D.
  Affiliation: Department of Economics and Management, University of Trento, Trento, Italy.
• Massimo Pizzato, Ph.D.
  Affiliation: Department of Cellular, Computational and Integrative Biology (CIBIO), University of Trento, Trento, Italy.
• Stefania Pighin, Ph.D.
  Affiliation: Center for Mind/Brain Sciences (CIMeC), University of Trento, Trento, Italy.
  ORCID: 0000-0002-9088-7201

4. Purpose

The primary purpose of the COVID-19 Risk Perception Measure is to provide a granular, theoretically robust assessment tool capable of capturing how human beings internally evaluate biological health threats when contemplating discretionary behavioral actions, with particular emphasis on mobility, leisure activities, and tourist travel. During major infectious disease outbreaks such as SARS-CoV-2, individuals are continually inundated with epidemiological data, varying communication formats (e.g., raw case numbers versus normalized incidence rates), and widespread institutional mandates. Standard health behavior instruments frequently collapse risk perception into a singular, unidimensional probability estimate. However, cognitive psychology demonstrates that risk appraisal is complex, nuanced, and multifaceted. Savadori et al. (2023) developed this instrument to decompose risk into its foundational affective, analytical, and experiential components, thereby enabling researchers to examine which cognitive-emotional mechanisms drive or inhibit preventive health behaviors and travel intentions.

In clinical and public health contexts, the tool serves to diagnose risk-miscalibration patterns, such as catastrophic dread or irrational optimism, which directly impact compliance with preventive behaviors (e.g., mask-wearing, vaccination, adherence to quarantine protocols, and voluntary mobility restrictions). Understanding whether health non-compliance or excessive behavioral avoidance stems from cognitive analytical miscalculation (underestimating or overestimating probability rates) or affective hypersensitivity (paralyzing anxiety, dread, and fear) allows epidemiologists and health communication officers to design tailored public interventions. For example, if analytical risk is the primary determinant of behavior, publishing transparent infection prevalence data will effectively adjust public actions; conversely, if affective dread dominates, narrative or reassurance-based emotional communication may be required.

From an applied research and socio-economic standpoint, particularly within hospitality management and tourism economics, the scale addresses the urgent operational need to predict consumer behavior under conditions of biological uncertainty. Destinations, airlines, and hospitality enterprises require valid metrics to quantify how perceived destination infection rates translate into travel hesitancy, booking cancellations, or risk mitigation adaptations. By offering distinct subscales for affective fear, analytical probability, and experiential availability, the measure facilitates structural equation modeling of consumer choice, providing empirical guidance for evidence-based risk communication, crisis management, and tourism recovery policies.

5. Psychological Construct

The COVID-19 Risk Perception Measure operationalizes subjective risk perception as a tripartite psychological construct, augmented by behavioral travel intentions. Rather than treating risk as a monolithic numerical calculation, the construct reflects the psychological interplay between deliberative logic, affective visceral reactions, and past episodic memory.

1. Affective Risk Perception

Affective risk perception represents the emotional, visceral, and intuitive apprehension elicited by a hazard. It is characterized by subjective feelings of dread, anxiety, fear, worry, and physiological discomfort when contemplating exposure to the pathogen. Unlike cold mathematical probability, affective risk operates through somatic markers and immediate emotional valence. For example, an individual scoring high on this dimension experiences pronounced psychological tension, dread, and worry at the mere thought of entering a virus-affected environment, regardless of whether statistical likelihood of exposure is low. Items measuring this construct focus on the degree of worry, anxiety, and dread evoked by traveling to a region with documented viral transmission.

2. Analytical Risk Perception

Analytical risk perception captures the deliberative, reflective, and cognitive appraisal of risk. This dimension reflects an individual's reasoned estimation of statistical probability, personal vulnerability, and the potential severity of negative physical health outcomes. It functions through conscious information processing, evaluating objective conditions (e.g., regional case numbers, personal protective behaviors, and immune vulnerabilities). Prototypical analytical appraisals involve answering questions such as: "What is the statistical likelihood that I will contract the virus if I spend seven days in this destination?" and "If infected, how severe will the clinical consequences be on my physical well-being?" Analytical risk reflects cold calculation rather than visceral terror, representing the classical expected-utility framework within health behavior models.

3. Experiential Risk Perception

Experiential risk perception bridges cognition and personal history, capturing the intuitive, heuristics-based mental availability of a hazard. Rooted in direct or vicarious past encounters with illness, this dimension measures the ease with which an individual can mentally simulate contracting the disease, the vividness of the threat, and personal feelings of susceptibility based on historical encounters. When an individual has previously fallen severely ill or witnessed close relatives suffer from a virus, the mental scenario of infection is highly accessible, leading to the intuitive reaction: "That could easily happen to me." Conversely, individuals lacking such experiences may maintain an optimistic illusion of invulnerability. Experiential items evaluate mental simulation ease, personal past encounters with infectious disease, and intuitive confidence in threat management.

4. Travel Intentions

Functioning as the focal behavioral outcome within the instrument's broader nomological network, travel intentions assess an individual's explicit willingness, plans, or deliberate avoidance regarding traveling to an area characterized by active pathogen transmission. This component captures the functional consequence of the risk appraisal process, manifesting either as approach behavior (willingness to travel) or avoidance behavior (canceling plans and avoiding endemic zones).

6. Theoretical Framework

The COVID-19 Risk Perception Measure is anchored in well-established paradigms of behavioral economics, cognitive science, and social psychology. Its theoretical foundation synthesizes the Risk-as-Feelings Hypothesis (Loewenstein et al., 2001), Dual-Process Theories of Cognition (Kahneman, 2011; Slovic et al., 2004), and classic models of health protection such as the Health Belief Model (Rosenstock, 1974) and Protection Motivation Theory (Rogers, 1975).

The Dual-Process Architecture and the Affect Heuristic

Traditional economic theories assumed that decision-makers assess risks logically by multiplying the objective probability of an adverse event by its expected severity (Subjective Expected Utility Theory). However, cognitive psychology demonstrates that human judgment is governed by two complementary cognitive systems: System 1 (fast, automatic, affective, associative, and experiential) and System 2 (slow, deliberative, logical, analytical, and rule-governed). Slovic et al. (2004) introduced the affect heuristic, proposing that individuals consult a readily accessible affective "pool" containing positive and negative feelings tagged to mental images. Loewenstein et al. (2001) expanded this with the Risk-as-Feelings model, demonstrating that emotional reactions to danger often diverge sharply from cognitive evaluations. When individuals face immediate or vivid threats, affective reactions (dread, fear) exert a direct and powerful influence on behavior, frequently overriding cognitive probability assessments. The Savadori et al. (2023) scale directly reflects this duality by separating analytical risk (System 2 deliberative calculations) from affective risk (System 1 visceral emotion).

Availability Heuristic and Experiential Learning

The experiential dimension of the scale is grounded in the foundational work of Tversky and Kahneman (1973) on the availability heuristic. According to this principle, people assess the frequency, probability, or danger of an event based on the ease with which relevant instances come to mind. Past personal experiences with disease, media exposure, and direct encounters with health crises populate episodic memory with salient, emotionally charged exemplars. When assessing personal risk in a travel setting, individuals do not merely review statistical tables; they simulate scenarios. If the image of becoming infected is cognitively available and vivid, subjective risk increases substantially. By explicitly incorporating experiential risk items, the scale models this heuristics-based cognitive processing.

Health Protective Behavior Paradigms

Finally, the analytical subscale incorporates foundational principles from the Health Belief Model (HBM; Rosenstock, 1974) and Protection Motivation Theory (PMT; Rogers, 1975). These theories postulate that protective behavioral adoption—such as travel avoidance, vaccination, or non-pharmaceutical interventions—is determined by two core cognitive assessments: perceived susceptibility (the perceived probability of contracting the health condition) and perceived severity (the evaluation of the seriousness of the consequences). The analytical dimension captures both elements by querying subjective infection probability alongside the severity of health consequences.

7. Validity

The validation of the COVID-19 Risk Perception Measure involved rigorous construct, convergent, discriminant, and predictive testing across two empirical studies conducted by Savadori et al. (2023). Although initial literature extractions noted early stages of structural confirmation, the complete peer-reviewed validation demonstrated robust psychometric validity across multiple domains.

Construct and Structural Validity

Construct validity was established through confirmatory factor analysis (CFA) performed on data gathered from adult Italian residents evaluating travel to COVID-19-affected destinations. The hypothesized three-factor risk structure was tested against alternative models (such as single-factor models where all risk dimensions collapse into one general factor). The tripartite structure separating affective, analytical, and experiential perceptions exhibited superior fit, confirming that respondents systematically differentiate visceral emotional dread from objective probabilistic calculations and episodic heuristic memories. Re-specification based on modification indices refined construct boundaries by removing items displaying cross-loadings or structural redundancy.

Convergent and Discriminant Validity

Convergent validity was verified by evaluating the magnitude and statistical significance of the standardized factor loadings of items onto their respective latent constructs. In the optimized CFA model, all retained indicators loaded significantly and strongly on their target dimensions (all standardized loadings > .60, p < .001). Average Variance Extracted (AVE) estimates confirmed that a substantial proportion of indicator variance was explained by the underlying latent factors rather than measurement error. Discriminant validity was evaluated using the Fornell-Larcker criterion; the square root of the AVE for affective, analytical, and experiential dimensions exceeded their mutual inter-construct correlations. This confirmed that while the three subscales share common variance as facets of general risk perception, each measures a distinct cognitive-emotional domain.

Predictive and Criterion Validity

Predictive validity was verified by examining the relationship between the three risk perception dimensions and behavioral travel intentions under varying experimental risk communication formats. In the original experiments, participants were exposed to official regional COVID-19 epidemiological data presented as either raw positive case counts or normalized incidence rates (rates per 100,000 residents). Analytical, affective, and experiential risk perceptions each predicted travel intentions in expected directions: higher perceived risk across all three dimensions correlated significantly with increased travel avoidance and decreased willingness to travel to the affected destination. Regression and structural path models revealed that affective risk perception mediated the effect of communication format on travel intentions, showing high criterion-related and predictive validity.

8. Reliability

The psychometric evaluation of the COVID-19 Risk Perception Measure demonstrates high internal consistency reliability across its constituent factors and behavioral outcome measures.

Internal Consistency Reliability

Savadori et al. (2023) evaluated internal consistency reliability using Cronbach's alpha (α) and Composite Reliability (CR) metrics across two independent experimental samples. In the preliminary 13-item measurement model and the final respecified 11-item model, all multi-item subscales demonstrated alpha values exceeding standard psychometric benchmarks (α > .70):

• Affective Risk Perception Subscale: Demonstrated strong internal consistency, with Cronbach's alpha coefficients typically ranging between α = .84 and α = .89, indicating that items capturing worry, dread, fear, and anxiety possess high internal coherence.
• Analytical Risk Perception Subscale: Yielded Cronbach's alpha values exceeding α = .78, confirming high reliability in evaluating perceived probability, personal vulnerability, and expected severity of health consequences.
• Experiential Risk Perception Subscale: Exhibited solid internal consistency, with alpha coefficients exceeding α = .72, demonstrating reliable measurement of mental simulation ease, past illness salience, and intuitive vulnerability.
• Travel Intentions Subscale: The behavioral intention items demonstrated high inter-item correlation and internal consistency (α > .82).

Composite Reliability and Scale Stability

In addition to Cronbach's alpha, composite reliability (CR) was calculated within structural equation modeling frameworks. All latent dimensions surpassed the recommended threshold of CR ≥ .70, confirming that the indicators reliably capture the latent constructs without undue inflation from scale length. While formal longitudinal test-retest reliability across multi-month intervals was naturally influenced by fluctuating pandemic conditions and shifting public health mandates, the scale's split-half and inter-study replicability demonstrated high structural stability across distinct experimental cohorts.

9. Factor Analysis

The internal dimensionality of the COVID-19 Risk Perception Measure was evaluated by Savadori et al. (2023) using rigorous Confirmatory Factor Analysis (CFA) to verify the theoretical tri-component architecture of health risk perception alongside travel behavioral intentions.

Initial Measurement Model Assessment

The original conceptualization incorporated thirteen items adapted from established medical and psychological literature (e.g., Dillard et al., 2012; Ferrer et al., 2016; Kaufman et al., 2020; Peters et al., 2011; Pighin et al., 2011, 2015). Confirmatory factor analysis was conducted on the full dataset using maximum likelihood estimation. The initial 13-item measurement model demonstrated acceptable global fit across incremental fit indices, but exhibited strain on absolute and parsimonious fit indices:

• Chi-Square (χ²): 375
• Degrees of Freedom (df): 62
• Normed Chi-Square (χ²/df): 6.05
• Comparative Fit Index (CFI): 0.951
• Tucker-Lewis Index (TLI): 0.938
• Root Mean Square Error of Approximation (RMSEA): 0.090

According to standard psychometric criteria (e.g., Hair, Black, Babin, & Anderson, 2010), an optimal model fit requires χ²/df < 3.00, CFI > 0.95, TLI > 0.95, and RMSEA < 0.07 (or ideally ≤ 0.06). While the CFI and TLI demonstrated acceptable incremental fit (> 0.90), the χ²/df ratio (6.05) and the RMSEA (0.090) exceeded conventional cut-offs, indicating structural strain and localized residual covariance.

Modification and Model Re-specification

To identify the sources of misfit, the authors examined modification indices (MI) and standardized residual covariances. The diagnostic evaluation revealed that two items introduced significant measurement error and structural cross-loadings: item Q2 (evaluating regional infection rate estimation: "How high do you rate the number of currently positive cases in this region?") and item Q3 (evaluating conditional risk perception dependent on individual health care behaviors: "Considering the way you take care of your health, in your opinion, the probability of you contracting coronavirus infection by going on vacation to this location for 7 days is…"). Item Q2 cross-loaded between objective cognitive estimation and affective reaction, while Q3 shared high residual variance with unconditional probability estimates (Q1 and Q5).

Following modification recommendations, these two items were systematically eliminated, and the structural model was re-specified. The refined measurement model demonstrated substantially improved goodness-of-fit parameters:

• Chi-Square (χ²): 126
• Degrees of Freedom (df): 41
• Normed Chi-Square (χ²/df): 3.07
• Comparative Fit Index (CFI): 0.984
• Tucker-Lewis Index (TLI): 0.978
• Root Mean Square Error of Approximation (RMSEA): 0.058

In this re-specified 11-item model, the CFI (0.984) and TLI (0.978) reflected excellent model fit, while the RMSEA (0.058) fell well within the stringent threshold for close approximation (< 0.06). The normed chi-square (χ²/df = 3.07) was reduced by approximately 50%, falling right at the conservative boundary (≤ 3.00). All standardized factor loadings for the retained items on their respective latent factors were high, positive, and statistically significant (p < .001), validating the three-factor structure of risk perception alongside the behavioral intention construct.

10. Instrument / Measurement Tool

The operational administration of the COVID-19 Risk Perception Measure follows a standardized experimental or survey protocol designed to gauge both cognitive-affective appraisals and behavioral intentions:

• Test Type: Standardized self-report rating scale / psychometric assessment questionnaire.
• Format: Administered digitally or via paper-and-pencil. Respondents are presented with a standardized scenario prompt: "Imagine you have to choose your summer vacation destination [or next vacation], and you are evaluating a specific location based on its safety with regard to coronavirus." In experimental paradigms, participants are presented with epidemiological data for a destination (e.g., active regional case counts or incidence rates per 100,000 residents) before completing the questionnaire.
• Item Count: 13 items total in the complete inventory (comprising 11 risk perception items and 2 travel intention items in the original full inventory; 11 items total in the CFA-optimized model after deleting Q2 and Q3).
• Dimensional Structure:
  • Affective Risk Perception (assessing fear, dread, worry, and subjective anxiety).
  • Analytical Risk Perception (assessing perceived infection probability, health consequence severity, and conditional susceptibility).
  • Experiential Risk Perception (assessing mental simulation ease, availability of prior disease experience, and intuitive vulnerability).
  • Travel Intentions (assessing planned mobility and intentional avoidance behaviors).
• Response Scale: 7-point Likert scale (1 = Strongly disagree to 7 = Strongly agree; or 1 = Not at all to 7 = Extremely, depending on the item).
• Target Population: Adult population (18 years and older); validated with adult general public samples.
• Administration Duration: Approximately 5 to 8 minutes for complete administration.
• Scoring and Transformation Rules:
  • Individual subscale scores are computed by calculating the arithmetic mean or sum of the items assigned to each respective dimension.
  • Items designated as reverse-scored (e.g., items measuring confidence in avoidance or intentional avoidance vs. approach) must be inverted prior to aggregation using the transformation formula: Score_Inverted = 8 - Original_Score.
  • In structural equation modeling and empirical research, latent variable modeling with CFA factor weights is recommended to account for item-specific measurement errors.
  • In psychometric analyses guided by modification indices, items Q2 and Q3 are omitted when computing optimized subscale indexes.

11. Permissions & Fee and Test Year

• Publication Year: 2023.
• Authors / Copyright Holders: Lucia Savadori, Oksana Tokarchuk, Massimo Pizzato, and Stefania Pighin. Original publication published by Elsevier Ltd. on behalf of the College of Hospitality and Tourism Management, Virginia Tech.
• Permissions: The measure is open and accessible for non-commercial academic research, pedagogical purposes, and institutional health assessment under academic fair-use conventions. Researchers are requested to cite the original validation article (Savadori et al., 2023).
• Commercial Use: Commercial applications, proprietary market research, or corporate deployment require formal licensing agreements and written permission from the copyright owners and publisher.
• Fee: There is no fee (free of charge) for non-commercial academic and clinical research use.
• Correspondence Information: Inquiries regarding the scale and its implementation in experimental health communication should be directed to Dr. Lucia Savadori, Department of Economics and Management, University of Trento, Via Inama 5, 38122 Trento, Italy (Email: [email protected]).

12. References

• Dillard, A. J., Ferrer, R. A., Ubel, P. A., & Fagerlin, A. (2012). Risk perception measures' associations with behavior and feelings of vulnerability in the context of colorectal cancer screening. Journal of Health Communication, 17(10), 1116–1130. https://doi.org/10.1080/10810730.2012.665424
• Ferrer, R. A., Klein, W. M., Persoskie, A., Avishai-Yitshak, A., & Sheeran, P. (2016). The tripartite model of risk perception (TRIRISK): Distinguishing deliberative, affective, and experiential components of perceived risk. Annals of Behavioral Medicine, 50(5), 653–663. https://doi.org/10.1007/s12160-016-9790-z
• Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2010). Multivariate data analysis: A global perspective (7th ed.). Pearson Education.
• Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux.
• Kaufman, A. R., Ellis, E. M., Waterman, L. A., & Ferrer, R. A. (2020). Beliefs about the nature of cancer and risk perceptions: An exploratory analysis. Journal of Health Psychology, 25(8), 1083–1094. https://doi.org/10.1177/1359105317748733
• Loewenstein, G. F., Weber, E. U., Hsee, C. K., & Welch, N. (2001). Risk as feelings. Psychological Bulletin, 127(2), 267–286. https://doi.org/10.1037/0033-2909.127.2.267
• Peters, E., Hart, P. S., & Fraenkel, L. (2011). Informing patients: The influence of numeracy, framing, and format of probability on comprehension and choice. Medical Decision Making, 31(3), 432–442. https://doi.org/10.1177/0272989X10391672
• Pighin, S., Bonnefon, J. F., & Savadori, L. (2011). Overcoming number-of-cases framing effects in medical risk communication: A test of iconic representations. Medical Decision Making, 31(2), 333–342. https://doi.org/10.1177/0272989X10385846
• Pighin, S., Savadori, L., & Bonnefon, J. F. (2015). Health risk communication: The effect of risk formats and frequency framing on cognitive, affective, and behavioral outcomes. Health Psychology Review, 9(2), 175–191. https://doi.org/10.1080/17437199.2013.863172
• Rogers, R. W. (1975). A protection motivation theory of fear appeals and attitude change. The Journal of Psychology, 91(1), 93–114. https://doi.org/10.1080/00223980.1975.9915803
• Rosenstock, I. M. (1974). Historical origins of the health belief model. Health Education Monographs, 2(4), 328–335. https://doi.org/10.1177/109019817400200403
• Savadori, L., Tokarchuk, O., Pizzato, M., & Pighin, S. (2023). The impact of infection risk communication format on tourism travel intentions during COVID-19. Journal of Hospitality and Tourism Management, 54, 65–75. https://doi.org/10.1016/j.jhtm.2022.12.004
• Sheeran, P., Harris, P. R., & Epton, T. (2014). Does heightening risk appraisals change intentions and behavior? A meta-analysis of the experimental evidence. Psychological Bulletin, 140(2), 511–543. https://doi.org/10.1037/a0033065
• Slovic, P., Finucane, M. L., Peters, E., & MacGregor, D. G. (2004). Risk as analysis and risk as feelings: Some thoughts about affect, reason, risk, and rationality. Risk Analysis, 24(2), 311–322. https://doi.org/10.1111/j.0272-4332.2004.00433.x
• Tversky, A., & Kahneman, D. (1973). Availability: A heuristic for judging frequency and probability. Cognitive Psychology, 5(2), 207–232. https://doi.org/10.1016/0010-0285(73)90033-9

13. Items of the Scale