Suspicious minds in times of crisis: determinants of Romanians’ beliefs in COVID-19 conspiracy theories

Employing data from a nationwide survey conducted in Romania in May, 2020, in this Research Note we examine the determinants of beliefs in several conspiracy theories (CTs) related to COVID-19. The main contributions of our study are as follows: First, our study adds to the growing body of work on CTs. Like most prior research (e.g. Golec de Zavala and Cichocka 2011; Miller et al. 2015; van Prooijen et al. 2015), we test in a single – country setting several hypotheses about the determinants of conspiracy thinking. However, our results depart in two important ways from those of prior research in regard to the effects of education and political preferences on endorsing CTs. As we will show, this could be due to the specific features of the CTs under scurtiny and of the local context. Second, while some of the studies on CTs cited throughout this paper have relied on availability samples of college students (e.g. Douglas and Sutton 2008), readers of popular magazines (e.g. van Prooijen 2017) or members of specific Web-based groups (e.g. Imhoff and Lamberty 2017), our research is based on a national representative sample of adult Internet users, which allows us to broaden our understanding of CTs. Third, we join current efforts of documenting CTs related to the COVID-19 crisis (see Cassese et al. 2020; Imhoff and Lamberty 2020; Miller 2020a, b; Pennycook et al. 2020; Uscinski et al. 2020). Embracing CTs has consequences for health choices (Lamberty and Imhoff 2018; Douglas et al. 2019) as conspiratorial beliefs affect vaccination intentions and COVID-19 social distancing (Biddlestone et al. 2020), may induce people to choose alternative therapies over biomedical medicine (Lamberty and Imhoff 2018), and may also build up potential for violent action (Jolley and Paterson 2020). Hence, examining conspiracy thinking offers important practical knowledge in the current pandemic context.

In the rest of this paper, we start with a short summary of previous studies on CTs, followed by a brief description of the Romanian context and our working hypotheses. Next, we present our data, measures, and methods. The results section summarizes the results of our multivariate analyses. Finally, we discuss some of the main findings of our exploratory study.

Conspiracy theories are ‘attempts to explain the ultimate causes of significant social and political events and circumstances with claims of secret plots by two or more powerful actors’ (Douglas et al. 2019: 4). These representations and narratives invoke covert, self-interested agencies which are detrimental to the wider society. The fascinating aspect of many popular CTs is their immunity to empirical evidence and rejection of seemingly better alternative explanations.

According to Huneman and Vorms’ (2018: 251) classification, some CTs refer to general or global plots (i.e. various secret societies controlling the world) while others target specific events (e.g. the assassination of JFK, the death of Princess Diana, The Romanian Revolution of 1989), they may refer to science or not (i.e. anti-vaxers, COVID-19), some but not all have an ideological bias, they may occur at institutional or informal/anti-institutional level. Some provide alternative explanations to official accounts, while others just deny them. Although CTs are ubiquitous, cross-cultural, and have great historical depth, the subject is still an emerging field in sociology. Notably, CTs have received more attention in psychology, social psychology, and political science.

Those who believe in one CT are likely to believe in others as well (Bruder et al. 2013); Miller (2020a) and Wood et al. (2012) found that CTs form a monological belief system in the US, even when CTs are mutually contradictory. Under conditions of uncertainty, CTs appeal more often to people who seek accurate and relevant explanations by seeking patterns in random events (van Prooijen et al. 2018) or seek a final and simplistic explanation for events. Narcissistic representations of belonging to a ‘special’ political or religious group (Golec de Zavala and Cichocka 2011) predict beliefs in CTs, especially when people think that their in-group is under threat, discriminated against, or feel a low level of control over the social and political world (Uscinski and Parent 2014).

While mental mechanisms are important for understanding CTs, they must be analyzed in a political and historical perspective. This is exactly what we do in this paper: we test in a single-country setting several hypotheses related to determinants of CTs related to the ongoing health crisis.

In Romania, the first COVID-19 case was confirmed on February 26, 2020. On March 16, 2020 Romania declared a state of emergency. People were allowed to leave their homes only to buy groceries, to go to work or for justified emergencies. Schools, universities, shopping malls, restaurants, some public institutions, and borders were closed, and a night-time curfew was imposed nationwide. The state of emergency ended on May 14, 2020, followed by a state of alert, which lifted many restrictions.

In dealing with the pandemic, Romania has faced numerous problems caused by lack of know-how, medical equipment and personnel, compounded by politicking and institutional stuttering. For instance, on February 27, 2020, Romania's President Klaus Iohannis stated that there are no real reasons to panic as the new virus triggers only mild, cold-like symptoms. Subsequently, he and his governing party (the National Liberal Party – PNL) pushed for observing strictly physical distancing and other rules. The main opposition party (the Social Democrats – PSD) did not initially challenge the reality of the pandemic but criticized the government's alleged chaotic policies to combat it. As time went on, individual members of PSD and other political parties started to publicly express doubts about the (severity of the) pandemic; some politicians refused to wear masks and endorsed various COVID-19 CTs. In July and August, 2020, two rallies were held in downtown Bucharest against wearing masks and in favor of a return to ‘normalcy.’ Both rallies were attended by several hundred people; one rally was organized by a new, fringe political movement (the Romanian Patriots’ Movement) founded by a local businessman who publicly endorsed COVID-19 CTs. Interestingly, participants at this rally wore T-shirts with the logo of QAnon – a CT en vogue in the US, which seems to be embraced by Donald Trump's supporters.

As of September 8, 2020, Romania – a country with a population of 19,220,188 individuals – had 97,033 confirmed cases, 3,967 deaths, and 40,838 recoveries (Grupul de Comunicare Strategică 2020).

Our data comes from a study that focused mainly on observing spatial distancing and physical hygiene recommendations, and on support for public policies aimed at containing the pandemic. The questionnaire included four items about COVID-19 CTs and only several variables previously employed in studies on CTs. Since CTs were a tangential topic, our attempt to test extensively their determinants has been limited by the measures included in this study's questionnaire. Yet, our sample's size, its selection method, target universe, and timely data collection compensate for lacking other and/or more detailed factors related to CTs. Specifically, we look at the effects of age, gender, residence (urban/rural), education, being in the labor force, social position, political preferences, social belonging, open- mindedness, collective narcissism, and analytical thinking on COVID-19 CTs.

We hypothesize that age is a positive predictor of beliefs in COVID-19 CTs since older Romanians tend to distrust official accounts due to their experiences under Ceaușescu's dictatorship. For gender, prior research has yielded mixed conclusions: while Miller et al. (2015) found no statistically significant effects of gender on endorsing left or right-wing CTs (for a similar non-significant effect on conspiracy thinking, see Uscinski and Parent 2014); Cassese et al. (2020) showed that men are more likely than women to endorse COVID-19 CTs in the US due to men's higher scores on learned helplessness and their propensity for conspiracy thinking. Lacking similar additional measures, we refrain from hypothesizing any specific effect of gender on COVID-19 CTs.

We further expect an inverse relationship between education level and beliefs in COVID-19 CTs. This is so because ‘education education may provide people with a set of cognitive and affective attributes that enable them to resist CTs’ (Douglas et al. 2019: 10). Given differences in the levels of socio-economic development (Sandu 1996), we also hypothesize that, compared to rural residents, urbanites are less likely to endorse COVID-19 CTs. Individuals who are in the labor force are expected to be less likely to endorse CTs about this pandemic since labor force participation exposes people to different points of view and sources of information, which can decrease an individual's tendency to believe in COVID-19 CTs. In line with previous studies (Douglas et al. 2019; Uscinski and Parent 2014), we anticipate that an individual's social position will be inversely related to beliefs in COVID-19 CTs as a sense of ‘losing out’ may induce suspicion in powerful entities rigging the social game. Also, based on prior research (Graeupner and Coman 2017), we hypothesize that higher levels of social integration predict decreased beliefs in COVID-19 CTs as social exclusion leads people to reject mainstream narratives and embrace fringe explanations.

Prior studies suggest that political polarization matters for conspiracy thinking (Moore et al. 2014; van Prooijen et al. 2015). Put another way, individuals holding extreme-left and those holding extreme-right political views are more likely to believe in CTs, albeit in different CTs. Yet, Miller (2020b) reports that, in the US, compared to Democrats, Republicans and Independents have higher chances to believe COVID-19 CTs. Similarly, Uscinski et al. (2020) found that Republican (partisanship) and holding conservative views (ideology) predict endorsing COVID-19 CTs. Against this backdrop, we also hypothesize that respondents with far-right political leanings are more likely to endorse COVID-19 CTs.

Furthermore, Swami et al. (2014) suggested that open-minded individuals have lower chances to endorse CTs, because open-mindedness implies flexibile and critical thinking, questioning one's own ideas, and rejecting simplistic views, which characterzie CTs (Swami et al. 2014: 575). We provide a test for this relationship, which has been infrequently addressed by prior research on CTs. Informed by other studies (Golec de Zavala and Cichocka 2011; Golec de Zavala et al. 2009), we expect that collective narcissism, a form of national self-victimization, will have positive effects on endorsing beliefs about COVID-19 CTs as these evoke powerful external enemies of one's group. Finally, in line with previous studies (Swami et al. 2014), we hypothesize that individuals with an analytical-rational cognitive style are less likely to embrace CTs as compared to people with a more intuitive thinking.

The data comes from an international project conducted in 67 countries and territories in April-May, 2020. The questionnaire and sample requirements were designed by a team led by Jay J. Van Bavel (NYU). Researchers who responded to this call had to secure the funds to conduct this survey locally, to translate, and pre-test the questionnaire. The data for all of the countries included in this project are not available yet as the international coordinating team is still working on merging individual country files. The nationwide sample we employ includes 500 respondents, with ages between 18 and 70 years. The data was collected during the lockdown, on-line, by Ipsos Interactive Services SRL Romania (7–11 May, 2020). Respondents were randomly selected from a nationwide panel of 2,000 Internet users aged 18–70 years.¹ In the following analyses, the achieved sample was weighted by education, gender, and residential milieu (urban and rural areas).

We ran a multiple linear regression model using the OLS method with the ‘Index of beliefs in COVID-19 conspiracy theories’ (ranging from 0 to 10) as our dependent variable. The latter is based on items that asked respondents to what extent they agree or disagree with statements that tap four CTs related to the current pandemic (see Table 1). ‘Index of beliefs in COVID-19 CTs’ is calculated as the arithmetic mean of answers to the four items under scrutiny. (See Table A1 in the online Appendix for correlations among these items.)

We employ the following independent variables: Gender (1 = female; 0 = males); age (in years), education level (three dummy variables); urban resident (1 = yes); occupational status (1 = in the labor force; 0 = otherwise); self-assessed social position (from 0 = bottom to 10 = top); political orientation (three dummy variables, based on self- assessed political leanings, measured on a scale from 0 = left to 10 = right); open-mindedness score (0 = minimum to 10 = maximum), which measures the inclination to consider different points of view and the capacity to change one's mind and admit mistakes (Swami et al. 2014); social belonging score (0 = minimum to 10 = maximum) that measures subjective feelings of social inclusion and acceptance; collective narcissism score (0 = minimum to 10 = maximum), which taps the perception that one's nation/group deserves but does not receive special recognition (Golec de Zavala et al. 2009). ‘Analytical thinking score’ (0 = minimum to 6 = maximum) aggregates a respondent's correct answers to the ‘Cognitive Reflection Test - CRT’ (Frederick 2005) which taps ‘the tendency to override a prepotent response alternative that is incorrect and to engage in further reflection that leads to the correct response’ (Toplak et al. 2011: 1275). Table 1 presents the CRT open-ended questions from this study; correct responses to each problem were coded 1, and 0 otherwise; larger scores indicate higher analytical thinking.

Table 2 presents the results of two nested models of the multiple regression of the ‘Index of beliefs in COVID-19 CTs’ on the independent variables discussed previously. In Model 1 we included only socio-demographic variables and in Model 2 we added several socio-psychological variables. F-test statistics values show that by adding these variables, the overall fit of the model improves significantly; the full model accounts for 34.3% of the variability in the dependent variable. Due to space limitations, we focus our discussion on the full model.

We already mentioned that prior research has yielded mixed conclusions about the relationship between gender and CTs. In our case, men and women do not differ in terms of endorsing COVID-19 CTs. Also, self-assessed social position, urban resident, and being in the labor force have no significant effects on endorsing CTs related to the current pandemic – three findings that infirm our hypotheses.

Age has positive significant effects on beliefs about COVID-19 CTs. Possible explanations might run as follows: First, older and younger individuals differ in respect to how they use and relate to the Internet (Ștefaniță and Ivan 2018). Older generations might be rather poorly equipped to employ alternative, competing sources to fact check the information vehiculated online. Unfortunately, this project lacked measures of media literacy, digital skills, and Internet use that would have allowed us to explore in-depth such issues.

Second, the positive effects of age on endorsing these CTs might be country specific. Ceaușescu's extremely authoritarian regime distinguished itself through the chasm between the official discourse (propaganda) and reality. In the mid to late-1980s, most Romanians distrusted the State and relied on ‘a widespread, fragmented system of rumors […] to make sense of their day-to-day lives.’ (Ely and Stoica 2004: 100) Shibutani (1966) defined rumors – incipient forms of conspiracy theories (Byford 2011: 139) – as ‘improvised news,’ which tend to proliferate when information from official sources is scarce and/or is distrusted. The pre-1990 experiences of older Romanians could make them particularly distrustful of information coming from official sources, hence more prone to endorse alternative, conspiracy views on the pandemic.

Contrary to previous studies (Douglas et al. 2015; van Prooijen 2017) and to our hypotheses regarding the inverse relationship between education and beliefs in CTs, we found that being a high school or a college graduate predicts increased beliefs in COVID-19 CTs. As we discuss in the next section, this finding might be related to the features of the national context.

As regards political leanings, holding far-left political views has no significant effects on embracing COVID-19 CTs. Far-wing political leanings, however, predict lower scores of the index of beliefs in CTs about the current crisis. As we discuss in the next section, this counterintuitive finding might be explained by the peculiar meanings of ‘left’ and ‘right’ in Central and Eastern Europe.

According to past research (Graeupner and Coman 2017), people who are less socially integrated may develop beliefs contrary to mainstream perspectives in their search for meaning. We also found that social exclusion (as measured by social belonging index) predicts increased beliefs in COVID-19 CTs – a fact that confirms our initial hypothesis.

As we hypothesized based on an idea advanced by Swami et al. (2014), open- mindedness predicts decreased beliefs in COVID-19 CTs. This suggests that epistemic humility makes people more trusting of expert knowledge which discredits CTs. Conversely, unjustified trust in one's opinions and intuitions creates the conditions for believing in fringe theories.

Collective narcissism signals ‘an excessive self-love or inflated, grandiose view of oneself that requires continual external validation’ (Golec de Zavala et al. 2009: 1074).

Collective narcissism is also associated with the tendency of viewing one's nation as a victim of conspiracies of various ‘others’ (Golec de Zavala and Cichocka 2011). Consistent with past research and our hypothesis, collective narcissism predicts increased beliefs in COVID- 19 CTs, which evoke threatening powerful entities outside one's nation.

Swami et al. (2014) showed that individuals more inclined towards an analytical, rational thinking-style are less likely to embrace CTs. Our results support such prior findings and hypotheses: higher analytical thinking scores are associated with decreased beliefs in COVID-19 CTs.

This Research Note focused on Covid-19 related CTs. The topic is not only timely but extremely important because, as prior studies have shown, adhering to CTs can have dire consequences for health-related choices (Biddlestone et al. 2020; Lamberty and Imhoff 2018; Jolley and Paterson 2020; Imhoff and Lamberty 2020). During the current global crisis, individuals who deem COVID-19 an invention can act irresponsibly, endangering themselves and other people. We tested in a single-country setting several hypotheses about the determinants of beliefs in CTs. Despite the fact that CTs and rumors were widespread under Ceaușescu, this is the first systematic study conducted in Romania not only on COVID-19 related CTs but on CTs in general.

Our findings depart from those of past studies on CTs in two important ways: First, prior studies (Uscinski and Parent 2014; van Prooijen 2017) reported a negative association between a respondent's education level and beliefs in (other) CTs. In our final model, education is a positive predictor of beliefs in COVID-19 CTs. This might be due to the features of the event under scrutiny (i.e. science-related), which might make high school and university graduates overconfident in their ability to evaluate the reasonableness of official accounts about COVID-19 (see Robson 2019 on the illusion of competence).

This result might also be country specific. At the onset of the COVID-19 crisis, local media featured extensively the opinions of various pundits, politicians, and ‘experts’ who either minimized the pandemic or deemed it a hoax. Even the President of the Romanian Academy had a controversial position on this issue. He was seconded by the Vice-President of the same Romanian Academy who publicly stated that ‘the SARS-CoV-2′s features are 95% similar to those of a bioweapon.’ (Gândul 2020).

The surprising effects of education on endorsing COVID-19 CTs might also be related to better-educated Romanians’ distrust of government. Romanian data from the World Values Survey (WVS) show that as education increases, the level of trust in government decreases. This relationship holds true irrespective of the political composition of the government at the time when different waves of WVS were conducted in Romania.² In this case, education might also capture lack of trust in government and its official stories. This hypothesis has important policy implications. A Gallup study conducted in 140 countries has found that ‘the world places more trust in health advice from medical professionals than from governments’ (Stevens and Ron-Levey 2020). Accordingly, this study advocates the inclusion of experts in government communication related to health crises. Additional research is needed to test if the negative relationship between education and CTs is period- dependent (i.e. an effect of the timing of the survey) and/or related to the specificity of the event under scrutiny (a science and health related issue [vs. other CTs]).

As regards political orientations, as mentioned previously, past research showed that individuals who hold extreme political views are more likely to believe in CTs (Moore et al. 2014; van Prooijen et al. 2015). Two recent studies conducted in the US (Miller 2020b; Uscinski et al. 2020) found that voting Republican and embracing conservative views positively predict beliefs in COVID-19 CTs. In contrast to these studies and our initial hypothesis, we found that far-right political views predict decreased beliefs in COVID-19 CTs. Holding far-left political views has no significant effects on endorsing such CTs but we note that the sign of the coefficient is positive.

These results, we claim, might be context specific. According to Aspelund et al. (2013), ‘left’ and ‘right’ have specific meanings in Eastern Europe. In Eastern Europe, although voters of left-wing parties favor redistributive economic policies like in the West, such voters tend to be less educated, and more attached to notions such as tradition, nation, order, and authority. These notions, however, are widely embraced in Western Europe by supporters of right-wing political parties. Left-leaning Romanians also tend to be more religious and less progressive, and support parties that are linked with the now defunct Communist Party. In this sense, these people are ‘conservative’ vis a vis the recent past (Sandu 1996).

In Eastern Europe, right-leaning individuals tend to be pro-market and pro-private sector, like many other Westerners with similar political orientations (Sandu 1996). At the same time, right-leaning Romanians (among others) are pro-EU, pro-globalization, less religious, less nationalistic, and – in Baker and Inglehart's (2000) parlance – more ‘modern.’

Such differences between Western and Eastern Europe might explain why holding far-right political views predicts decreased beliefs in COVID-19 CTs in this case. The relationship between political preferences and CTs should be further investigated in the Romanian context by taking into account other variables related to trust in government, partisanship, ideology, and social class.

We thank the following institutions and people for their support and/or feedback: Ipsos Interactive Services Romania, Nicoleta Constantinescu and George Gervec (from Ipsos), Jay van Buel, Oana Lup, Laura-Maria Ilie, John F. Ely, and David Mathers-Winn. We are also greatful to Dr. Yuliya Kosyakova, Editor of this issue, and to three anonymous reviewers for their valuable suggestions.

No potential conflict of interest was reported by the author(s).

Radu Umbreș (Ph.D. in Anthropology, University College London) is an Assistant Professor of Anthropology at the National University for Political Studies and Public Administration (Romania). He focuses on cognitive anthropology, with a special emphasis on trust and cooperation, folk epistemology, cognitive biases, and cultural evolution.

Cătălin Augustin Stoica (Ph.D. in Sociology, Stanford Univeristy) is an Associate Professor of Sociology at the National University for Political Studies and Public Administration (Romania). His research interests center on collective memory, social change, post-communist transtion, conspiracy theories, survey methods, economic sociology, and social stratification.