Explicit discrimination and ingroup favoritism, but no implicit biases in hypothetical triage decisions during COVID-19.

Disturbingly realistic triage scenarios during the COVID-19 pandemic provide an opportunity for studying discrimination in moral reasoning. Biases and favoritism do not need to be explicit and overt, but can remain implicit and covert. In addition to assessing laypeople's propensity for engaging in overt discrimination, the present study examines whether they reveal implicit biases through seemingly fair random allocations. We present a cross-sectional online study comprising 6 timepoints and a total of 2296 participants. Each individual evaluated 19 hypothetical scenarios that provide an allocation dilemma between two patients who are in need of ventilation and differ only in one focal feature. Participants could either allocate the last ventilator to a patient, or opt for random allocation to express impartiality. Overall, participants exhibited clear biases for the patient who was expected to be favored based on health factors, previous ethical or caretaking behaviors, and in-group favoritism. If one patient had been pre-allocated care, a higher probability of keeping the ventilator for the favored patient indicates persistent favoritism. Surprisingly, the absence of an asymmetry in random allocations indicates the absence of covert discrimination. Our results demonstrate that laypeople's hypothetical triage decisions discriminate overtly and show explicit biases.

How to convince the vaccine-hesitant? An ease-of-access nudge, but not risk-related information increased Covid vaccination-related behaviors in the unvaccinated.

In this study, we contrast how different benefit and harm information formats and the presence or absence of an ease-of-access nudge may facilitate COVID vaccination uptake for a sample of 620 unvaccinated Dutch adults at a timepoint when the vaccine had been widely available for more than a month. Using a 2 × 2 between-subjects factorial design, we varied the information format on mRNA COVID vaccination statistics (generic text vs. facts box) and an affirmative nudge emphasizing the ease of making a vaccination appointment (absent vs. present). We assessed the acceptance of the vaccination information provided, perceptions on the vaccination, and whether participants directly visited a COVID vaccination appointment website. Whereas the facts box did not significantly affect participants' information acceptance, vaccination attitudes, intentions, and link clicking, the affirmative nudge alongside an online link systematically increased the likelihood of clicking on the link to make a vaccination appointment. A verbal nudge emphasizing the ease of vaccine accessibility is more likely to increase vaccination uptake in an unvaccinated population than informational campaigns on vaccine effectiveness.

The Lure of Beauty: People Select Representations of Statistical Information Largely Based on Attractiveness, Not Comprehensibility.

OBJECTIVE: People differ in whether they understand graphical or numerical representations of statistical information better. However, assessing these skills is often not feasible when deciding which representation to select or use. This study investigates whether people choose the representation they understand better, whether this choice can improve risk comprehension, and whether results are influenced by participants' skills (graph literacy and numeracy). METHODS: In an experiment, 160 participants received information about the benefits and side effects of painkillers using either a numerical or a graphical representation. In the "no choice" condition, the representation was randomly assigned to each participant. In the "choice" condition, participants could select the representation they would like to receive. The study assessed gist and verbatim knowledge (immediate comprehension and recall), accessibility of the information, attractiveness of the representation, as well as graph literacy and numeracy. RESULTS: In the "choice" condition, most (62.5%) chose the graphical format, yet there was no difference in graph literacy or numeracy (nor age or gender) between people who chose the graphical or the numerical format. Whereas choice slightly increased verbatim knowledge, it did not improve gist or overall knowledge compared with random assignment. However, participants who chose a representation rated the representation as more attractive, and those who chose graphs rated them as more accessible than those without a choice. LIMITATIONS: The sample consisted of highly educated undergraduate students with higher graph literacy than the general population. The task was inconsequential for participants in terms of their health. CONCLUSIONS: When people can choose between representations, they fail to identify what they comprehend better but largely base that choice on how attractive the representation is for them. HIGHLIGHTS: People differ systematically in whether they understand graphical or numerical representations of statistical information better. However, assessing these underlying skills to get the right representation to the right people is not feasible in practice. A simple and efficient method to achieve this could be to let people choose among representations themselves.However, our study showed that allowing participants to choose a representation (numerical v. graphical) did not improve overall or gist knowledge compared with determining the representation randomly, even though it did slightly improve verbatim knowledge.Rather, participants largely chose the representation they found more attractive. Most preferred the graphical representation, including those with low graph literacy.It would therefore be important to develop graphical representations that are not only attractive but also comprehensible even for people with low graph literacy.

How Do People Process Different Representations of Statistical Information? Insights into Cognitive Effort, Representational Inconsistencies, and Individual Differences.

BACKGROUND: Graphical representation formats (e.g., icon arrays) have been shown to lead to better understanding of the benefits and risks of treatments compared to numbers. We investigate the cognitive processes underlying the effects of format on understanding: how much cognitive effort is required to process numerical and graphical representations, how people process inconsistent representations, and how numeracy and graph literacy affect information processing. METHODS: In a preregistered between-participants experiment, 665 participants answered questions about the relative frequencies of benefits and side effects of 6 medications. First, we manipulated whether the medical information was represented numerically, graphically (as icon arrays), or inconsistently (numerically for 3 medications and graphically for the other 3). Second, to examine cognitive effort, we manipulated whether there was time pressure or not. In an additional intervention condition, participants translated graphical information into numerical information before answering questions. We also assessed numeracy and graph literacy. RESULTS: Processing icon arrays was more strongly affected by time pressure than processing numbers, suggesting that graphical formats required more cognitive effort. Understanding was lower when information was represented inconsistently (v. consistently) but not if there was a preceding intervention. Decisions based on inconsistent representations were biased toward graphically represented options. People with higher numeracy processed quantitative information more efficiently than people with lower numeracy did. Graph literacy was not related to processing efficiency. LIMITATIONS: Our study was conducted with a nonpatient sample, and the medical information was hypothetical. CONCLUSIONS: Although graphical (v. numerical) formats have previously been found to lead to better understanding, they may require more cognitive effort. Therefore, the goal of risk communication may play an important role when choosing how to communicate medical information. HIGHLIGHTS: This article investigates the cognitive processes underlying the effects of representation format on the understanding of statistical information and individual differences therein.Processing icon arrays required more cognitive effort than processing numbers did.When information was represented inconsistently (i.e., partly numerically and partly graphically), understanding was lower than with consistent representation, and decisions were biased toward the graphically represented options.People with higher numeracy processed quantitative information more efficiently than people with lower numeracy did.

Risk-Adjusted Cancer Screening and Prevention (RiskAP): Complementing Screening for Early Disease Detection by a Learning Screening Based on Risk Factors.

Background: Risk-adjusted cancer screening and prevention is a promising and continuously emerging option for improving cancer prevention. It is driven by increasing knowledge of risk factors and the ability to determine them for individual risk prediction. However, there is a knowledge gap between evidence of increased risk and evidence of the effectiveness and efficiency of clinical preventive interventions based on increased risk. This gap is, in particular, aggravated by the extensive availability of genetic risk factor diagnostics, since the question of appropriate preventive measures immediately arises when an increased risk is identified. However, collecting proof of effective preventive measures, ideally by prospective randomized preventive studies, typically requires very long periods of time, while the knowledge about an increased risk immediately creates a high demand for action. Summary: Therefore, we propose a risk-adjusted prevention concept that is based on the best current evidence making needed and appropriate preventive measures available, and which is constantly evaluated through outcome evaluation, and continuously improved based on these results. We further discuss the structural and procedural requirements as well as legal and socioeconomical aspects relevant for the implementation of this concept.

Conformist social learning leads to self-organised prevention against adverse bias in risky decision making.

Given the ubiquity of potentially adverse behavioural bias owing to myopic trial-and-error learning, it seems paradoxical that improvements in decision-making performance through conformist social learning, a process widely considered to be bias amplification, still prevail in animal collective behaviour. Here we show, through model analyses and large-scale interactive behavioural experiments with 585 human subjects, that conformist influence can indeed promote favourable risk taking in repeated experience-based decision making, even though many individuals are systematically biased towards adverse risk aversion. Although strong positive feedback conferred by copying the majority's behaviour could result in unfavourable informational cascades, our differential equation model of collective behavioural dynamics identified a key role for increasing exploration by negative feedback arising when a weak minority influence undermines the inherent behavioural bias. This 'collective behavioural rescue', emerging through coordination of positive and negative feedback, highlights a benefit of collective learning in a broader range of environmental conditions than previously assumed and resolves the ostensible paradox of adaptive collective behavioural flexibility under conformist influences.

When it comes to making decisions, like choosing a restaurant or political candidate, most of us rely on limited information that is not accurate enough to find the best option. Considering others' decisions and opinions can help us make smarter choices, a phenomenon called "collective intelligence". Collective intelligence relies on individuals making unbiased decisions. If individuals are biased toward making poor choices over better ones, copying the group's behavior may exaggerate biases. Humans are persistently biased. To avoid repeated failure, humans tend to avoid risky behavior. Instead, they often choose safer alternatives even when there might be a greater long-term benefit to risk-taking. This may hamper collective intelligence. Toyokawa and Gaissmaier show that learning from others helps humans make better decisions even when most people are biased toward risk aversion. The experiments first used computer modeling to assess the effect of individual bias on collective intelligence. Then, Toyokawa and Gaissmaier conducted an online investigation in which 185 people performed a task that involved choosing a safer or risker alternative, and 400 people completed the same task in groups of 2 to 8. The online experiment showed that participating in a group changed the learning dynamics to make information sampling less biased over time. This mitigated people's tendency to be risk-averse when risk-taking is beneficial. The model and experiments help explain why humans have evolved to learn through social interactions. Social learning and the tendency of humans to conform to the group's behavior mitigates individual risk aversion. Studies of the effect of bias on individual decision-making in other circumstances are needed. For example, would the same finding hold in the context of social media, which allows individuals to share unprecedented amounts of sometimes incorrect information?

How do we raise media bias awareness effectively? Effects of visualizations to communicate bias.

Media bias has a substantial impact on individual and collective perception of news. Effective communication that may counteract its potential negative effects still needs to be developed. In this article, we analyze how to facilitate the detection of media bias with visual and textual aids in the form of (a) a forewarning message, (b) text annotations, and (c) political classifiers. In an online experiment, we randomized 985 participants to receive a biased liberal or conservative news article in any combination of the three aids. Meanwhile, their subjective perception of media bias in this article, attitude change, and political ideology were assessed. Both the forewarning message and the annotations increased media bias awareness, whereas the political classification showed no effect. Incongruence between an articles' political position and individual political orientation also increased media bias awareness. Visual aids did not mitigate this effect. Likewise, attitudes remained unaltered.

The four weeks before lockdown during the COVID-19 pandemic in Germany: a weekly serial cross-sectional survey on risk perceptions, knowledge, public trust and behaviour, 3 to 25 March 2020.

BackgroundDuring the COVID-19 pandemic, public perceptions and behaviours have had to adapt rapidly to new risk scenarios and radical behavioural restrictions.AimTo identify major drivers of acceptance of protective behaviours during the 4-week transition from virtually no COVID-19 cases to the nationwide lockdown in Germany (3-25 March 2020).MethodsA serial cross-sectional online survey was administered weekly to ca 1,000 unique individuals for four data collection rounds in March 2020 using non-probability quota samples, representative of the German adult population between 18 and 74 years in terms of age × sex and federal state (n = 3,910). Acceptance of restrictions was regressed on sociodemographic variables, time and psychological variables, e.g. trust, risk perceptions, self-efficacy. Extraction of homogenous clusters was based on knowledge and behaviour.ResultsAcceptance of restrictive policies increased with participants' age and employment in the healthcare sector; cognitive and particularly affective risk perceptions were further significant predictors. Acceptance increased over time, as trust in institutions became more relevant and trust in media became less relevant. The cluster analysis further indicated that having a higher education increased the gap between knowledge and behaviour. Trust in institutions was related to conversion of knowledge into action.ConclusionIdentifying relevant principles that increase acceptance will remain crucial to the development of strategies that help adjust behaviour to control the pandemic, possibly for years to come. Based on our findings, we provide operational recommendations for health authorities regarding data collection, health communication and outreach.

Determinants of information diffusion in online communication on vaccination: The benefits of visual displays.

OBJECTIVE: Social media are an increasingly important source of information on the benefits and risks of vaccinations, but the high prevalence of misinformation provides challenges for informed vaccination decisions. It is therefore important to understand which messages are likely to diffuse online and why, and how relevant aspects-such as scientific facts on vaccination effectiveness-can be made more comprehensible and more likely to be shared. In two studies, we (i) explore which characteristics of messages on flu vaccination facilitate their diffusion in online communication, and (ii) whether visual displays (i.e., icon arrays) facilitate the comprehension and diffusion of scientific effectiveness information. METHODS: In Study 1, 208 participants each rated a random sample of 15 out of 63 messages on comprehensibility, trustworthiness, persuasiveness, familiarity, informativeness, valence, and arousal, and then reported which information they would share with subsequent study participants. In Study 2 (N = 758), we employed the same rating procedure for a selected set of 9 messages and experimentally manipulated how scientific effectiveness information was displayed. RESULTS: Study 1 illustrated that scientific effectiveness information was difficult to understand and thus did not diffuse well. Study 2 demonstrated that visual displays improved the understanding of this information, which could, in turn, increase its social impact. CONCLUSIONS: The comprehensibility of scientific information is an important prerequisite for its diffusion. Visual displays can facilitate informed vaccination decisions by rendering important information on vaccination effectiveness more transparent and increasing the willingness to share this information.

Influence of stress on physiological synchrony in a stressful versus non-stressful group setting.

Physiological synchrony (PS) is defined as the co-occurrence and interdependence of physiological activity between interaction partners. Previous research has uncovered numerous influences on the extent of PS, such as relationship type or individual characteristics. Here, we investigate the influence of acute stress on PS. We do so in a setting in which PS was not promoted, but contact between group members was explicitly minimized. We reanalyzed cortisol, alpha-amylase, and subjective stress data from 138 participants (mean age = [Formula: see text], 47.1% female) who previously underwent the Trier Social Stress Test for groups (TSST-G) or a non-stressful control task together, collected as part of a larger project by Popovic et al. (Sci Rep 10: 7845, 2020). Using a stability and influence model, an established method to test for synchrony, we tested whether individuals' cortisol and alpha-amylase concentrations could be predicted by group members' levels. We found cortisol PS in participants who were in the same group, the extent of which was stronger in the non-stressful control condition. For alpha-amylase, participants were synchronized as well; furthermore, there was an interaction between previous stress levels and PS. This suggests that while synchrony of both stress markers can occur in group settings even with spurious interaction, stressor exposure might attenuate its extent. We argue that if PS occurs in a sample where interaction was minimal, the phenomenon might be more widespread than previously thought. Furthermore, stressor exposure might influence whether a situation allows for PS. We conclude that PS should be investigated within group settings with various degrees of social interaction to further expose mechanisms of and influence on PS.

Contagious Health Risk and Precautionary Social Distancing.

BACKGROUND: Since humans are social animals, social relations are incredibly important. However, in cases of contagious diseases such as the flu, social contacts also pose a health risk. According to prominent health behavior change theories, perceiving a risk for one's health motivates precautionary behaviors. The "behavioral immune system" approach suggests that social distancing might be triggered as a precautionary, evolutionarily learned behavior to prevent transmitting contagious diseases through social contact. This study examines the link between personal risk perception for an infectious disease and precautionary behavior for disease-prevention in the context of social relationships. METHODS: At 2-week intervals during the first semester, 100 Psychology freshmen indicated their flu risk perception, whether they had been ill during the previous week, and their friendships within their freshmen network for eight time points. RESULTS: Social network analysis revealed that participants who reported a high flu risk perception listed fewer friends (B = -0.10, OR = 0.91, p = 0.026), and were more likely to be ill at the next measuring point (B = 0.26, OR = 1.30, p = 0.005). Incoming friendship nominations increased the likelihood of illness (B = 0.14, OR = 1.15, p = 0.008), while the reduced number of friendship nominations only marginally decreased this likelihood (B = -0.07, OR = 0.93, p = 0.052). CONCLUSION: In accordance with the concept of a "behavioral immune system," participants with high flu risk perception displayed a social precautionary distancing even when in an environment, in which the behavior was ineffective to prevent an illness.

Current Challenges When Using Numbers in Patient Decision Aids: Advanced Concepts.

BACKGROUND: Decision aid developers have to convey complex task-specific numeric information in a way that minimizes bias and promotes understanding of the options available within a particular decision. Whereas our companion paper summarizes fundamental issues, this article focuses on more complex, task-specific aspects of presenting numeric information in patient decision aids. METHODS: As part of the International Patient Decision Aids Standards third evidence update, we gathered an expert panel of 9 international experts who revised and expanded the topics covered in the 2013 review working in groups of 2 to 3 to update the evidence, based on their expertise and targeted searches of the literature. The full panel then reviewed and provided additional revisions, reaching consensus on the final version. RESULTS: Five of the 10 topics addressed more complex task-specific issues. We found strong evidence for using independent event rates and/or incremental absolute risk differences for the effect size of test and screening outcomes. Simple visual formats can help to reduce common judgment biases and enhance comprehension but can be misleading if not well designed. Graph literacy can moderate the effectiveness of visual formats and hence should be considered in tool design. There is less evidence supporting the inclusion of personalized and interactive risk estimates. DISCUSSION: More complex numeric information. such as the size of the benefits and harms for decision options, can be better understood by using incremental absolute risk differences alongside well-designed visual formats that consider the graph literacy of the intended audience. More research is needed into when and how to use personalized and/or interactive risk estimates because their complexity and accessibility may affect their feasibility in clinical practice.

Current Best Practice for Presenting Probabilities in Patient Decision Aids: Fundamental Principles.

BACKGROUND: Shared decision making requires evidence to be conveyed to the patient in a way they can easily understand and compare. Patient decision aids facilitate this process. This article reviews the current evidence for how to present numerical probabilities within patient decision aids. METHODS: Following the 2013 review method, we assembled a group of 9 international experts on risk communication across Australia, Germany, the Netherlands, the United Kingdom, and the United States. We expanded the topics covered in the first review to reflect emerging areas of research. Groups of 2 to 3 authors reviewed the relevant literature based on their expertise and wrote each section before review by the full authorship team. RESULTS: Of 10 topics identified, we present 5 fundamental issues in this article. Although some topics resulted in clear guidance (presenting the chance an event will occur, addressing numerical skills), other topics (context/evaluative labels, conveying uncertainty, risk over time) continue to have evolving knowledge bases. We recommend presenting numbers over a set time period with a clear denominator, using consistent formats between outcomes and interventions to enable unbiased comparisons, and interpreting the numbers for the reader to meet the needs of varying numeracy. DISCUSSION: Understanding how different numerical formats can bias risk perception will help decision aid developers communicate risks in a balanced, comprehensible manner and avoid accidental "nudging" toward a particular option. Decisions between probability formats need to consider the available evidence and user skills. The review may be useful for other areas of science communication in which unbiased presentation of probabilities is important.

Maximizing as satisficing: On pattern matching and probability maximizing in groups and individuals.

Distinguishing meaningful structure from unpredictable randomness is a key challenge in many domains of life. We examined whether collaborating three-person groups (n = 81) outperform individuals (n = 81) in facing this challenge with a two-part repeated choice task, where outcomes were either serially independent (probabilistic part) or fixed in a particular sequence (pattern part). Groups performed as well as the best individuals in the probabilistic part but groups' accuracy did not credibly exceed that of the average individual in the pattern part. Qualitative coding of group discussion data revealed that failures to identify existing patterns were related to groups accepting probability maximizing as a "good enough" strategy rather than expending effort to search for patterns. These results suggest that probability maximizing can arise via two routes: recognizing that probabilistic processes cannot be outdone (maximizing as optimizing) or settling for an imperfect but easily implementable strategy (maximizing as satisficing).

Acute Stress Reduces the Social Amplification of Risk Perception.

Risk perceptions typically underlie a complex social dynamic: Risk-related information is transmitted between individuals, this information influences risk perceptions, and risk perceptions influence which information is transmitted. This can lead to a social amplification of risk. We test how stress, a widespread affective state, influences the social dynamics of risk perception. Participants (N = 146) read articles about the controversial antibacterial agent Triclosan and were then asked to inform another person about Triclosan. Before and after reading the articles, participants reported their concern about Triclosan. Stress exposure before the task led to a smaller increase in concern in response to the articles. The stronger the increase in cortisol, the smaller the increase in concern. Furthermore, participants in the stress group transmitted less negative information about Triclosan to others. In contrast, participants' subjective feelings of stress were associated with higher concern and more alarming risk communication. We conclude that feeling stressed can amplify risk perception, whereas the endocrine stress reaction can attenuate risk perception when information about risk is exchanged in a social context.

"I was seen by a radiologist, but unfortunately I can't remember the name and I still have questions. What should I do?" Radiologists should give thoughts to improve service professionalism and patient esteem.

BACKGROUND: The aim of the study is to investigate how well patients remember the radiologist's name after a radiological examination, and whether giving the patient a business card improves the patient's perception of the radiologist's professionalism and esteem. METHODS: In this prospective and randomized two-centre study, a total of 141 patients with BI-RADS 1 and 2 scores were included. After screening examination comprising mammography and ultrasound by a radiologist, 71 patients received a business card (group 1), while 70 received no business card (group 2). Following the examination, patients were questioned about their experiences. RESULTS: The patients in group 1 could remember the name of the radiologist in 85% of cases. The patients in group 2, in contrast, could only remember the name in 7% of cases (p < 0.001). 90% of the patients in group 1 believed it was very important that they are able to contact the radiologist at a later time, whereas only 76% of patients in group 2 felt that this was a very important service (p < 0.025). A total of 87% of the patients in group 1 indicated that they would contact the radiologist if they had any questions whereas 73% of the patients in group 2 would like to contact the radiologist but were not able to do so, because they could not remember the name (p < 0.001). All questions were analysed with a Cochran-Mantel-Haenszel (CMH) test that took study centre as stratification into account. In some cases, two categories were collapsed to avoid zero cell counts. CONCLUSIONS: Using business cards significantly increased the recall of the radiologist's name and could be an important tool in improving the relationships between patients and radiologists and enhancing service professionalism. TRIAL REGISTRATION: We have a general approval from our ethics committee. The patients have given their consent to this study.

When Does the Incremental Risk Format Aid Informed Medical Decisions? The Role of Learning, Feedback, and Number of Treatment Options.

Background. Informed medical decisions require understanding the benefits and risks of treatments. This entails comparing treatment outcomes to a control group. The incremental risk format has been recommended as it directly visualizes the differences between treatment and control group in 1 graph, whereas they have to be calculated from 2 separate graphs in the total risk format. We investigated when the incremental risk format aids understanding. Methods. In 2 experiments, participants received information about medical treatments, either as incremental or total risk format. We assessed verbatim knowledge (precise quantitative knowledge), gist knowledge (knowledge of essential meaning), and evaluations of the formats. Study 1 (N = 99) consisted of only 1 trial with medical information and also assessed recall. Study 2 (N = 222) assessed learning across multiple trials and also varied the presence of feedback and the number of treatment options. Results. In study 1, the incremental risk format (v. total risk format) led to worse knowledge, recall, and evaluations. In study 2, participants learned to understand the incremental risk format over time, resulting in comparable verbatim knowledge and evaluations as in the total risk format, as well as in even better gist knowledge. Feedback and number of treatment options did not moderate the effect of risk format. Limitations. The studies were conducted with nonpatient samples, and study 2 employed hypothetical treatments. Conclusions. The incremental risk format was initially less understandable than the total risk format. After a short learning period, however, the incremental risk format resulted in better gist knowledge and was comparable otherwise, which suggests that participants had to get used to that format. This has important implications for the study of new formats.

Perspectives on the 2 × 2 Matrix: Solving Semantically Distinct Problems Based on a Shared Structure of Binary Contingencies.

Cognition is both empowered and limited by representations. The matrix lens model explicates tasks that are based on frequency counts, conditional probabilities, and binary contingencies in a general fashion. Based on a structural analysis of such tasks, the model links several problems and semantic domains and provides a new perspective on representational accounts of cognition that recognizes representational isomorphs as opportunities, rather than as problems. The shared structural construct of a 2 × 2 matrix supports a set of generic tasks and semantic mappings that provide a unifying framework for understanding problems and defining scientific measures. Our model's key explanatory mechanism is the adoption of particular perspectives on a 2 × 2 matrix that categorizes the frequency counts of cases by some condition, treatment, risk, or outcome factor. By the selective steps of filtering, framing, and focusing on specific aspects, the measures used in various semantic domains negotiate distinct trade-offs between abstraction and specialization. As a consequence, the transparent communication of such measures must explicate the perspectives encapsulated in their derivation. To demonstrate the explanatory scope of our model, we use it to clarify theoretical debates on biases and facilitation effects in Bayesian reasoning and to integrate the scientific measures from various semantic domains within a unifying framework. A better understanding of problem structures, representational transparency, and the role of perspectives in the scientific process yields both theoretical insights and practical applications.

The echo in flu-vaccination echo chambers: Selective attention trumps social influence.

BACKGROUND: Online discussions may impact the willingness to get vaccinated. This experiment tests how groups of individuals with consistent and inconsistent attitudes towards flu vaccination attend to and convey information online, and how they alter their corresponding risk perceptions. METHODS: Out of 1859 MTurkers, we pre-selected 208 people with negative and 221 people with positive attitudes towards flu vaccinations into homogeneous or heterogeneous 3-link experimental diffusion chains. We assessed (i) which information about flu vaccinations participants conveyed to the subsequent link, (ii) how flu-vaccination related perceptions were altered by incoming messages, and (iii) how participants perceived incoming information. RESULTS: Participants (i) selectively conveyed attitude-consistent information, but exhibited no overall anti-vaccination bias, (ii) were reluctant to alter their flu-vaccination related perceptions in response to messages, and (iii) evaluated incoming information consistent with their prior attitudes as more convincing. DISCUSSION: Flu-vaccination related perceptions are resilient against contradictions and bias online communication. Contrary to expectations, there was no sign of amplification of anti-vaccine attitudes by online communication.