Do unbiased people act more rationally?—The more unbiased people assessed their own risk of COVID-19 compared to that of others, the less willing they were to be vaccinated

Do unbiased people act more rationally?—The case of comparative realism and vaccine intention. Kamil Izydorczak, Dariusz Dolinski, Oliver Genschow, Wojciech Kulesza, Pawel Muniak, Bruno Gabriel Salvador Casara and Caterina Suitner. Royal Society Open Science, February 1 2023. https://doi.org/10.1098/rsos.220775

Abstract: Within different populations and at various stages of the pandemic, it has been demonstrated that individuals believe they are less likely to become infected than their average peer. This is known as comparative optimism and it has been one of the reproducible effects in social psychology. However, in previous and even the most recent studies, researchers often neglected to consider unbiased individuals and inspect the differences between biased and unbiased individuals. In a mini meta-analysis of six studies (Study 1), we discovered that unbiased individuals have lower vaccine intention than biased ones. In two pre-registered, follow-up studies, we aimed at testing the reproducibility of this phenomenon and its explanations. In Study 2 we replicated the main effect and found no evidence for differences in psychological control between biased and unbiased groups. In Study 3 we also replicated the effect and found that realists hold more centric views on the trade-offs between threats from getting vaccinated and getting ill. We discuss the interpretation and implication of our results in the context of the academic and lay-persons' views on rationality. We also put forward empirical and theoretical arguments for considering unbiased individuals as a separate phenomenon in the domain of self–others comparisons.

5. General discussion

Comparative optimism is a robust phenomenon. The bias proved to be present inter-contextually [46], and since the first theoretical works in the 1980s, it is still considered a replicable and practically significant effect. Furthermore, the bias has been successfully discovered by multiple research teams in many settings during the COVID-19 pandemic [49–51]. But do social psychologists have a firm understanding of why this bias occurs and its consequences?

As with many other collective irrationalities, we can too often be taken in by the ‘rational = desirable’ narrative. In such a narrative we implicitly or explicitly assume that the most desirable state would be ‘unbiased’, and, if the examined population fails to adhere to this pattern, we conclude that the cognitive processes we examine are somewhat ‘flawed’. In the presented studies, we concluded that those who are ‘unbiased’ more often abstain from taking one of the most (if not the most) effective, evidence based and affordable actions that could protect them from deadly threat. A seemingly ‘rational’ mental approach to the issue of COVID-19 contraction is related to a more irrational response to that threat—namely not getting vaccinated.

In the mini meta-analysis and two pre-registered studies, we discovered that those who express either comparative pessimism or optimism have a higher intention to get vaccinated for COVID-19 than those who are unbiased. The relationship of comparative pessimism to pro-health behaviour seems more intuitive, and the positive relationship of comparative optimism comes as a surprise, but our discovery is not isolated in that regard [52].

In Study 2, we found no evidence of a relationship between psychological control and comparative optimism with vaccine intention.

In Study 3 we found a common denominator of people who are realists and who have a lower vaccine intention. It turned out that both phenomena are related to lower COVID-19 ThreatDifference (ThreatDisease − ThreatVaccine). Furthermore, in line with the extended protection motivation theory (PMT [47,48]), the trade-off between risks of the disease and risks of the vaccine proved to predict being unbiased, and this relationship is partly mediated by vaccine intention.

Our studies present evidence that counters the ‘rational = desirable’ narrative, but that could lead into another trap: assuming that it is irrationalities and biases that help us cope more effectively. We think that such a narrative can be an equally false over-simplification and our studies offer more compelling explanations.

Collective irrationalities, such as comparative optimism may neither enhance nor hamper our coping abilities. They may, in turn, be a by-product of ongoing coping processes, possibly leading to greater protection (in the case of our studies, vaccination against COVID-19). From the perspective of our studies, it is clear that we might wrongfully ascribe a causal role to these biases.

While one might think that comparative optimism may cause reckless behaviour, such as refusal to vaccinate, Study 3 suggests another plausible alternative mechanism: ThreatDifference might be the reason for stronger or weaker vaccine intention (along with many other factors; see [43,53]) and comparative optimism might be a result of knowing one's own efforts, such as vaccination. In fact, a recent experimental study [52] provides evidence that being more aware of one's own self-protective effort enhances comparative optimism.

It is also noteworthy that comparative biases may arise in part from a lack of information about the comparative target, and that providing people with information about the comparative target diminishes the bias [54]. Accordingly, the comparative optimists in our study may have lacked information about the preventive behaviour of others.

The case of the relationship between comparative optimism and constructive pro-health behaviour is complex. On the one hand, we have evidence for both the benefits and drawbacks of CO [55]. On the other hand, CO may be the result rather than the cause of pro-health behaviour. Clearly there are many contextual factors involved and we should discard the overly simplistic view of an inherently beneficial or inherently harmful nature of comparative optimism (which also might be the case for many other collective irrationalities).

Our paper presents a pre-registered and high-powered line of research, which addresses differences between comparative optimists and the ‘unbiased’—a category of individuals that has most often been either left undiscussed or barely mentioned in previous studies regarding CO. Examining the bias from the perspective of the unbiased and using a mixed method approach that combined theory-driven hypotheses with a bottom-up strategy, thus giving a voice to participants, offered the opportunity to enrich theoretical knowledge on comparative bias and led to the surprising discovery that being unbiased can be related to a less pro-health attitude.

5.1. Limitations and future directions

The main limitation of our study is the lack of behavioural measures. This was a result of an early stage of our research project, which took place before COVID-19 vaccines were available. For that reason, we gathered data only about vaccine intention. In follow-up studies the vaccines were available but we decided to examine the intention of the yet unvaccinated to ensure the direct comparability of follow-up studies with the studies from a mini meta-analysis. This limitation leads to another one—at the time of Study 2 and especially Study 3, the number of unvaccinated was shrinking and we can expect that they might differ from the general population in many ways (for example, from study to study, we observed the diminishing share of ‘realists’). This constitutes a limit for the generalization of our conclusions.

The future direction of research regarding the differences between unbiased and comparative optimists should concentrate on actual behaviours rather than intentions or declarations. Moreover, future studies should enhance the scope of generalization by investigating more representative samples.

Another limitation is the possibility of an alternative explanation of our results. We interpret the results of Study 3 in the light of the extended PMT theory, assuming that the relationship between predicted outcomes of falling ill and getting vaccinated leads to engagement or disengagement with vaccination, which it turn results in them feeling superior (comparatively optimistic) or similar (comparatively realistic) to others.

But an alternative is probable. Following Gigerenzer's theory of ‘fast and frugal heuristics' [56], people can often make more ecologically valid decisions when they follow heuristics, without engaging in deep, analytical processes.

Perhaps people who chose the ecologically rational option to take the vaccine did so because they followed their intuition/shortcuts when making the decision. By doing so, they estimated the trade-offs between the disease and vaccine in line with the mainstream message (media, experts and authorities). If these individuals followed intuition in this respect, they may also be more prone to the default bias, namely optimistic bias. On the other hand, people who engage in processing the information more reflectively might end up being more sceptical towards vaccination and also less prone to the optimistic bias.

These alternative explanations could be empirically tested—if pro-vaccine attitudes could be ascribed to using more ‘fast and frugal heuristics’, people more sceptical of the vaccines should be able to recall more information about vaccines (regardless of their epistemic status) and provide more elaborate explanations for their stance.

As a general direction for future research on comparative biases, we advocate for considering a categorical approach to measuring biases—individuals who do not exhibit a bias should be treated as a separate category, especially when empirical results would indicate a substantial inflation of scores signalling a lack of bias (a similar inflation has been identified in the case of dehumanization—see [57], p. 12). Alternatively, if one decides to treat comparative bias as a continuous scale, a nonlinear relationship should be investigated. If comparative biases can have two directions, it is reasonable to expect that different directions might have different correlations.