How people decide who is correct when groups of scientists disagree.

Uncertainty that arises from disputes among scientists seems to foster public skepticism or noncompliance. Communication of potential cues to the relative performance of contending scientists might affect judgments of which position is likely more valid. We used actual scientific disputes-the nature of dark matter, sea level rise under climate change, and benefits and risks of marijuana-to assess Americans' responses (n = 3150). Seven cues-replication, information quality, the majority position, degree source, experience, reference group support, and employer-were presented three cues at a time in a planned-missingness design. The most influential cues were majority vote, replication, information quality, and experience. Several potential moderators-topical engagement, prior attitudes, knowledge of science, and attitudes toward science-lacked even small effects on choice, but cues had the strongest effects for dark matter and weakest effects for marijuana, and general mistrust of scientists moderately attenuated top cues' effects. Risk communicators can take these influential cues into account in understanding how laypeople respond to scientific disputes, and improving communication about such disputes.

Trust and trustworthy artificial intelligence: A research agenda for AI in the environmental sciences.

Demands to manage the risks of artificial intelligence (AI) are growing. These demands and the government standards arising from them both call for trustworthy AI. In response, we adopt a convergent approach to review, evaluate, and synthesize research on the trust and trustworthiness of AI in the environmental sciences and propose a research agenda. Evidential and conceptual histories of research on trust and trustworthiness reveal persisting ambiguities and measurement shortcomings related to inconsistent attention to the contextual and social dependencies and dynamics of trust. Potentially underappreciated in the development of trustworthy AI for environmental sciences is the importance of engaging AI users and other stakeholders, which human-AI teaming perspectives on AI development similarly underscore. Co-development strategies may also help reconcile efforts to develop performance-based trustworthiness standards with dynamic and contextual notions of trust. We illustrate the importance of these themes with applied examples and show how insights from research on trust and the communication of risk and uncertainty can help advance the understanding of trust and trustworthiness of AI in the environmental sciences.

Information effects on lay tradeoffs between national regulatory costs and benefits.

A novel stated-preference "macro-risk" approach introduced to estimate the life-prolonging benefits of proposed environmental, health, and safety regulations may answer questions unasked or wrongly answered by conventional revealed-preference (e.g., "wage premiums" for high occupational risks) and stated-preference methods (e.g., willingness to pay for tiny reductions in one's own premature death risk). This new approach asks laypeople to appraise directly their preferred tradeoffs between national regulatory costs and lives prolonged nationwide (regulatory benefits). However, this method may suffer from incomplete lay understanding of national-scale consequences (e.g., billions of dollars in regulatory costs; hundreds of lives prolonged) or tradeoffs (e.g., what are lives prolonged worth?). Here we (1) tested effects of numerical contextual examples to ground each hypothetical regulatory tradeoff, and (2) explored why some people implicitly offer "implausible" values (< $10,000 or > $1 billion) for the social benefit of prolonging one life. In Study 1 (n = 356), after testing their separate effects, we combined three contextual-information aids: (1) comparing hypothetical regulatory costs and benefits to real-life higher and lower values; (2) reframing large numbers into smaller, more familiar terms; and (3) framing regulatory costs as having diffuse versus concentrated impacts. Information increased social benefits values on average (from $4.5 million to $13.8 million). Study 2 (n = 402) found that the most common explanations for "implausible" values included inattention, strong attitudes about regulation, and problems translating values into responses. We discuss implications for this novel stated-preferences method, and for comparing it to micro-risk methods.

Trust, confidence, familiarity, and support for land-based recirculating aquaculture facilities.

Surveys in three U.S. localities (n = 523) with proposed or existing land-based aquaculture facilities probed trust's relationship with perceived net benefits and public intentions to cooperate with siting of this novel technology. The trust, confidence, and cooperation (TCC) model posits that shared values shape willingness to be vulnerable to others (trust), while past performance shapes certainty that others will behave as expected (confidence). Trust affects confidence given moral outweighs performance information, possibly varying by familiarity. Other research suggests that trust shapes benefit and risk perceptions, which drive cooperation (defined here by potentially observable behavior: voting on siting, trying to influence government decisions directly or through citizen groups, and buying or eating facility fish). Confirmatory factor analyses suggested that a two-factor model fit the trust/confidence measures better than a one-factor model or a two-factor model without inter-factor correlation, indicating (despite a strong association of trust and confidence) that they are empirically distinct. Path analyses suggested that trust had stronger direct effects on cooperation than did confidence, reflecting the TCC notion that moral information underlying trust judgments is more influential, and stronger indirect effects through benefit-risk judgments. Model fit was better than if the benefit-risk mediator was omitted. Trust in government had a small direct effect on cooperation and confidence, but a large effect on trust in the corporation, and model fit was much worse if any of these paths was omitted. Low familiarity with the project lowered both model fit and trust-confidence association. We discuss implications for risk analysis theory and practice.

Evaluating the Effects of News-following, Volume and Content of News Coverage on Americans' Risk Perceptions during the 2014-2016 Ebola Outbreak.

We used the 2014-2016 Ebola outbreak to examine the relationships between risk perceptions and media coverage (volume and content). We analyzed how public opinion from longitudinal U.S. panel data related to the number of published news articles and the proportion that discussed risks. News following, volume and risk content were positively related to U.S. and global risk perceptions. Perceptions of U.S. risk declined at different rates, depending upon news attention and potential exposure to risk content. Both media volume and content were significant factors, suggesting scholars should focus more on combined effects of news media volume and content.

A Longitudinal Analysis of Americans' Media Sources, Risk Perceptions, and Judged Need for Action during the Zika Outbreak.

The objective of this study is to better understand the effects of media attention on Americans' perceptions of risk by analyzing the different media sources and outlets, or "repertoires," reported as used during the small 2016-2017 Zika outbreak in the U.S. We analyzed survey data from a four-wave longitudinal panel study over nine months - July 19, 2016 through April 24, 2017 (n = 743) - using an online panel of American adults. Media attention related to ratings of personal risk, U.S. risk, and need for action. Personal risk was enhanced more by reported attention to international coverage, reduced by certain reported website attention, but enhanced by reported attention to public health agency websites. U.S. risk was enhanced by reported attention to both domestic and international coverages, reduced by television. Judged need for U.S. action was enhanced more by exposure to domestic coverage, reduced by reported attention to television and local newspapers, but enhanced by reported exposure to BBC and CNN. Our results demonstrate how the use of different media outlets and sources are related to different perceptions of risk and need for action during 2016-2017 Zika outbreak.

Cultural Theory's Contributions to Risk Analysis: A Thematic Review with Directions and Resources for Further Research.

Cultural theory (CT) developed from grid/group analysis, which posits that different patterns of social relations-hierarchist, individualist, egalitarian, and fatalist-produce compatible cultural biases influencing assessment of which hazards pose high or low risk and how to manage them. Introduced to risk analysis (RA) in 1982 by Douglas and Wildavsky's Risk and Culture, this institutional approach to social construction of risk surprised a field hitherto focused on psychological influences on risk perceptions and behavior. We explain what CT is and how it developed; describe and evaluate its contributions to the study of risk perception and management, and its prescriptions for risk assessment and management; and identify opportunities and resources to develop its contributions to RA. We suggest how the diverse, fruitful, but scattered efforts to develop CT both inside and outside the formal discipline of RA (as exemplified by the Society for Risk Analysis) might be leveraged for greater theoretical, methodological, and applied progress in the field.

Americans' Views of Voluntary Protective Actions Against Zika Infection: Conceptual and Measurement Issues.

Understanding factors affecting decisions by people to protect themselves, or not, is critical to designing supportive communications. Here, threat, protective-action, and stakeholder perceptions were evaluated for effects on mainland Americans' behavioral intentions regarding Zika in April 2017, as postulated by the Protective Action Decision Model. Although all three perception types (including a novel resource sufficiency measure) affected intentions, these relationships varied widely depending upon the method used to measure adoption of actions (e.g., total count of all behaviors adopted vs. behavior-specific analyses), the behaviors involved, and whether analysis focused on the full sample or only on people who had a reasonable opportunity to enact the behavior or who believed it relevant to their lives. There was a general contrast between mosquito control actions (removal of mosquito breeding areas and pesticide spraying) and travel-related behaviors (avoiding travel to areas of local transmission of the virus, protecting oneself from mosquito bites after potential exposure, and practicing safe sex after potential exposure). Reported action or inaction during the 2016 mosquito season, and stages of behavior change, were both elicited in January-February 2017; both drove intentions in April 2017 for the upcoming season, although direct and indirect effects varied widely. Collectively these findings present theoretical, measurement, and practical implications for understanding, tracking, and promoting voluntary protective actions against hazards.

Experiments in Lay Cues to the Relative Validity of Positions Taken by Disputing Groups of Scientists.

Risk analysis and hazard management can prompt varied intra-scientific disputes, some which have or will become public, and thus potentially available for lay judgments of the relative validity of the positions taken. As attentive laypeople may include elites as well as the general public, understanding whether and how cues to credibility of disputing groups of scientists might shape those lay judgments can be important. Relevant literatures from philosophy, social studies of science, risk analysis, and elsewhere have identified potential cues, but not tested their absolute or relative effects. Two experiments with U.S. online panel members tested multiple cues (e.g., credentials, experience, majority opinions, research quality) across topics varying in familiarity subject to actual intra-science disputes (dark matter, marijuana, sea-level rise). If cues supported a position, laypeople were more likely to choose it as relatively more valid, with information quality, majority "vote," experience, and degree source as the strongest, and interest, demographic, and values similarity as the weakest, cues. These results were similar in overall rankings to those from implicit rankings of cue reliability ratings from an earlier U.S. online survey. Proposed moderators were generally nonsignificant, but topic familiarity and subjective topic knowledge tended to reduce cue effects. Further research to confirm and extend these findings can inform both theory about citizen engagement with scientific and risk disputes, and practice in communication about science and risk.

Residential Location and Psychological Distance in Americans' Risk Views and Behavioral Intentions Regarding Zika Virus.

Two 2017 experiments with a U.S. national opportunity sample tested effects of location, psychological distance (PD), and exposure to location-related information on Americans' Zika risk views and behavioral intentions. Location-distance from mosquito transmission of the virus in Florida and Texas; residence within states with 100+ Zika infections; residence within potential mosquito vector ranges-had small, inconsistent effects. Hazard proximity weakly enhanced personal risk judgments and concern about Zika transmission locally. It also increased psychological proximity, and intentions of mosquito control, avoiding travel to Zika-infected areas, and practicing safe sex. PD-particularly social and geographical distance, followed by temporal distance, with few effects for uncertainty-modestly and inconsistently decreased risk views and intentions. Exposure to location-related information from the U.S. Centers for Disease Control and Prevention website-naming states with 100+ Zika cases; maps of potential mosquito vector habitat-increased risk views and psychological closeness, but not intentions; maps had slightly stronger if inconsistent effects versus prevalence information. Structural equation modeling (SEM) of a location > PD > risk views > intention path explained modest variance in intentions. This varied in degree and kind (e.g., which location measures were significant) across behaviors, and between pre- and postinformation exposure analyses. These results suggest need for both theoretical and measurement advances regarding effects of location and PD on risk views and behavior. PD mediates location effects on risk views. Online background information, like that used here, will not enhance protective behavior without explicitly focused communication and perhaps higher objective risk.

Effects of Acknowledging Uncertainty about Earthquake Risk Estimates on San Francisco Bay Area Residents' Beliefs, Attitudes, and Intentions.

We test here the risk communication proposition that explicit expert acknowledgment of uncertainty in risk estimates can enhance trust and other reactions. We manipulated such a scientific uncertainty message, accompanied by probabilities (20%, 70%, implicit ["will occur"] 100%) and time periods (10 or 30 years) in major (≥magnitude 8) earthquake risk estimates to test potential effects on residents potentially affected by seismic activity on the San Andreas fault in the San Francisco Bay Area (n = 750). The uncertainty acknowledgment increased belief that these specific experts were more honest and open, and led to statistically (but not substantively) significant increases in trust in seismic experts generally only for the 20% probability (vs. certainty) and shorter versus longer time period. The acknowledgment did not change judged risk, preparedness intentions, or mitigation policy support. Probability effects independent of the explicit admission of expert uncertainty were also insignificant except for judged risk, which rose or fell slightly depending upon the measure of judged risk used. Overall, both qualitative expressions of uncertainty and quantitative probabilities had limited effects on public reaction. These results imply that both theoretical arguments for positive effects, and practitioners' potential concerns for negative effects, of uncertainty expression may have been overblown. There may be good reasons to still acknowledge experts' uncertainties, but those merit separate justification and their own empirical tests.

Public Perceptions of Regulatory Costs, Their Uncertainty and Interindividual Distribution.

Public perceptions of both risks and regulatory costs shape rational regulatory choices. Despite decades of risk perception studies, this article is the first on regulatory cost perceptions. A survey of 744 U.S. residents probed: (1) How knowledgeable are laypeople about regulatory costs incurred to reduce risks? (2) Do laypeople see official estimates of cost and benefit (lives saved) as accurate? (3) (How) do preferences for hypothetical regulations change when mean-preserving spreads of uncertainty replace certain cost or benefit? and (4) (How) do preferences change when unequal interindividual distributions of hypothetical regulatory costs replace equal distributions? Respondents overestimated costs of regulatory compliance, while assuming agencies underestimate costs. Most assumed agency estimates of benefits are accurate; a third believed both cost and benefit estimates are accurate. Cost and benefit estimates presented without uncertainty were slightly preferred to those surrounded by "narrow uncertainty" (a range of costs or lives entirely within a personally-calibrated zone without clear acceptance or rejection of tradeoffs). Certain estimates were more preferred than "wide uncertainty" (a range of agency estimates extending beyond these personal bounds, thus posing a gamble between favored and unacceptable tradeoffs), particularly for costs as opposed to benefits (but even for costs a quarter of respondents preferred wide uncertainty to certainty). Agency-acknowledged uncertainty in general elicited mixed judgments of honesty and trustworthiness. People preferred egalitarian distributions of regulatory costs, despite skewed actual cost distributions, and preferred progressive cost distributions (the rich pay a greater than proportional share) to regressive ones. Efficient and socially responsive regulations require disclosure of much more information about regulatory costs and risks.

Modeling retrospective attribution of responsibility to hazard-managing institutions: an example involving a food contamination incident.

Perceptions of institutions that manage hazards are important because they can affect how the public responds to hazard events. Antecedents of trust judgments have received far more attention than antecedents of attributions of responsibility for hazard events. We build upon a model of retrospective attribution of responsibility to individuals to examine these relationships regarding five classes of institutions that bear responsibility for food safety: producers (e.g., farmers), processors (e.g., packaging firms), watchdogs (e.g., government agencies), sellers (e.g., supermarkets), and preparers (e.g., restaurants). A nationally representative sample of 1,200 American adults completed an Internet-based survey in which a hypothetical scenario involving contamination of diverse foods with Salmonella served as the stimulus event. Perceived competence and good intentions of the institution moderately decreased attributions of responsibility. A stronger factor was whether an institution was deemed (potentially) aware of the contamination and free to act to prevent or mitigate it. Responsibility was rated higher the more aware and free the institution. This initial model for attributions of responsibility to impersonal institutions (as opposed to individual responsibility) merits further development.

Cross-temporal relations of conditional risk perception measures with protective actions against COVID-19.

Two decades ago a research team clarified that cross-sectional associations of risk perceptions and protective behavior can only test an "accuracy" hypothesis: e.g., people with higher risk perceptions at Ti should also exhibit low protective behavior and/or high risky behavior at Ti. They argued that these associations are too often interpreted wrongly as testing two other hypotheses, only testable longitudinally: the "behavioral motivation" hypothesis, that high risk perception at Ti increases protective behavior at Ti+1, and the "risk reappraisal" hypothesis, that protective behavior at Ti reduces risk perception at Ti+1. Further, this team argued that risk perception measures should be conditional (e.g., personal risk perception if one's behavior does not change). Yet these theses have garnered relatively little empirical testing. An online longitudinal panel study of U.S. residents' COVID-19 views across six survey waves over 14 months in 2020-2021 tested these hypotheses for six behaviors (hand washing, mask wearing, avoiding travel to infected areas, avoiding large public gatherings, vaccination, and [for five waves] social isolation at home). Accuracy and behavioral motivation hypotheses were supported for both behaviors and intentions, excluding a few waves (particularly in February-April 2020, when the pandemic was new in the U.S.) and behaviors. The risk reappraisal hypothesis was contradicted-protective behavior at one wave increased risk perception later-perhaps reflecting continuing uncertainty about efficacy of COVID-19 protective behaviors and/or that dynamic infectious diseases may yield different patterns than chronic diseases dominating such hypothesis-testing. These findings raise intriguing questions for both perception-behavior theory and behavior change practice.

Climate change communication: a provocative inquiry into motives, meanings, and means.

The deliberately provocative theme of this article is that perceived difficulties in climate change communication (CCC)--e.g., indifference about or denial of climate change's reality, negative consequences, anthropogenic causes, or need to mitigate or adapt to it-are partly the fault of climate change communicators. Fischhoff's model of risk communication development is used to demonstrate that CCC to date has tended to stress persuasion, rather than social movement mobilization or deliberation, and with a focus on the model's early stages. Later stages are not necessarily better, but a more diverse strategy seems superior to a focus perhaps narrowed by empathic, ideological, psychological, and resource constraints. Furthermore, even within persuasion, emphasizing a wider set of values, consequences, and audiences could be fruitful. Social movement mobilization has its own set of weaknesses, but usefully complements persuasion with a focus on developing power, subverting mainstream assumptions, and engaging people in collective action. Deliberation similarly has its drawbacks, but unlike the other two approaches does not define the solution-or even, necessarily, the problem-in advance, and thus offers the chance for people of contending viewpoints to jointly develop concepts and action agendas hitherto unimagined. Simultaneous pursuit of all three strategies can to some degree offset their respective flaws, at the potential cost of diffusion of energies and contradictory messages. Success in CCC is by no means guaranteed by a more diverse set of strategies and self-reflection by communicators, but their pursuit should better reveal CCC's limits.

Experience with urban air pollution in Paterson, New Jersey and implications for air pollution communication.

Communication about air pollution can help reduce health risks, but a scattered, largely qualitative literature on air pollution beliefs, attitudes, and behaviors raises questions about its effectiveness. A telephone survey of Paterson, New Jersey (USA) residents tested four hypotheses aimed toward integrating these findings. Self-reported sheltering indoors during high pollution, the recommended strategy, was predicted by perceived air quality and self-reported "sensitivity" to air pollution. Nearly a quarter of the sample reported mandatory outdoor activity (e.g., work) that might increase their exposures, but this factor did not significantly affect self-reported sheltering. Perceptions of air quality did not correlate strongly with official monitoring data (U.S. Air Quality Index (AQI)); even people who regularly sought AQI data relied upon sensory cues to high pollution, and secondarily upon health cues. Use of sensory and health cues, definitions of what makes someone sensitive to air pollution, and (less strongly) definitions of vulnerability to air pollution varied widely. The minority aware of the AQI were more likely to seek it if they had illnesses or saw themselves in the targeted AQI audience, yet less likely if they believed themselves sensitive to pollution. However, their sense of the AQI's match to their own experience was driven by whether they used sensory (yes) or health (no) cues, not by illness status. Some urban residents might not have access to AQI data, but this barrier seems outweighed by need to bridge interpretive gaps over definitions of air pollution, sensory perception, vulnerability, and health consequences.

Acculturation, ethnicity, and air pollution perceptions.

A globalizing world increases immigration between nations, raising the question of how acculturation (or its lack) of immigrants and their descendants to host societies affects risk perceptions. A survey of Paterson, New Jersey, residents tested acculturation's associations with attitudes to air pollution and its management, and knowledge of and self-reported behaviors concerning air pollution. Linguistic and temporal proxy measures for acculturation were independent variables along with ethnicity, plus controls for gender, age, education, and income in multivariate analyses. About one-fifth of contrasts between non-Hispanic whites, non-Hispanic blacks, English-interviewed Hispanics, and Spanish-interviewed Hispanics were statistically significant (Bonferroni-corrected) and of medium or higher affect size, with most featuring the Spanish-interviewed Hispanics. Knowledge variables featured the most significant differences. Specifically, Spanish-interviewed Hispanics reported less concern, familiarity with pollution, recognition of high pollution, and vigorous outdoor activity, and greater belief that government overregulates pollution than English-interviewed Hispanics (and than the other two groups on most of these variables too). English-interviewed Hispanics did not differ from non-Hispanic whites, but did on several variables from non-Hispanic blacks. Temporal proxies of acculturation among the foreign-born were far less significant, but concern and familiarity with air pollution increased with time spent in the United States, while belief in overregulation and a positive trend in New Jersey pollution increased with time in the nation of origin. Implications of these acculturation and ethnicity findings for risk perception/communication research and practice are discussed.

Lay beliefs about scientists' relations with their employers.

Lay beliefs about scientist-employer relations may affect public attitudes toward science. A representative sample of US residents characterized scientists' relations with one of four employers: federal government agency, large business corporation, advocacy group (nonprofit seeking to influence policy), or university. Overall, they held moderate views of how much scientists and employers shared motivations, interests, and values, and of whether employers tried to change-and succeeded in changing-how scientist employees did their scientific work. Judgments differed little across employers. Best predictors of these views were belief in scientific positivism, subjective knowledge of science, and age. These findings suggest scientific authority in the United States is not immediately threatened by public beliefs that employers skew their scientific employees' work, although that might differ for specific topics or demographic sub-groups.

Conservatives' Moral Foundations Are More Densely Connected Than Liberals' Moral Foundations.

We use network psychometrics to map a subsection of moral belief systems predicted by moral foundations theory (MFT). This approach conceptualizes moral systems as networks, with moral beliefs represented as nodes connected by direct relations. As such, it advances a novel test of MFT's claim that liberals and conservatives have different systems of foundational moral values, which we test in three large datasets (NSample1 = 854; NSample2 = 679; NSample3 = 2,572), from two countries (the United States and New Zealand). Results supported our first hypothesis that liberals' moral systems show more segregation between individualizing and binding foundations than conservatives. Results showed only weak support for our second hypothesis, that this pattern would be more typical of higher educated than less educated liberals/conservatives. Findings support a systems approach to MFT and show the value of modeling moral belief systems as networks.

Trust and terrorism: citizen responses to anti-terrorism performance history.

The "intuitive detection theorists" model of trust posits greater trust for correctly distinguishing danger from safety and an activist response under uncertainty about danger. An American sample evaluated U.S. Department of Homeland Security (DHS) performance after two possible terrorism events in which DHS has the same activist or nonactivist response bias. Outcomes were two successes (bombing prevented or lack of threat accurately foretold), two failures (bombing or DHS action against high school prank leads to student deaths), or a mix. Hindsight empathy (a belief one would have made the same decision) differed across treatments but trust less so; contrary to a similar one-event experiment in Germany, an active but incorrect response did not raise trust relative to passive incorrect action. Political conservatives were much more trusting and empathetic than liberals, and all ideological groups (including moderates) exhibited little internal variation reflecting experimental conditions. Consistently accurate outcomes rated significantly higher in empathy than either inconsistent results or consistent inaccuracy (the lowest rated); trust exhibited no significant differences. Results in this study show actual (experimentally manipulated) performance being trumped by the interpretive screen of political ideology, but this seemed less the case in the earlier German study, despite its finding of a strong moderating effect of right-wing authoritarianism. Trust scholars need to attend more to effects of performance history (i.e., a sequence of events) and their limiting factors. More systematic testing of effects of ideology and performance history would enhance future research on trust.