A paper in npj vaccines in September 2024 explores the significance of “biased processing” of information relating to vaccines in addressing vaccine hesitancy during the COVID-19 pandemic. Although vaccine hesitancy is influenced by various factors, the research suggests that “deliberate ignorance” was more closely associated with vaccine refusal than “typically investigated demographic variables”. This emphasises the importance of tailored information to meet individual information-processing needs.
Hesitancy and interventions
Vaccine hesitancy, “the reluctance or refusal to get vaccinated despite the availability of vaccines”, was recognised as a top ten global health threat by WHO in 2019. The authors acknowledge that it is a “complex phenomenon”, determined by “historical, political, and socio-cultural factors” as well as “individual knowledge and risk perception”.
Recent research has highlighted the significance of concerns about side effects and effectiveness for vaccine hesitancy. Therefore, many responses to vaccine hesitancy use information on vaccine evidence, including possible harms or potential benefits. However, there is evidence that “transparent communication of the evidence” does not influence vaccination intentions and neglects factors such as experiences of racism or mistreatment by medical professionals or distrust of the pharmaceutical industry.
The study
The authors consider “how (if at all) people use information about vaccine evidence”. They combine theoretical and analytical ideas with methodological tools from cognitive and behavioural science to investigate how individuals with different attitudes towards COVID-19 vaccines process vaccine evidence information. They characterise and measure how people process “commonly provided information about vaccine evidence” and compare the influence of “extraneous factors” on vaccination decisions.
With a process-tracing methodology and computational modelling, the authors examine the extent to which people may engage in “deliberate ignorance”. In this context, deliberate ignorance signifies “choosing not to inspect a piece of information on a vaccine’s side effects, benefits, and their probabilities in the pre-decision phase”. Deliberate ignorance is distinguished through three levels:
- Full deliberate ignorance – people abstain from inspecting any information on vaccine evidence; their decisions may be based on other factors instead.
- Partial deliberate ignorance – people ignore some, but not all, vaccine evidence information. The research focusses on ‘probability neglect’, in which a vaccination outcome is inspected, but its probability is not.
- No deliberate ignorance – people inspect all information on vaccine evidence and consider it in their decision. However, even in this category this information may be processed in a distorted fashion, deviating from a “rational way” of processing information.
As vaccination decisions are conceptualised as “instances of risky choice”, the authors use a framework that considers two types of cognitive distortions: (nonlinear) probability weighting and loss aversion:
- Probability weighting – people make risky decisions as if they processed probabilities nonlinearly, with low and high probabilities being over- and underweighted.
- Loss aversion – people make risky decisions as if the psychological impact of losses is greater than that of gains.
The online study involved 1200 United States citizens who self-reported as having anti- (365), neutral (373), or pro- (462) COVID-19 vaccine attitudes. Participants made a series of decisions concerning their willingness to get vaccinated with each of eight internationally licensed COVID-19 vaccines. For each of these vaccines, participants could choose to inspect information on vaccine evidence, including side effects and benefits. Information inspection behaviour was recorded through Mouselab, a process-tracing tool.
Findings
61.9% of participants in the anti-vaccination group, 11.7% of participants in the neutral group, and 0.4% of participants in the pro-vaccination group refused all eight vaccines. On average, participants accepted one (anti-vaccination), three (neutral), and five (pro-vaccination) of the eight vaccines. Notably, the non-zero acceptance rate in the anti-vaccine group was influenced by almost 30% of participants indicating willingness to accept the Bharat Biotech vaccine.
The strongest predictors of vaccine acceptance were vaccination attitude, the number of COVID-19 vaccinations a participant had received by the time of the study, and vaccine brand. Political orientation and education level were related to vaccination decisions in raw data, but these relationships “vanished” in the full statistical model.
Statistical models were used to investigate the relationship between deliberate ignorance of vaccine evidence and vaccination decisions. Mouselab data facilitated an analysis of participants’ information inspection behaviour. This found that anti-vaccination, neutral, and pro-vaccination groups exhibited full deliberate ignorance in 18%, 9%, and 7% of decisions respectively. The level of deliberate ignorance was “strongly related” to vaccination decisions; probability of vaccine refusal was highest when participants exhibited full deliberate ignorance and lowest when they exhibited no deliberate ignorance.
“In the anti-vaccine group, full deliberate ignorance was almost always followed by vaccine refusal; in the pro-vaccination group, by contrast, full deliberate ignorance was associated with a higher probability of vaccine acceptance than partial deliberate ignorance.”
Probability neglect was defined as cases in which there was at least one instance where a participant inspected an outcome but not its probability. Participants in the anti-vaccination, neutral, and pro-vaccination groups exhibited probability neglect for side effects in 15%, 13%, and 9% of vaccination decisions respectively. For benefits, they exhibited this in 8%, 6%, and 4% of decisions respectively.
To account for the possibility that the effect of probability neglect on vaccination decisions depended on side effect severity, the authors distinguished whether the probability neglect occurred for an “extreme, severe, or mild” side effect or for a benefit. This revealed that vaccine refusal was “much more likely” in trials where the probability of an extreme side effect was neglected, and vaccine refusal was “much less likely” in trials where the probability of a mild side effect was neglected.
“How participants inspected and ignored information about vaccine evidence seemed to be a key predictor of their decision to get vaccinated with given vaccine or not.”
The authors then used computational modelling to explore cognitive distortions in the processing of the inspected vaccine evidence and its effect on vaccination decisions. For quantitative measures of each participant’s subjective valuation of a vaccine’s possible outcomes, participants were asked to rate the emotion they would feel due to each effect. The anti-vaccination group gave the most negative affect ratings for side effects and the least positive affect ratings for benefits; the pro-vaccination group gave the least negative affect ratings for side effects and the most positive affect ratings for benefits.
To investigate how vaccination decisions were driven by individual decision biases, vaccine-specific effects, and subject distortion of vaccine evidence, the authors developed a computational model to capture paths to a decision. This identified a decision bias in most of the anti-vaccination group to refuse the vaccine; this was strong that the effects of the vaccine’s properties and valuations “rarely pushed the probability of acceptance” higher than 50%.
Neutral group participants showed a “weak a priori propensity to refuse a vaccine” but vaccination decisions were driven by vaccine-specific effects and consideration of vaccine evidence information. Most participants in the pro-vaccination group showed a bias towards vaccine acceptance, but this was not as pronounced as the refusal bias in the anti-vaccination group.
“The subjective valuations of the vaccine’s effectiveness, side effects, and probabilities drove the vaccination decisions, particularly among the neutral and pro-vaccination participants.”
Implications for vaccine interventions
An insight that the authors highlight is the importance of “tailoring interventions” to specific target groups. If a person is asked to self-assess their general vaccination attitude, the content and format of vaccine information could be adjusted. Although the deliberate ignorance of vaccine evidence among the anti-vaccination group is a “practical barrier to the approach of risk communication that is meant to inform but not persuade”, health communicators and health authorities should not abandon their goal of informing.
“Risk evidence communicators need to be realistic about their expectations. It also means that they must consider other aspects of their efforts, such as the relationship between the communicator and the audience.”
Communicators must also explicitly address the “major concerns” of vaccine sceptical people, such as “what science does not know”; this must be communicated in “understandable, nontechnical, and transparent language”. However, once trust is re-established among people with anti-vaccination attitudes, the “tendency to close one’s eyes to probabilities” presents a challenge. Targeted interventions that address this “disregard of probabilities” would be useful. For example, interactive simulations could be used to convey vaccine evidence, imitating the “sequential and experiential mechanisms by which people naturally encounter risk information”.
The observed side-effect aversion in all groups might lead communicators to avoid disclosing side effects. This could initially decrease vaccine hesitancy but “at a huge cost”. Full transparency is critical for maintaining trust. Again, interactive simulations might be a solution, or targeting the strong negative emotions associated with side effects.
Concluding that behavioural scientists have the “task” of understanding the reasons for vaccine refusal, the authors emphasise the need for effective evidence communication to take “new and innovative paths”.
“Societies can be fully prepared for future pandemics only when technological ingenuity is coupled with cognitive and behavioural insights.”
If the ideas explored in the study are of interest to you, you could participate in the pre-Congress Vaccine Equity Workshop in Barcelona next month; you will hear from experts on the importance of effective vaccine outreach and how social and cultural factors contribute to vaccine inequities. Don’t forget to subscribe to our weekly newsletters here for the latest vaccine news.
