Every year, traffic safety agencies launch campaigns reminding drivers to wear seatbelts, stop texting behind the wheel, and slow down in school zones. The information is clear, the statistics are public, and the risks are well documented. Yet millions of drivers continue to engage in exactly the behaviors they know are dangerous. A commuter who checks a phone at every red light, a teenager who drives without a seatbelt, a homeowner who ignores hurricane evacuation orders — each of these people has some awareness of the risk. The question is not whether they know. The question is why knowing is not enough.
The Health Belief Model (HBM) was designed to answer precisely this kind of question. Originally developed to explain why people failed to participate in free tuberculosis screening programs in the 1950s, the HBM has become one of the most widely applied frameworks for understanding protective behavior — and the failure to adopt it.
Why Protective Behavior Matters
Protective behaviors in transportation and planning contexts include wearing seatbelts, using helmets, obeying speed limits, avoiding phone use while driving, preparing for natural disasters, and complying with evacuation orders. These behaviors share a common structure: they involve a person weighing a perceived threat against the perceived cost and benefit of taking protective action.
The stakes are high. Seatbelt use alone prevents an estimated 15,000 deaths per year in the United States. Distracted driving contributes to roughly 3,000 fatalities annually. Disaster preparedness — having supplies, knowing evacuation routes, maintaining communication plans — reduces mortality and displacement after hurricanes, floods, and earthquakes.
Yet adoption of these behaviors is far from universal. Understanding why requires a model that goes beyond knowledge and focuses on how people perceive risk, evaluate trade-offs, and respond to triggers.
Core intuition: The Health Belief Model says that people adopt protective behavior when they believe they are personally susceptible to a serious threat, they believe the protective action will work, they believe the costs of action are manageable, and something triggers them to act. Remove any one of these elements, and behavior often does not change.
The Model in Plain Language
The Health Belief Model proposes that health-related and safety-related behavior is a function of two broad assessments: threat perception and behavioral evaluation.
Threat perception asks: How likely is this bad thing to happen to me, and how bad would it be? Behavioral evaluation asks: If I take the recommended action, will it help — and what will it cost me?
These two assessments combine with external triggers (cues to action) and internal confidence (self-efficacy) to determine whether a person takes protective action.
The model is deliberately individual-level and cognitive. It treats behavior as the product of a cost-benefit calculation performed under subjective risk assessment. It does not claim that people perform this calculation consciously or rationally — only that these perceptual variables reliably predict behavioral outcomes.
Core Constructs
The HBM includes six main constructs, each representing a distinct dimension of how people evaluate threats and responses.
Belief about the likelihood of experiencing a threat.
Belief about how serious the consequences would be.
Belief that the recommended action will reduce the threat.
Belief about the costs, inconveniences, or difficulties of acting.
Triggers that activate readiness: events, reminders, symptoms, social influence.
Perceived Susceptibility
This is the person’s belief about how likely they are to experience the negative outcome. A driver who believes “I’m a careful driver — crashes happen to other people” has low perceived susceptibility. A driver who recently witnessed a crash on their daily commute may have higher perceived susceptibility.
Susceptibility is subjective and often biased. Research consistently shows that people underestimate their personal risk for familiar hazards. This optimism bias is one of the most robust findings in risk psychology and a major reason why information campaigns alone fail.
Perceived Severity
Even if someone acknowledges that a threat is possible, they may not believe it is serious. Perceived severity captures how bad the person thinks the consequences would be. A driver might acknowledge that texting while driving increases crash risk but believe that any crash would be minor — a fender bender rather than a fatality.
Together, perceived susceptibility and perceived severity form perceived threat. The model predicts that behavior change requires a threshold level of perceived threat. Too little of either component, and motivation to act remains low.
Perceived Threat:
Perceived Threat = f(Perceived Susceptibility × Perceived Severity)
Both components must be elevated for threat to motivate action. High severity with low susceptibility ("crashes are deadly, but it won't happen to me") produces complacency. High susceptibility with low severity ("I might get a ticket, but it's just a fine") produces tolerance.
Perceived Benefits
This construct captures whether the person believes the recommended action will actually reduce the threat. If a driver does not believe that putting down the phone will meaningfully reduce crash risk — perhaps because they think they can multitask safely — then perceived benefits are low.
Perceived benefits must be specific to the recommended behavior, not to threat reduction in general. A person might believe that crashes can be prevented but not believe that seatbelts specifically make a difference. The benefit must be linked to the action.
Perceived Barriers
This is often the strongest predictor in HBM studies. Perceived barriers include any cost, inconvenience, difficulty, discomfort, or social consequence associated with the protective behavior. For seatbelt use, barriers might include physical discomfort, forgetting, or a belief that seatbelts restrict movement. For disaster preparedness, barriers include cost, time, storage space, and the belief that preparation is futile against large-scale events.
The model predicts that even when perceived threat is high and perceived benefits are clear, high perceived barriers can prevent action.
Cues to Action
Cues to action are the triggers that activate a person’s readiness to act. They can be internal (experiencing a near-miss while driving, feeling anxiety during a storm warning) or external (seeing a public service announcement, receiving an emergency alert, witnessing a crash, or hearing a friend describe a close call).
Cues are the least well-developed construct in the HBM. They are difficult to measure retrospectively and hard to isolate from other influences. However, their inclusion acknowledges that threat perception alone is insufficient — something must activate the decision process.
Self-Efficacy
Added to the model later by Rosenstock, Strecher, and Becker (1988), self-efficacy is the person’s confidence in their ability to successfully perform the protective behavior. A person might believe that emergency preparedness is important and beneficial, but if they do not believe they can assemble a kit, learn evacuation routes, or convince their family to participate, they will not act.
Self-efficacy is borrowed from Bandura’s Social Cognitive Theory and has proven to be a powerful predictor across many health and safety domains.
Causal Logic
The HBM proposes a specific causal sequence, though it is more accurately described as a set of conditions that jointly determine behavior rather than a strict temporal chain.
Susceptibility × Severity create motivation to consider action.
Expected benefits minus perceived barriers determine net attractiveness of action.
An external or internal trigger activates the decision process.
Confidence in one's ability to carry out the behavior.
The person adopts (or fails to adopt) the recommended action.
The core prediction is:
Behavioral prediction:
P(Behavior) = f(Threat Perception, Benefits − Barriers, Cues, Self-Efficacy)
Behavior is most likely when perceived threat is high, perceived benefits exceed perceived barriers, a cue activates readiness, and self-efficacy is sufficient.
Importantly, the model also explains inaction. A person who perceives no susceptibility will not be motivated regardless of how beneficial the action is. A person who perceives high susceptibility and severity but also high barriers and low self-efficacy may experience fear without action — a state sometimes described as fatalism.
The person believes: "This threat is real and could happen to me. The consequences would be serious. This action will help. The costs are manageable. I can do it. Something just reminded me."
The person believes any of: "It won't happen to me. It wouldn't be that bad. This action won't help much. The costs are too high. I can't do it. Nothing has prompted me to act."
Data Needed
Applying the HBM requires measuring each construct, typically through self-report surveys.
Likert-scale items: "I am likely to be involved in a crash caused by distracted driving." "My neighborhood is at risk of flooding."
Likert-scale items: "A crash caused by phone use would result in serious injury or death." "Flood damage to my home would be financially devastating."
Paired scales: "Putting my phone away while driving would reduce my crash risk" vs. "I would miss important calls or messages if I stopped using my phone while driving."
SUSCEPTIBILITY
Q1: "I could be involved in a crash if I use my phone while driving."
[Strongly Disagree] 1 — 2 — 3 — 4 — 5 [Strongly Agree]
SEVERITY
Q2: “A crash caused by distracted driving would have life-altering consequences.”
[Strongly Disagree] 1 — 2 — 3 — 4 — 5 [Strongly Agree]
BENEFITS
Q3: “Not using my phone while driving would significantly reduce my risk of a crash.”
[Strongly Disagree] 1 — 2 — 3 — 4 — 5 [Strongly Agree]
BARRIERS
Q4: “It is difficult to ignore incoming calls and messages while driving.”
[Strongly Disagree] 1 — 2 — 3 — 4 — 5 [Strongly Agree]
CUES TO ACTION
Q5: “I have seen news reports about crashes caused by distracted driving.”
[Never] 1 — 2 — 3 — 4 — 5 [Very Often]
SELF-EFFICACY
Q6: “I am confident I can keep my phone put away for an entire drive.”
[Not at all confident] 1 — 2 — 3 — 4 — 5 [Very confident]
Additional data sources that can complement survey measures include crash records (to validate severity perceptions against actual outcomes), observational studies of phone use at intersections, focus groups and interviews to explore barrier perceptions in depth, and GPS or smartphone sensor data to link self-reported constructs with actual driving behavior.
Methods
The HBM is typically analyzed using quantitative methods that link construct scores to behavioral outcomes.
Most common. Predicts binary behavior (wears seatbelt or not, prepared for disaster or not) from HBM construct scores. Allows identification of which constructs are significant predictors and their relative strength.
Tests the full model simultaneously, including latent variables for each construct, measurement error, and indirect effects. Provides fit indices that evaluate overall model adequacy.
Enters constructs in blocks — demographics first, then threat perception, then benefit-barrier assessment, then self-efficacy — to examine incremental explanatory power at each stage.
Qualitative methods are also valuable. Semi-structured interviews can reveal barrier perceptions that survey scales might miss, and focus groups can explore how cues to action operate in specific communities.
Transportation Example: Distracted Driving Prevention
Consider a state transportation department evaluating its distracted driving campaign. The department has run billboard, radio, and social media messages for two years, but phone-use-while-driving rates remain stubbornly high among drivers aged 18–34.
An HBM-based study would survey drivers in that age group and measure all six constructs. Hypothetical findings might look like this:
This pattern reveals the problem. Young drivers acknowledge that distracted driving crashes are severe (high severity), but they do not believe it will happen to them (low susceptibility). They see some benefit to putting the phone away, but the perceived barriers — fear of missing messages, social pressure to respond quickly, habitual phone checking — are very high. Self-efficacy is moderate: many drivers want to stop but doubt their ability to resist checking.
Diagnostic insight: The campaign focused on severity ("Texting kills"), but severity was already high. The real problems were low susceptibility, high barriers, and insufficient self-efficacy. The HBM diagnosis suggests redirecting the campaign toward personalized susceptibility messages, barrier-reduction strategies (phone holders, Do Not Disturb modes, passenger norms), and efficacy-building content ("Here's how to go phone-free on your commute").
A logistic regression might confirm the diagnosis:
Outcome: Phone-free driving (0 = uses phone, 1 = phone-free)B SE OR pPerceived Susceptibility 0.41 0.12 1.51 .001 ** Perceived Severity 0.08 0.14 1.08 .572 Perceived Benefits 0.29 0.11 1.34 .008 ** Perceived Barriers −0.53 0.10 0.59 <.001 *** Self-Efficacy 0.47 0.09 1.60 <.001 ***
Nagelkerke R² = .34
In this hypothetical result, severity is not a significant predictor — because nearly everyone already agrees crashes are serious. The strongest predictors are barriers (negative), self-efficacy (positive), and susceptibility (positive). This is a classic HBM finding: the information gap is not about facts but about personal relevance and practical feasibility.
Application to Other Domains
The HBM framework applies broadly across transportation safety and disaster preparedness:
Low susceptibility ("I'm a safe driver") and barriers (discomfort, forgetting) are common predictors of non-use. Click-it-or-ticket campaigns work partly because they add a cue to action (enforcement visibility) and increase susceptibility (likelihood of being caught).
Many cyclists acknowledge severity of head injuries but perceive low susceptibility on familiar routes. Barriers include discomfort, appearance, storage, and cost. Self-efficacy for consistent use varies by cycling experience.
Homeowners in flood-prone areas often show high severity but low susceptibility ("we haven't flooded in 20 years"). Barriers to preparation include cost of supplies, time, and fatalism. Cues to action — a neighbor's flood experience, an emergency alert — are powerful triggers.
Strengths
The Health Belief Model has several important strengths that have sustained its use for over six decades.
Strong on risk perception. The model directly addresses how people perceive threats, which is central to safety behavior. The susceptibility-severity framework captures why people remain passive even when they have accurate factual knowledge.
Intuitive and practical. The constructs map naturally onto survey questions, and findings translate into actionable recommendations. If barriers are the main obstacle, reduce barriers. If susceptibility is low, increase personalization. If self-efficacy is weak, build skills.
Widely applied and validated. The HBM has been applied across dozens of health and safety domains, generating a large empirical base. Janz and Becker’s (1984) review found that perceived barriers was consistently the strongest predictor across studies, and perceived susceptibility was particularly important for preventive behaviors.
Diagnostic power. The model does not just predict behavior — it diagnoses which component is failing, allowing targeted intervention design rather than generic information campaigns.
Limitations
Despite its longevity, the HBM has well-documented limitations that researchers should consider before selecting it as a primary framework.
Weak on social influence. The HBM is an individual cognitive model. It does not account for social norms, peer pressure, cultural expectations, or community-level influences. A teenager’s decision to text while driving is heavily influenced by what friends do, but the HBM has no construct for social norms.
No temporal dimension. The model is essentially static. It does not describe how beliefs change over time, how behavior develops through stages, or how past behavior feeds back into current perceptions. The Transtheoretical Model (Post 8) and Protection Motivation Theory (Post 7) address temporality more directly.
Habit and automaticity are missing. Many safety-relevant behaviors — seatbelt use, phone checking, speeding — become habitual. Once a behavior is automatic, conscious cost-benefit calculation is not the primary driver. The HBM assumes deliberative processing.
Structural constraints are invisible. The model focuses on individual perceptions but does not account for infrastructure, policy, income, or access. A low-income family may not prepare for a disaster because they cannot afford supplies, not because they misjudge the threat. The HBM might classify this as a “barrier,” but that conflates perceived barriers with real material constraints.
Weak specification of relationships. Unlike the Theory of Planned Behavior or Structural Equation Models, the HBM does not specify the exact functional form of relationships between constructs. Different researchers operationalize the model differently, making cross-study comparison difficult.
Cues to action remain underdeveloped. Despite being a named construct, cues to action have received less empirical attention than other constructs, partly because they are difficult to measure retrospectively and vary greatly in type and timing.
Best Use Case
Best use case: The Health Belief Model is most useful when the research question is: "Why do people fail to adopt a specific protective behavior despite knowing the risk?" It is particularly strong for diagnosing which perceptual component — susceptibility, severity, benefits, barriers, or self-efficacy — is the bottleneck, and for designing targeted interventions that address the right gap.
It is less suitable when the behavior is habitual, socially driven, structurally constrained, or when the research question requires modeling behavior change over time.
Key Takeaway
People do not protect themselves simply because they know a threat exists. They protect themselves when they believe the threat is personally relevant, the protective action works, the costs are manageable, and they have the confidence to act. The Health Belief Model identifies exactly which of these beliefs is failing.
Key References
- Rosenstock, I. M. (1974). "Historical Origins of the Health Belief Model." Health Education Monographs, 2(4), 328–335. — The original formulation that established the model's theoretical foundations and core constructs.
- Janz, N. K., & Becker, M. H. (1984). "The Health Belief Model: A Decade Later." Health Education Quarterly, 11(1), 1–47. — A landmark review of 46 HBM studies that established perceived barriers as the most powerful predictor across behaviors.
- Champion, V. L., & Skinner, C. S. (2008). "The Health Belief Model." In K. Glanz, B. K. Rimer, & K. Viswanath (Eds.), Health Behavior and Health Education: Theory, Research, and Practice (4th ed., pp. 45–65). Jossey-Bass. — The definitive textbook chapter that synthesizes decades of HBM research and integrates self-efficacy into the framework.
- Şimşekoğlu, Ö., & Lajunen, T. (2008). "Social psychology of seat belt use: A comparison of theory of planned behavior and health belief model." Transportation Research Part F, 11(3), 181–191. — Applies both TPB and HBM to seatbelt use, demonstrating the diagnostic value of HBM constructs in a transportation safety context.
- Abraham, C., & Sheeran, P. (2005). "The Health Belief Model." In M. Conner & P. Norman (Eds.), Predicting Health Behaviour (2nd ed., pp. 28–80). Open University Press. — A thorough critical review that evaluates the model's empirical record, measurement approaches, and theoretical limitations.
- A city runs a campaign showing graphic images of crashes caused by distracted driving, but phone use rates do not change. Using the HBM, identify at least two construct-level explanations for why the campaign might be failing, and propose modifications that target different constructs.
- Compare how the HBM would explain low seatbelt use among ride-hailing passengers versus low seatbelt use among daily commuters in personal vehicles. Which constructs would differ between the two groups, and why?
- The HBM treats perceived barriers as an individual cognitive variable. But some barriers to disaster preparedness — such as poverty, lack of transportation, or inaccessible emergency information — are structural, not perceptual. How would you modify an HBM-based study to account for both perceived and actual barriers? What does this limitation tell us about the boundary conditions of the model?