Beyond General Political Attitudes: Conspiracy Mentality as a Global Belief System Predicts Endorsement of International and Local Conspiracy Theories

Conspiracy mentality is a general tendency to attribute significant events to the actions of malevolent actors, without referencing to a specific event. In two independent representative surveys of adult Serbian citizens (N1 = 1194; N2 = 1258) we validated Serbian version of the conspiracy mentality questionnaire (CMQ), a reasonably content-free tool designed to capture global conspiratorial beliefs. We successfully validated the adapted CMQ and replicated findings on two national representative samples. In both studies the results demonstrated: good psychometric properties of the CMQ and its predictive capacity for endorsing the international Conspiracy Theories (CTs) (Study 1) and the locally specific CTs (Study 2) over and above the measures of perceptions of political climate (trust in institutions, corruption perception, feeling of insecurity – Study 1), and generalized political attitudes (right-wing authoritarianism, social dominance orientation, political cynicism – Study 2). The study presents a unique adaptation and implementation of the CMQ in the non-English speaking country with very active and widespread conspiracy beliefs and a long history of conflicts.

We examined population in this region outside of the English-speaking Western world, with a long history of war and violence, to examine the conspiracy mentality. Arguably, identifying processes that are lying behind conspiracy belief becomes a much more existential task in this setting. Thus, an extension of research to Serbian territory is very important, because being the target of a CT can become a lethal threat not only on the basis of the actions of some individual acting in the CT, but as a matter of minority suppression. In addition, we tested the replicability of the results in two separate studies on high quality representative samples.
To capture individual differences in conspiratorial tendencies, psychologists have developed two different sets of measures: (a) endorsement of different real-world CTs, and (b) endorsement of a more generic 'conspiracy mentality', a general tendency to attribute significant events to the actions of malevolent actors, without referencing to a specific event (Swami et al., 2017). The former usually refers to a singular event (e.g. moon-landing is staged; the assassinations of popular figures happened differently than media presented it) or classes of events (e.g. HAARP system controls the weather; "chemtrails" are toxic to people and environment). Having in mind historical specifics of the region, we developed two different types of specific CTs that include global plots controlled by secret centers of power and specific CTs, mostly related to the breakup of Yugoslavia.
The need for a general measure of conspiracy belief was twofold: (a) when measuring endorsement of specific CTs, there is always noise coming from their content that affects their selective endorsement (e.g. hardly any are politically neutral), and (b) the events and actors that constitute specific CTs are always dependent on the context and moment in history. To enable important cross-cultural comparisons and significant relations with relevant correlates, a wide-ranging approach was called for. In the recent overview by Goreis and Voracek (2019), thirteen questionnaires have been used to measure conspiracy beliefs in the last decade, and two of most-used scales measured generic beliefs without specific conspiracies (GCBS; Brotherton, French, & Pickering, 2013;and CMQ;Bruder et al., 2013). However, other scales with specific content about conspiracies also exist, like Conspiracy Theory Questionnaire (CTQ, Darwin, Neave, & Holmes, 2011) and Generic Conspiracist Belief Scale (GCBS, Brotherton et al., 2013). Here, we used a short version of Conspiracy Mentality Questionnaire (CMQ, a 12-item and 5-item version) in both studies. The CMQ was conceived and validated by Bruder and colleagues (Bruder et al., 2013;Imhoff & Bruder, 2014). Conspiracy mentality is typically described as a general political attitude, different from similar constructs, such as right-wing authoritarianism (RWA; Altemeyer, 1988) and social dominance orientation (SDO; Pratto, Sidanius, Stallworth, & Malle, 1994). To prove it is a distinct ideological belief system, researchers demonstrated that it relates differently to prejudice than these two measures -whilst the conspiracy mentality predicted prejudice towards high power groups, SDO and RWA lead to more prejudice towards low power ones (Imhoff & Bruder, 2014).
The CMQ is widely used and conceived as a measure that would guarantee more invariance in cross-cultural research. However, it has been adapted to a relatively modest number of languages (to our knowledge English, German, Turkish and French). In addition, its measurement characteristics have not been always proven robust (see, for example, Swami et al., 2017). Namely, one-factor solution fitted in some samples (German and English speaking), but proved inadequate in other samples (e.g. Turkish). Additionally, reported internal consistency also varied widely. This could partially be attributed to the lack of construct validity: whilst some statements clearly reflect conspiratorial ideation (e.g. I think that events which superficially seem to lack a connection are often the result of secret activities), others can be viewed as statements of fact, especially in a light of current findings regarding unauthorized gathering of personal data by institutions and corporations (e.g. I think that government agencies closely monitor all citizens).
Having all this in mind, drawing from two independent representative samples of adult Serbian citizens, we set out to explore: a. The internal consistency and uniqueness of the CMQ instrument compared with similar constructs; we think it is of high relevance to further explore construct validity of CMQ and its predicting power to predict specific CTs, due to constantly evolving nature of CTs. In the area of CTs, specific theories (at least those with immediate political relevance) are so fluid, that a generic measure is called for. In some studies, strong claims have been made about comparative viability of different measures based on a not so strong empirical data (Swami et al., 2017). Therefore, it is critical that further research is being brought into the public domain, testing whether different scales stand up to scrutiny, especially on a high quality sampling from the underrepresented non-English speaking region, such as Balkans. b. How generic conspiracist ideation (CMQ) relates to belief in concrete CTs -international ones adapted from Belief in Conspiracy Theory Inventory (Study 1) or the local CTs, typical for the Balkan region, designed specifically for the study (Study 2); c. If the predictive power of CMQ for belief in concrete CTs goes over and above measures of perceptions of political climate -trust in institutions, corruption perception, feeling of insecurity (Study 1), and over and above the measures of generalized political attitudes -right-wing authoritarianism, social dominance orientation and political cynicism (Study 2).
We conducted both studies with a full sampling procedure (including random walk procedures, random selection of households and household members) resulting in two representative samples of the Serbian population. This procedure allows limiting self-selection bias and ensuring representation much more, than almost all other studies on conspiracy mentality in the literature using samples of participants recruited online or focused on targeted student population. For example, the recent meta-analysis by Goreis and Voracek (2019) states only two studies using representative samples (Hollander, 2018, in the US;Mancosu et al., 2017 in Italy). However, both of mentioned studies focus on specific CTs, while we included the generic belief in CTs and two types of specific CTs, local and global. Other authors found heterogeneity in designs and operationalization in the study of conspiracy beliefs, while asking for consensus. Our study is in line with this call. Therefore, this manuscript provides further evidence for the viability of one of the two most often-used questionnaires to assess conspiracy mentality. Additional strength is that issue is reported from two studies that are conceptual replications of one another. We expect that the instrument will show good internal reliability and that measurement model will show it to be a distinct construct in comparison to other measures of political climate (Study 1) and other measures of generalized political attitudes (Study 2) even in the non-English speaking region with a strong conflict history (H1). We also hypothesized that it will predict both internationally (Study 1) and locally relevant (Study 2) specific CTs (H2). Finally, we expected it to provide a unique contribution to belief in concrete CTs, above the measures of political climate (Study 1) and climate coupled with the political attitudes (Study 2) (H3).

Respondents and Sampling
A total of 1194 adult citizens of Serbia were recruited in a stratified three-stage probability sampling procedure in No vember 2016 by the Institute of Political Studies, Belgrade. The sampling frame was based on the data from 2011 Census. The first stage sampling was done by polling station territory (212 sampling points). The second stage were households selected by random route technique starting from the randomly selected addresses (6 households by sampling points). The third were respondents randomly selected within households. The average length of the interview was 30 min. Research procedures adhered to the APA ethical guidelines. Table 1 summarizes the demographic characteristics of the both samples.

Instruments
We added full questionnaire in English and Serbian language into the Supplementary Materials. Conspiracy mentality was assessed with a five-item version of CMQ (Bruder et al., 2013) with an 11-point scale, from 1 (completely false) to 11 (completely true) (α = .81).
Questionnaire for specific CTs consisted of international CTs (e.g. Moon landing was staged; Vapor trails left by aircraft are actually chemicals that pollute the environment and are toxic to people; People are deceived about vaccine safety) anchored 1 (I do not agree at all) to 7 (I completely agree) (α = .88). The items were chosen based on several validated instruments, such as Belief in Conspiracy Theory Inventory (Swami et al., 2011), Vaccination Conspiracy Scale (Shapiro, Holding, Perez, Amsel, & Rosberger, 2016), Conspiratorial narratives (Oliver & Wood, 2014).
Respondents reported how much they (dis)trust nine public institutions (e.g. federal government, parliament, police, army, president, and judiciary), on a scale from 1 (I completely trust) to 4 (I do not trust at all) (α = .91).
Respondents reported about corruption perception in the society in general on six items (Sampford, 2016) (e.g. Public employees use their positions for personal gain; Contracts are given to the companies that offer highest bribes to public officials) with a scale from 1 (never) to 7 (always) (α = .93).
Respondents assessed how secure they perceive the world in general, the region, and their personal surroundings using a scale from 1 (I don't feel secure at all) to 5 (I feel completely secure) (α = .69).

Results
To test the independence of constructs employed as predictors in Study 1 (Conspiracy mentality, Feeling of security, Trust in government institutions, Corruption perception), we ran a confirmatory factor analysis in AMOS (Version 18). Since we tested predictive power of CMQ for perceptions of political climate we ran only four factor model and did not include in the model the similar concept to CMQ (such as global CT). A proposed four-factor model fitted the data well, as the indices demonstrate: Chi-square = 1429.8, Chi-square/df = 1429.8/201 = 7.11, Comparative fit index (CFI = .92), Bentler-Bonett Normed Fit Index (NFI = .90), Root Mean Square Error Of Approximation (RMSEA) = .07. All the items loaded on the designated factors with loadings ranging from .66 to .83 for conspiracy mentality; from .62 to .82 for trust in institutions; from 0.85 to 0.90 for corruption perception; and from .66 to .89 for the feeling of security. This allowed us to treat the conspiracy mentality as an empirically different construct from the other three.
As the CMQ explores global conspiratorial tendencies, it should not be particularly sensitive to any preexisting knowledge, whilst endorsement of concrete CT depends on their familiarity. Therefore, all respondents assessed the truthfulness of the CMQ items, and only those who previously heard about a specific CT assessed its truthfulness. Due to this restriction, we did not impute missing data for these questions, but instead, we run the analysis on the subsamples of those familiar with specific CTs.
The average agreement with five statements from conspiracy mentality questionnaire was M = 7.88, on an elev en-point scale. Most endorsed item was "Politicians usually do not tell us the true motives for their decisions" (M = 8.60, SD = 2.80), and least endorsed was "Government agencies closely monitor all citizens (M = 6.67, SD = 3.08)". In exploratory factor analysis, all items loaded on one factor, explaining 57.98% variance.
The average agreement with eight statements from the specific CT scale was M = 4.86 (out of seven). The agreement with the specific CTs varied from M = 4.40, SD = 2.12 for "The moon landing by the US astronaut was staged" to M = 5.44, SD = 1.72 for "Big corporations are working towards systematically eradicating minority populations throughout the world". All descriptive measures are detailed in Table 2. Note. CMQ = Conspiracy Mentality Questionnaire; CT = Conspiracy theory.
As expected, conspiracy mentality and endorsement of specific CTs were highly correlated, r(619) = .49, p < .001. The conspiracy mentality also highly positively correlated with corruption perception, r(929) = .42, p < .001, and slightly positively with distrust in institutions, r(880) = .08, p = .021, whilst it was unrelated with the perception of security (Table 3). To test the predictive power of conspiracy mentality for endorsement of specific CTs over and above perceptions of political climate, we conducted a hierarchical regression in which we entered the feeling of security, distrust in institutions and perception of corruption in the first step, and CMQ in the second. Whilst the first set of predictors explained around 9% of variance, adding the second resulted with 28% of variance explained (ΔR 2 = .19). Regression models are presented in Table 4. Additionally, to analyze the CMQ predicative value for endorsement of specific international CTs one by one, we conducted a series of regression analyses (Table 5). The CMQ proved to be a significant predictor of all nine specific CTs, from "Big business" conspiracy, to "Vaccine safety" conspiracies. In order to explore further confirmation about the validity and predicting value of CMQ instrument in Serbia, we kept investigating those issues using similar methodology in Study 2. We tested CMQ as oppose to local CTs from the Balkan region, and its predictive value for general political attitudes (right-wing authoritarianism, social dominance orientation and political cynicism).

Respondents and Sampling
A total of 1258 adult citizens of Serbia were recruited in a stratified three-stage probability sampling procedure in May 2017. The sampling procedure was the same as previously described for Study 1. The sample's demographic breakdown is presented in Table 1. All research procedures adhered to the APA ethical guidelines.

Instruments
We used the same instruments as in Study 1 to assess conspiracy mentality (α = .83), distrust in institutions (α = .93), corruption perception (α = .93) and security (α = .76). For this study, we used a set of six regionally widespread CTs, all related to the dissolution of former Yugoslavia. They included the actions of local and international politicians during and after the conflict (e.g. Presidents Milošević and Tuđman agreed in Karađorđevo how to divide Bosnia and Herzegovina; The crimes against the civilian population in action Flash and Storm were part of the systematic policy of ethnic cleansing devised beforehand by the Croatian authorities; World Bank, MMF and other international institutions economically and financially colonize Serbia for several decades). When choosing the specific CTs, we relied upon the ones validated on Croatian sample (Blanuša, 2013), as well as Serbian forums, blogs and social media groups formed around various CTs. Respondents expressed their endorsement on a scale from 1 I do not agree at all to 5 I completely agree (α = .71).
Right-wing authoritarianism was measured with a nine-item version of Altemeyer's RWA scale such as the most important virtue that a child should learn is obedience and respect for authority (Altemeyer, 1988(Altemeyer, , 1998 adapted for Serbian sample (as in Todosijević, 2013;Enyedi & Todosijević, 2003) (α = .92).
Social dominance orientation scale comprised of eight items (Ho et al., 2015) adapted for Serbian (as in Jovanović et al., 2017;Žeželj & Pratto, 2017), with typical statements such as "Some groups of people are simply inferior to other groups", anchored with 1 (I completely disagree) to 5 (I completely agree) (α = .70).
Political cynicism was assessed by eight items derived from a widely-used scale (Dekker & Meijerink, 2012;Schyns & Nuus, 2007) adapted for Serbian (Pavlović, 2013), with statements like "Political parties are only interested in my vote and not my opinion", anchored with 1 (I completely disagree) to 5 (I completely agree) (α = .94).

Results
To test if conspiracy mentality can be viewed as the empirically distinct construct in comparison to measures of generalized political attitudes, we ran a confirmatory factor analysis in AMOS (Version 18). In the Study 2 we tested predictive power of CMQ for general political attitudes and did not include in the model other similar concepts (such as regional CT). Again we ran four-factor model with only new additional variables of general political attitudes. A four-factor model fitted the data well: Chi-square = 1849.4, Chi-square/df = 1849.4/368 = 5.02, Comparative fit index (CFI = .92), Bentler-Bonett Normed Fit Index (NFI = .90), Root Mean Square Error of Approximation (RMSEA) = .06. All items loaded on the designated factors with loadings ranging from .72 to .83 for conspiracy mentality; and from .69 to .81 for RWA; from .73 to .87 for political cynicism; and loadings from .68 to .82 for social dominance orientation.
In the Study 2, the average agreement with all items in CMQ was even higher than in the Study 1 (M = 8.14, SD = 2.74 out of eleven). The ranking of items mirrored the ranking from the first study exactly. Similarly, all items loaded on one factor that explained 60.72% variance.
The average endorsement of regional CTs was M = 3.47, SD = 1.38 on a five-point scale. Item loadings and averages are detailed in Table 6.

Regional CTs (min 1 max 5)
The European Union is a conspiracy of big business whose goal is the destruction of nation states. Even though we used a different set of concrete CTs compared to the first study (regional versus international), the average endorsement is still quite highly correlated with the CMQ, r(1228) = .33, p < .001. Similarly, out of political climate variables that were the same we used in the first study, CMQ was correlated the most with the perception of corruption, r(1235) = .34, p < .001, followed by distrust in institutions, r(1218) = .19, p < .001. When related measures of individual differences in general political attitudes were assessed, they were most highly correlated with political cynicism, r(1232) = .28, p < .001, significantly albeit weekly with RWA, r(1227) = .07, p = .01, and unrelated to the social dominance orientation. The full correlation matrix is presented in Table 7. Specific CT (1) -.33*** .14*** .03 .02 .10*** .13*** .20*** CMQ (2) -.34*** .19*** .08*** .05 .28*** .07*** Corruption perception (3) -.33*** .12*** -.10*** .40*** -.04 Distrust in institutions (4) --.19*** .01 -.26*** .25*** Feeling of security (5) -.11*** -.02 -.03 Social dominance (6) --.09*** .08*** Political cynicism (7 Following the same logic as in the first study, we conducted a hierarchical regression analysis, but this time with three steps. Measures of political climate (trust in institutions, perceived corruption, and feeling of security) explained 3% of the variance in local CTs, adding a set of general political attitudes (RWA and SDO) resulted with the total of 11%. Finally, adding CMQ significantly improved the predictive power, with additional 7.5% -in total 18.5% (Table 8). Finally, to analyze the CMQ predictive power for endorsement of singular locally specific CTs, we conducted a set of regression analysis, similar to Study 1 ( Table 9). The CMQ predicted endorsement of five out of six CTs -all except the one that blamed the ingroup (Serbs) and the Serbian institutions for initiating the Yugoslav civil wars, which was probably saturated with other constructs (such as attribution of guilt, ingroup bias, etc.) more than actual conspiratorial thinking.

Discussion
In two independent representative samples (total N = 2452) the CMQ demonstrated excellent internal consistency and stable average scores by items over time. Exploratory factor analysis yielded one factor in both studies, indicating it is a unidimensional construct. Measurement models in both studies also supported the conspiracy mentality as a distinct construct, different from indicators of global political climate and generalized political attitudes. Firstly, taken together, the two studies demonstrate that conspiracy mentality, as a global measure of tendency for conspiratorial thinking, predicts belief in the combined index of endorsement of international CTs (e.g. chemtrails, moon landing, 9/11) tested in Study 1 and the combined index of endorsement of more locally oriented CTs (e.g. IMF/World bank, Operation Storm), tested in Study 2. In separate regression analyses, it also predicted the endorsement of every single CT (apart from one) regardless of the fact that their content was very different. As the CMQ is a relatively content-free instrument, that captures susceptibility to doubt the official explanations of events and true intents of political actors, it might be viewed as a test of global worldview that may incorporate specific CTs. However, the rela tionship between specific CTs and conspiracy mentality was stronger in the first study which measured international conspiracies, than in the second which measured local ones. This could point out to some content-sensitivity.
Secondly, conspiracy mentality was also meaningfully related to the trustworthiness of political climate: higher conspiracy mentality was related to less trust in institutions and higher perception of corruption. Its relation with generalized political attitudes was also in line with the initial expectations: individuals who were more politically cyni cal and more authoritarian endorsed CM more. This is in line with previous studies demonstrating that authoritarian personalities tend to be more prone to CTs. This can be true for endorsing specific CTs (e.g. Abalakina-Paap, Stephan, Craig, & Gregory, 1999;Grzesiak-Feldman & Irzycka, 2009, but see also Swami, 2012), and for general conspiratorial attitudes (e.g. Bruder et al., 2013; for a comprehensive review see Grzesiak-Feldman, 2015). Political cynicism, insecurity, impaired trust in institutions and perception of corruption may be viewed as an indicator of anomie, and there is empirical base supporting the idea that CTs flourish in such circumstances, and are typical of distrustful and cynical individuals (Imhoff & Bruder, 2014;Swami et al., 2011). There is a large body of research that investigated political climate fostering conspiracy mentality linking it to the low levels of trust, both interpersonal and institutional (Goertzel, 1994;Putnam, 2000). In a study that explored predictive power of anomie for endorsement of 20 specific CTs, anomie proved to be predictive over and above other measures of individual differences (Abalakina-Paap et al., 1999). In a similar study on a French-speaking Swiss sample, Wagner-Egger and Bangerter (2007) proved that anomie, distrust and belief in the malevolent world are powerful predictors of beliefs in CTs. More specifically, some studies focused on relation between political ideologies and believe in CTs. In the study on a US sample, left-and right-wing extremists participate in the CTs equally, but the patterns are different (Uscinski & Parent, 2014). This was confirmed by another study on US and Netherlands samples (Van Prooijen, Krouwel, & Pollet, 2015). There is evidence that the level of anomie and confidence in authority is higher in post-communist European societies in comparison to the Western European ones due to the rapid sociopolitical transition (Zhao & Cao, 2010). Here, we explored the relations between these indicators, conspiratorial thinking and belief in specific CTs and have confirmed statistically significant correlation with distrust in institutions and corruption perception (Study 1 & Study 2) and political cynicism and right-wing authoritarianism (Study 2). In our study, we did not register the relation of conspiracy mentality with social dominance orientation, which measures the value that people attribute to hierarchical relations among social groups (Sidanius & Pratto, 1999), although some previous studies did find that higher social dominance orientation is related to more CTs endorsement (e.g. Grzesiak-Feldman, 2015). It might be due to the fact that these studies measured so-called "conspiracy stereotypes", beliefs about malevolent minority groups, which is more likely to be typical of people embracing the social hierarchy, whilst general conspiratorial attitudes regarding the global events can be weakly related or completely unrelated to social dominance orientation, as our results demonstrated.
Additionally, the predictive validity of the conspiracy mentality for specific CTs was consistent in the two studies. It predicted specific CTs over and above a set of measures that addressed the global climate of distrust and insecurity (Study 1), and even when measures of individual differences in political attitudes were included in the second step, it still added to the prediction (Study 2). These results clearly demonstrated that conspiracy mentality is a distinct psychological construct which cannot be equated with the other generalized political attitudes and that it can be a strong predictor of belief in CTs constructed around very diverse social events.
Lastly, we should point to several pluses of this study -the value of careful sampling methodology done in this study in a specific under-represented region. To our knowledge it is one of the very few studies that investigate CTs on the representative samples using face-to-face procedures, in a region with conflict history.

Limitations
We compared the predictive power of CM for both local and international CTs across studies. However, it would be even more informative to compare it in a single design. In addition, to have clearer insight about which instrument best captures the variance in global conspiratorial thinking, one would need to compare the CMQ to the other relevant measures. This leads us to another limitation of the study, that we have only one instrument for assessing CM. It would be valuable to use additional alternative instruments for assessing conspiracy mentality in our survey. Some of the statements from the CMQ that we used, lately have been a subject of academic criticism (being more factual statements, or likely to vary massively between different contexts). Results demonstrated divergent validity of one specific questionnaire in two high quality studies, besides acknowledging as a limitation the absence of alternative measurement for CMQ. Thus we believe this limitation does not constitute a substantive weakness of this report.
In the future studies, the Serbian version of the CMQ might be easily adapted for use in other countries in the region, the topic that warrants further studies. Lastly, future studies might include not only event-related, but also group-related CTs (so-called conspiracy stereotypes), and compare their relation to CM. This could be particularly relevant for the post-conflict Balkan region in which intergroup relations are fragile and full of mistrust.

Conclusions
There is a growing scholarly interest in the psychology of CTs (Butter & Knight, 2020), and researchers have developed a broad range of different scales to measure individual differences in conspiracist ideation (Swami et al., 2017). Valida tion of CMQ, as a main tool in these efforts, has been the subjected of this report: factorial and convergent validity, and internal consistency. The ongoing criticisms argues that CMQ only partially captures variance in the conspiratorial thinking, as some items can be seen as factual rather than conspiratorial, and its distribution is typically heavily skewed. However, our research asserted the CMQ to be structurally stable in two Serbian representative samples, as well as predictive of belief in concrete CTs, even the ones very specific to the context (the Balkans). These results indicate it can be a reliable tool for cross-cultural research. Our studies benefit from large, representative, culturally-homogeneous samples, with replication in two studies conducted in different period, confirming that scholars working on conspiracist ideation, may use CMQ for further investment.