Tolerating the Intolerant: Does Realistic Threat Lead to Increased Tolerance of Right-Wing Extremists?

Previous research suggests that threat can bolster anti-immigration attitudes, but less is known about the effects of threat on ideological tolerance. We tested the hypothesis that realistic threats — tangible threats to e.g., the safety or financial well-being of one’s group — bolster support for right-wing extremists. In Experiment 1, participants (N = 200) learned that crime and unemployment rates were either increasing (high threat condition) or remaining the same (low threat condition). Consistent with our hypothesis, higher threat lead to a significant increase in tolerance for right-wing, but not left-wing, extremists. In a second, pre-registered extended replication experiment (N = 385), we added a baseline (no threat) condition. Additionally, attitudes to immigrants were examined as a mediator. This experiment produced non-significant threat effects on tolerance of right-wing extremists. Overall, the current research provides weak support for the hypothesis that realistic threats have asymmetric effects on tolerance of political extremists. However, consistent with previous research, people were more tolerant of extremists within their own ideological camp.

power, resources, or general welfare.More specifically, feelings of realistic threat emerge when people start associating immigrants with competition over scarce resources and with making neighborhoods less safe (Stephan, Ybarra, & Bachman, 1999).
In line with this theorizing, minority groups are often rejected or derogated in times of extreme threats, such as terrorist attacks, health epidemics, or times of high financial strain (Huddy et al., 2007).Moreover, increased unemployment rates have been linked to immigration restriction attitudes (Strabac & Listhaug, 2008); expected declines in economic wellbeing and fear of unemployment can indeed evoke prejudicial responses (Citrin, Green, Muste, & Wong, 1997); and perceived realistic threats from a growth in the immigrant population lead to subsequent prejudice (Stephan, Renfro, Esses, Stephan, & Martin, 2005).A recent study by Schmuck and Matthes (2015) found that young people who were exposed to advertisements signaling symbolic and economic threat (similar to advertisements that right-wing populist parties typically use) displayed an increase in negative attitudes toward immigrants, and further, that these effects occurred irrespective of individual differences in political party affiliation.
The research reviewed above suggests that perceived economic and safety-related threat can foster negative attitudes toward minority groups.An interesting question that emerges is if the effects of realistic threat extend to ideological tolerance, or more specifically, people's tolerance of political right-wing extremists.Although increases in tolerance of political groups, in the form of support for their rights to express their ideas and participate in public life, do not necessarily imply increased support for these groups, there may be empirical links between tolerance and support.It is possible that threat leads to greater tolerance of right-wing extremists by virtue of shifting attitudes in their direction, and if this is indeed the case, it could bring greater opportunities for extreme right-wing populist parties to spread their message, which in turn might bolster their support from the electorate.
Recent studies have suggested that perceived symbolic threat from an ideological (left-or right-wing) target predicts intolerance of that target (Brandt, Reyna, Chambers, Crawford, & Wetherell, 2014;Crawford & Pilanski, 2014).However, there is also research suggesting that while cultural or symbolic threats posed by immigration often polarize public opinion, realistic threats, such as concern for physical safety, tend to produce more unified reactions, shifting people in the same direction (Feldman & Stenner, 1997).If this is the case in the domain of ideological tolerance as well, then we should expect that general realistic threats of crime or unemployment have asymmetrical effects on ideological tolerance, leading to higher tolerance of right-wing but not necessarily left-wing extremist groups.In other words, if general realistic threats intensify anti-immigrant attitudes, then they should have a stronger effect on tolerance of extreme-right wingers than on tolerance of extreme left-wingers.We tested this hypothesis in Sweden, where the gap in unemployment rates between natives and non-natives is high compared to other European countries (Statistics Sweden, 2019).Despite the fact that Swedes score high on ethnic tolerance compared to other Europeans (Isaksen, Jakobsen, Filindra, & Strabac, 2016), the right-wing populist party "Sverigedemokraterna" has rapidly gained support in recent years and is the third largest party in recent opinion polls (Statistics Sweden, 2021).Because previous studies suggest that left-and right-wingers may differ in their reactivity to threat (Nail, McGregor, Steele, Drinkwater, & Thompson, 2009), we also controlled for individual differences in left-right ideological self-placement in the present study.
Most studies assessing the role of threat for prejudice are correlational (e.g., Caricati, Mancini, & Marletta, 2017;Cohrs & Stelzl, 2010;Crawford, 2014).Although some studies have experimentally manipulated threat, the majority of these have framed or portrayed immigration in ways that were intended to evoke realistic threat (e.g., Lucassen & Lubbers, 2012;Manzi, Roccato, & Russo, 2015;Schmuck & Matthes, 2015;Stephan et al., 2005).We were instead interested in whether it would suffice to present a threatening image without explicitly blaming immigration, that is, whether the threat itself is sufficient to influence attitudes toward right-wing extremism.
Recent research conducted in the United States also suggests that conservatives tend to display higher intolerance of a left-wing target than an opposing right-wing target (e.g., pro-gay vs. anti-gay rights activists), while liberals tend to be more intolerant of a right-wing target than an opposing left-wing target (Crawford & Pilanski, 2014).Besides effects of threat on tolerance, we thus also expected to find general asymmetries in ideological tolerance; such that people should be more tolerant of extremists within their own ideological camp.
In sum, intergroup threat theory suggests that threat stemming from increased unemployment rates and less safe neighborhoods will awaken negative attitudes to immigrants (Stephan & Renfro, 2002;Stephan, Ybarra, & Bachman, 1999).Extreme right-wingers in particular offer a means to ease these feelings of threat by blaming immigrants for societal problems.In times of perceived societal threat, people may therefore exhibit increased tolerance of extreme right-wing (more so than extreme left-wing) political extremists.By tolerating the views of right-wing extremists, one likely increases the probability that concrete action is taken to mitigate these threats, although indirectly so via the actions of other people.Supporting left-wing extremists would do less to mitigate these threats, as these groups are more supporting of immigration.

Participants
The data were collected at public places such as train stations and libraries in different Swedish cities.We tried to recruit a diverse sample with regards to age, gender, and socioeconomic background.Two-hundred participants (47% women, M Age = 38.72,SD = 12.87), of whom 82.5% were employed, 9% students, 8% retired, and 0.5% "other", were randomly assigned to a high threat or a low threat condition.A power analysis conducted in G*power 3.1 showed that this sample enabled the detection of effect sizes (two-tailed) in the small to medium range, d = .40,with 80% power.On average, participants rated themselves 5.72 (SD = 2.30) in terms of left-right ideological self-placement on a scale ranging from 1 = left to 10 = right.

Materials and Procedure
After giving their informed consent to take part in a study on political beliefs, the participants were randomly assigned to a high or low threat condition.Our threat manipulation was designed to elicit different levels of realistic threat pertaining to degrees of safety and economic concerns.In the high threat condition, participants were informed that crime and unemployment rates had increased: This study concerns political beliefs.Recent statistics reveal that people's living standards have in general deteriorated in Sweden, and that both crime and unemployment rates are increasing.This study aims at surveying people's views on current political issues.
In the low threat condition, participants were informed that crime and unemployment levels remain unchanged: This study concerns political beliefs.Recent statistics reveal that people's living standards in general remain unchanged in Sweden, and that both crime and unemployment rates are stable.This study aims at surveying people's views on current political issues.
We measured tolerance for extreme right-wingers with two items, one of which was directly about immigration: "I think that activists who demand that all begging EU-migrants shall be deported, should be allowed to demonstrate outside the Swedish parliament", whereas the other was more general: "I think that political messages from extreme right-wing groups should be blocked or censored from the public" (reverse coded; 1 = do not agree at all, 7 = completely agree, r = .55).The more specific item reflects an issue that has been on the agenda for anti-immigrant parties in Western European countries.The two items were collapsed.Higher ratings reflect higher tolerance of extremists.Tolerance of extreme left-wingers was measured with the following items: "I think that activists who demand that all begging EU-migrants have the right to receive Swedish citizenship and welfare benefits should be allowed to demonstrate outside the Swedish parliament" and "I think that political messages from extreme left-wing groups should be blocked or censored from the public" (reverse coded; 1 = do not agree at all, 7 = completely agree, r = .50).The order was counter balanced for tolerance of right-and left-wingers.
As a manipulation check, we asked the participants what the opening information had stated about people's living standards, with the response alternatives deteriorated, unchanged, and don´t know.Although 10% of participants failed to answer this question correctly, we included them in the statistical analyses but confirmed that excluding them did not change the results.Last, the participants answered the political self-placement item and other demographic questions, where after they were debriefed.

Results and Discussion
Means, standard deviations, and correlations between continuous variables are displayed in Table 1.To test the hypothe sis that realistic threat leads to increased tolerance of extreme right-wingers while controlling for left-right ideological self-placement, we conducted a mixed Analysis of Covariance (ANCOVA) with threat condition (2: High threat vs Low threat; between groups) x tolerance (2: Right-wing vs Left-wing extremists; within groups) while treating left-right ideological self-placement as a covariate.First, there was an effect of threat on average tolerance, F(1, 197) = 5.32, p = .022,η p 2 = .026(η p 2 = .022without the covariate), suggesting that participants who were exposed to high threat became more tolerant of extremists overall.There was also a significant effect of type of tolerance, F(1, 197) = 80.02, p < .005,η p 2 = .29,showing that tolerance was greater for extreme left-than right-wingers when we controlled for left-right ideological self-placement.When we did not control for this variable, there was no difference in tolerance for right-compared to left-wingers (p = .55).
Importantly, the expected interaction between threat (High vs. Low) and type of tolerance (multivariate tests, Wilks' Lambda) was significant, F(1, 197) = 4.28, p = .04,η p 2 = .021,showing that the effect of the experimental manipulation differed depending on type of tolerance.This effect was, however, reduced when participants' left-right ideological self placement was not controlled for, F(1, 198) = .93,p = .34,η p 2 = .005.We proceeded by conducting follow-up ANCOVAs for tolerance of right-and left-wingers separately.First, we conducted a One-way Analysis of Covariance (ANCOVA) with threat condition (2: High threat vs. Low threat; between groups) as the independent variable and tolerance of right-wing extremists as the dependent variable, while controlling for left-right ideological self-placement.The covariate left-right ideological self-placement showed a significant relationship with tolerance of right-wing extremists, F(1, 197) = 35.79,p < .001,η p 2 = .154.As expected, greater self-placement to the right was associated with greater tolerance of right-wing extremists.More importantly, and consistent with the hypothesis, participants in the high threat condition (M = 5.94, SD = 1.62) became more tolerant of extreme right-wingers compared to those in the low threat condition (M = 5.41, SD = 1.81),F(1, 197) = 8.52, p = .004,η p 2 = .04.This effect became somewhat weaker but remained significant without the covariate, F(1, 198) = 4.66, p = .03,η p 2 = .023,d = .30.Next, we ran the same analysis but with tolerance of extreme left-wingers as the dependent variable.The covariate left-right ideological self-placement showed a significant relationship with tolerance of left-wing extremists F(1, 197) = 8.16, p = .005,η p 2 = .04.As expected, greater self-placement to the right was associated with less tolerance of extreme left-wing extremists.Participants in the high threat condition (M = 5.89, SD = 1.48) did not differ significantly from the low threat condition (M = 5.61, SD = 1.53) in their tolerance of extreme left-wingers, F(1, 197) = 1.15, p = .29,η p 2 = .006.This difference remained small and non-significant without the covariate, F(1, 198) = 1.78, p = .18,η p 2 = .009,d = .19.Because our outcome measure of tolerance consisted of only two items, we confirmed that ANCOVAs with each individual item as the dependent variable produced statistically significant effects on tolerance of right-wing extremists (ps = .02;.007),and non-significant effects for tolerance of left-wing extremists (ps = .56;.22).
In sum, the results of our first experiment suggest that people who are exposed to realistic threat show more tolerance of extreme right-wing, but not extreme left-wing, groups.
Finally, although we expected effects of threat to shift tolerance regardless of participants´ ideological viewpoint, it is possible that threat fosters support for far-right extremism particularly among individuals who are already on the right side of the political spectrum, or reversely, that those leaning towards the left are more reactive to threat (Nail et al., 2009).We therefore conducted a series of exploratory analyses to examine whether ideological self-place ment moderated effects of condition (High threat vs. Low threat) and target (Left-wing vs. Right-wing extremists) on tolerance through linear mixed models with Maximum Likelihood estimation.There was a significant two-way interaction between left-right self-placement and target (unstandardized fixed effect = .487[.357, .617],p < .001)but no significant three-way interaction between left-right self-placement, target, and condition (unstandardized fixed effect = -.099[-.284, .086],p = .29)and no significant two-way interaction between left-right self-placement and condition (unstandardized fixed effect = .066[-.128, .261],p = .51).In other words, ideological position moderated the effect of target on tolerance (i.e.participants were more tolerant of extremists from their own ideological group), but we found no evidence that left-and right-wingers responded differently to the threat manipulation used in this study.

Experiment 2
The main aim of Experiment 2 was to replicate the effect from Experiment 1 while adding a baseline control group where participants were not exposed to any message concerning living standards, crime or unemployment, and using a more extensive measure of political tolerance that we developed for this study.The experiment was preregistered on the Open Science Framework (see Supplementary Materials).
The participants were randomly assigned to either the same high threat or low threat condition as in Experiment 1, or to a baseline (no threat) condition. 1Once again, we hypothesized that realistic threat leads to increased tolerance of right-wing extremists in particular.In the context of our three conditions, we predicted that participants in the high threat condition would show greater tolerance toward right-wing (but not left-wing) extremists than participants in the low threat and baseline/no threat conditions.We also wanted to look into the possibility that realistic threat leads to increased tolerance of extreme right-wing groups through a more negative view of immigrants.A second aim was thus to test the hypothesis that attitudes to and perceptions of immigrants mediate the effect of realistic threats on tolerance of right-wing extremists.

Method Participants
The data were collected at public places in Swedish cities.We aimed for a similar sample size as in Experiment 1, with n = 100 per experimental condition, for a total of 300 participants.However, we decided to collect data until we reached this sample size for participants who correctly answered the manipulation check to be able to conduct well-powered tests of our hypotheses both with and without participants who failed the manipulation check.We excluded one participant who had responded only to two questions and one who had selected multiple responses to all questions.This left us with a total sample of 385 participants, of which 309 correctly answered the manipulation check.The gender distribution was very even (50.8%women, 47.8% men, and 1.0% who identified as "other").The mean age (M Age = 38.72,SD = 12.87) was also fairly close to the national average (41.2 years, Statistics Sweden).Moreover, 54.4% were employed, 35.5% were students, 6.5% described their occupation as "other", and 3.6% did not respond.On average, participants rated themselves 4.91 (SD = 1.92) in terms of left-right ideological self-placement on a scale ranging from 1 = left to 10 = right.

Materials and Procedure
After giving their informed consent, the participants in the high threat and low threat conditions received information about crime and unemployment rates that was identical to that presented to participants in Experiment 1, while the participants in the baseline/no threat condition received no information at all.
We extended our dependent variables of tolerance of extreme right-and left-wingers by including eight items each (i.e., 16 items in total), which we pretested with a sample of 54 people.Specifically, we kept the two items from Experiment 1, and constructed six more items in order to cover a broader spectrum of contexts relevant for tolerance, e.g., "I think that newspapers should censor authors who express extreme right-wing political opinions", "I think that right-wing extremists should be allowed to take part in cultural events, even though a lot of people may feel offended by it", and "I think that products that are manufactured by individuals or groups with politically extreme right-wing opinions should be boycotted" (1 = do not agree at all, 7 = completely agree).We measured tolerance of left-wing groups with the same items replacing "right-wing" with "left-wing".See Table 2 for the full set of items along with factor loadings based on exploratory factor analysis with principal axis factoring, oblimin rotation, and a two-factor solution (identified through the scree plot).Both measures showed acceptable reliability with Cronbach's α = .84for tolerance of extreme right-wingers and .82for tolerance of extreme left-wingers.The fact that the new tolerance measures displayed improved internal consistency compared to the shorter measures used in Experiment 1, coupled with their significant relationship with attitudes toward immigrants (see Table 3), attest to the appropriateness of these measures.As in Experiment 1, we counter balanced the order of items for tolerance of right-and left-wingers.Note.Factor loadings were derived from the pattern matrix.The first eight items (translated from Swedish to English) measured tolerance of right-wing extremists and the remaining eight items measured tolerance of left-wing extremists.The two-factor solution accounted for a total of 47.5% of the variance, with Factor 1 contributing 35.0% and Factor 2 contributing 12.5%.
We measured attitudes toward immigrants with four items that were adapted from Schwartz, Caprara, and Vecchione (2010): "People who come to live here from other countries generally make our country a better place to live" (reverse coded), "People who come to live here from other countries generally take jobs away from the workers of our country", "People who come to live here from other countries make our country's cultural life richer" (reverse coded), and "People who come to live here from other countries undermine harmony in our country" (1 = strongly disagree, 7 = strongly agree), Cronbach's α = .82.We also asked the participants to estimate the number of crimes that are committed by immigrants as opposed to natives (1 = Swedes commit almost all crimes, 7 = Immigrants commit almost all crimes).We used the same manipulation check as in Experiment 1, but we added an explicit request that participants answer this question without returning to check the information that they had been provided with at the beginning of the study (i.e., the manipulation).Although 20% of the participants failed to answer this question correctly, we included them in the statistical analyses after confirming that excluding them did not alter the results.Last, the participants answered the political self-placement item and demographic questions and were debriefed.

Results and Discussion
Descriptive statistics and correlations are provided in Table 3.To test the hypothesis that realistic threat leads to increased tolerance of extreme right-wingers while controlling for left-right ideological self-placement, we conducted a mixed Analysis of Covariance (ANCOVA) with experimental condition (3: High threat, Low threat, and Baseline/no threat control; between groups) x tolerance (2: Right-wing vs Left-wing extremists; within groups) while treating left-right ideological self-placement as a covariate.The results revealed no significant effect of threat on average tolerance, F(2, 365) = .66,p = .52,η p 2 = .004(η p 2 = .005without the covariate).The predicted interaction between threat and type of tolerance was weak but statistically significant, F(2, 365) = 3.32, p = .04,η p 2 = .018(multivariate tests, Wilks' Lambda), indicating that the effect of the experimental manipulation differed somewhat depending on type of tolerance (see Figure 1).This effect was, however, noticeably reduced when we removed the covariate, F(2, 381) = .36,p = .70,η p 2 = .002.As in Experiment 1, there was a significant effect of type of tolerance, F(1, 365) = 288.26,p < .001,η p 2 = .44,showing that overall, tolerance was higher for extreme left-than right-wingers, when controlling for left-right ideological self-placement.When we did not control for this variable, the effect of type of tolerance was weaker (η p 2 = .24)but remained significant (p < .001).
We proceeded by conducting follow-up ANCOVAs for tolerance of right-and left-wingers separately.First, we conducted a One-way Analysis of Covariance (ANCOVA) with the three experimental conditions (High vs. Low.vs. Baseline/no threat) as the independent variable and tolerance of right-wing extremists as the dependent variable, while controlling for left-right ideological self-placement.
In sum, the results of Experiment 2 did not provide any clear support for the hypothesis that realistic threat leads to increased tolerance of extreme right-wingers.
We proceeded to test the hypothesis that attitudes to immigrants mediate effects of threat on political tolerance.We investigated mediation by estimating unstandardized indirect effects with bias-corrected 95% bootstrap confidence intervals (10000 resamples) through the PROCESS macro v. 3.4 (Hayes, 2017) for SPSS.We tested one model with high threat condition vs. low threat condition as the independent variable and one with high threat condition vs. baseline/no threat condition as the independent variable; both models contained two mediators (attitudes to immigrants and estimated proportion of crimes committed by immigrants) and one outcome variable (tolerance of right-wing extremists).Neither model yielded any evidence of indirect effects through attitudes to immigrants (High threat vs. Low threat condition: indirect effect = -.010[-.104, .087];High threat vs. Baseline: indirect effect = -.070[-.191, .039])or perceived criminality of immigrants (High threat vs. Low threat condition: indirect effect = -.008[-.056, .018];High threat vs. Baseline: indirect effect = .008[-.036, .064]).All of the indirect effects remained non-significant when we controlled for left-right self-placement.The hypothesis that attitudes to immigrants mediate effects of threat on political tolerance was thus not supported.
Finally, we conducted exploratory (non-preregistered) analyses through mixed linear models (with Maximum Like lihood estimation) to investigate whether ideological self-placement moderated the effects of experimental condition (High threat, Low threat, and Baseline/no threat control) and target (Right-wing vs Left-wing extremists) on tolerance.Similar to Experiment 1, there was a significant two-way interaction between left-right self-placement and target (un standardized fixed effect = .364[.182, .546],p < .001),indicating that participants showed greated tolerance of extremists within their own ideological camp.There was no significant three-way interaction between left-right self-placement, target, and being in the high-threat condition (unstandardized fixed effect < .001[-.242, .242],p = 1.00), but there was a significant interaction between left-right self-placement and high-threat condition (unstandardized fixed effect -.217 [-.389, -.046], p = .013)indicating that there was a slightly stronger association among left-wingers than right-wingers between high threat and tolerance of extremists overall, irrespective of whether the targets were left-or right-wing extremists.Excluding participants who failed to answer the manipulation check correctly produced almost identical results.

General Discussion
Previous research suggests that feelings of realistic threat emerge when people start associating immigrants with competition over scarce resources and with making neighborhoods less safe (Stephan & Renfro, 2002;Stephan, Ybarra, & Bachman, 1999), which in turn fuels increased hostility toward immigrants and minority groups (Canetti, Halperin, Hobfoll, Shapira, & Hirsch-Hoefler, 2009;Huddy, Feldman, & Weber, 2007;Sari, 2007).In the present research program, we sought to extend previous work on the effects of realistic threats by focusing on tolerance of ideological groups rather than prejudice or intolerance of minority groups per se.Specifically, we examined whether a general realistic threat has effects on tolerance of political extremists.Previous research has shown that right-wing populist parties tend to blame societal problems on immigrants, and using immigrants as a scapegoat may help people cope with feelings of threat (Rothschild et al., 2012;Schmuck & Matthes, 2015).We therefore hypothesized that the realistic threat evoked by the presentation of a pessimistic view of crime and unemployment rates leads to increased tolerance of extreme right-wing, but not necessarily left wing, groups.However, the results were mixed.Although a first experiment provided support for the hypothesis, a second pre-registered experiment with an extended design that included a baseline condition produced weak and inconclusive findings.Bearing in mind that both experiments were designed with sufficient statistical power to detect medium-sized effects, it seems unlikely that there are strong or moderate effects of realistic threat on tolerance of extremists in the social and cultural context within which this research was conducted.
Although the findings from our two experiments combined were inconclusive with regards to the size of threat effects, we did find clear indications of general asymmetries in ideological tolerance.Specifically, even though there was greater tolerance of extreme left-wingers than extreme right-wingers overall in both studies, persons who leaned more to the right were consistently more tolerant of extreme right-wingers and less tolerant of extreme left-wingers than were persons who leaned more to the left.In other words, people tended to be more tolerant of extremists within their own ideological camp.The findings are very clear in this regard, and they are also in line with findings of research con ducted in the United States, suggesting that conservatives tend to display higher intolerance of a left-wing target than an opposing right-wing target (e.g., pro-gay vs. anti-gay rights activists), while liberals tend to be more intolerant of a right-wing target than an opposing left-wing target (Crawford & Pilanski, 2014).This general ideological asymmetry in tolerance of extremists might help explain the controversies over the boundaries of tolerance and free speech for the intolerant that have cropped up recently all over the Western world -an issue that has been debated by philosophers for decades (e.g., Rawls, 1971;Popper, 1945).Undoubtedly, the defense of extremists' rights to protest, demonstrate, and voice their opinions is in part motivated by legitimate liberal concerns with free speech and democracy, rather than advocacy of extremist opinions or prejudice.However, our results suggest that such tolerance is not evenly dispersed, but awarded to certain extremists while at the same time denied to others, depending on the ideological beliefs of the observer.It is important to be aware of how this mechanism may be affecting the ways that contemporary controversies over tolerance of the intolerant unfold -controversies about, for example, whether neo-Nazi or radically left-wing groups should have unrestricted opportunities to spread their message in political forums.

Limitations and Further Research
Arguably, realistic threat should be most relevant for low-skilled workers, who compete more directly with immigrants on the labor market (Schmuck & Matthes, 2015).Some findings suggest that prejudiced attitudes increase primarily in poorer regions with the size of the immigrant population (Isaksen et al., 2016), while others suggest that when immigrants are perceived as an economic burden, opposition to immigration increases regardless of the perceiver's skill level (Hainmueller & Hiscox, 2007).We did not examine socio-economic status as a potential moderator of threat effects on tolerance for political extremists, but researchers may wish to probe this potentially moderating variable in future studies.
A related limitation is that we did not investigate the extent to which the participants de facto perceived the information about crime and unemployment rates that we presented to them as threatening, as our manipulation check focused on the cognitive accessibility of threat rather than its affective component, nor did we investigate whether there are threat effects on tolerance that persist over time.It is possible that the manipulation would be stronger if for example newspaper articles about current economy issues and crime rates were provided.
Furthermore, a general weakness of the experimental manipulation is that we could not know whether it is the mentioning of crime or economic woes, or both, that would drive potential threat effects with the manipulation we used.Future studies should also seek more knowledge about the relative impact of different threats suggested by intergroup threat theory (Stephan & Renfro, 2002), by pitting the effects of realistic threat against those of symbolic threat on tolerance of extremists.Finally, researchers might want to assess other mediators of threat effects, such as perceptions of immigrants as the "cause" of economic problems, or perceptions of extremist parties offering a solution to these problems, as this could be a more precise test of the proposed mechanism.
On a positive note, a strength of this research is that instead of investigating our hypotheses for tolerance of right-wing extremists only, we also measured tolerance of left-wing extremists, enabling us to study the relative impact of threat on tolerance of right-vs.left-wing extremists.Although asymmetries in the characteristics of left-and right-wingers have generated research for decades (for a review, see Jost, 2017), much less is known about asymmetries (and symmetries, see e.g.Brandt, 2017;Crawford & Pilanski, 2014;Graham, Nosek, & Haidt, 2012) in the domain of prejudice and tolerance directed at different ideological targets.Continuing to investigate these asymmetries is important for understanding contemporary political controversies as well as promoting civil discourse and democracy.On the other hand, we did not measure tolerance of specific ideological targets.Investigating tolerance across a wide range of left-wing and right-wing targets in a systematic way could help researchers take individual differences in the perception of extremism on the left and the right into account.
Both our samples were diverse with regards to occupation, age, gender, and political orientation, and larger than those used in the majority of previous studies on effects of realistic threat.Nevertheless, a critical limitation of our research is that it focused on one single country.It is possible, for example, that chronic levels of threat matter for the impact of realistic threat on tolerance, and these may be particularly low in Sweden, which has a long history of relative peace and prosperity (Kent, 2008).The strong general levels ot tolerance, support for democratic principles, and secular attitudes in Sweden (Inglehart & Welzel, 2010) may also play an important role.It is possible, for instance, that tolerance of violations of basic democratic principles, such as freedom of speech, reflect beliefs and ideals that are more stable compared to attitudes to minorities in the Swedish context, and therefore less likely to be influenced by exposure to threatening information.It is worth noting, furthermore, that Sweden has a multiparty system, a diverse ideological landscape with several distinct groups of voters with different ideological worldviews on the left and the right (Nilsson et al., 2020), and a highly politically knowledgeable population (Milner, 2002).It is possible that effects of threat on tolerance of left-and right-wing extremists specifically are stronger or more consistent in countries with a biparty system or a more bipolar political discourse (e.g., the United States).These and other idiosyncratic features of the Swedish context, in combination with the unclear findings, make it difficult to extrapolate the findings to other cultural contexts, particularly beyond the Western European cultural sphere.Systematic cross-cultural studies are ultimately needed to shed light on contextual moderators of the link between realistic threat and political tolerance.Considering that this (to the best of our knowledge) was the first attempt to study the consequences of realistic threat on tolerance of political extremists, more research, perhaps using other methods, is called for.

Figure 1
Figure 1 Effects of Threat on Tolerance of Right-Wing and Left-Wing Extremists (Experiment 2)

Table 1
Descriptive Statistics and Pearson Correlations for the Continuous Variables in Experiment 1 Note.Left-right ideological self-placement is measured on a scale from 1-10, tolerance is measured on a scale from 1-7. **p < .01.

Table 2
Exploratory Factor Analysis for the Items That Measured Tolerance of Right-Wing and Left-Wing Extremists in Experiment 2I think that activists who demand that all begging EU-migrants have the right to receive Swedish citizenship and welfare benefits should be allowed to demonstrate outside the Swedish parliament.
I think that representatives of the Sweden Democrat party should be allowed to visit high schools and talk to students about the party.I think that extreme left-wing activists/groups (e.g.Antifascistisk Aktion) should be allowed to participate in the Book I think that members of left-wing extreme groups should be allowed to make a speech in public places.

Table 3
Descriptive Statistics and Pearson Correlations for the Continuous Variables in Experiment 2 Note.Left-right ideological self-placement is measured on a scale from 1-10; tolerance, attitudes toward immigration, and perceptions of immigrants as criminals are measured on a scale from 1-7. *p < .05. **p < .01.