Introduction

Different scholars from both individual psychology, social psychology, and culture studies have made basically the same observation: that fear and danger can move a population in the direction of authoritarianism, conservatism, nationalism, xenophobia, intolerance, and strict religiosity. Many different theories are describing psychological and cultural effects of fear and danger. The so-called regality theory adds a deeper level of explanation to this well-known effect based on evolutionary psychology. Regality theory predicts that individual danger and collective danger have very different effects, as explained below.

This article presents an exploratory study aiming to examine the psychological and cultural effects of different kinds of danger and to test whether individual and collective danger have different effects.

First, we will briefly review a number of theories from different fields of science that all describe responses to threat and danger. In the research section, we will map the correlations of different kinds of danger with different psychological and cultural outcomes. Finally, we will discuss how well the statistical findings fit the different theories.

Psychological and cultural theories

Authoritarianism was originally regarded by psychologists as a mainly fixed part of a person's personality that is formed in childhood (Adorno, Frenkel-Brunswik, Levinson, & Sanford, 1950). Later studies contradict this and find that authoritarianism is influenced by threats, danger, and uncertainty (Van Hiel, Pandelaere, & Duriez, 2004). The different studies cited here use different definitions of authoritarianism, but common elements include hierarchy, obedience, traditionalism, nationalism, xenophobia, intolerance, and punitiveness.

A major improvement in the understanding of authoritarianism was provided by theories of motivated social cognition. A meta-analysis has found that conservatism and authoritarianism can be explained as responses to a number of different perceived threats (Jost, Glaser, Kruglanski, & Sulloway, 2003). A further development of the motivational theory is distinguishing between two subtypes of authoritarianism called right-wing authoritarianism (RWA) and social dominance orientation (SDO), triggered by two different world views. A view of the world as a dangerous place is motivating RWA and leads to authoritarian submission and conformity. A view of the world as a competitive jungle where might is right provides the motive behind SDO which leads to authoritarian dominance and discrimination against groups with lower status (Duckitt & Sibley, 2009). Many psychologists regard authoritarianism as an individual phenomenon, but some argue that it may be viewed as a group phenomenon (Stellmacher & Petzel, 2005).

Another theory with a group focus is realistic group conflict theory. Animosity will result if two or more groups are competing for the same limited resources or otherwise have conflicting goals. This will result in prejudice, stereotyping, and hostility against outgroups, as well as increased ingroup identification and solidarity, according to the realistic group conflict theory (Jackson, 1993).

The view of authoritarianism as a group phenomenon is common in culture theories. A group in danger is likely to develop a culture of collectivism and rally around a strong leader. The connection between intergroup conflict and centralization of power was noticed already by Georg Simmel who wrote in 1908 that “war needs a centralistic intensification of the group form, and this is guaranteed best by despotism.” Such an authoritarian organization does not only have a defensive function, it also inspires offensive fighting, as Simmel observes: “Once despotism exists and that centralized form has materialized, the energies thus accumulated and pushed close together very easily strive after a natural relief, war with the outside.” Simmel also offers an example of the opposite extreme: The Greenland Eskimos (Inuit) have no war, no chieftainship, no authority, and no means of discipline (Simmel, 1908).

Recent sociological theories have focused on cultural values that change as the life conditions of people change. The modernization theory of Ronald Inglehart and Christian Welzel describes how socioeconomic development has changed the conditions of life from precariousness to a situation where survival is taken for granted. This has led to a change from what they call survival values to self-expression values. Survival values are reactions to a situation where the survival of the social group is threatened. Survival values include intolerance, xenophobia, and accept of an authoritarian government. Improved collective security, on the other hand, leads to a change to self-expression values such as tolerance, trust, democracy, and individual autonomy. This process was preceded by industrialization and a change from traditional values, including strong religiosity and traditional family values, to secular/rational values based on a scientific world view (Inglehart & Welzel, 2005). Further studies have confirmed that the secularization has been connected with increasing existential security (Norris & Inglehart, 2011). Existential threats make people prioritize in-group solidarity, while they will prioritize freedom over security, autonomy over authority, diversity over uniformity, and creativity over discipline in a safe world where existential security is taken for granted (Welzel, 2013). These so-called emancipative values are reduced by violent conflict. Echeverría, Hemmerechts, and Kavadias (2019) find evidence that emancipative values are lower in countries with a history of violent conflict.

A somewhat similar theory has been put forth by cultural psychologists. Michele Gelfand et al. (2011) make a distinction between tight and loose cultures. Tight cultures develop as a response to societal threats, including ecological and human-made threats according to this theory. Tight cultures are characterized by strong social norms, autocratic governing systems, punishment of deviant behavior, intolerance of dissent, and strong religiosity. Loose cultures develop in the absence of severe threats. Loose cultures are characterized by weak norms and a high tolerance of deviant behavior (Gelfand, et al., 2011).

The similarity between these psychological and cultural theories is notable. Threats and existential insecurity lead to a society characterized by strong hierarchy, intolerance, discipline, and strong religiosity, while the absence of threats leads to more tolerance and egalitarianism. The different scientific paradigms have made similar observations about human responses to existential threats, and invented different terms, such as authoritarianism, survival values, or cultural tightness for the same or closely related phenomena.

Distinctions between individual and collective threats

The psychological and cultural theories mentioned above are not very clear about what kinds of threats and dangers are leading to authoritarianism, survival values, cultural tightness, etc. Early psychological theories have often focused on intrapsychic conflicts and internal threats as explanations of authoritarianism, while later studies find that RWA is predicted by external threats rather than by internal threats (Onraet, Dhont, & Van Hiel, 2014; Shaffer & Duckitt, 2013).

The so-called terror management theory (TMT) is also focusing on intrapsychic conflicts. TMT is based on the hypothesis that people who fear their own death try to suppress this fear by adhering to conventional or conservative world views when reminded about their own mortality (Burke, Kosloff, & Landau, 2013). This theory is at odds with modern evolutionary psychology. From an evolutionary point of view, we would expect people to react to danger by actions that reduce the actual danger rather than actions that merely reduce the fear of the danger. Kirkpatrick and Navarrete (2006) extend this criticism by arguing that the psychological responses reported by TMT theorists might be better explained as mechanisms that facilitate the formation of alliances in accordance with coalitional psychology. This coalitional explanation is supported by the finding of Ullrich and Cohrs (2007) that the response of system justification postulated by TMT theorists was increased by salience of terrorism, not by salience of individual mortality.

Several recent studies find that threats to the social group or nation have a particularly strong effect on RWA and intolerance, while threats to the individual have little or no such effect (Huddy, Feldman, Capelos, & Provost, 2002; Hutchison & Gibler, 2007; Miller, 2017; Onraet & Van Hiel, 2013; Shaffer & Duckitt, 2013; Tir & Singh, 2015). One study makes the opposite finding that RWA is increased by fear of personal consequences of terrorism rather than collective consequences in a lab experiment with imagined scenarios (Asbrock & Fritsche, 2013).

Responses to individual threats are also studied in the context of evolutionary theories of human threat management. Neuberg, Kenrick, and Schaller (2011) find that responses to threatening stimuli are functionally distinct and highly domain-specific. Responses to certain threats are moderated by individual differences in disgust sensitivity. Disgust sensitivity has been found to be correlated with various differences in attitudes including political conservatism, discrimination against homosexuals, and general demands for government protection (Kam & Estes, 2016; Smith, Oxley, Hibbing, Alford, & Hibbing, 2011). These patterns become clearer when different kinds of disgust are distinguished. Tybur and de Vries (2013) distinguish between pathogen disgust, sexual disgust, and moral disgust. They find that different kinds of disgust sensitivity are correlated with different personality traits. A later study finds that disgust sensitivity depends more on genes than on environment, and that the different kinds of disgust sensitivity differ genetically (Sherlock, Zietsch, Tybur, & Jern, 2016).

Economic threats

Studies of the effects of economic threats have reported mixed results. A study of historical voting patterns in the German Weimar Republic shows that voting for the anti-democratic Nazi party was increased among white-collar employees who feared the consequences of the economic crisis, while the blue-collar workers who were the most affected by unemployment tended to vote for the communist party (Falter, Link, Lohmöller, de Rijke, & Schumann, 1983). Fritsche, Jonas, and Kessler (2011) discuss various explanations to these findings. One possible explanation is that a perceived global threat of economic crisis fosters authoritarian voting, while individual threats do not. Another possibility is that the crisis makes people vote for the party that best serves the interests of their social class (Fritsche et al., 2011). A recent study by Brandt and coworkers finds that individual economic threats are associated with political support for equality, not social conservatism. Collective economic threats were not included in their study (Brandt et al., 2021).

An indication of the effects of collective economic threats may be found in cultural studies. Socioeconomic development has changed the life conditions from a source of threats to a source of opportunities. This has been followed by a change in people's priorities from survival values to self-expression values, emancipative values, and increased individualism (Beugelsdijk & Welzel, 2018).

Environment threats

There is a growing body of research on the connections between the natural environment of people and their political ideology. A harsh climate, natural disasters, and infectious diseases are factors that seem to foster ideological conservatism and prejudice against minorities (Conway, Chan, & Woodard, 2020; Jackson et al., 2019). Some studies are pooling together a range of different kinds of environmental threats, thus making the distinction between different threats difficult. Murray, Schaller, and Suedfeld (2013) find that famine predicts authoritarianism. Kusano and Kemmelmeier (2018) find no effects of climate or natural disasters on socio-political freedom.

A large number of studies have focused on one particular type of environmental threats, namely infectious diseases. Proponents of the so-called pathogen stress theory (often using the misleading term parasite stress theory) have presented statistical studies suggesting that a high incidence of infectious diseases is causing more solidarity, strong family ties, more religiosity, less democracy, more sexual restrictiveness, less individualism, and more cultural tightness in a culture (Fincher & Thornhill, 2012; Gelfand et al., 2011). Two different evolutionary explanations for this effect have been proposed: (1) Avoidance of strangers who might carry diseases leads to xenophobia. (2) People in primeval times did not know how diseases spread. They had to rely on certain rituals and food taboos that might prevent the spreading of diseases. Cultural conformity was needed to make people obey arbitrary taboos. Tybur et al. (2016) present evidence against explanation (1) and in favor of explanation (2).

The pathogen stress theory has caused a lot of debate. Several investigators argue that the correlations between disease incidence and cultural variables may be spurious. Bromham and coworkers find that the correlations can just as well be explained by geographic proximity, biodiversity, climate, or population size (Bromham, Hua, Cardillo, Schneemann, & Greenhill, 2018). Currie and Mace (2012) argue that the correlations become insignificant when controlling for GDP and geographic latitude.

Hruschka and Henrich (2013) demonstrate that countries with low and high incidence of infectious diseases have different colonial histories leading to differences in government institutions that may explain the cultural differences between these countries. Murray et al. (2013) disagree and find that the effect of pathogen prevalence remains significant when controlling for state-level authoritarian governance, while Pollet (2014) finds that the results reported by Murray and coworkers become insignificant when a different statistical method is used. Horita and Takezawa (2018) find significant effects of disease incidence only in restricted global regions when controlling for modern institutions. Kusano and Kemmelmeier (2018) find significant effects of both pathogen stress and government effectiveness only in specific regions of the world, but these effects were not stable across the measurements.

Hackman and Hruschka (2013) argue that many of the studies are based on North American culture where the disease data are dominated by sexually transmitted diseases. They find that differences in life history strategies is a likely confounding factor that is influencing disease incidence as well as family ties and religiosity. Thornhill and Fincher (2013) complain that Hackman and Hruschka's analysis is eliminating most of the relevant variance in disease stress.

Direction of causality

While the connection between threat and authoritarianism, xenophobia, religiosity, etc. is well established, the direction of causality is being debated. Does threat cause authoritarianism, or are authoritarian persons more sensitive to threat? Does the causation go both ways, or is there perhaps a third factor that causes both authoritarianism and an increased perception of threat? The direction of causality cannot be established from simple cross-sectional correlations. We have to determine the direction of causality in other ways.

One way to determine the direction of causality is laboratory experiments. There are many published studies where threatening stimuli have been manipulated in laboratory experiments. For example, Mirisola, Roccato, Russo, Spagna, and Vieno (2013) found that people with low RWA scores increased their RWA score after reading threatening information.

Lavine, Lodge, Polichak, and Taber (2002) found that authoritarian people were more sensitive to threatening information, which implies the reverse causality.

Cohrs and Asbrock (2009) found that threatening information about an outgroup increased prejudice more in persons with high RWA than in low RWA persons. A further study found that authoritarians reacted more strongly to perceived threat (Cohrs & Ibler, 2009). The authors discussed three possible explanations for this observation: (1) Authoritarians have greater sensitivity to perceive threat; (2) Authoritarians have greater motivation to react to perceived threat; and (3) differences in cognitive style explain differences in the perception of threat as well as differences in authoritarianism (Cohrs & Ibler, 2009). Thus, we can find support for both forward causality, reverse causality, and third factor causality in lab tests. Choma and Hodson (2017) have made similar observations and argued that the causality is bidirectional.

An alternative to lab tests is natural experiments. Observing a population before and after a threatening event can give an indication of the psychological effect of such events. Echebarria-Echabe and Fernández-Guede (2006) observed that a terrorism event in Madrid in 2004 increased people's prejudice not only against the culprits (Arabs), but also against an unrelated group (Jewish). Likewise, Nail and McGregor (2009) observed a general political shift in the conservative direction following the terrorism events of Sept. 11, 2001 in USA. These observations support the theory that threat causes authoritarianism. A meta-analysis suggests that the effect of terrorism depends on the actors and on the political discourse (Godefroidt, 2022).

A more detailed insight may be obtained by following a population over time. Doty, Peterson, and Winter (1991) followed the US population over a 10-year period and observed that 13 out of 20 indicators of authoritarianism showed increasing trends following threatening events. Mirisola et al. (2013) followed a panel of test persons over a three-year period. They found that increases in perceived threat in one year led to increases in RWA in the following year for persons with an initially low RWA score. These observations indicate that the dominating causality is from threat to authoritarianism.

Inglehart and Welzel (2005) have made a particularly enlightening study based on the World Values Survey that provides longitudinal data from many countries around the world. They are using cohort analyses and path analyses to study the worldwide trend of democratization. Their important finding is that increasing socioeconomic resources have improved existential security which has led to increasing self-expression values, which in turn have led to increasing tolerance and democratization. They have found little or no evidence of reverse causation (Inglehart & Welzel, 2005).

A historical study of large empires finds that empires have grown through a self-amplifying cycle of increasing militarism and increasing authoritarianism. This implies a cyclic causality in the social dynamics (Turchin, 2007).

There is yet another way of determining the direction of causality, namely to use instrumental variables. An instrumental variable is a variable that is known for technical reasons to influence the predictor variables without any direct interaction with the target variables. A detected correlation between an instrumental variable and a target variable is an indication of the causal relationship: instrumental variable → predictor variable → target variable. Hutchison (2014) made such a study based on the observation that a rugged or mountainous terrain increases the likelihood of violent civil conflict. Using ruggedness as an instrumental variable, Hutchison found that civil conflict causes political intolerance and a preference for conformity over civil liberties.

A similar study by Fog (2017) used various environmental features that enable or impede violent conflict as instrumental variables in non-modern societies. Environmental factors that make intergroup conflict likely include large, defendable food sources and a geography and technology that allow easy travel of warriors to neighbor territory. Factors that impede war include geographic isolation and ecological niche specialization. Using these factors as instrumental variables, Fog found that violent intergroup conflict leads to a hierarchical political system, high punitiveness, religiosity, group identification, high fertility, strict sexual morals, and child labor in ancient, non-industrial societies.

The direction of causality is a particularly important question for the well-known correlation between democracy and peace. There are two competing theories relating to this correlation. The democratic peace theory posits that democracy causes peace (Russett & Oneal, 2001), while the territorial peace theory maintains the opposite claim that peace causes democracy (Hutchison & Starr, 2017). If war can cause authoritarianism and autocracy then we should expect the absence of war to be a prerequisite for democracy. The causation may be bidirectional, but recent studies seem to indicate that the strongest effect is from peace to democracy, and not vice versa (Gibler & Owsiak, 2017; Gibler & Miller, 2021). Similarly, reversals from democracy to autocracy are observed to be caused by perceived territorial threats (Karreth, Tir, & Gibler, 2021). Obviously, this question of causality has important consequences for international security politics.

Regality theory

Recently, it has been suggested that authoritarianism, strict hierarchy, strict discipline, intolerance, and xenophobia have an evolutionary explanation (Fog, 2017; Grigoryev, Batkhina, van de Vijver, & Berry, 2019; Kessler & Cohrs, 2008; Sinn, 2019). The regality theory of Fog (2017) contends that violent intergroup conflict has been a strong evolutionary force in human prehistory. Prehistoric hunter-gatherer tribes were not always as peaceful as the romantic picture early anthropologists have painted of them (Allen & Jones, 2014; Hames, 2019; Kiblinger, 2020). Varying levels of conflict has led to the evolution of a flexible psychology that allows human groups to adjust their social structure to violent or peaceful environments.

A hierarchical social structure with strict discipline and a strong leader can be an efficient means for suppressing free riding in case of intergroup conflict or war, and for making sure that everybody is fighting in a coordinated way (Fog, 2017; Sinn & Hayes, 2017). The leader has a strong incentive to reward brave warriors and punish cowards and defectors because he has more at stake than his followers (Gavrilets & Fortunato, 2014; Johnson, 2015). Such a hierarchical structure was likely to be optimal in a dangerous environment with frequent wars or violent conflicts in primeval societies.

The situation is very different in a safe and peaceful environment where there is less need for collective action. A powerful and despotic leader in this situation is likely to take advantage of everybody else without providing enough collective benefit to justify his power. This is the reason why a flexible psychology is advantageous. People will be likely to support a strong leader and to show psychological preferences for strict discipline in case of war or other collective danger that requires collective action. In case of peace and security, however, people will prefer an egalitarian society and a tolerant culture because this gives them more personal freedom and frees them from the tyranny of a powerful leader (Fog, 2017). This flexibility is an example of what biologists call phenotypic plasticity (Bateson & Gluckman, 2011). People will have different psychological preferences depending on the perceived level of conflict and collective danger in their environment.

Regality theory offers an evolutionary explanation for collective fighting without relying on the controversial theory of group selection. Regality theory describes a psychological flexibility that allows humans and their culture to adapt to varying needs for collective action. The combined effect of the preferences of all the individual members of a society results in an emergent effect on the social and cultural structure of the whole group. The causal chain can be outlined like this:

Perceived collective danger → Preference for a strong leader and strict discipline → Strong leader → Hierarchical and authoritarian social structure → Efficient collective fighting.

Fog argues that the psychological preference for a strong leader in case of collective danger is an evolutionarily stable strategy because it suppresses the free riding of others and makes sure that nobody has to fight alone. Each individual warrior is likely to suffer a high risk that is not sufficiently compensated for by his own share of the group-level benefit he provides. But by supporting a strong leader, he makes sure that everybody else is also fighting so that he can benefit not only from his own fighting but also from the fighting of his fellow warriors.

A social group in a dangerous and conflict-filled environment will develop in a direction called regal, typically including a hierarchical political system with a strong leader, strict discipline, intolerance of deviants, xenophobia, strict religion, strict sexual morals, and a high birthrate. The opposite situation is seen in safe and peaceful environments. A society in the latter situation will develop in a direction called kungic, typified by egalitarianism, tolerance, and a low birthrate. The characteristics of a regal society are all features that increase the social cohesion and the capacity for collective action, while the characteristics of a kungic society give higher weight to individual freedom and welfare.

The terms regal and kungic, defined by regality theory, may also be used for describing the psychological responses of the individual members of a society. Persons who see their tribe or social group as threatened will develop regal psychological traits, such as the desire for a strong leader, strict discipline, xenophobia, and conformity. The prediction of regality theory is that the regal effects will appear in the case of collective danger, but not individual danger. Regality develops when people perceive a danger that requires collective action and a powerful leader to avert (Fog, 2017).

A note on terminology

Integrating traditional psychological and sociological theories with evolutionary psychology requires an objective terminology that makes sense when applied to the environment of evolutionary adaptation.

The term right-wing authoritarianism (RWA) is problematic because it is applicable only in a specific cultural setting. The right wing/left wing distinction did not exist prior to the French Revolution. The term RWA is in fact misleading because the characteristic symptoms of RWA are found in persons who follow the ideologies of dominant leaders, whether these are left wing or right wing. In fact, characteristics analogous to RWA have been documented in persons with left wing opinions in communist and ex-communist countries and even in the USA (Conway, 2021; Costello et al., 2022; De Regt, Mortelmans, & Smits, 2011; Grigoryev, Batkhina, Conway, & Zubrod, 2022). The situation is far from symmetrical, though. Threat leads to political and economic conservatism in the sense of resistance to change, acceptance of inequality, and trust in powerful authority figures. The paradox of “left-wing authoritarianism” stems from the use of culturally relative terms (Jost, Stern, Rule, & Sterling, 2017; Nilsson & Jost, 2020). This flawed terminology is likely to lead to political bias in the research (Ray, 1989).

The terms liberal and conservative are often used synonymously with left wing and right wing, especially in US literature. This terminology is no less confusing because the term liberal has very different meanings in different parts of the world. In Europe, for example, political parties that label themselves liberal are focusing on economic liberalism in the sense of reliance on unregulated market forces. The same policy is associated with conservatism in the US (Caprara, 2020; Haute & Close, 2019).

Some authors find that the unidimensional left/right or liberal/conservative spectrum is inadequate. They recommend a distinction between ideological conservatism/liberalism, economic conservatism/liberalism, cultural conservatism, and status quo conservatism in research on the psychological effects of threats (Conway et al., 2020; Federico & Malka, 2018; Malka, Soto, Inzlicht, & Lelkes, 2014). In the context of evolutionary psychology, however, we should prefer even more specific and culture-independent concepts, such as hierarchical vs. egalitarian, punitive vs. permissive, tolerant vs. intolerant and xenophobic, warlike vs. peaceful, or regal vs. kungic.

Investigating the effects of different threats

The above review shows that several different theories originating from different scientific paradigms have described somewhat similar psychological and cultural responses to various kinds of threat and danger. Terror management theory predicts an effect of individual danger, while regality theory predicts an effect of collective danger. Pathogen stress theory predicts an effect of infectious diseases while ignoring other kinds of danger. RWA theory makes little distinction between different types of danger. Cultural modernization theory describes cultural changes as responses to decreasing existential threats, including economic threats, while making little or no distinction between individual and collective threats. The theory of tight and loose cultures makes similar descriptions about responses to threats, which include both man-made threats such as resource scarcity and war, and natural threats including natural disasters and infectious diseases.

The different theories have different lists of responses to threats, with many similarities and overlaps between these lists. It is relevant to ask whether all these theories are in fact reflecting the same or closely related underlying psychological mechanisms, and whether it is possible to combine these theories into a common synthesis.

The statistical studies described below aim to explore the effects of different kinds of threats to see if different threats do in fact have different effects. In particular, we want to investigate whether individual threats and collective threats have different effects, as predicted by regality theory.

Research design and variables

The World Values Survey is a large research project based on surveys of representative national samples of individuals in many countries. Cultural values, norms, and attitudes have been measured in seven waves of surveys spaced 5–10 years apart (Haerpfer, et al., 2022; Inglehart, et al., 2014). Only data from survey wave 6 and 7 are used here because earlier waves did not include the relevant measures of fear and danger. These two survey waves have been carried out in 2010–2014 and 2017–2021, respectively. The countries included in each wave are listed in Table 1.

The variables are selected based on theory only. The predictor variables are indicators of different kinds of perceived or real danger. The outcome variables are all variables that are predicted from regality theory or any other theory mentioned here to be influenced by some kind of fear or danger.

The survey variables used in the present study are listed in Table 2. All variables are normalized to unit variance. The predictor variables are indicators of individual danger and collective danger. A measure of individual danger is composed of 17 survey items with equal weight covering worries related to living in an insecure neighborhood, access to vital resources, and education of the respondent's children. What these items have in common is that they affect the individual respondent but not the whole society.

Collective danger is composed of three survey items covering worries about war, civil war, and terrorism (see Table 2). Additional predictor variables are sex, age, and education. Outcome variables are 37 single and composite variables listed in Table 2. Cronbach's alphas for the composite variables are also listed in Table 2.

Some additional country-level variables are used as predictors and control in between-countries statistics. It is important to control for the level of development which has a strong influence on many variables. The human development index (HDI) is obtained from the United Nations Development Programme (2022). The national values of HDI for 2014 and 2019 are used for statistics on wave 6 and 7, respectively.

Pathogen stress is a measure of the incidence of infectious diseases, including bacteria, viruses, and parasites. The variable named “combined parasite stress” calculated by Fincher and Thornhill (2012: supplement 2) is used as a measure of pathogen stress for each country.

To get an objective measure of violent conflict, we used the number of deaths due to organized violence, including war, civil war, and terrorism, that have occurred within the territory of the country in a 10-year period prior to the survey wave, i.e. 2001–2010 for wave 6, and 2008–2017 for wave 7. These values are calculated from the database maintained by the Uppsala Conflict Data Program (2018). The conflict data are log-transformed in order to get an approximately normal distribution. A value of 0.1 is added to avoid taking the logarithm of zero.

It is assumed that the outcome variables can be described by a linear two-level model where the psychological responses of each individual are influenced by the experiences and life conditions of the individual as well as by country-level effects that are different for each country. The predictor variables at the individual level are perceived individual danger, perceived collective danger, sex, age, and education level. Predictor variables in country-level statistics include human development index (HDI), pathogen stress, and deaths by violent conflicts in the country's territory.

Two studies are carried out based on these data. Study 1 is testing individual-level effects in the two-level model, and study 2 is testing correlations between countries. The individual level effect and the between-country effect may or may not be congruent.

Data are analyzed using the R packages psy, lme4, lmerTest, miscor, testcorr, ridge, EnvStats, and mutoss (R Core Team, 2022).

Study 1

Study 1 is testing the individual-level effects of perceived individual danger, perceived collective danger, sex, age, and education in the two-level model where the effects of country-level influences are unique for each country and for each outcome variable. The outcome variables are listed in Table 2. The results are listed in Table 3 as coefficients in the two-level model. These coefficients are used instead of partial correlation coefficients.

The tests are calculated for each survey wave separately. This makes it possible to evaluate the reproducibility of the results. A few of the outcome variables are only available in one of the two survey waves (see Table 2).

Most of the individual-level coefficients are weak, yet highly significant with p values often lying below 10⁻⁴ due to the high N.

While the reproducibility between the two survey waves is reasonably good (see Table 3), the reproducibility between countries is somewhat inferior. There are several cases where a partial correlation coefficient is significantly positive in one country and significantly negative in another country (correlations calculated for each country separately are not listed here). A poor reproducibility between countries may be due either to country-specific peculiarities or measurement error and noise. It is concluded that noise is a serious reason since extreme country values are not reproduced from survey wave 6 to wave 7.

It is a common observation that the measurement noise is high in representative population surveys, which leaves most of the variance unexplained (Akaliyski, Welzel, Bond, & Minkov, 2021). A high proportion of noise or measurement error will lead to lower correlation coefficients (Ostroff, 1993), which is indeed what we observe here. Therefore, we cannot make reasonable estimates of effect sizes in study 1, but we can still make qualitative conclusions about the effects of individual and collective danger.

Table 3 lists the coefficients representing the individual-level effects of each predictor variable on each outcome variable in the sample of all the countries. Standard deviations are listed in parentheses. Results are listed for survey wave 6 and 7 separately. Levels of significance are indicated with 1–3 asterisks for p values less than 0.05, 0.01, and 0.001, respectively (two-tailed).

There are two general types of errors that must be taken into account in tests of statistical significance. A type I error, or false positive, is the error of classifying a correlation as significant when in fact it is merely the result of random variation. A type II error, or false negative, is the classification of a correlation as non-significant when in fact there is a true relationship. Both types of errors are undesired in an exploratory study.

In study 1, we can check for type I errors by comparing the results for survey wave 6 and 7. If a highly significant correlation in wave 6 is reproduced in wave 7 then it can be regarded as reliable. If the significance of a correlation is weak (few asterisks) or not reproduced between the waves, then the interaction should be regarded as possible, but uncertain. The high number of survey respondents leads to low p values which increases the risk that spurious correlations are classified as significant.

Type II errors can be expected because of the high level of noise, but this is compensated for by the high number of survey respondents which increases the statistical power.

Study 2

Study 2 is based on variation between countries using the country averages of each variable. Survey waves 6 and 7 are combined in order to maximize the number of countries. The values from the two waves are combined into an average for countries that are represented in both waves, while values from a single wave are used for countries that are included only in one of the two waves (see Table 1).

Ridge regression is used for estimating the coefficients of the multiple regression model with automatic choice of ridge parameter (Cule & Iorio, 2013). Ridge regression is a method that is useful for estimating the most likely coefficients when the predictor variables are correlated with each other. The dependent variables or outcome variables are listed in Table 2. The predictor variables are the country averages of perceived individual danger, perceived collective danger, pathogen stress, territory deaths by organized violence, and HDI. The results are listed in Table 4. The correlations between countries are higher than the individual-level coefficients, as one can expect when the random within-country variation is high (Ostroff, 1993).

A set of partial regressions of the same data is calculated as a robustness check. The partial correlation of each outcome variable against each predictor variable is calculated while controlling for the other predictor variables. The two predictor variables collective danger and territory deaths are regarded as perceived and real values of the same underlying danger and therefore not controlled for each other. In other words, the partial correlation of an outcome variable against collective danger is controlled for all other predictor variables except territory deaths, and vice versa. Likewise, partial correlations against pathogen stress should not be controlled for HDI because this includes health data in the form of life expectancy. Instead, correlations against pathogen stress are controlled for a modified HDI where health data are removed.

The results of the between-country partial regressions are listed in Table 5. Re-calculating the regressions with different timespans than 10 years for the territory deaths made very little difference in the results. Other measures for the level of violent conflict gave somewhat different results. The “conflict score” calculated by Echeverría et al. (2019) gave stronger correlations for xenophobia and somewhat weaker correlations for religiosity and national pride. For the present study we preferred the data from the Uppsala Conflict Data Program (2018) because they are the most objective.

We cannot check for type I errors by comparing the results for wave 6 and wave 7 in study 2 because the data for the two waves are not independent at the country level. Instead, the method of Benjamini, Krieger, and Yekutieli (2006) is used for adjusting the false discovery rate, i.e. the proportion of type I errors. The adjusted p values are indicated in Tables 4 and 5 with 1–3 black stars (★) for p < 0.05, 0.01, and 0.001, respectively. The number of black and white stars (☆) combined indicate the unadjusted p values, using the same limits. For example, ★★☆ means adjusted p < 0.01 and unadjusted p < 0.001. A coefficient is less likely to be a type I error the more stars it is marked with, where black stars have higher weight than white stars. Type II errors are more likely in study 2 than in study 1 because the statistical power is lower in study 2.

Outliers are detected using Rosner's test (EPA, 2006) with an alpha value of 0.05. The number of outliers in the between-country statistic is low. Removing the outliers had little effect on the results. The coefficients listed in Tables 4 and 5 are calculated without removing outliers. A few cases where outliers have significant effects on the results are mentioned in the next section.

Results

Many of the outcome variables are correlated with collective danger at the individual level in study 1 (Table 3) or the between-country level in study 2 (Tables 4 and 5) or both. Most of the coefficients at the individual level are small but significant with low p values. The coefficients at the between-country level are higher, but only around one third of these have sufficiently low p values to be marked as significant. The coefficients for collective danger that are marked as significant are mostly in the direction predicted by theory, except for a few cases at the individual level discussed below.

The results for individual danger are mixed. Many of the correlations of outcome variables with individual danger have opposite signs of the correlation with collective danger and opposite of the expectations in both studies. The raw correlations of the outcome variables against individual danger without control for collective danger or other confounding variables are generally in the same direction as the correlation with collective danger (The raw correlations are not listed here because they are misleading). This indicates that the correlations of outcome variables against individual danger, as reported by other authors, are in fact nullified or even reversed when we control for the effects of collective danger and other confounding variables in many of the cases. The correlations of collective danger, on the other hand, are not eliminated when controlling for individual danger. This means that collective danger is a stronger predictor than individual danger, in accordance with regality theory.

The correlation between country averages of individual danger and collective danger is 0.42 (p = 0.0001). It is reduced to 0.09 (n.s.) when controlled for HDI. Within-country correlations between individual and collective danger is 0.19 on average with a SD between countries of 0.13 which makes it nonsignificant at the global level. The correlation between country averages of perceived collective danger and real collective danger as measured by territory deaths is 0.51 (p < 0.0001). It is reduced to 0.18 (n.s.) when controlled for HDI.

A more detailed look at the different outcome variables gives many interesting observations (Tables 3 and 4). Religiosity is strongly increased by collective danger, while there is no reproducible effect of individual danger. Interestingly, religious behavior is different from religious beliefs. Religious behavior is increasing with both individual and collective danger.

Xenophobia has a non-significant positive correlation with collective danger and a marginally significant positive correlation with territory deaths in study 2 (Tables 4 and 5). In study 1, xenophobia is increased by individual danger while the effect of collective danger is not reproducible (Table 3).

Family ties and respect for tradition are strongly increased by collective danger, but deceased by individual danger.

A desire for more respect for authorities is increased by collective danger, but deceased by individual danger in study 1. The correlation with collective danger in study 2 becomes marginally significant (r = 0.16☆) when the low outlier Japan is removed.

Support for a strong leader is increased by collective danger as predicted. Support for a strong leader is also increased by individual danger in study 1 (Table 3). This possible effect of individual danger may be explained by the theory that people seek protection under a powerful leader in case of danger (Eibl-Eibesfeldt, 1989; Mayseless & Popper, 2007).

National pride is increased by collective danger as predicted, but decreased by individual danger.

Sexual morals are remarkably correlated with other cultural variables, as several studies show (Inglehart, 2018; Jackson et al., 2019). The acceptance of homosexuality, divorce, and abortion are reduced by collective danger as predicted, but increased by individual danger.

The acceptance of domestic violence, i.e. wife beating and child beating, is decreasing with collective danger and increasing with individual danger in study 1 (Table 3). Domestic violence may in fact be condemned by persons with high moral strictness regardless of whether it is common in the country.

The requirements for children to be obedient, responsible, and to show good manners are increased by collective danger, as expected, according to study 1. Children are less independent under collective danger at the level of between-country correlations in study 2. Children's imagination is encouraged under individual danger, but discouraged under collective danger in accordance with the prediction of conformity.

The feeling of having freedom of choice and being in control of one's own life is decreased by individual danger in study 1 as expected, but increased by collective danger contrary to previous findings (Smith, Trompenaars, & Dugan, 1995). There is no effect at the between-country level in study 2.

Trust in other people is decreased by both kinds of danger in both studies.

The feeling of a personal identity connected to one's town, district, or country is increased by collective danger, as expected, but decreased by individual danger.

The acceptance of war as a means to obtain justice has a positive association with both individual and collective danger in study 1.

The willingness to fight for one's country is increasing with collective danger, as expected, and decreasing with individual danger. The correlation with collective danger is significant in study 1. It becomes significant also in study 2 when the low outliers Haiti and Japan are removed (r = 0.23★★☆).

Confidence in the government and armed forces is increased by collective danger and decreased by individual danger, as expected. The perceived level of corruption is increasing with both kinds of danger. Satisfaction with the political system is decreasing with both individual and collective danger.

Happiness is decreased by individual danger as expected. Happiness is increasing with collective danger at the individual level in study 1. It is a paradox that collective danger appears to increase happiness since the World Happiness Report lists the safe North European welfare states as the happiest countries (Helliwell et al., 2021). We may speculate that the positive effect of collective danger on happiness at the individual level is mediated by stronger family ties, solidarity, and a sense of purpose in life.

The support for democracy is decreased by collective danger, as predicted, according to study 1. Study 2 shows no significant effect. The support for democracy is increased by individual danger, which is logical because democracy benefits the disadvantaged through redistribution of resources (Acemoglu & Robinson, 2005).

People tend to prioritize security over freedom in case of collective danger, as expected. Support for economic income equality is increased by individual danger and decreased by collective danger, as expected, according to study 1.

The left wing/right wing scale is culture-specific as argued above. It is included here only for the sake of comparison with existing literature. Right-wing attitudes are increased by collective danger as predicted, while there is no reproducible effect of individual danger (Table 3).

The so-called post-materialism values are calculated based on the prioritization of 12 items by the survey respondents (World Values Survey, 2021). Post-materialism values show a positive association with individual danger and a negative association with collective danger, as expected.

The effects of pathogen stress are weak, and few of the coefficients are statistically significant (Tables 4 and 5). There is some indication that pathogen stress may increase religious behavior. The coefficient is increased to 0.05☆ when the low outlier China is removed. It is likely that people use religious rituals for the purpose of warding off diseases. The statistically significant correlation of pathogen stress with child beating is unexplained. The pathogen stress theory predicts that a high incidence of infectious diseases will lead to increased group solidarity, authoritarianism, xenophobia, and religiosity (Fincher & Thornhill, 2012; Murray et al., 2013; Tybur et al., 2016). These predictions are not confirmed by the results in Tables 4 and 5. The partial correlation of pathogen stress with religiosity is near zero, the correlation with xenophobia is negative (nonsignificant), and the solidarity as reflected in feeling close to town, district, or country have significantly negative correlations.

The effects of territory deaths are also weak in Tables 4 and 5, but in the expected direction. The figures indicate that violent conflicts may increase religiosity, xenophobia, confidence in the armed forces, and right-wing attitudes. The weakness of statistical significance here is either a result of low statistical power or an indication that perceived collective danger is a stronger predictor than real collective danger.

The possible effects of economic threats, ecological threats, and natural disasters are not tested in the current study.

There are significant sex differences in many variables as Table 3 shows. Women are more religious than men and attach more importance to family and tradition than men do. Women are more tolerant of abortion, homosexuality, and divorce, but less tolerant of domestic violence and less tolerant of war.

Age differences are quite significant too. Religiosity, respect for tradition and authority, nationalism, feeling of closeness to home town, district, and country, and confidence in government increase with the age of the respondents. Tolerance of abortion, homosexuality, divorce, domestic violence, and war decrease with age. Older people are less supportive of democracy and post-materialist values.

A high education level is associated with less fear of individual danger, while there is no reproducible effect on perceived collective danger. Respondents with higher education are less religious, but they do not show less religious behavior. A high education level is associated with less xenophobia, less respect for authority, less support for a strong leader, more tolerance of abortion, homosexuality, and divorce, more happiness, and more post-materialist values. Contrary to our expectation, respondents with high education show less support for democracy and income equality even though they have less right-wing attitudes (see Table 3).

The effects of sex, age, and education on the results may be due to differences in life experiences, cultural influences, perceptions, knowledge, priorities, or evolved strategies.

Discussion and conclusion

This study confirms the findings of previous studies that collective danger has a stronger effect than individual danger on authoritarianism, intolerance, religiosity, etc (Huddy et al., 2002; Hutchison & Gibler, 2007; Miller, 2017; Onraet & Van Hiel, 2013; Shaffer & Duckitt, 2013; Tir & Singh, 2015).

Perceived collective danger is associated with increased religiosity, regard for family and tradition, respect for authorities, support for a strong leader, nationalism, strict sexual morals, disapproval of domestic violence, less trust in other people, feeling of closeness to home town, district, and country, willingness to fight for one's country, confidence in armed forces and government (even though they are perceived as corrupt), prioritization of security over freedom, less support for democracy, acceptance of economic inequality, and low scores on post-materialism values. These observations are all in agreement with the predictions of regality theory and several other theories. Perceived collective danger is also positively correlated with happiness and freedom of choice in the present study contrary to the findings of previous studies.

Objective measures of collective danger are positively correlated with religiosity, xenophobia, confidence in armed forces, and right-wing attitudes. These correlations are only marginally significant. Perceived collective danger seems to be a better predictor than objective collective danger as measured by deaths by violent conflict.

The effects of individual danger are very different from collective danger. Several of the outcome variables have correlations with individual danger that go in the opposite direction when the statistic is controlled for collective danger. Thus, individual danger increases religious behavior, while it has little or no effect on religious beliefs. Individual danger is associated with less regard for family, tradition, and authorities, less nationalism, more permissive morals, less allegiance to home town, district, or country, less confidence in government, less satisfaction with the political system, less happiness, more support for democracy, more support for economic equality, and more post-materialism values. Individual danger and collective danger go in the same direction with respect to religious behavior, support for a strong leader, distrust, and perceived corruption.

We may put forth the hypothesis that responses to individual danger are mainly domain-specific (Neuberg et al., 2011), while collective dangers have a broader range of effects insofar as leadership and collective action are needed for averting these dangers. More research is needed to elucidate the psychological and cultural effects of different kinds of danger.

The remarkable differences between the effects of individual danger and collective danger lead to the suspicion that many previous studies of the effects of danger may be misleading if different kinds of danger are confounded. The different theories reviewed above describe possible psychological and cultural responses to fear and danger. These theories are not always clear about what kinds of dangers have the purported effects. Some theories explain the current findings better than others. Regality theory predicts that perceived collective danger will increase religiosity, xenophobia, family ties, respect for authority, support for a strong leader, nationalism, solidarity, strict sexual morals, strict discipline, bellicosity, tradition, accept of inequality, and less support for democracy. These predictions are all supported by the present study, except for the effect on xenophobia which is not reproducible in study 1 and not statistically significant in study 2.

The predictions of right-wing authoritarianism (RWA) theory are generally in agreement with regality theory and with the current findings, though it is not very clear about what types of danger cause RWA. The validity of RWA theory is limited to a particular cultural and historical setting.

Terror management theory predicts that individual danger will increase conservatism and traditionalism. These predictions are contradicted by the present study where the observed effects of individual danger are generally in the opposite direction of what would be expected from terror management theory. A problem with many previous studies of terror management theory is that different kinds of danger are confounded.

The realistic group conflict theory has an explicit focus on collective danger and conflict with external groups. This theory is in very good agreement with the finding that authoritarianism is caused by collective danger rather than individual danger.

Modernization theory predicts that increasing existential security is driving a cultural change from traditional values to secular/rational values and from survival values to self-expression values. This theory makes no clear distinction between individual and collective security, but the theory is in agreement with the current findings if we narrow down the focus to collective security.

The theory of tight and loose cultures is partially supported by the present study. The characteristics of tight cultures are very similar to regal cultures in regality theory and to traditional values and survival values in modernization theory. The kinds of dangers that are assumed to lead to cultural tightness include man-made as well as natural dangers, including infectious diseases and natural disasters (Gelfand, et al., 2011). The present study indicates that the types of dangers that have these effects are limited to collective danger, and in particular violent intergroup conflict. The purported effect of infectious diseases is not supported by the present study, while the effects of natural disasters are not tested here.

The connection between peace and democracy is confirmed by the present study. Two competing theories are trying to explain this connection, namely democratic peace theory and territorial peace theory. These two theories differ in the assumed direction of causality. The present study cannot determine the direction of causality, but this study supports regality theory which provides a plausible theoretical explanation for the territorial peace theory. Other studies that analyze the direction of causality are discussed in the introduction.

The present study was prompted by the prediction of regality theory that individual danger and collective danger have different effects. This prediction is confirmed with regard to many different outcome variables. Regality theory is useful because it provides a deeper level of theoretical explanation based on evolutionary mechanisms, while several other theories are based more on empirical observations and proximate causes. However, regality theory cannot predict the effects of individual danger. This question must be dealt with by other theories.

The surprising observation that individual danger and collective danger can have opposite effects calls for an explanation. We may propose the hypothesis that people who have to defend themselves against individual danger pay less attention to collective danger. People who live in a poor and unsafe neighborhood may be less authoritarian because they are less informed and pay less attention to threatening national issues (Federico & Malka, 2018). If perceived collective danger is a precursor of authoritarianism, then individual danger may lead to less authoritarianism by making collective danger less salient.

Another possible explanation is that people regard it as the responsibility of the state and government to protect them against individual danger. If people experience that the state apparatus fails to protect them against danger then they will have less trust in the government and regard it as corrupt. This leads to less support for an authoritarian government and less nationalism.

Authoritarian governments often use fearmongering as a strategy for creating a rallying effect. They are inciting support and rallying for the government by exaggerating the danger of external enemies (Fog, 2017). Individuals who have less trust in the government and regard it as corrupt will be less likely to believe the fearmongering rhetoric that would otherwise make them more authoritarian. Tables 3 and 4 show that individual danger is associated with less confidence in government, more perceived corruption, and dissatisfaction with the political system. These factors have significant correlations with several of the other outcome variables, but controlling for these factors has little effect on the calculated correlations with individual and collective danger.

The observed effects of infectious diseases are weak. There is a possible positive correlation with confidence in government (Tables 4 and 5) but the supposed increase in solidarity is contradicted by a negative correlation with attachment to home town, district, and country. There is a non-significant positive correlation with religious behavior, but not with religiosity. It is difficult to test pathogen stress theory because sample sizes are limited by the number of available countries or regions and because the incidence of infectious diseases is confounded by other variables.

The measures of collective danger in the present study are limited to violent conflicts and infectious diseases. Other kinds of perceived collective danger such as economic threats and natural disasters cannot be tested with the present data. This study is limited by the available items in the World Values Surveys. The distinction between individual and collective danger is not always clear. For example, when respondents fear a terrorist attack, do they fear that terrorism will endanger them personally or do they fear that it will endanger their country? More research is needed to find the specific characteristics that distinguish the kinds of dangers that have the effects that we have ascribed to collective danger here. Is it only violent conflicts that increase authoritarianism, intolerance, etc. or can other collective dangers have the same effects? Does it matter whether people believe that a strong leadership is necessary for solving the problems, as the evolutionary theory suggests? Further research on natural disasters, for example, may throw light on this question. Does it make a difference whether people feel helpless and believe that they need a strong government to avert the dangers, or whether they believe that they can best meet the dangers with personal efforts.

Does it make a difference whether a crisis is blamed on external enemies or internal culprits? For example, the great depression in the 1930's was blamed on Jews and other foreigners by the Nazi party in Germany, while the financial crisis in 2008 was blamed on “greedy bankers” by US commentators. It has been suggested that the two economic crises had markedly different cultural effects because of the different attributions of blame (Fog, 2017).

It may be enlightening to compare different countries with respect to their level of collective danger and its effects. Inglehart and Welzel's renowned cultural map of the world shows such an effect clearly. Poor and war-torn countries in Africa and parts of Asia are positioned in one corner of the map with high levels of traditional values and survival values, while the opposite corner is populated by the rich North European welfare states with high levels of secular values and self-expression values (Inglehart-Welzel World Cultural Map, 2022). A similar map provided by Fog (2022) is rotated differently so that the effects of collective danger versus security are aligned with the main axis. Again, we are seeing poor and war-torn countries displaying the cultural effects of a high level of danger, while the North European welfare states are reflecting a high level of security and safety.

The direction of causality cannot be determined from the current data, but a review of literature from different fields of study suggests a strong causal effect from perceived collective danger to various psychological and cultural variables related to authoritarianism etc., and a weaker causal effect in the opposite direction from authoritarianism to a heightened perception of dangers.

The conclusion is that authoritarianism and related variables are increased by perceived collective danger as predicted, but not by individual danger. The observed effects of collective danger are in the same direction as predicted by regality theory as well as by several other psychological and cultural theories that do not distinguish between individual and collective danger. The findings are not in agreement with terror management theory and pathogen stress theory.

Limitations of the study

This study is limited by the available data in the World Values Survey. The study could have benefitted from additional survey questions about other threats and fears, such as diseases, economic crises, natural disasters, etc.

The assumption of linear relationships between variables is a possible source of error. The assumption that variables are normally distributed is reasonable in the between-country statistic where each value is calculated as an average of individual values in each country. This assumption is less certain in the two-level level statistic of study 1.

The predictor variables are all highly correlated with each other. This makes it difficult to separate the effects of each predictor variable. Testing with large sample sizes is the best we can due to get sufficient statistical power in a natural setting.

The use of countries as cultural units ignores variation within each country. Fortunately, cultural variables have been observed to cluster around countries (Akaliyski et al., 2021).

Surveys with representative population samples are known to have high levels of measurement error. The high level of statistical noise is a big problem in the present study. The noise is reducing correlation coefficients to a level where type II errors are likely and effect sizes cannot be estimated.