Cultural values, risk characteristics, and risk perceptions of controversial issues: How does cultural theory work?

Abstract

Cultural theory and the psychometric paradigm are two frameworks proposed to explain risk perceptions, mostly used independently of each other. On the one hand, psychometric research identified key characteristics of hazards responsible for their level of perceived riskiness. On the other hand, cultural studies provided evidence that different worldviews lead to divergent perceptions of risk in a way supportive of individuals’ cultural values. The purpose of this research was to combine both approaches into mediational models in which cultural values impact risk perceptions of controversial hazards through their influence on the characteristics associated with those hazards. Using data from an online survey completed by 629 French participants, findings indicated specific associations between cultural values and risk characteristics, both of them exhibiting effects on risk perceptions that depend largely on hazardous issues. More specifically, we found that people confer specific characteristics on hazards (common or dreadful, beneficial or costly, affecting few or many people), depending on whether they are hierarchists–individualists, egalitarians, or fatalists; in turn, such characteristics have an impact on the perceived riskiness of hazards such as cannabis, social movement, global warming, genetically modified organisms, nuclear power, public transportation, and coronavirus. Finally, this article discussed the interest of addressing the mechanisms that explain how cultural values shape individuals’ perceptions of risk.

1. INTRODUCTION

People respond to the hazards they perceive in a way that varies considerably across hazards (Slovic, 1987, 2016) as well as individuals (Chauvin, 2018, for a review). Over the past 40 years, a significant part of work in the field of risk research has arisen from two influential approaches: the psychometric paradigm (Slovic et al., 1985) and the cultural theory of risk (Douglas & Wildavsky, 1982). On the one hand, psychometric studies identified cognitive and affective dynamics used by people on average to characterize hazardous activities or technologies and differentiate them as more or less risky (Slovic, 2000). Within this framework, social and cultural influences on risk perception have been largely neglected; risk has been “depoliticized” (Douglas, 1997). On the other hand, cultural studies provided valuable insights on why different (culturally‐oriented) people perceive the same activity as risky or conversely safe and report on the whole distinctive patterns of risk perception (e.g., Dake, 1991, 1992; see also Chauvin & Chassang, 2022). Within this perspective, the psychological mechanisms through which cultural values shape risk perceptions have been largely ignored, until Kahan et al. cultural cognition pioneering work focused on how social psychological factors mediate cultural influences on risk perception (more later on; see Kahan, 2012, for synthesis). Note that when Wildavsky and Dake (1990) claimed that cultural theory best explained who fears what and why, they did not mention how.

The current research aimed to bring together the psychometric paradigm and the cultural theory of risk to understand how risk perception can vary through hazards and individuals. We hypothesized a mediation mechanism in which cultural values—as orienting dispositions (Dake, 1991; Peters & Slovic, 1996)—generate discrepancies about the risks of various hazards through specific dimensions characterizing those hazards. Working within the logic of cultural worldview[ing], we were interested in controversial issues and selected hazards that are likely to generate debates about risk seen as threatening specific worldviews while complying with others (environmental issues, technological hazards, disruptive and deviant activities, security‐related hazards, and health threats).

1.1. The psychometric study of risk perception

The basic assumption underlying the psychometric paradigm is that laypersons do not reduce riskiness to a probability of harm (as experts do) but have a broad, qualitative, and complex conception of risk (Slovic, 1987). Research following the psychometric paradigm has led to a taxonomy of hazards useful for understanding public responses to risks: Every hazard has a unique pattern of qualities that appears to be related to its perceived risks (Slovic, 1992). Specifically, psychometric studies showed that laypersons incorporate qualitative characteristics such as dread, catastrophic potential, (un)controllability, (in)equity in distributing risks and benefits, (lack of) familiarity, newness, or exposure, which—when combined—generate (up to) three factors respectively labeled Dread risk (defined by the extent to which the risk evokes a feeling of dread and is threatening), unknown risk (that reflects the degree to which the risk is understood), and number of people exposed (that reflects how large the risk is) (Slovic et al., 1985). As a whole, research reported that a substantial part of the variance of risk assessments can be explained by a combination of the three factors—in the United States (e.g., Slovic et al., 1980) as in numerous countries worldwide (see Boholm, 1998, or Chauvin, 2014, for reviews). Most important is usually the factor Dread risk. The higher a hazard's score on this factor, the higher its perceived risk and the more people need to reduce risk (Slovic, 1987). A few but noticeable variants on this structure have been introduced. One is the use of the Controllable risk label as a substitute for either classic factor label (e.g., Lai & Tao, 2003; Zhang, 1994), considering this dimension as an important determinant in risk perception and protective decision making (Brun, 1992; Savadori et al., 1998). Another important variation is the addition of a fourth factor termed evaluative factor (Mullet et al., 1993). Defined as an evaluation of personal acceptability of the risk that depends in part on the extent to which the risk is beneficial to society and economically justified, this evaluative factor emerged as a major determinant in judgments of riskiness (after the Dread factor) and as the first one in demands for risk regulation (Mullet et al., 1993).

Psychometric studies grew out of a concern to understand why the risks from some hazardous activities appear to be treated differently from the risks of other activities (Slovic et al., 1986). Early work examined how people process with hazards and how they combine relevant dimensions for estimate the riskiness of those hazards (Slovic, 1987). Given their main cognitive emphasis, these studies focused on identifying the risk characteristics responsible for differences in risk level between hazards but neglected differences in risk perception between individuals—apart from descriptive comparisons between experts and laypeople regarding the meaning of the concept “risk” (Marris et al., 1998; Siegrist & Arvai, 2020). Slovic (2000, 2010, 2016) was fully aware of these limitations, and with other researchers, he had gone beyond the psychometric approach to develop ambitious research programs allowing them to deal with such shortcomings. Interestingly, a specific approach highlighted that social processes like trust (e.g., Slovic, 1999; see Siegrist, 2021, for a review) and cultural factors like worldviews (e.g., Dake, 1991; see Xue et al., 2014, for a review) could play an important role in determining risk perception. It was gradually recognized that social, political, and cultural factors—that is, value‐laden issues—should be considered in risk perception. The dominant approach to sociopolitical and cultural construction of risk is the cultural theory of risk developed by Douglas and Wildavsky (1982) (Johnson & Swedlow, 2019a).

1.2. The cultural theory of risk

The cultural theory of risk posits that risk perception cannot be studied only as a matter of individual cognition occurring outside the social world (Douglas & Wildavsky, 1982; Thompson et al., 1990). Risk perception should also be considered oriented by worldviews that provide powerful cultural lenses, encouraging individuals to selectively highlight some hazards and disregard others (Wildavsky & Dake, 1990). Cultural worldviews are defined as shared beliefs and values regarding society and how the world works and should work, including ideas about the functioning of nature (Dake, 1991; Schwarz & Thompson, 1990). Stemming from the orthogonal combination of two dimensions of sociality posited by Douglas (1978)—one named grid (referring to the degree of social prescriptions) and another named group (referring to the degree of allegiance to a social group) (Thompson et al., 1990)—four cultural worldviews have generally been conceptualized. Following the most conventional terms, they are labeled hierarchy (high‐grid, high‐group), individualism (low‐grid, low‐group), egalitarianism (low‐grid, high‐group), and fatalism (high‐grid, low‐group) (Mamadouh, 1999). Persons with a hierarchic orientation are part of the establishment (Douglas & Wildavsky, 1982; Thompson et al., 1990). As such, they are committed to maintaining existing power structures that protect their interests and go along with their values of authority and social order. Their favored way to get there is to privilege trust in experts (who are also members of the establishment). Providing technology is certified safe by experts; they approve it and regard nature as being tolerant to its adverse effects—viewing technology as a way to improve the quality of life (Dake, 1991, 1992). Empirical research showed that persons of hierarchical orientation express low concern about environmental (e.g., air pollution—Xue et al., 2014) and technological issues (e.g., nuclear power—Peters & Slovic, 1996) but high concern about any activity that could undermine the established (hierarchical) procedures (e.g., civil disobedience, demonstrations—Dake, 1991). Persons who have an individualistic orientation are also part of the establishment; along with the hierarchists, they form the establishment coalition (Wildavsky & Dake, 1990; see Mamadouh, 1999, for details). They want to maintain the system in power because it fits their values well: it allows market relationships and ensures the freedom to bid and bargain as well as to benefit from such activities (Dake, 1991; Jenkins‐Smith & Herron, 2009). They are thus sensitive to the cost‐benefit trade‐off concerning risk issues. Consequently, individualists hold a positive view of technological innovation and see the nature as being resilient to exploitation—regarding benefits as more than compensating for any environmental damage that is created (Dake, 1992; Thompson et al., 1990). Individualists, as hierarchists, are “technologically optimistic” (Wildavsky & Dake, 1990, p. 169). They show low concern about environmental and technological risks (e.g., ozone depletion, nuclear power—Marris et al., 1998) but high concern about situations that are likely to unsettle the established (economic) rules (e.g., inflation, debt—Dake, 1991). People who hold an egalitarian orientation value fairness and social justice. They exhibit strong support for participatory democracy and consensus‐based decision making (Tansey & O'Riordan, 1999; Wildavsky & Dake, 1990). Accordingly, egalitarians are critical of every putative source of social inequality like free commerce and industry and are suspicious of the system in power because they consider it as working for its own ends at the expense of the majority of people. An emblematic case for egalitarians is about technologies (a) whose hazards are often imposed on the mass of citizens by a few decision‐makers, (b) which create or maintain inequalities between citizens, and (c) that are likely to be harmful for society as a whole by provoking large‐scale environmental damage to nature—viewed by egalitarians as being fragile, vulnerable, and limited in resources (Steg & Sievers, 2000). Global exposure to hazards thus appears to be a key concern for egalitarians (Ellis & Thompson, 1997; Van de Graaff, 2016). This profile also says that the benefit‐risk issue is relevant for egalitarians, for whom opportunities and outcomes should be distributed in a balanced way. Persons with more egalitarian values are indeed worried about environmental issues (e.g., global warming—Thaker et al., 2020) and technological hazards (e.g., nuclear power stations, chemical installations—Brenot et al., 1998), but more relaxed about civil disturbances (e.g., Chauvin & Chassang, 2022). Persons with a fatalistic orientation are resigned and disengaged (Thompson et al., 1990). They believe they have only limited control over their own life and the outcomes they face, viewed largely as random and beyond human control (Dake, 1992). Psychological studies indicated that fatalists tend to display a low perceived self‐efficacy (Sun et al., 2022) while they exhibit a high external locus of control leading them to perceive outcomes as the result of external factors such as luck or fate (Corcoran et al., 2011). Beyond their own perceived helplessness, they tend to be very skeptical, or even mistrustful, about the ability of authorities to manage hazardous activities properly (e.g., Ripberger et al., 2014; Van de Graaf, 2016). Based on this profile, some authors decided to leave fatalists out of the risk story (e.g., Kahan, 2012), whereas others opted for omitting them from their study of risk (environmental ones, for example) (e.g., Leiserowitz, 2006). Among work including fatalism, some have shown that fatalists are often indifferent to risk (e.g., Xue et al., 2014) or inconsistent in their risk perceptions (e.g., Steg & Sievers, 2000; Verweij et al., 2011). Recent findings, however, showed that fatalists exhibit high personal risk ratings on a variety of hazards, particularly in relation to personal health (e.g., abortion, viruses—Johnson et al., 2019) as well as security issues (e.g., public transportation—Chauvin & Chassang, 2022; mugging—Marris et al., 1998)—thus possibly reflecting the great relevance of those risks to fatalists (Johnson & Swedlow, 2019b). All together, this suggests that fatalists may adopt a number of different responses to risk issues (Perri 6 et al., 2002). They tend to cope with social context and capricious nature, staying largely out of societal and environmental risk issues while using their little agency to take care of themselves through a strategy of personal survival (Entwistle, 2021; Thompson et al., 1990, 1999). Such a strategy thus disposes persons who are fatalistic to be particularly worried about hazards which affect them personally and inevitably. As a whole, this pattern reveals the great relevance of the (lack of) control over risk to fatalists, leading them to be disengaged from “out of control” large‐scale hazards (e.g., climate change—Johnson & Swedlow, 2019b; earthquakes—Sun et al., 2022) but receptive to those (just as out of control) they personally suffer (e.g., vaccines—Kiss et al., 2020).

It is quite clear from this review that there is an empirical proximity between hierarchy and individualism regarding risk perception. Persons of hierarchical and individualistic orientations exhibit strong similarity in their pattern of perceived risk. Furthermore, although conceptually distinct, both cultural orientations have a theoretical affinity in such a way that an “opposites attract” mechanism of cooperation applies, especially when it is embedded in a specific cultural context that promotes an establishment alliance (Mamadouh, 1999; Wildavsky, 2006). Theoretically, “market and hierarchy make a formidably stable combination” (Douglas & Wildavsky, 1982, p.181), by providing functional services for each other: Hierarchists receive economic growth from individualistic markets to advance their collectivist projects and enhance capacity for entrepreneurial innovation, while individualists gain stability in property rights and protection against competitors (Sotirov & Winkel, 2016). As such, both perceive their strategic alliance as a “winning coalition” offering them an opportunity to meet their ends, whereas otherwise either hierarchists or individualists are likely to break up such a coalition (Ripberger et al., 2011; Sotirov & Memmler, 2012). Combined with accumulating evidence that hierarchy and individualism are positively and highly correlated with each other (e.g., Brenot et al., 1998; Johnson & Swedlow, 2019b; Kim & Kim, 2019), some authors proposed to combine hierarchy and individualism into a single hierarchical–individualistic worldview in their study of risk perception (e.g., Chauvin & Chassang, 2022; Kahan, 2012). Given the relevance of such a combination in the current French political context where hierarchists and individualists presently form an ad hoc alliance (see Chauvin & Chassang, 2022, for details), we opted for this combination in the present study conducted in France. Therefore, throughout the article, we will use the hierarchical–individualistic orientation as a whole.

The proponents of the cultural theory of risk have presented a functionalist account focused on the contending ways of life (not on the society as a whole) to explain the way preferences about risk are formed (Thompson et al., 1999). This functional explanation suggests that individuals form risk perception congruent with their favorite way of life precisely because holding these positions about risk promotes and reinforces their idealized way of life (Thompson et al., 1990). In this view, individual preferences can be understood only as being defined by (and maintained within) specific social contexts (Thompson et al., 1999). Other researchers, relying on the psychometric theory of risk and regarding cultural worldviews as stable features of individuals, have posited that culture is connected to perceptions of risk through a set of psychological processes (Kahan, 2012, 2011). In this view, individual preferences are based on a cultural appraisal of risk issues that leads people to impute specific meaning to risk situations, thus processing information in a manner that confirms their cultural orientation (Jenkins‐Smith, 1993, 2001; Kahan et al., 2017; Peters et al., 2004). In the same vein, cultural settings have also been used to shed some light on the formation of people's preferences through the somatic marker mechanism, insofar as they participate in tagging the specific situation as either attractive or aversive (Verweij & Damasio, 2019; Verweij et al., 2015). Functional explanation operates at the social level, whereas psychological processes operate at the individual level. So these are largely complementary explanations, not rival ones. This view is fully in line with Slovic's stance that “dread [for example] appears to be both psychological and cultural, and it does not seem worthwhile [to him] to attempt to disentangle these various aspects” (Slovic, 1992, p.150).

1.3. The psychometric paradigm and the cultural theory of risk

From their launch in the early 1980s, the psychometric paradigm and the cultural theory of risk have been mostly used independently of each other to explain risk perception (Marris et al., 1998). Only few have considered both approaches together in order to compare them as explanations for perceived risk. On the one hand, using individual data, Marris et al. found some variability across respondents in their ratings of the same risk issue on the same risk characteristics. Thus they demonstrated that risk characteristics can be interpreted and rated differently depending on cultural orientation (Marris et al., 1997, 1998). This is an important finding because it encouraged to consider the qualitative risk characteristics not (only) as inherent attributes of the hazards themselves but (also and especially) as constructs of the individuals used as vectors of their views about hazardous activities. On the other hand, Sjöberg claimed that cultural theory as a whole should be dismissed and that cultural worldviews are not core factors in risk perception (e.g., Sjöberg, 1997, 2003). He also estimated the psychometric approach to risk perception as “clearly insufficient” (Sjöberg, 1996, p. 224). Sjöberg's stance about cultural theory and psychometric approach—although defensible—has been largely criticized by many scholars who find both approaches useful in explaining risk perceptions (e.g., Slovic, 2016; Slovic & Peters, 1998; see Johnson & Swedlow, 2019a, for a recent review).

None of these studies, yet, really combined the psychometric paradigm with the cultural theory of risk into an integrative model, enabling both approaches to help each other and shedding light on their relationship with perceived risk. The first attempt to fuse the cultural theory of risk with the psychometric paradigm has been undertaken by Kahan et al. (2011) into what they call the cultural cognition theory. Cultural cognition has endeavored to combine both approaches into a coalition framework that allows to shed light on the risk perception issue by means of both psychological and cultural explanations. This conceptualization provided empirical support for a series of individual‐level mechanisms showing how cultural theory works; it also yielded insights into how one and the same process can generate individual differences in risk perception between people with distinct cultural orientations (see Kahan, 2012, for details). In the last decade, additional work on cultural cognition has made some clarification about the conditions in which such mechanisms are triggered, mainly identity‐protective cognition. Identity‐protective cognition involves conforming assessments of information to culturally congruent affective appraisals of a risk source. Individuals thus can credit or react dismissively to (the same) risk information, depending on its consonance with their cultural values and the resulting affective reactions. Nevertheless, such a cultural division seems to occur only when people are exposed to communications incorporating antagonistic memes (Kahan et al., 2017). So called, it is easy to view that the mechanisms of cultural cognition mostly refer to heuristic processes. Here is our point. This line of work is based on the cognitive mindset of the psychometric theory, but it does not really make use of the psychometric paradigm in itself. Of particular interest here is that cultural cognition does not use any qualitative risk characteristic featured in the psychometric paradigm, although some of them have proven to be key determinants of perceived risk (e.g., Slovic, 1987; see also Siegrist & Arvai, 2020). In other words, in line with the cultural cognition project, we aimed to explore the psychological mechanisms through which cultural values influence risk perceptions. Unlike Kahan et al., however, we were not interested in heuristic processes but in qualitative risk characteristics that might mediate the impact of culture on risk perception.

As a reminder, worldviews are viewed as orienting dispositions that guide the evaluation of hazardous events in a manner supportive of individuals’ values. According to us, this dynamics not only leads culturally diverse individuals to hold distinct views about how much risky is one hazard but also drives their allocation of attributes for characterizing it¹. We then posited that people confer specific characteristics on hazardous activities or technologies (common or dreadful, known or unknown, beneficial or costly, affecting few or many people, under or out of control) depending on their cultural orientation and associated values; in turn, such characteristics have an impact on the perceived riskiness of these hazards. As such, this mechanism can help explain how individual differences in the perceived riskiness of a given hazard are generated from different worldviews. Based on this assumption, we made several more specific, testable predictions. As “technologically optimistic persons” (Wildavsky & Dake, 1990, p.169), hierarchists–individualists are expected to express risk skepticism about environmental and technological issues (e.g., genetically modified organisms [GMO], nuclear power); that is, they should see such hazards as low risk because they perceive them as not very dreadful, highly beneficial, and well understood to experts. As part of the establishment, they are hypothesized to exhibit high risk sensitivity toward any form of disruption or deviance (e.g., social movement and cannabis); that is, they should worry more about such dreadful activities that clearly challenge the system in power. Egalitarians believe that “an inegalitarian society is likely to insult the fragile environment just as it exploits the poor” (Dake, 1991, p.67). Therefore, they are expected to be very sensitive to technological and environmental hazards (e.g., global warming, GMO, and nuclear power); that is, they should be most concerned about risk issues they view as very dreadful, with no social benefit but affecting a large number of people. They might be more skeptical and less concerned, however, about disruptive activities (e.g., social movement), seeing them as a good way to make radical changes. Fatalists tend to see the world as “lottery‐controlled” (Entwistle, 2021; Thompson et al., 1990). They are also characterized by little solidarity and believe that they have little political efficacy—leading them to consider that other people or organizations should be responsible for managing public risk issues such as societal or environmental hazards (Trousset et al., 2015; Yuan et al., 2022). As a result, fatalists are likely to display risk neutrality for global hazards they judge as being “not their business” (e.g., global warming, GMO, and social movement). While still seeing the world in terms of randomness and powerlessness, [some] fatalists would be receptive to hazards confronting them personally—pushing them to develop a bespoke understanding of such threats and a will to guard against them in order to defend their interests (Entwistle, 2021). As a result, fatalists should report risk sensitivity to security and health‐related hazards that fall under their private scope (e.g., public transportation, cannabis, and coronavirus); that is, they should see them as high risk because they fear those hazards with immediate and uncontrollable outcomes for themselves. See Table 1 for a summary of hypotheses regarding how cultural values are associated with risk characteristics and risk perception as a function of risk issue.

TABLE 1.

Summary of hypothesized associations between cultural values and risk‐related variables.

	Risk‐related variables
Cultural values	Benefit	Dread	Unknown to experts	Number of exposed	Lack of control	Perceived risk
	Cannabis
Hierarchy–individualism	(−)	(+)				(+)
Egalitarianism
Fatalism		(+)			(+)	(+)
	Coronavirus
Hierarchy–individualism
Egalitarianism
Fatalism		(+)			(+)	(+)
	Global warming
Hierarchy–individualism
Egalitarianism	(−)	(+)		(+)		(+)
Fatalism
	GMO
Hierarchy–individualism	(+)	(−)	(−)			(−)
Egalitarianism	(−)	(+)		(+)		(+)
Fatalism
	Nuclear power
Hierarchy–individualism	(+)	(−)	(−)			(−)
Egalitarianism	(−)	(+)		(+)		(+)
Fatalism
	Public transportation (by plane/train)
Hierarchy–individualism
Egalitarianism
Fatalism		(+)			(+)	(+)
	Social movement
Hierarchy‐individualism	(−)	(+)				(+)
Egalitarianism	(+)	(−)				(−)
Fatalism

Note: For each risk issue, a (−) sign signifies a predicted negative association between worldview and risk‐related variables, whereas a (+) sign signifies a predicted positive association.

Abbreviation: GMO, genetically modified organisms.

2. METHOD

2.1. Participants and procedure

A total of 887 French citizens were asked to participate voluntarily in this study. After removing the responses from participants with significant missing data (over 10% of total responses), 629 participants remained for analysis (31% men; mean age = 33.8 years, range 16–77). A sensitivity power analysis revealed that our study is suitably powered to detect at least correlations exhibiting small effects (|r| = 0.11) with a two‐tailed alpha of 0.05 and a power of 0.80 (G*Power 3.1.9.2; Faul et al., 2009).

Participants were recruited online through social media during the first French lockdown due to the coronavirus pandemic (March–April, 2020). They received an invitation e‐mail with a brief description of the survey and an embedded survey link to complete the questionnaire battery online. We used the Qualtrics survey platform to collect our data. The data set is available here: https://osf.io/x46qb/.

2.2. Measures

Cultural values were measured first through an 18‐item scale of worldviews recently developed in Chauvin and Chassang (2022). Participants were asked to rate each item on a 5‐point scale varying from strongly disagree to strongly agree. Higher scores indicated stronger endorsement of cultural values. This scale was used because it is a valid and reliable tool for combining hierarchy and individualism into a single hierarchical–individualistic worldview, thus reflecting the cultural alliance of hierarchists and individualists in the current French society (see Johnson & Swedlow, 2024, about alternative cultural theory survey measures). Confirmatory factor analysis and reliability indices provided support for a three‐factor solution reflecting hierarchy–individualism, egalitarianism, and fatalism, respectively. Accordingly, a mean score was computed for each factor by averaging the scores of the six corresponding items. The model is detailed in Table 2.

TABLE 2.

Standardized factor loadings and reliability values for the three‐factor model of worldviews.

	Worldviews
Worldview indicators	Hierarchy–individualism	Egalitarianism	Fatalism
HI1—Respect for authority is one of the most important things that children should learn	0.75
HI2—One of the problems with people is that they challenge authority too often	0.57
HI3—A free society can only exist by giving companies the opportunity to prosper	0.51
HI4—Continued economic growth is the answer to improved quality of life	0.45
HI5—The police should have the right to listen to private phone calls when investigating crime	0.43
HI6—I like to plan carefully so that financial risks are not taken	0.27
E1—If people in this country were treated more equally we would have fewer problems		0.69
E2—The world could be a more peaceful place if its wealth were divided more equally among nations		0.68
E3—The difference between rich and poor nations is not right		0.65
E4—What this country needs is a “fairness revolution” to make the distribution of goods more equal		0.62
E5—Those who get ahead should be taxed more to support the less fortunate		0.55
E6—I would support a tax change that made people with large incomes pay more		0.53
F1—Most people make friends only because friends are useful to them			0.57
F2—A person is better off if he or she doesn't trust anyone			0.55
F3—Cooperating with others rarely works			0.45
F4—There is no use in doing things for other people—you only get it in the neck in the long run			0.45
F5—The future is too uncertain for a person to make serious plans			0.42
F6—It seems to me that, whoever you vote for, things go on pretty much the same			0.32
Reliability values	0.67	0.79	0.62

Note: Indicators are ordered according to the size of their factor loadings and numbering has been added to make reading easier. Fit indices were comparative fit index (CFI) = 0.90, Tucker–Lewis Index (TLI) = 0.88 (values close to 0.90 refer to an acceptable fit), root mean square error of approximation (RMSEA) = 0.05 (value below .06 refers to a good fit), and standardized root mean square residual (SRMR) = 0.06 (value below 0.08 refers to a good fit) (Hu & Bentler, 1999). All loadings were statistically significant (p < 0.001). Reliability values are Rho coefficients (cut‐off value for acceptable reliability is 0.60; Raines‐Eudy, 2000). N = 629.

Risk characteristics and risk perception were then assessed for seven hazards selected to represent controversial issues: global warming, GMO, nuclear power, cannabis, social movement (yellow vests, strikes), public transportation (plane/train), and coronavirus. To assess risk characteristics, respondents were asked to rate each hazard on five 7‐point scales drawn from psychometric literature (Mullet et al., 1993; Slovic et al., 1985): benefit (To what extent is the risk beneficial to society? Not at all beneficial—Very beneficial), dread (Is this a risk that people have learned to live with and can think about reasonably calmly, or is it one that people have great dread for? Common—Dread), number of people exposed (How many people are exposed to the risk in the country? Few—Many), knowledge to experts (To what extent is the risk known to science? Risk level not known—Risk level known precisely), and control over risk (To what extent is the risk associated with uncontrollable adverse effects? Risk cannot be controlled—risk can be controlled). The sequence of the psychometric scales was counterbalanced to avert potential order effects. Higher scores, respectively, indicated greater benefit, higher level of dread, many people exposed, lower level of knowledge (after inverting score), and fewer control over risk (after inverting score). These scales have been selected primarily because they best operationalize the psychometric dimensions hypothesized to be important to hierarchists–individualists, egalitarians, and fatalists in judging risk; in addition, the benefit, dread, and number of people exposed scales most contribute to those dimensions in a French context (see Mullet et al., 1993). To measure perceived risk, participants were asked to express the level of risk (for all people in the whole world) of being seriously ill, wounded, or dying from each of the seven hazards (listed alphabetically). They used an 11‐point scale ranging from 0 (no risk) to 100 (extremely severe risk). Higher scores indicated greater concern about risk.

2.3. Data analytic strategy

All the analyses were performed using Mplus (version 8.7, Muthén & Muthén, 1998–2017). We hypothesized a mediation mechanism in which cultural values (as predictors) impact risk perceptions (as outcomes) through multiple risk characteristics (as mediators). In statistical terms, it means that a parallel multiple mediator model—that allows for multiple processes to be at work simultaneously—needs to be assessed for each risk issue². For building such complex models, we relied on a stepwise procedure. The first step involved running bivariate correlation analyses among the variables to evaluate the plausibility of hypothesized models and refine them in order to establish the most appropriate pattern of predictor–mediator–outcome relationships (thus avoiding potential problems with omitted mediators or with collinearity between mediators that could muddle the results; Hayes, 2022; Muthén et al., 2016). The second step consisted of conducting several multiple regression analyses to test direct effects and offer definitive guidelines in defining the specific mediation model to be retained and assessed for each risk issue. As a third step, we moved into a path analysis framework to examine mediation mechanisms by testing indirect and total effects. A bootstrap method was used to provide the nonsymmetric 95% confidence intervals for inference about (the significance of) indirect effects (Hayes, 2022; Williams & MacKinnon, 2008).

3. RESULTS

3.1. Associations between risk characteristics and risk perception

Table 3 shows the correlations between risk characteristics and risk perception. For all hazards, the level of perceived risk was negatively associated with the level of benefits while positively related to dread, number of people exposed, and lack of control. The more an individual perceived a hazard as low in benefit, dreadful, affecting a high number of people, and out of control, the more he/she perceived a high level of risk from this hazardous activity. A pattern of negative, relatively small or no significant correlations was observed between the dimension unknown to experts and perceived risk.

TABLE 3.

Correlations between risk characteristics and risk perception (N = 629).

	Perceived risk
Risk characteristics	Cannabis	Coronavirus	Global warming	GMO	Nuclear power	Public transportation	Social movement
Benefit	−0.45 ****	−0.24 ****	−0.25 ****	−0.56 ****	−0.50 ****	−0.41 ****	−0.35 ****
Dread	0.64 ****	0.58 ****	0.55 ****	0.72 ****	0.68 ****	0.61 ****	0.62 ****
Unknown to experts	−0.19 ****	−0.10 *	−0.17 ****	−0.10 *	0.03	−0.09	−0.18 ****
Number of exposed	0.54 ****	0.63 ****	0.55 ****	0.65 ****	0.62 ****	0.44 ****	0.51 ****
Lack of control	0.35 ****	0.23 ****	0.15 ****	0.34 ****	0.35 ****	0.19 ****	0.21 ****

Abbreviation: GMO, genetically modified organisms.

^* p < 0.05.

^**** p < 0.001.

3.2. Associations between cultural values, risk characteristics, and risk perception

Table 4 shows the correlations between cultural values and risk‐related variables (risk characteristics and perceived risk). We observed that associations were largely a function of risk issues. Hierarchy–individualism was significantly associated with most of risk characteristics and risk ratings for cannabis and coronavirus, but it showed very low or no significant correlation with the global warming, GMO, nuclear power, or public transportation risk issues. Specifically, compared to those who scored low, individuals scoring high on hierarchy–individualism were more likely to perceive cannabis and coronavirus as low in benefit, [but] dreadful, affecting a high number of people, and high in risk. Conversely, egalitarianism had moderate associations with perceived risk ratings and most of risk characteristics related to global warming, GMO, nuclear power, and public transportation, but it was weakly correlated with cannabis and uncorrelated with coronavirus. In other words, the more the respondents’ worldview was egalitarian, the more they viewed global warming, GMO, nuclear power, and public transportation as low in benefit, [but] dreadful, affecting a high number of people, out of control (except for GMO), and the more was their perceived risk of these issues. The fatalistic worldview showed associations mainly to cannabis, coronavirus, nuclear power, and public transportation. Individuals reporting high levels of fatalism were more likely than individuals who were low in fatalism to associate cannabis with high dread, high number of people exposed, strong lack of control and high risk, as well as to perceive public transportation as being dreadful and risky, and coronavirus and nuclear power as being risky. Finally, social movement was an interesting issue insofar as it led to opposing views over risk from people holding different worldviews. Specifically, the more the respondents’ worldview was hierarchical–individualistic, the more was their perceived risk of social movement, and the more they considered this issue as low in benefit, dreadful, affecting a large number of people, [but] relatively well‐known to experts. The exact opposite pattern was found for egalitarians (except for the unknown to experts dimension, which was not significantly associated with egalitarianism). As for fatalists, individuals scoring high on fatalism were more likely than those who scored low to perceive social movement as affecting a high number of people and high in risk.

TABLE 4.

Correlations between cultural values, risk characteristics and risk perception (N = 629).

	Risk‐related variables
Cultural values	Benefit	Dread	Unknown to experts	Number of exposed	Lack of control	Perceived risk
	Cannabis
Hierarchy–individualism	−0.23 ***	0.29 ****	0.01	0.16 ****	0.08	0.32 ****
Egalitarianism	0.13 ***	−0.10 *	0.05	−0.07	−0.08 *	−0.11 *
Fatalism	−0.03	0.16 ****	0.08	0.10 *	0.09 *	0.16 ****
	Coronavirus
Hierarchy–individualism	−0.10 *	0.23 ****	−0.05	0.18 ****	0.02	0.28 ****
Egalitarianism	0.02	0.01	−0.02	0.01	−0.00	0.01
Fatalism	−0.02	0.04	−0.01	0.02	−0.02	0.12 *
	Global warming
Hierarchy–individualism	0.05	−0.06	0.04	−0.08 *	−0.07	−0.09 *
Egalitarianism	−0.08 *	0.22 ***	−0.05	0.18 ***	0.09 *	0.21 ****
Fatalism	0.11 *	−0.04	0.01	−0.12 *	−0.06	−0.01
	GMO
Hierarchy–individualism	−0.03	−0.02	0.01	−0.03	−0.06	0.01
Egalitarianism	−0.09 *	0.16 ****	−0.02	0.16 ****	0.06	0.17 ****
Fatalism	0.01	−0.01	0.06	−0.01	−0.02	0.03
	Nuclear power
Hierarchy–individualism	0.05	−0.01	0.05	−0.05	−0.13 ***	−0.02
Egalitarianism	−0.27 ****	0.21 ****	−0.03	0.25 ****	0.10 *	0.26 ****
Fatalism	−0.02	0.04	0.04	0.04	−0.08	0.10 *
	Public transportation (by plane/train)
Hierarchy–individualism	0.03	0.08	0.05	−0.00	−0.04	0.04
Egalitarianism	−0.17 ****	0.13 ***	−0.02	0.17 ****	0.08 *	0.17 ****
Fatalism	−0.00	0.10 *	0.07	0.06	0.00	0.12 **
	Social movement
Hierarchy–individualism	−0.36 ****	0.30 ****	−0.15 ***	0.21 ****	−0.01	0.31 ****
Egalitarianism	0.43 ****	−0.22 ****	0.05	−0.09 *	−0.06	−0.22 ****
Fatalism	0.01	0.05	−0.01	0.10 *	−0.02	0.09 *

Note: Partial correlation analyses between cultural values and risk perception controlling for participant gender were performed. Although partial correlations (not shown) were, logically, slightly lower in magnitude than full correlations (Table 3), both exhibited a very similar pattern.

Abbreviation: GMO, genetically modified organisms.

^* p < 0.05.

^** p < 0.01.

^*** p < 0.005.

^**** p < 0.001.

3.3. Direct effects of cultural values on risk characteristics and direct effects on cultural values combined with risk characteristics on risk perception

Based on observed associations, risk characteristics (the mediators) were regressed on cultural values (the predictors), and then riskiness (the outcome) was regressed on both cultural values and risk characteristics³ (see Table 5). For cannabis, coronavirus, and social movement, hierarchy–individualism had a direct negative impact on the level of perceived benefit (and experts’ lack of knowledge for social movement only) and a direct positive impact on feelings of dread and number of people exposed (see Table 5, model 1, for details). Irrespective of hierarchy–individualism scores, benefit (but not the unknown to expert dimension) was significantly negatively related to the perceived risk of the three hazards, whereas dread and number of people exposed were significantly positively related to risk perception. We also observed a significant, positive, direct effect of hierarchy–individualism on risk perception independent of the effect of the three risk characteristics (see Table 5, model 2, for details). Hierarchy–individualism was not significantly associated with any of other risk characteristics and hazards.

TABLE 5.

Results of multiple regression analyses testing direct effects of cultural values on risk characteristics (Model #1) and direct effects of cultural values combined with risk characteristics on risk perception (Model #2).

			Direct effects
			Model 1	Model 2
Cultural values (P)	Risk characteristics (M)	Perceived risk (C)	P on M	M on C	P on C
Hierarchy–individualism	Benefit	Cannabis	−0.19 (0.04) ****	−0.14 (0.04) ****	0.11 (0.03) ****
	Dread		0.25 (0.04) ****	0.38 (0.04) ****
	Number of exposed		0.13 (0.04) ***	0.31 (0.03) ****
Egalitarianism	Benefit		0.07 (0.04)
	Dread		−0.03 (0.04)
	Lack of control		−0.08 (0.04)	0.11 (0.03) ***
Fatalism	Dread		0.12 (0.04) ***		0.04 (0.03)
	Number of exposed		0.08 (0.04)
	Lack of control		0.10 (0.04) *
Hierarchy–individualism	Benefit	Coronavirus	−0.10 (0.04) *	−0.09 (0.03) **	0.11 (0.03) ****
	Dread		0.23 (0.04) ****	0.33 (0.05) ****
	Number of exposed		0.19 (0.04) ****	0.44 (0.05) ****
Hierarchy–individualism	Number of exposed	Global warming	−0.02 (0.03)
Egalitarianism	Benefit		−0.08 (0.04) *	−0.13 (0.05) ***	0.06 (0.03)
	Dread		0.22 (0.05) ****	0.36 (0.06) ****
	Number of exposed		0.19 (0.06) ***	0.34 (0.06) ****
	Lack of control		0.09 (0.04) *	0.08 (0.03) *
Egalitarianism	Benefit	GMO	−0.09 (0.04) *	−0.21 (0.04) ****	0.03 (0.03)
	Dread		0.16 (0.04) ****	0.45 (0.04) ****
	Number of exposed		0.16 (0.04) ****	0.32 (0.04) ****
Egalitarianism	Benefit	Nuclear power	−0.27 (0.04) ****	−0.18 (0.03) ****	0.04 (0.03)
	Dread		0.21 (0.04) ****	0.40 (0.04) ****
	Number of exposed		0.25 (0.04) ****	0.30 (0.04) ****
	Lack of control		0.10 (0.04) *	0.11 (0.03) ****
Fatalism					0.09 (0.03) ***
Egalitarianism	Benefit	Public transportation	−0.17 (0.04) ****	−0.18 (0.04) ****	0.05 (0.03)
	Dread		0.13 (0.04) ****	0.43 (0.04) ****
	Number of exposed		0.17 (0.04) ****	0.22 (0.04) ****
	Lack of control		0.08 (0.04) *	0.06 (0.03)
Fatalism	Dread		0.07 (0.04) *		0.06 (0.03) *
Hierarchy–individualism	Benefit	Social movement	−0.28 (0.04) ****	−0.11 (0.04) **	0.08 (0.03) *
	Dread		0.26 (0.04) ***	0.42 (0.05) ****
	Unknown to experts		−0.15 (0.04) ***	−0.06 (0.03)
	Number of exposed		0.19 (0.04) ****	0.25 (0.04) ****
Egalitarianism	Benefit		0.37 (0.04) ****		−0.05 (0.04)
	Dread		−0.15 (0.04) ****
	Number of exposed		−0.06 (0.04)
Fatalism	Number of exposed		0.07 (0.04)

Note: Standardized regression weights (beta values) are reported. Figures in parentheses are standard errors (SE). Attentive readers might be surprised to find an estimation of the direct effect of F on nuclear power absent any (significant) mediators. The reason is that the association between F and nuclear power is remaining after accounting for the effect of E on nuclear power.

Abbreviations: C, criterion; GMO, genetically modified organisms; M, mediator; P, predictor.

^* p < 0.05.

^** p < 0.01.

^*** p < 0.005.

^**** p < 0.001.

For global warming, GMO, nuclear power, and public transportation, egalitarianism had a direct negative impact on the level of perceived benefit and a direct positive impact on feelings of dread, number of people exposed, and lack of control (except for GMO). Regardless of egalitarianism differences, benefit was negatively related to the perceived risk of the four hazards, whereas dread, number of people exposed, and lack of control (except for public transportation) were positively related to risk perception. Regarding social movement, egalitarianism had a direct positive impact on the level of perceived benefit and a direct negative impact on feelings of dread. Benefit and dread were significantly related to the perceived risk of social movement, no matter of egalitarianism scores. Controlling for risk characteristics, egalitarianism did not directly impact the perception of risk associated with global warming, GMO, nuclear power, public transportation, and social movement. No further significant associations were observed between egalitarianism and other risk issues or characteristics.

For cannabis and public transportation, fatalism directly yielded an increase in feelings of dread and perceived lack of control (for cannabis only). No matter how fatalism scores were, dread and lack of control were positively related to risk perception. Controlling for risk characteristics, fatalism did not directly impact the perception of risk associated with cannabis but directly impacted public transportation risk perception. Results also showed a significant positive direct impact of fatalism on nuclear power risk perception. Fatalism was not significantly associated with any of other risk issues or characteristics.

In brief, hierarchy–individualism was a significant predictor of higher levels of concern (characterized mostly by low benefit, high dread, great number of people exposed, and high risk) about cannabis, coronavirus, and social movement. The egalitarian worldview was a significant predictor of higher levels of concern (characterized mostly by low benefit, high dread, great number of people exposed, lack of control, and—in turn—high risk) regarding global warming, GMO, nuclear power, and public transportation. Egalitarianism was also found to predict that social movement is beneficial, common, and—in turn—not risky. The fatalistic worldview was a significant predictor of higher levels of concern for cannabis (perceived as dreadful, out of control, and—in turn—as risky), public transportation (perceived as dreadful and risky), and nuclear power (perceived as high in risk).

3.4. Mediating effects of risk characteristics on the relationship between cultural values and risk perception

Based on regression analysis results, we developed path models for each risk issue. Figure 1 displays these models. Fit indices were excellent for each model. Table 6 shows the standardized total, specific indirect, and total indirect effects (and their significance), as well as the standardized nonsymmetric 95% confidence intervals of the total indirect effects. Regarding cannabis, the final path model (diagram A) showed that hierarchy–individualism contributed both directly and indirectly, whereas fatalism contributed only indirectly to explain the variance in risk perception (54% in total). Precisely, focusing on indirect effects, both cultural values had a significant total indirect effect on cannabis risk perception through benefit, dread, and number of people exposed for hierarchy–individualism, and through dread and lack of control for fatalism. Both effects were significant and positive, meaning that those higher in hierarchy–individualism or fatalism were higher in their perception of risk associated with cannabis as a result of the effect of these cultural values on the risk characteristics just mentioned, which in turn influenced risk perception (see Table 6). Within the coronavirus domain, the final model (diagram B) indicated that hierarchy–individualism had direct and indirect paths to risk perception, accounting for 53% of the variance in risk perception. Inspection of the indirect effects revealed that benefit, dread, and number of people exposed mediated the relationship between hierarchy–individualism and coronavirus risk perception. The total indirect effect was significant and positive, suggesting that higher levels of hierarchy–individualism were associated with greater feelings of dread and number of people exposed along with lower perceived benefits from coronavirus, which, in turn, were related to higher levels of perceived risk. The final path models for global warming (diagram C), GMO (diagram D), and nuclear power (diagram E) explained, respectively, 43%, 65%, and 61% of the variance in risk perception. They revealed indirect paths connecting egalitarianism to risk perception via benefit, dread, number of people exposed, and lack of control (except for GMO)⁴. All three total indirect effects were significant and positive, meaning that persons with high scores on the egalitarian dimension obtained high scores of perceived risk associated with those hazards as a result of the effect of egalitarianism on the risk characteristics above mentioned, which in turn had an effect on risk perception. Regarding public transportation (by plane or train), the final path model (diagram F) showed that egalitarianism contributed indirectly, whereas fatalism contributed directly and indirectly to explaining the variance in risk perception (45% in total). Both cultural values had a significant total indirect effect on risk perception through benefit, dread, and number of people exposed for egalitarianism and through dread for fatalism. Both effects were significant and positive, meaning that those higher in egalitarianism or fatalism were higher in their perception of risk associated with public transportation, as a result of the effect of these cultural values on the risk characteristics just mentioned, which in turn impacted risk perception. Within the social movement domain, the final model (diagram G) accounted for 47% of the variance in total in risk perception. This model displayed indirect paths from egalitarianism to risk perception through benefit and dread. Higher levels of egalitarianism were associated with lower levels of dread and higher levels of perceived benefit from social movement, which in turn were associated with lower levels of perceived risk—consistent with the negative total indirect effect. This model also indicated that hierarchy–individualism had direct and indirect paths to risk perception through benefit, dread, and number of people exposed. The total indirect effect was significant and positive, suggesting that higher levels of hierarchy–individualism were associated with greater feelings of dread and number of people exposed along with lower perceived benefits from social movement, which in turn were associated with higher levels of perceived risk.

FIGURE 1.

Path models of risk perception for all risk issues. The values above one‐headed arrows are standardized regression coefficients. The values at double‐headed arrows are correlations (residual covariances between the mediator variables were accounted for in models but they were not mentioned for the sake of readability). Single‐headed arrows without originating variables represent residuals. Numbers above the rectangles reflect the amount of explained variance of the corresponding variables. Fit indices are reported below each path model. Path diagrams of the final models for the cannabis (diagram A), coronavirus (diagram B), global warming (diagram C), genetically modified organisms (GMO) (diagram D), nuclear power (diagram E), public transportation (diagram F), and social movement (diagram G) risk issues.

TABLE 6.

Standardized effects (total and indirect), significance of the standardized effects (total and indirect), and standardized nonsymmetric confidence intervals of the indirect effects.

	Total effects on	Indirect effects						Nonsymmetric confidence intervals
		Specific indirect					Total indirect
Cultural values	Perceived risk	Benefit	Dread	Unknown to experts	Number of exposed	Lack of control		Lower	Upper
	Cannabis
Hierarchy–individualism	0.29 ****	0.03 ***	0.09 ****		0.05 ***		0.17 ****	0.12	0.22
Fatalism	0.05 ***		0.04 **			0.01 †	0.05 ***	0.02	0.08
	Coronavirus
Hierarchy–individualism	0.28 ****	0.01 †	0.08 ****		0.08 ****		0.17 ****	0.11	0.22
	Global warming
Egalitarianism	0.16 ****	0.01	0.08 ****		0.06 **	0.01	0.16 ****	0.09	0.24
	GMO
Egalitarianism	0.14 ****	0.02 *	0.07 ***		0.05 ***		0.14 ****	0.07	0.21
	Nuclear power
Egalitarianism	0.23 ****	0.05 ****	0.09 ****		0.08 ****	0.01 †	0.23 ****	0.16	0.29
Fatalism	0.09 ***
	Public transportation (by plane/train)
Egalitarianism	0.13 ****	0.03 ***	0.06 ***		0.04 ***		0.13 ****	0.08	0.18
Fatalism	0.10 **		0.03 *				0.03 *	0.002	0.06
	Social movement
Hierarchy–individualism	0.30 ****	0.04 ***	0.12 ****		0.05 ****		0.21 ****	0.16	0.26
Egalitarianism	−0.10 ****	−0.05 ***	−0.05 ***				−0.10 ****	−0.15	−0.06

Note: Confidence intervals have been obtained using bootstrapping. They are reported for the total indirect effects. Confidence intervals do not include 0 for any specific indirect effect except for the two non‐marked ones.

Abbreviation: GMO, genetically modified organisms.

^† p < 0.10.

^* p < 0.05.

^** p < 0.01.

^*** p < 0.005.

^**** p < 0.001.

4. DISCUSSION

In an innovative way, the goal of the study was to examine how culturally diverse individuals tend to form their perceptions of risk. To address this issue, we were interested in qualitative risk characteristics as key features of the mediation mechanism through which cultural orientation could impact risk perception. The main hypothesis was that people assign specific characteristics on various hazards depending on their cultural worldviews; in turn, these characteristics have an influence on the level of perceived risk associated with those hazards.

Collectively, our findings provide strong support for such a mechanism [by] showing the major role of dread, number of people exposed, and benefit as mediators of most of the relationships between cultural orientation and risk perception. In our view, this is an important contribution to the research on risk perception because (1) it meets a strategic expectation from Douglas (1997) to “marry” cultural theory and the psychometric theory of risk in order to better understand the perception of risk (see also Kahan, 2012) and (2) it allows to answer yes to the crucial question “is it possible to identify the hazard characteristics that moderate the association between people's values and risk perceptions?” that Siegrist and Arvai (2020, p. 2202) recently asked. By providing some new insight into how individuals’ cultural worldviews contribute to their perceptions of risk, our work both joins and enriches Kahan's research path, as it sheds light on mechanisms connecting culture to risk perceptions (e.g., Kahan et al., 2010, 2007, 2011). This study is also in line with psychometric research pioneered by Slovic (1987). Indeed, we found that people systematically include dread, number of people exposed, as well as perceived benefit in their risk assessment, thus expanding the broad spectrum of evidence accumulated since four decades regarding the importance of such dimensions (see Chauvin, 2014, for a review). A significant difference with prior work, however, is that the level of knowledge does not have a noticeable impact on riskiness in the present study. Such a minor role has been previously observed in a variety of studies (e.g., Mullet et al., 1993; Sjöberg, 2000; Slovic et al., 1985). Specifically, contrary to our expectation, we did not find that hierarchists–individualists rely on the degree to which hazards are known or unknown to experts for evaluating their riskiness. A plausible explanation could be that trust in experts, rather than scientific knowledge (or lack thereof) per se, really matters to hierarchists–individualists in their perception of risk. Trust in institutions or scientists has indeed been reported to influence risk/benefit perception of technological hazards (e.g., Siegrist, 2000; Siegrist et al., 2000). Recent work displayed the mediating role of trust in various information sources (e.g., government and industry) between cultural worldviews and public responses to environmental hazards in the United States (e.g., Tumlison & Song, 2019) or health hazards in China (Yuan et al., 2024).

Focusing on worldviews, support was partially found for our hypotheses (see Table 7 for a summary).

TABLE 7.

Summary of observed associations between cultural values and risk‐related variables.

	Risk‐related variables
Cultural values	Benefit	Dread	Unknown to experts	Number of exposed	Lack of control	Perceived risk
	Cannabis
Hierarchy–individualism	(−) Y	(+) Y		(+) (Ø)		(+) Y
Egalitarianism
Fatalism		(+) Y			(+) Y	(+) Y
	Coronavirus
Hierarchy–individualism	(−) (Ø)	(+) (Ø)		(+) (Ø)		(+) (Ø)
Egalitarianism
Fatalism		(+) N			(+) N	(+) N
	Global warming
Hierarchy–individualism
Egalitarianism	(−) Y	(+) Y		(+) Y	(+) (Ø)	(+) Y
Fatalism
	GMO
Hierarchy–individualism	(+) N	(−) N	(−) N			(−) N
Egalitarianism	(−) Y	(+) Y		(+) Y		(+) Y
Fatalism
	Nuclear power
Hierarchy–individualism	(+) N	(−) N	(−) N			(−) N
Egalitarianism	(−) Y	(+) Y		(+) Y	(+) (Ø)	(+) Y
Fatalism						(+) (Ø)
	Public transportation (by plane/train)
Hierarchy–individualism
Egalitarianism	(−) (Ø)	(+) (Ø)		(+) (Ø)		(+) (Ø)
Fatalism		(+) Y			(+) N	(+) Y
	Social movement
Hierarchy–individualism	(−) Y	(+) Y		(+) (Ø)		(+) Y
Egalitarianism	(+) Y	(−) Y				(−) Y
Fatalism

Note: For each risk issue, Y indicates that the expected negative (−) or positive (+) association was observed, whereas N indicates that the expected association was not observed. A (Ø) symbol means that no prediction was made but a negative (−) or positive (+) association was observed, whereas a blank cell means no prediction and no observed association.

Abbreviation: GMO, genetically modified organisms.

First, as expected, high scores in hierarchy–individualism were associated with high risk perception for cannabis and social movement both directly and indirectly, through the level of dread, the number of people exposed, and the benefit associated with each issue. Persons with a hierarchical–individualistic orientation try to maintain the system in power because it matches and protects their values (Chauvin & Chassang, 2022; Thompson et al., 1990). Accordingly, transgressive or disruptive activities (such as cannabis consumption or social movement)—ones that are likely to challenge this system—are frightening and penalizing to hierarchists–individualists, leading them to see those activities are risky (see Dake, 1991, or Kiss et al., 2020, for similar results). Further, an increase in hierarchy–individualism scores was associated with an increase in perceptions of risk from coronavirus, partially due to hierarchists–individualists’ tendency to see coronavirus as dreadful, economically detrimental, and impacting a large number of people. Although we did not anticipate it when setting up our assumptions, this relationship is no longer a surprise with accumulating knowledge recently gained about the virus. Coronavirus during the pandemic indeed posed a serious threat to political stability (e.g., loss of trust in governments—Cori et al., 2020; science skepticism—Rutjens et al., 2021), global economy (Nelson et al., 2020), and public health (World Health Organization, 2020); that is, the pandemic went clearly against what hierarchists–individualists value and support, in turn leading them to perceive coronavirus as highly risky. In addition, this result seems to be inconsistent with findings of existing COVID studies, which found a negative association between individualistic values and risk sensitivity toward coronavirus. A large part of the divergence is likely to be related to the way individualism was measured (e.g., using a single item or using survey items emphasizing more the “government over‐regulation facet” than the “economic facet” of individualism—Dryhurst et al., 2020; Siegrist & Bearth, 2021) and/or to the outcome variables being the primary interest of studies (e.g., protective behaviors like social distancing and mask‐wearing—Bazzi et al., 2020; Yuan & Swedlow, 2022). This discrepancy may also reflect between‐country differences as suggested by the recent finding that Chinese individualists perceive the risk from COVID‐19 as high while American individualists assess it as low (Yuan et al., 2024). Contrary to expectation, hierarchy–individualism had no effect on nuclear power and GMO risk perceptions. Further, persons who hold this worldview show no risk skepticism about any environmental and technological issue considered here (including global warming), unlike previous work (e.g., Kahan et al., 2007, 2011). This inconsistent result may well be due to the time of data collection. Both nuclear power and global climate change have been at the center of controversies over the last few decades (Van de Graaff, 2016; Van der Linden, 2015). What happened recently, however—from the 2011 Fukushima nuclear disaster to the 2022 worldwide mega‐fires, through the global temperature rise or the increase of droughts around the world (for details, see the global issue of climate change United Nations, 2023)—resulted undoubtedly in strengthening social conflict and public debates surrounding such issues. This may have led hierarchists and individualists to adopt a more balanced view of costs and benefits, which are now considered hot topics, which is in line with recent findings showing no significant or very weak relationship between hierarchists, individualists, and climate change (Johnson & Swedlow, 2019b), nuclear energy, or GMOs (Kiss et al., 2020). This hypothesis calls for further research.

Second, as we had hypothesized, egalitarians express greater concern about the risk of nuclear power, GMO, and global warming, likely due to their propensity to see those hazards as dreadful, low in benefit, and having high catastrophic potential for the most part. This result is in line with past literature, which repeatedly found that higher scores on egalitarianism are associated with higher levels of perceived technological and environmental risk (e.g., Johnson & Swedlow, 2019b; Kahan et al., 2009, 2012; Kim & Kim, 2019; Yuan et al., 2022). More importantly, this result contributes to explaining how an egalitarian orientation predisposes individuals to risk sensitivity toward such issues. Egalitarians have a strong sense of social solidarity and equality, along with a strong aversion to tampering with nature (Thompson et al., 1990; Van de Graaff, 2016). As a consequence, the more egalitarian respondents are, the more worried they are that technological and environmental issues would (a) threaten present as well as future generations, (b) lead to more benefits for a few while making the majority more vulnerable, and (c) have harmful consequences for the environment (see Ellis & Thompson, 1997; Leiserowitz, 2006, or Thaker et al., 2020, for similar interpretations). On the other hand, we expected and found that higher scores on egalitarianism are associated with lower concern about social movement, due to egalitarians’ view of this activity as being both common and beneficial. Egalitarians value community‐driven decision making and blame the top–down system currently in power (Mamadouh, 1999; Wildavsky & Dake, 1990). They therefore fight for more democratic processes and public participation to risk management strategies (Steg & Sievers, 2000; Tansey & O'Riordan, 1999). Accordingly, any activity allowing egalitarian people to reach this goal is not fearful in their view, but it is good for them and, in turn, not risky (see Chauvin & Chassang, 2022, for a similar view). Unexpectedly, we found that egalitarians credit public transportation with no benefit but view it as a dreaded activity with potential large‐scale consequences, leading to a high level of perceived risk. Perhaps this result is attributable to the fact that people who hold an egalitarian orientation focus primarily on the adverse effects from transport activities. For example, egalitarians may assess transport by plane by thinking about the growing (and increasingly salient) climate impact of aviation (Grewe et al., 2021), the massive energy consumption of trains (railroads as a sector is the biggest industrial consumer of electrical energy in France), or unfortunate events like railroad or airplane accidents (Johnson & Swedlow, 2019b). Future research can provide evidence in support of this assumption.

Third, as expected, we observed that fatalists are indifferent to hazardous activities such as global warming or GMO but tend to be afraid of public transportation and cannabis, thus rating both as high risk. As fatalism scores increase, perceived risks increase as well, partially due to fatalists’ high level of dread for such activities (see Chauvin & Chassang, 2022, or Peters & Slovic, 1996, for similar results). In other words, we found that fatalists display mostly risk neutrality but also some risk sensitivity derived from their feelings of dread, depending on which hazards are involved. Specifically, in accordance with the core of fatalistic orientation, fatalists tend to be resigned toward global hazards like public health issues or environmental threats, all of which overwhelming their personal (little) agency to deal with them (Tansey & O'Riordan, 1999; Thompson et al., 1990; Xue et al., 2014). They therefore can be disengaged from such issues, being refractory for acting and letting “others decide” (Howell et al., 2019; Ripberger et al., 2014; Swedlow et al., 2020, p.45). It could furnish a plausible explanation for why fatalism scores, contrary to expectation, are not related to perceptions of risk for coronavirus. They probably view coronavirus as a pandemic that represents a massive global health crisis overriding, by far, any personal concern (Cori et al., 2020; Van Bavel et al., 2020). Acknowledging clearly their lack of control over such a pandemic, fatalists are likely to rely on others or luck or destiny to remain healthy and avoid becoming infected (e.g., Kwiringira et al., 2019). At the same time, fatalists tend to be susceptible to risk issues, which they perceive as running counter to their strategy of personal survival (Entwistle, 2021). That is, they have particular concern about fearful activities that affect them personally, thus adopting a negative view about them. Regarding cannabis, for instance, we might argue that fatalists are mostly aware of health adverse effects when they assess the riskiness of this substance, focusing on potential effects on [their] neurocognitive development or unintentional overdosing related to the poor regulation of edibles (Williams, 2015). As such, the current study provides support for the idea that “fatalists desire the right to be left alone, to stay out of harm's way, or not, as they so choose” (Dake, 1992, p.30). It echoes the finding that active fatalism can combine an acceptance of the “divine” control of events with a belief in individual agency for coping with health problems (Shahid et al., 2020). At the conceptual level, such a fatalistic pattern fits very well with protective fatalism, whereby some fatalists have developed a situated account of the risks they face and the ameliorative measures that make sense for them (Entwistle, 2021). As a whole, it also helps understand how fatalists should be expected to respond to various risk issues, thus addressing some uncertainty about the relationship between fatalism and risk perception (as pointed by Kahan, 2012, or more recently by Yuan et al., 2022, regarding environment). Somewhat surprisingly, lack of control played no role for fatalistic persons in their risk ratings, perhaps because personal skills were not mentioned in our measure of this dimension. The fatalism literature has indeed suggested that fatalists could respond to risk information as long as they perceive its relevance (Entwistle, 2021). Future work utilizing another operationalization will be beneficial to further examine this assumption (see Slovic et al., 1985, who provide various ways for operationalizing control over risk).

Of course, this study is not without limitations. First, although large enough, our sample of participants is not representative of the French population but consists predominantly of female and young participants, thus limiting the possibility of generalizing our results. Our recruitment through social media could be responsible for this representativeness bias. Future work should aim to replicate our findings in a representative sample of French derived from a random stratified sampling technique. Second, although some risk characteristics such as dread turn out to be key dimensions, other characteristics, such as unknown to experts or (lack of) control over risk, tend to be less significant in the account of risk perception. Relatedly, we opted for a limited set of five risk characteristics, which may not encompass all relevant dimensions. By and large, there was indeed around half of unexplained variance in any mediation model. This may imply that other risk characteristics or another operationalization thereof, such as preventive control over risk by personal skills or diligence (Slovic et al., 1985), may be efficient to drive people's risk perceptions. Further, it would also be beneficial for future studies to include other key factors in the relationship between cultural values and risk perceptions, like trust (e.g., Siegrist, 2000; Tumlison & Song, 2019; see also Slovic, 1999) or value congruence (i.e., individuals’ perceptions of how their values are advanced or undermined by a specific protective measure—Yuan & Swedlow, 2022). Another limitation relates to our worldview measure. We selected a scale assessing three cultural values (hierarchy–individualism, egalitarianism, and fatalism). As a result, it was not possible to conduct analyses for hierarchy and individualism separately, even though it has recently been shown that this distinction is of little relevance in a French‐speaking population (Chauvin & Chassang, 2022). Nevertheless, future studies might use another operationalization of worldviews such as measures developed by Jenkins‐Smith et al. (see Swedlow et al., 2020, for an extensive review) to allow for comparisons to other cultural theory studies of risk perception (including COVID studies; see Yuan & Swedlow, 2022). Last, although not the scope of the current article, it would be a fruitful avenue of future research to take into account other values, together with cultural ones, that are also likely to dispose individuals to confer specific meanings to risk issues (e.g., the comfort vs. concern about tampering with nature, Raimi et al., 2020; see also Siegrist & Arvai, 2020).

Despite these limitations, we believe that the present study contributes to a better understanding of how cultural orientation impacts risk perception by offering some insight about the mediating role of key risk characteristics in the relationship between cultural values and individuals’ perceptions about risk.

AUTHOR CONTRIBUTIONS

The third author is the lead and corresponding author. We describe contributions to the paper using the CRediT taxonomy. Conceptualization: B. C., O. R, and I. C.; Methodology: B. C., O. R., and I. C.; Software: B. C.; Validation: B. C., O. R., and I. C.; Formal analysis: B. C.; Investigation: I. C.; Resources: B. C. and I. vvxC.; Data Curation: B. C.; Writing ‐ Original Draft: B. C. and I. C.; Writing ‐ Review & Editing: B. C, O. R., and I. C.; Visualization: B. C. and I. C.; Supervision: B. C.; Project administration: B. C. and O. R.

This research was funded by the Grand Est Region, Contract 19_GE6_195. The authors are thankful to master students Bastien Julie, Muller Manon, Schmitt Alexandre, Sgambati Léa and Touel Sabrina for their help in participant recruitment.

Chassang, I. , Rohmer, O. , & Chauvin, B. (2025). Cultural values, risk characteristics, and risk perceptions of controversial issues: How does cultural theory work? Risk Analysis, 45, 682–700. 10.1111/risa.17636