Vaccination Hesitancy

Abstract

Vaccination hesitancy, which the Medical Subject Headings vocabulary classifies under vaccination refusal, is the delay in acceptance or outright refusal of vaccination despite the availability of vaccination services. The classification is an indexing artefact rather than a conceptual claim: hesitancy is in fact the broader, graded state of which flat refusal is only the extreme pole. It sets out the World Health Organization's SAGE Working Group definition and its matrix of contextual, individual, and vaccine-specific determinants; the instruments that turn hesitancy into a measurable quantity, the Vaccine Hesitancy Scale, the 5C psychological-antecedents scale, and the Parent Attitudes about Childhood Vaccines survey; the psychological antecedents those scales recover; and the large-scale monitoring of vaccine confidence that tracks hesitancy across countries and time. A worked example scores a respondent on the Vaccine Hesitancy Scale, and three interactive demonstrations follow.

Keywords: vaccination hesitancy, vaccine confidence, 5C model

A health worker who meets a parent uncertain about a recommended vaccine is not usually meeting a committed refuser. Far more often the parent is somewhere in the wide middle ground between wholehearted acceptance and flat rejection: willing in principle, but delaying, asking for one vaccine and declining another, or agreeing only after visible doubt. That graded, conditional, often vaccine-specific uncertainty is what the field means by vaccination hesitancy, and it is the state that actually governs population coverage, because the hesitant are both numerous and movable in a way that the small pole of absolute refusers is not. Making hesitancy tractable has therefore required two things the study of outright refusal did not: an agreed definition precise enough to measure, and instruments that convert a diffuse attitude into a number a clinic or a country can act on.

This article is about those two achievements and what they revealed. It sets out the SAGE definition and determinants matrix that gave the field a shared vocabulary, the validated scales that operationalize hesitancy, the psychological antecedents those scales recover, and the global confidence monitoring that watches hesitancy move. Throughout, the organizing fact is that hesitancy is a measurable position on a continuum, not a fixed type of person.

What Vaccination Hesitancy Is

Vaccination hesitancy is a delay in acceptance, or a refusal, of vaccination despite the availability of vaccination services. The definition is the World Health Organization's SAGE Working Group's, and every word in it does work: delay and refusal mark the two ends of a range, despite availability excludes the person who would vaccinate but cannot reach a clinic, and the clause that follows in the original, that hesitancy is complex and context-specific, varying across time, place, and vaccines, denies that it is a stable trait of a kind of person (MacDonald, 2015). A parent can be hesitant about one vaccine and not another, hesitant this year and not last, hesitant in one country and confident in a neighbor.

In the Medical Subject Headings vocabulary, hesitancy is filed beneath vaccination refusal as a narrower descriptor. That placement is an artefact of how the indexing tree was built rather than a statement about the concepts, because hesitancy is in substance the broader state and refusal its extreme instance: the hesitant may vaccinate late, selectively, or with doubt, so hesitancy captures the uncertain majority that outright refusal does not. The practical consequence of treating hesitancy as the primary object is that the target of measurement and intervention becomes the whole continuum, not just its refusing tail, and the people who most affect coverage are the far larger group who have not refused anything (Larson et al., 2014). The stakes of attending to that majority are concrete: hesitancy that erodes coverage below the threshold for herd immunity has been tied to the resurgence of controlled diseases, which is why the field has framed it as a population-health problem demanding a coordinated response rather than a matter of individual choice alone (Salmon et al., 2015).

The SAGE Definition and the Determinants Matrix

Before 2015 the field lacked a common definition, and studies measured different things under the same word. The SAGE Working Group's contribution was to fix both the definition and a structure for its causes. Alongside the definition it published a matrix of determinants sorting the influences on hesitancy into three categories: contextual influences (historical, sociocultural, political, and media environments), individual and social-group influences (personal experience, beliefs about health, perceived risk, and the views of one's social network), and vaccine- and vaccination-specific influences (the evidence on a particular vaccine, the mode of delivery, the schedule, and the reliability of the supply) (MacDonald, 2015). The matrix is not a theory of mechanism; it is a checklist that forces an investigator to ask where, in a specific setting, the hesitancy is coming from.

That categorization matters because it separates causes that call for entirely different responses. A contextual driver such as a political controversy is not addressed by the same means as a vaccine-specific concern about a new adjuvant, and neither is reached by fixing an individual's access. Ève Dubé and colleagues, surveying the determinants from the clinical side, reached the same conclusion: hesitancy arises from the interaction of these layers rather than from any single cause, so a diagnosis precedes a remedy (Dubé et al., 2013). The matrix also anticipates the measurement problem the next section takes up, because a construct with determinants on three levels cannot be captured by any one question.

Measuring Hesitancy: From Attitude to Score

A definition becomes scientifically useful when it can be measured, and the decade after 2015 produced several validated instruments that turn hesitancy into a score. The most direct is the Vaccine Hesitancy Scale (VHS), developed by Heidi Larson and colleagues out of the SAGE work as a short survey tool intended to be comparable across countries: a set of Likert items that resolve, on analysis, into two factors, a lack of confidence in vaccines and their delivery, and a perception of risks in vaccination (Larson et al., 2015). Gilla Shapiro and colleagues then put the scale through a formal psychometric validation, confirming the two-factor structure and establishing its reliability and construct validity so that a VHS score could be used as a dependent measure rather than an ad-hoc tally (Shapiro et al., 2018).

A second instrument measures the antecedents rather than the attitude itself. Cornelia Betsch and colleagues built and validated the 5C scale, which assigns a respondent five sub-scores, one for each of confidence, complacency, constraints (convenience), calculation, and collective responsibility, on the reasoning that these psychological antecedents, not a single hesitancy score, are what an intervention must target (Betsch et al., 2018). A third, older instrument is specific to the clinic: Douglas Opel and colleagues developed and validated the Parent Attitudes about Childhood Vaccines (PACV) survey, a short questionnaire that predicts a parent's subsequent vaccination behavior from their expressed attitudes, so a provider can identify a hesitant parent before any refusal is voiced (Opel et al., 2011). The three instruments differ in grain, a single hesitancy score, a five-part antecedent profile, a clinical risk flag, but they share the achievement that made the modern field possible: a diffuse attitude rendered as a reproducible number.

Answer the nine items from 1 (strongly disagree) to 5 (strongly agree). The seven confidence items are reverse-scored (6 − response) so that a higher index always means more hesitant; the two risk items are kept as-is. The index is the mean of all nine, and the two subscale means say where the hesitancy sits.

The presets reproduce the article’s Worked Example: Parent A scores 3.78 (high hesitancy, driven by a 4.50 risk subscale against a 3.57 confidence subscale), Parent B scores 1.44 (accepting). Notice that two parents can reach the same index by different routes — which is why the subscales, not the single number, are what an intervention needs.

The Psychological Antecedents

What the scales recover, when they are applied, is a small and stable set of antecedents. The SAGE group's 3C model names the first three: confidence (trust in the safety and effectiveness of vaccines, in the providers who deliver them, and in the institutions that recommend them), complacency (a low perceived risk of the vaccine-preventable disease, which makes vaccination feel unnecessary), and convenience (the physical availability, affordability, and accessibility of services) (MacDonald, 2015). Complacency is the vaccination instance of a much older idea, the Health Belief Model's claim that preventive action depends on the perceived susceptibility to and severity of a threat; when a disease is unseen because vaccination suppressed it, its perceived risk falls and the motive to act falls with it (Rosenstock, 1974).

The 5C model extends the three with calculation (the extent of deliberate information-seeking and risk weighing, which can amplify the effect of any perceived risk) and collective responsibility (the willingness to protect others through one's own vaccination), and it is this five-part structure that the Betsch scale measures (Betsch et al., 2018). The antecedents are not interchangeable. Confidence is consistently the strongest and most general predictor of uptake, so that where it is intact, convenience and complacency largely govern behavior, and where it has collapsed, no amount of accessibility or evidence substitutes for it (Larson et al., 2014). Separating the antecedents is what lets a measurement say not merely how hesitant a person is but why, which is the information an intervention actually needs.

Set a person’s standing on the five antecedents. Each is converted to a barrier — how much it pushes toward hesitancy — and the largest barrier is the diagnosed driver. The point is not a single hesitancy number but the profile: the same overall hesitancy calls for different responses depending on which determinant produced it.

The three presets reach similar overall hesitancy through different determinants, and each selects a different remedy: trust-building for the distrustful parent, risk-salience for the complacent one, and structural access for the constrained one. A campaign that offered all three the same leaflet would reach, at most, one of them — which is why the diagnosis precedes the prescription.

Monitoring Vaccine Confidence at Scale

If hesitancy can be measured in an individual, it can be monitored in a population, and the public-health payoff of the measurement program is an early-warning system. Heidi Larson's Vaccine Confidence Project built exactly this, running repeated cross-national surveys of public confidence in vaccines. Alexandre de Figueiredo and colleagues used that data in a large retrospective modelling study, mapping global trends in vaccine confidence across 149 countries and showing that confidence is volatile, falls sharply in response to local controversies, and that a decline in confidence is associated with, and tends to precede, a decline in uptake (de Figueiredo et al., 2020). Monitoring confidence is thus not a descriptive luxury; it is a leading indicator that can flag a coming coverage problem while there is still time to act.

The monitoring frame also reframes what counts as a determinant at the population level. Larson's foundational systematic review had already established that hesitancy is context-specific, varying by vaccine, place, and time rather than tracking any fixed national character (Larson et al., 2014). Seen at scale, the volatility is the signal: because confidence responds to events, a sudden fall identifies where trust has been damaged and often why, directing a response to the specific contextual determinant, from the SAGE matrix, that is actually driving the change. The instruments of the previous section are what make the monitoring possible, since a comparable score is the precondition of a comparable trend.

A controversy knocks public confidence down; uptake follows, but later. Because the confidence dip arrives first, a monitoring survey can flag the coming coverage problem while there is still time to act. The navy curve is confidence, the gold curve is uptake, lagged behind it.

Raise the severity and the uptake trough deepens; shorten the lag and the warning window shrinks toward nothing. The open methodological problem is latency: a survey that reports the confidence fall weeks after it begins reports it too late, which is why the field is pushing toward faster, higher-frequency signals.

From Hesitancy to Behavior

Measuring and monitoring hesitancy is in service of changing behavior, and here the evidence has overturned the intuitive remedy. The deficit model assumes the hesitant lack facts and will accept once informed, but supplying facts frequently fails. Brendan Nyhan and colleagues found in a randomized trial that standard pro-vaccine messages did not raise the intent to vaccinate, and that correcting the autism myth actually reduced it among the parents already least favorable (Nyhan et al., 2014). The reason is that a hesitancy score is driven by trust and perceived risk, not by a knowledge gap, so a message aimed at the gap misses the determinant.

The constructive turn has been to target the measured antecedent rather than argue with the attitude. Noel Brewer and colleagues, synthesizing the evidence, distinguished interventions aimed at what people think from those aimed at what people do, and found the latter far more effective: reminders, standing orders, convenient access, and presumptive provider recommendations raise uptake more reliably than belief-change campaigns (Brewer et al., 2017). The same logic has been carried into policy design, where behavioral-insight levers such as default appointments and eased access are argued to lift coverage more dependably than persuasion aimed at the attitude (Betsch et al., 2015). A systematic review of hesitancy interventions reached a compatible verdict, finding that multicomponent, dialogue-based strategies tailored to the specific determinant at work outperform any single information-giving approach (Jarrett et al., 2015). A review of published reviews drew the program together: there is no universal fix, and the effective response is the one matched to the hesitancy's measured cause (Dubé et al., 2015). Where communication is used, its design matters more than its volume, and the counter to hesitancy is credible messengers and transparent engagement rather than the correction of individual claims (Razai et al., 2021).

Worked Example: Scoring the Vaccine Hesitancy Scale

Consider two parents who complete a nine-item Vaccine Hesitancy Scale. Seven items are confidence statements worded so that agreement signals confidence (for example, “Childhood vaccines are important for my child's health” and “I trust the information I receive about vaccines”), and two are risk statements worded so that agreement signals concern (for example, “I worry about serious adverse effects of vaccines”). Each item is answered on a five-point Likert scale from 1 (strongly disagree) to 5 (strongly agree). To produce a single hesitancy index on which higher means more hesitant, the seven confidence items are reverse-scored as `6 − response`, the two risk items are kept as-is, and the mean of all nine is taken. A clinic reads below 2.5 as accepting, 2.5 to 3.5 as moderate hesitancy, and above 3.5 as high hesitancy (Larson et al., 2015; Shapiro et al., 2018).

- Parent A answers the seven confidence items `2, 3, 2, 2, 3, 2, 3` (mostly disagreeing that they are confident) and the two risk items `5, 4` (agreeing they are worried). - Parent B answers the confidence items `5, 4, 5, 4, 5, 4, 5` (agreeing they are confident) and the risk items `2, 1` (disagreeing they are worried).

Parent A's reverse-scored confidence items are `6−2, 6−3, 6−2, 6−2, 6−3, 6−2, 6−3 = 4, 3, 4, 4, 3, 4, 3`, summing to 25; the risk items sum to `5 + 4 = 9`; the total is 34, and `34 ÷ 9 = 3.78`, above the 3.5 threshold, so the scale flags high hesitancy. Parent B's reverse-scored confidence items are `1, 2, 1, 2, 1, 2, 1`, summing to 10; the risk items sum to 3; the total is 13, and `13 ÷ 9 = 1.44`, below 2.5, so the scale reads acceptance. The two subscales are informative in themselves: Parent A's confidence subscale mean is `25 ÷ 7 = 3.57` and risk subscale mean `9 ÷ 2 = 4.50`, locating the hesitancy specifically in low confidence and high perceived risk, which is the diagnostic information an intervention needs (Betsch et al., 2018). The scale does not say Parent A is irrational or ignorant; it says, in a reproducible number, where on the continuum the parent sits and which antecedent is driving the position.

Discussion

The study of vaccination hesitancy is, at bottom, the story of turning an unhelpfully vague worry, that people are “losing faith” in vaccines, into a defined, measurable, and monitorable construct. The SAGE definition supplied the object, fixing hesitancy as a context-specific delay-or-refusal rather than a trait (MacDonald, 2015); the validated scales supplied the measurement, so that hesitancy could be a score with known reliability rather than an impression (Larson et al., 2015; Shapiro et al., 2018; Betsch et al., 2018); and the confidence-monitoring program supplied the population-level instrument, converting those scores into trends that warn of trouble before coverage falls (de Figueiredo et al., 2020).

That measurement program is what distinguishes hesitancy from its extreme pole, outright vaccination refusal. Refusal is a visible act; hesitancy is a latent position that only an instrument reveals, and most of the people whose uncertainty determines coverage never refuse anything. The reason the deficit model keeps failing is visible in the measurements themselves: what the scales load on is confidence and perceived risk, not knowledge, so a message aimed at knowledge addresses a determinant the data say is not the one at work (Nyhan et al., 2014; Brewer et al., 2017). The field's maturation has been to let the measurement dictate the response, diagnosing the specific antecedent before prescribing a remedy, rather than assuming every hesitant person is a missing fact away from acceptance.

Cognitive Implications

Vaccination hesitancy is a case study in the difference between measuring a attitude and explaining a behavior. The validated scales are, in cognitive terms, attempts to externalize a latent evaluative disposition, and their two-factor structure, confidence versus perceived risk, recovers a familiar pattern: an attitude is not a single quantity but a composite of belief-based and affect-based components that can pull in different directions (Shapiro et al., 2018). A respondent can rate vaccines effective and still feel the pull of perceived risk, and a measure that collapsed the two would miss the structure that predicts behavior.

The gap between attitude and behavior is the second implication. Ajzen's theory of planned behavior holds that behavior follows from an intention assembled out of attitude, subjective norm, and perceived control (Ajzen, 1991), and the hesitancy instruments can be read as measuring exactly those inputs, which is why a PACV score predicts a later vaccination decision (Opel et al., 2011). But the measurement program also exposes the limits of the rational-intention model: the backfire of corrective information shows that new evidence is filtered through a prior commitment rather than weighed against it, so an intention is not always revised by the facts that a rational-updater account predicts would revise it (Nyhan et al., 2014; Hornsey et al., 2018). The practical lesson for cognitive psychology is that a good measurement of an attitude is necessary but not sufficient to change the behavior it predicts, because the attitude is doing affective and identity work an argument does not touch.

Current Directions

The COVID-19 pandemic was the largest stress test the hesitancy-measurement program has faced, and it is reshaping the field in two ways. First, it moved confidence monitoring from a specialist activity to a real-time instrument of policy, as the cross-national surveys that de Figueiredo and colleagues built were turned on a novel vaccine in real time and used to track where confidence was falling fast enough to threaten a rollout (de Figueiredo et al., 2020). The open methodological problem is latency: a survey that reports a confidence decline weeks after it begins may report it too late, and the drive is toward faster, higher-frequency signals that can flag a problem closer to its onset.

Second, the pandemic sharpened the question of what hesitancy measurements should include now that the information environment is a determinant in its own right. Sahil Loomba and colleagues showed experimentally that brief exposure to online misinformation measurably lowered the intent to vaccinate, especially among the already-wavering, which implies that a static hesitancy score misses a dynamic that an exposure measure would capture (Loomba et al., 2021). The emerging consensus on response, articulated by Razai and colleagues, is that trust-centered, community-embedded engagement outperforms fact-correction, and the measurement agenda is following it toward instruments that capture trust and its erosion rather than knowledge alone (Razai et al., 2021). The field's center of gravity has shifted from asking how hesitant a population is to asking how fast its confidence is moving and why.

Common Misconceptions

Hesitancy is just a polite word for being anti-vaccine.

No. Hesitancy is a graded continuum from delay and selective acceptance through doubt; the committed anti-vaccine refuser is only its extreme pole, and the large, movable middle is where most coverage is won or lost (MacDonald, 2015; Larson et al., 2014).

Hesitancy is a fixed trait of certain people.

No. The SAGE definition makes it explicitly context-specific, varying across time, place, and vaccine; a person can be hesitant about one vaccine and confident about another, or confident one year and hesitant the next (MacDonald, 2015).

Hesitancy is too vague to measure.

Not any more. The Vaccine Hesitancy Scale, the 5C scale, and the PACV survey are validated instruments that convert the attitude into reliable scores, and those scores predict behavior and support cross-national monitoring (Larson et al., 2015; Shapiro et al., 2018; Betsch et al., 2018).

A high hesitancy score means a person lacks information.

Usually not. The scales load on confidence and perceived risk, not knowledge, so a high score typically signals low trust rather than a fact deficit, which is why information-only interventions so often fail (Nyhan et al., 2014; Brewer et al., 2017).

Glossary

3C model.

The SAGE Working Group's framework attributing hesitancy to confidence, complacency, and convenience.

5C model.

Betsch and colleagues' extension of the 3C model, adding calculation and collective responsibility, with a validated scale assigning a sub-score to each of the five psychological antecedents.

Attitude.

A favorable or unfavorable evaluation of an object or action; the hesitancy scales externalize a vaccination attitude as a composite of belief-based and affect-based components.

Calculation.

In the 5C model, the extent to which a person engages in extensive information-seeking and risk weighing before deciding; high calculation can amplify the effect of perceived risks.

Collective responsibility.

In the 5C model, the willingness to protect others by one's own vaccination, reflecting the social externality of immunity.

Complacency.

In the 3C/5C models, a low perceived risk of the vaccine-preventable disease, which makes vaccination seem unnecessary.

Confidence.

Trust in the safety and effectiveness of vaccines, in the providers who deliver them, and in the institutions that recommend them; the most consistent predictor of uptake.

Convenience.

In the 3C/5C models, the physical availability, affordability, and accessibility of vaccination services, which govern whether a willing person actually vaccinates.

Determinants matrix.

The SAGE Working Group's classification of hesitancy's influences into contextual, individual/social-group, and vaccine-specific categories.

Health Belief Model.

The classic account holding that preventive action depends on perceived susceptibility to and severity of a threat; the source of complacency as a determinant.

PACV.

The Parent Attitudes about Childhood Vaccines survey, a validated short instrument that predicts a parent's vaccination behavior from their expressed attitudes.

SAGE Working Group.

The World Health Organization's Strategic Advisory Group of Experts working group on vaccine hesitancy, which fixed the standard definition and the determinants matrix.

Vaccination refusal.

The declining of a recommended immunization; the extreme pole of the hesitancy continuum and, in MeSH, the descriptor beneath which hesitancy is indexed.

Vaccine confidence.

Trust in vaccines, providers, and the institutions that recommend them, monitored across countries and time as a leading indicator of uptake.

Vaccine Hesitancy Scale (VHS).

Larson and colleagues' short survey tool, validated by Shapiro and colleagues, that measures hesitancy as two factors: lack of confidence and perceived risks.

Key Researchers

Cornelia Betsch

(University of Erfurt). Health-communication psychologist who developed the 5C model of the psychological antecedents of vaccination and built and validated the scale that measures all five, giving the field an instrument that reports not only how hesitant a person is but which antecedent drives it. ORCID · Wikidata

Ève Dubé

(Université Laval / Institut national de santé publique du Québec). Medical anthropologist whose overview and review-of-reviews work mapped the contextual, individual, and vaccine-specific determinants of hesitancy and catalogued the interventions designed to address them. ORCID · Faculty

Heidi J. Larson

(London School of Hygiene & Tropical Medicine). Anthropologist who founded the Vaccine Confidence Project, led development of the Vaccine Hesitancy Scale, and whose 2014 systematic review established hesitancy as a context-specific continuum, setting the modern research agenda. ORCID · Wikipedia · Faculty

Noni E. MacDonald

(Dalhousie University). Pediatric infectious-disease physician who chaired the WHO SAGE Working Group whose 2015 paper gave the field its working definition of vaccine hesitancy and its matrix of determinants. Wikipedia · Wikidata

Gilla K. Shapiro

(Princess Margaret Cancer Centre / University of Toronto). Psychologist and clinician-scientist who led the formal psychometric validation of the Vaccine Hesitancy Scale, confirming its two-factor structure and establishing it as a usable research measure. ORCID · Google Scholar · Faculty

Frequently Asked Questions

What is vaccination hesitancy?

It is a delay in acceptance, or a refusal, of vaccination despite the availability of vaccination services. The WHO SAGE definition stresses that it is complex and context-specific, varying across time, place, and vaccine, rather than a fixed trait of a kind of person (MacDonald, 2015).

How is hesitancy different from refusal?

Refusal is the act of declining a vaccine; hesitancy is the broader, graded state of which refusal is only the extreme pole. Most hesitant people never refuse outright, and because they are numerous and movable, the hesitant middle governs coverage more than the refusing pole does (Larson et al., 2014).

Can vaccine hesitancy actually be measured?

Yes. Validated instruments turn the attitude into a reliable score: the Vaccine Hesitancy Scale measures it as two factors, lack of confidence and perceived risk; the 5C scale profiles five antecedents; and the PACV survey predicts a parent's behavior from their attitudes (Larson et al., 2015; Shapiro et al., 2018; Opel et al., 2011).

What is the SAGE determinants matrix?

It is the WHO SAGE Working Group's classification of the influences on hesitancy into three categories: contextual (the surrounding environment), individual and social-group (personal experience and beliefs), and vaccine-specific (features of the vaccine and its delivery). It is a diagnostic checklist, not a theory of mechanism (MacDonald, 2015).

What are the 3C and 5C antecedents?

The 3C model names confidence, complacency, and convenience. The 5C model adds calculation (deliberate risk weighing) and collective responsibility (willingness to protect others), with a validated scale measuring each. Confidence is consistently the strongest predictor of uptake (MacDonald, 2015; Betsch et al., 2018).

Why does vaccine confidence get monitored across countries?

Because a falling confidence trend is a leading indicator: a large modelling study across 149 countries showed that confidence is volatile and that declines tend to precede falls in uptake, so monitoring flags a coming coverage problem while there is still time to act (de Figueiredo et al., 2020).

Does giving hesitant people more facts reduce their hesitancy?

Often not. A randomized trial found standard pro-vaccine messages did not raise intent to vaccinate, and correcting the autism myth reduced it among the least favorable parents. Because hesitancy loads on trust and perceived risk rather than knowledge, information-only approaches frequently miss the determinant (Nyhan et al., 2014; Brewer et al., 2017).

What actually reduces hesitancy and raises uptake?

Interventions matched to the measured determinant, and especially those aimed at behavior rather than belief: reminders, convenient access, and presumptive provider recommendations, plus tailored, dialogue-based engagement from credible messengers (Brewer et al., 2017; Jarrett et al., 2015; Razai et al., 2021).

References