A “Conspiracy Theory” Conspiracy? A Mixed Methods Investigation of Laypeople’s Rejection (and Acceptance) of a Controversial Label

Nowadays, it seems that conspiracy theories (CTs) are everywhere, as the term has made its way into public discourse and has become a popular topic in the mainstream media. This label is used to qualify a variety of beliefs, some of which are endorsed only by few people (e.g., ‘the earth is flat and that fact is being concealed by the evil elites’), others by large portions of the population (e.g., according to a recent poll, 33% of a representative sample of the French population ‘somewhat’ or ‘fully’ agrees with the statement ‘The health ministry collaborates with pharmaceutical companies to hide the truth regarding the dangers of vaccines’, Ifop, 2020, our translation). We could extrapolate and reasonably speculate that nowadays every significant social event generates its share of CTs. Usually defined as explanations of events based on the concealed role of an evil group (e.g., Keeley, 1999), CTs have also become a prominent topic of research in social psychology, as well as in social sciences in general (for reviews, see Douglas et al., 2019; van Prooijen & Douglas, 2018). However, while social psychology research has investigated the antecedents and consequences of endorsing both specific and generic conspiracist beliefs, it has scarcely examined how the label ‘conspiracy theory’ itself is perceived (see however Lantian et al., 2018; Wood, 2015; Wood & Douglas, 2013). Similarly, the rejection of the label ‘conspiracy theory’ by advocates of beliefs labeled as such has been the focus of little interest (Franks et al., 2017; Harambam & Aupers, 2016). Whether the label ‘conspiracy theory’ is accepted or rejected in the general population, and by whom, is a crucial question, however, as social psychology – as well as other social sciences – is building an increasingly large body of knowledge based on a generalist view of CTs (i.e., the assumption that CTs can be studied as a general, relatively homogeneous phenomenon). Hence, the perception of the expression ‘conspiracy theory’ by the general public constrains the credibility of – publicly funded – academic research on this concept in this population. Nera, K., et al. (2020). A “Conspiracy Theory” Conspiracy? A Mixed Methods Investigation of Laypeople’s Rejection (and Acceptance) of a Controversial Label. International Review of Social Psychology, 33(1): 13, 1–18. DOI: https://doi.org/10.5334/irsp.401

Usually defined as explanations of events based on the concealed role of an evil group (e.g., Keeley, 1999), CTs have also become a prominent topic of research in social psychology, as well as in social sciences in general (for reviews, see Douglas et al., 2019;van Prooijen & Douglas, 2018). However, while social psychology research has investigated the antecedents and consequences of endorsing both specific and generic conspiracist beliefs, it has scarcely examined how the label ' conspiracy theory' itself is perceived (see however Lantian et al., 2018;Wood, 2015;Wood & Douglas, 2013). Similarly, the rejection of the label ' conspiracy theory' by advocates of beliefs labeled as such has been the focus of little interest (Franks et al., 2017;Harambam & Aupers, 2016). Whether the label ' conspiracy theory' is accepted or rejected in the general population, and by whom, is a crucial question, however, as social psychology -as well as other social sciences -is building an increasingly large body of knowledge based on a generalist view of CTs (i.e., the assumption that CTs can be studied as a general, relatively homogeneous phenomenon). Hence, the perception of the expression ' conspiracy theory' by the general public constrains the credibility of -publicly funded -academic research on this concept in this population. Nera, K., et al. (2020). A "Conspiracy Theory" Conspiracy? A Mixed Methods Investigation of Laypeople's Rejection (and Acceptance) of a Controversial Label. International Review of Social Psychology, 33(1): 13, 1-18. DOI: https://doi.org/10. 5334/irsp.401 Grounding our rationale in social representations theory (Moscovici, 1961(Moscovici, , 1988, our main contention is that the label ' conspiracy theory' is associated with polemical social representations (i.e., mutually exclusive representations grounded in antagonistic relations between groups), anchored in individuals' pre-existing beliefs. We operationalize this theoretical contention by hypothesizing that the more one holds generic conspiracist beliefs, the more one will tend to endorse two forms of rejection of the label named and defined as follows: 1) The belief in what we called the 'meta-conspiracy theory', that is, the idea that the label ' conspiracy theory' was created by the elite to discredit their opponents; 2) The endorsement of a particularist view of CTs, that is, the idea that CTs cannot be considered as a homogeneous phenomenon, and that it is therefore an invalid (or at least, deceptive) label.
We first test our hypotheses using correlational methods. Second, we further our understanding of the results using lexicometric analyses of open-ended questions. Specifically, we explore participants' hypotheses regarding how the label appeared in the first place, and the arguments they mobilize to justify, or disqualify, the use of the label ' conspiracy theory'.

Polemical Social Representations of the Label 'Conspiracy Theory'
Social representations theory (Moscovici, 1961(Moscovici, , 1988 offers an adequate framework for our research question, as it bears on collectively constructed lay knowledge that enables individuals to deal with new and/or threatening situations. The idea that CTs might assume such a social function is not new (e.g., Douglas, Sutton, & Cichocka, 2017;Franks, Bangerter, & Bauer, 2013;Moscovici, 1987;van Prooijen & Douglas, 2017;van Prooijen & van Lange, 2014). Moreover, social representations are typically derived from expert knowledge that is transformed through the processes of anchoring (i.e., interpretation through the lens of pre-existing beliefs) and objectification (i.e., making the abstract concrete). Franks and colleagues (2013) proposed that these processes might be at play in the formation of CTs. Indeed, on the one hand, CTs tend to interpret elements of reality (e.g., the existence of the Bilderberg Group) in such a way that these elements are perceived as evidence of a conspiracy. On the other hand, CTs simplify complex realities (e.g., expert discourses, historical events) into more concrete, easily understandable concepts (e.g., a conspiracy of malevolent individuals). In a 2006 lecture, Moscovici (2020) further suggested that CTs can be apprehended as polemical representations, that is, mutually exclusive representations of the same social object (e.g., an historical event), grounded in antagonistic intergroup relations (Moscovici, 1988). In this research, we take a step back and hypothesize that the label ' conspiracy theory' itself might be associated with such polemical social representations. Specifically, we expect individuals rejecting conspiracist beliefs to accept the label as relevant and being an accurate description of a social phenomenon, and individuals endorsing such beliefs to reject the label as unfounded and illegitimate.

Two Forms of Rejection of the Label: (Meta) Conspiratorial and Conceptual
Even though the rejection of the label ' conspiracy theory' has been scarcely investigated, the existing literature suggests the existence of at least two forms of rejection, namely, belief in what we called 'meta-conspiracy theory' and a particularist view of CTs. While distinct, both forms of rejection seem to be related to one's own conspiracy beliefs, which is congruent with the proposed social representations framework.
Meta-conspiracy theory: A derogatory label created by the elites A recurring criticism of the label ' conspiracy theory' is that it is frequently weaponized by powerholders (e.g., Buenting & Taylor, 2010;Dentith & Keeley, 2018;Husting & Orr, 2007). Indeed, labeling an utterance as a ' conspiracy theory' amounts to disqualifying this statement as a legitimate point, and it has happened that allegations of misconduct were discarded as ' conspiracy theories' by the officials called into question (e.g., Benen, 2014;Buenting & Taylor, 2010;Helm & Boffey, 2011). As a consequence, some authors have warned against what they called a ' contemporary anti-conspiracy theory panic' (Basham & Dentith, 2016; for critical accounts of this view, see Dieguez et al., 2016;Wagner-Egger et al., 2019).
This debate is not limited to the academic world, as a seemingly common feature to every belief labelled as ' conspiracy theory' is that its advocates reject this qualification, which is perceived as stigmatizing (Franks et al., 2017;Giry, 2017;Harambam & Aupers, 2016). Ironically (but unsurprisingly), such rejection can take the form of a CT. For example, a relatively common CT is that the label ' conspiracy theory' was coined and disseminated by the CIA to discredit those who questioned the conclusions of the Warren commission after the JFK assassination (e.g., Corey's Digs, 2018;Tracy, 2013;Unz, 2019;Zero Hedge, 2015). Besides, conspiratorial rhetoric often identifies academics who criticize CTs as accomplices of the conspirators (e.g., Franks et al., 2017;Wood & Douglas, 2015). We propose to call such beliefs 'meta-conspiracy theory', as it is the belief that the label ' conspiracy theory' is itself an instrument created by the (evil) elites to silence dissent.
Despite the considerable amount of empirical support this view has gathered over the years, the very idea that CTs can be apprehended as a relatively unified phenomenon has been criticized, often by authors who also expressed concerns about the weaponization of the label (e.g., Basham & Dentith, 2016;Buenting & Taylor, 2010;Dentith & Keeley, 2018;Hagen, 2017). According to Buenting and Taylor (2010), there seems to be a schism within social sciences between 'generalist' and 'particularist' views of CTs. Proponents of the generalist view (e.g., Sunstein & Vermeule, 2009;Uscinski & Parent, 2014) consider that CTs can be studied as a relatively homogeneous social phenomenon. Most social psychologists working on the topic subscribe to such a view, arguing that even though CTs can widely differ in their content and overall plausibility, they have many common psychological underpinnings (Douglas, Sutton, & Cichocka, 2017;Klein & Nera, 2020;van Prooijen, 2018). Besides, as psychologists, they tend to be interested in how individuals perceive the world, rather than in how the world actually is. Thus, distinguishing between 'real conspiracies' and 'unwarranted conspiracy theories' (Keeley, 1999) is considered as a secondary matter (e.g., van Prooijen, 2018).
By contrast, proponents of the 'particularist' view of CTs argue that since conspiracies happen in the real world, there is no a priori criterion to disqualify accusations of conspiracy (e.g., Basham & Dentith, 2016;Buenting & Taylor, 2010;Dentith & Keeley, 2018). As a result, each claim of conspiracy should be judged on its own merits. The main criticism of the 'particularists' addressed to the generalist view is that it pathologizes suspicion towards powerholders and epistemic institutions (i.e., official experts), and implies an underestimation of the occurrence of real conspiracies in politics (Bale, 2007;Basham & Dentith, 2016;Dentith & Keeley, 2018). Hence, even though the relation might not be as straightforward as for belief in meta-conspiracy theory, perceiving the label ' conspiracy theory' to be a(n) (in)valid category also seems to be related to one's own beliefs about the existence of conspiracies.

The Interest of Mixed Methods to Investigate Conspiracy Theories
Up to now, social psychologists have mostly relied on classic quantitative methods (i.e., confirmatory null hypothesis significance tests) to investigate CTs (Klein & Nera, 2020). However, in addition to the two forms of rejection defined above, we seek to investigate individuals' hypotheses regarding the origin of the label ' conspiracy theory', as well as the arguments participants spontaneously mobilize to justify their acceptance -or rejection -of the label. In this regard, allowing participants to express themselves without being constrained by rating scales and items selected by researchers seemed appropriate. Thus, the use of a more qualitative method, namely, analyses of participants' answers to open ended questions, appears more suited than classic quantitative methods to investigate these aspects of the present research.
The use of more qualitative approaches is justified for theoretical and practical reasons. On the theoretical ground, qualitative methods can be used to refine the interpretation of quantitative data (e.g., Duffy, 1987;Rossman & Wilson, 1985;Spillman, 2014), which is what we do in this research. Conversely, they can be used as an exploratory tool to design quantitative hypotheses. The few existing researches mobilizing qualitative methods to investigate CTs have brought light to important and neglected aspects of CTs (e.g., Franks et al., 2017;Harambam & Aupers, 2016;Lukić, Žeželj, & Stanković, 2019). For example, based on interviews of people attending a ' conspiracy' convention, Franks and colleagues (2017) found that their participants tended to identify with a positive ingroup (i.e., the 'Truth Seekers' community), and to devalue those who do not believe (e.g., refer to them as 'sheep'). Such identification with a positive ingroup is surely an important aspect of CTs to investigate, as most research conducted up to now has focused on the adverse correlates of those beliefs.
Another example is the idea that people drawn to CTs have no problem endorsing contradictory beliefs (e.g., Brotherton et al., 2013;Irwin, Dagnall, & Drinkwater, 2015;Wood, Douglas, & Sutton, 2012). In a recent contribution, Lukić and colleagues (2019) have interviewed participants who endorsed mutually incompatible CTs in a survey. This seemingly irrational stance was explained by the fact that participants often considered the mutually contradictory CTs as two potential scenarios of the same event. In other situations, participants focused on the common features of the contradictory items (e.g., 'the HIV was created in a laboratory', p. 98) and did not pay attention to the incompatible elements (e.g., 'HIV is deliberately made in the USA laboratories […]' versus 'HIV is deliberately made in the laboratories of European pharmaceutical corporations […]'). Hence, it turned out that they were not acting as irrationally as it might have seemed first. These two examples show that far from being incompatible, qualitative and quantitative methods can be complementary.
On the practical side, an approach focusing on both understanding (i.e., the reconstruction of individuals' subjective experience of the investigated phenomenon, Weber, 2003) and explanation (i.e., investigation of the causes of a phenomenon, as practiced in natural sciences) might be particularly appropriate to properly tackle the question of the acceptance of research on CTs by the general public. By combining classic correlational analyses (which enable to identify generic trends in the population) and more qualitative analyses of participants' representations surrounding the label 'conspiracy theory', we may be able to better identify obstacles to such acceptance. Doing so may enable to adapt social psychologists' communication about CTs. This might also highlight discrepancies between lay perceptions of the label 'conspiracy theory' and how researchers define this notion. Last, it might potentially raise legitimate questions regarding how the term is defined within research, and ultimately lead to improvements in the definitions we use.

Overview of the Research
Across two studies, using correlational analyses, principal components and confirmatory factor analyses, we examined how the belief in the 'meta-conspiracy theory' relates to generic conspiracist beliefs. In Study 1, we sought to recruit a sample of participants highly distrustful of political and media authorities. Indeed, while research on the topic primarily seeks to study belief in CTs, populations actually endorsing such beliefs are hard to recruit for research (Franks et al., 2017;Wood & Douglas, 2015). As a result, convenience samples in studies usually yield low levels of conspiracist beliefs, which might lead to erroneous conclusions (e.g., drawing conclusions about the impact of endorsing conspiracy beliefs, while the results might only hold for people giving mild credit to these beliefs, Klein & Nera, 2020). In the second study, which aimed at recruiting a large and ideologically diversified sample, we also examined how generic conspiracist beliefs were correlated with the endorsement of a particularist view of CTs. Participants were also asked 1) how they thought that the label ' conspiracy theory' appeared in the first place, and 2) their opinion about the debate over the particularist versus generalist views of CTs. We expected that the first question would help identify the variety of representations regarding who created the label ' conspiracy theory', and for what purpose. The second question sought to highlight the arguments participants spontaneously mobilized in favor and/or against the generalist view of CTs.

Study 1
In Study 1, we hypothesized that generic conspiracist beliefs would be positively correlated to belief in metaconspiracy theory. The data, questionnaires, and all supplementary materials used in both studies are available on the following OSF page: https://osf.io/jq4pk/.

Participants
Two hundred and fifty-one Belgian residents completed the questionnaire, out of which 221 remained after removing participants who did not answer correctly to the two attention checks (M age = 47.5, SD = 12.5, min = 18, max = 72). One hundred and seventeen participants identified as women (54%), six as non-binary (2.7%), and the remaining as men. The mean political orientation was 4.33 (SD = 1.79) on a scale ranging from 1 (far left) to 9 (far right).

Procedure and Materials
Participants were recruited using a sponsored Facebook ad that stated that the study was interested in people who distrust authorities and the mainstream media. They were informed that by continuing, they gave their consent to have their data used for research and uploaded in an anonymized format on an open-access platform. Participants first answered to questions pertaining to per-ceived discrimination of people who distrust authorities and the mainstream media (these variables are treated in a separate paper in preparation). They then reported their endorsement of generic conspiracist beliefs and belief in meta-conspiracy theory on a Likert scale ranging from 1 (certainly not true) to 5 (certainly true) on the following scales.
Eleven items Generic Conspiracist Beliefs Scale (GCBS, Brotherton et al., 2013, α = 0.93). We used this scale to measure participants' endorsement of conspiracy beliefs, because it is currently one of the most widely used measure of this construct (Drinkwater et al., 2020) and that a validated French translation is available . We decided not to include in the questionnaire the alien contacts subscale (3 items), nor the item pertaining to mind control technologies, for three reasons. First, these items were found to stick out as a distinct factor in a research examining the validity of conspiracy beliefs measures (Swami et al., 2017). Even though this last reference discourages the use of the GCBS because of its alleged inconsistent factor structure (however, the scale was recently rehabilitated by Drinkwater et al., 2020), we still used it because there was no entirely satisfying alternative available with a validated French translation. Second, removal of these items was motivated by the fact that we sought to recruit participants who would strongly endorse conspiracist beliefs, rather than people scoring low on the conspiracist beliefs scale. The first author had past experiences of negative feedback regarding the ' alien contacts' items (e.g., some participants accused the author of equating ' questioning the government' with 'believing in aliens'). Given our limited recruitment resources, these items were removed to reduce potential attrition among 'highly conspiracist' participants. Last, these items seem rather specific to the United States context and appeared of limited relevance in a study conducted on a Belgian sample. Following the recommendations of past researchers who expressed concern about the factor structure of the GCBS (Atari, Afhami, & Swami, 2019;Swami et al., 2017), we ran an exploratory factor analysis on the remaining items. The single dimension structure was not corroborated. However, since it is common practice and that the scale yielded an excellent internal reliability, it was treated as a single construct.
Belief in meta-conspiracy theory (α = 0.83). Four items were designed to capture the belief that the label ' conspiracy theory' is but a rhetorical weapon of powerholders to silence dissent (e.g., 'the expression "conspiracy theory" was created in order to discredit views that question those in power'; 'those who criticize "conspiracy theories" often do so because they want to hide things'). Confirmatory factor analysis returned an excellent fit of a single factor structure, Chisq/df = 2.29, CFI = 0.99, TLI = 0.97, RMSEA = 0.08, confirming that these items capture a single construct.

Results and Discussion
Correlations and descriptives are displayed in Table 1. Congruent with our hypothesis, we observed a strong correlation between belief in meta-conspiracy theory and generic conspiracist beliefs, r = 0.77, p < 0.001. Furthermore, Principal Components Analysis on 11 items of the Generic Conspiracist Beliefs Scale combined with the meta-conspiracy theory items revealed a single underlying dimension (see Table 2). The 15 items combined yielded a strong internal consistency (α = 0.94).
These results show that individuals position themselves differently towards the label ' conspiracy theory' depending on their own conspiracist beliefs, which is congruent with the hypothesis that the label is associated with polemical representations. According to these results, the belief that the label ' conspiracy theory' is a weapon created by the elite is not only associated with generic conspiracist beliefs, but it appears that the measures of the two constructs are empirically barely distinguishable.
However, this first study relied exclusively on rating scales. As a result, we were not able to identify different forms of meta-conspiracy theories. The items expressed an extreme instance of meta-conspiracy theory and we can expect that participants might have more nuanced opinions regarding the weaponization of the label by powerholders. Second, since the sample size was limited, we did not carry out confirmatory factor analysis to compare the fit of a single factor structure to the fit of a two factors structure. Indeed, since this requires randomly splitting the sample in half to conduct the PCA and the CFA on different participants, the sample size would have been too small to return reliable results (Kline, 2016). Last, Study 1 only investigated the relation between generic conspiracist beliefs and belief in metaconspiracy theory. As we have discussed in the introduction, there is also a distinct, conceptual discussion about whether the label is legitimate. All these limitations are addressed in Study 2.

Study 2
In Study 2, we sought to replicate the relation between conspiracist beliefs and belief in meta-conspiracy theory and examined how conspiracist beliefs relate to a particularist view of CTs. We tested two hypotheses: H 1 : Belief in meta-conspiracy theory will be positively correlated with generic conspiracist beliefs, to the point that the items will fit in a single factor structure.  Table 2: Principal components analysis with Oblimin rotation for the GCBS and the meta-conspiracy theory items.

Items (English translation) Component 1
The government is involved in the murder of innocent citizens and/or well-known public figures, and keeps this a secret 0.75 The power held by heads of state is second to that of small unknown groups who really control world politics 0.72 The spread of certain viruses and/or diseases is the result of the deliberate, concealed efforts of some organization 0.78 Groups of scientists manipulate, fabricate, or suppress evidence in order to deceive the public 0.70 The government permits or perpetrates acts of terrorism on its own soil, disguising its involvement 0.79 H 2 : Participants scoring higher on the GCBS will tend to endorse a particularist view of CTs, that is, to consider that the label ' conspiracy theory' is illegitimate.
We moreover investigated the content of participants' responses to open-ended questions to investigate 1) participants' hypotheses regarding the origins of the label ' conspiracy theory' and 2) their opinions regarding the validity of the label ' conspiracy theory'. Doing so, we sought to examine if different levels of conspiracist beliefs were associated with different types of vocabulary, different views regarding the origin of the label, and distinct arguments to justify one's perception of the label.

Participants
One thousand one hundred and sixty-nine participants completed our online questionnaire, out of which 1076 remained (M age = 37, SD = 12.7) after excluding those who did not answer correctly to the attention or seriousness checks. Seven hundred and fifty six participants identified as male (70.3%), 295 as female (27.4%), and the remaining 25 as non-binary (2.3%). The mean political orientation was 3.56 (on a scale ranging from 1 'radical left' to 9 = 'radical right', SD = 1.70). Regarding professional situation, the sample consisted of 518 employees (48.1%), 171 students (15.9%), 137 self-employed (12.7%), 94 unemployed (8.7%), 42 retired (3.9%), and 29 people in situation of work incapacity (2.7%). One hundred and thirtyseven participants came out of the Belgian education system, while the remaining 939 went to school in France.

Procedure and Materials
In order to maximize the ideological diversity of the sample, the questionnaire was disseminated on Facebook through two means. It was first posted on several Frenchspeaking pages related to a variety of topics (' alternative information' pages, skeptics/debunkers pages, etc.). The link to the questionnaire was introduced by the following paragraph: 'Nowadays, we hear a lot about "conspiracy theories". But what is a "conspiracy theory"? This questionnaire allows you to freely express your views on this question and takes about 15 minutes (or more if you have a lot to say).' Nine hundred and forty-seven valid participants (i.e., participants who answered correctly to the attention and seriousness checks) were recruited through this means. However, given that the average endorsement of generic conspiracist beliefs in the sample was low (M GCBS = 2.16, SD = 0.82, on a scale ranging from 1 to 5) and that only a few participants scored high on the generic conspiracist beliefs scale, we created a sponsored ad to specifically target people who would potentially score higher on the conspiracist beliefs scale. This ad was introduced by the aforementioned paragraph, but it was preceded by this sentence: 'Have you ever been accused of defending a "conspiracy theory"?' We targeted an audience interested by the following keywords: Alternative medicine, Conspiracy theory, Homeopathy, Paranormal and Religion. The sponsored ad enabled us to recruit 129 valid partici-pants (75.2% identifying as males, 23.3% as females, and the remaining as non-binary) who scored significantly higher on the GCBS than participants recruited via Facebook pages, t(1075) = 13.1, p < 0.001, d = 1.23 (M GCBS = 3.14, SD = 0.94). Participants recruited via the sponsored ad were also on average older, t(1075) = 6.39, p < 0.001, d = 0.60 (M age = 43.59, SD = 13.21), and less left-wing, t(1075) = 4.09, p < 0.001, d = 0.38 (M political orientation = 4.13, SD = 1.61), than participants recruited via Facebook pages. Gender balance did not differ between the subsamples, χ 2 = 1.71, p = 0.19. Analyses were carried out on the separate subsamples to check if they returned different conclusions than the complete sample. This was the case for none of the analyses.
In an introductory paragraph, participants were told that their participation was anonymous, and that by carrying on they were giving their consent for the publication of their anonymized data. They were then told that they should answer the questionnaire with their current knowledge and personal opinions, and not use any help (e.g., internet, dictionary, or a friend).
Participants were first asked to propose an explanation of how the expression appeared in the first place ('How do you think the expression "conspiracy theory" appeared?'). Participants were then introduced to the academic debate opposing the 'generalists' to the 'particularists': There is a controversy among researchers about the term ' conspiracy theories'. Some believe that despite their diversity, it is acceptable to talk about ' conspiracy theories' in general. 'Conspiracy theories' share common characteristics that make them comparable. Thus, for these researchers, the use of the label ' conspiracy theories' is legitimate. Conversely, other researchers believe that talking about ' conspiracy theories' in general makes no sense. 'Conspiracy theories' are too diverse to draw any general conclusions. Every accusation of conspiracy should be judged on its own merits, without bias. For these researchers, the label ' conspiracy theories' is illegitimate.

Participants then answered an open-ended question:
'What do you think of this controversy?' After that, they were asked to position themselves ('With regard to this controversy, with your knowledge and opinions on the subject, where would you stand?') on a scale ranging from 1 ('the label "conspiracy theory" is legitimate') to 9 ('the label "conspiracy theory" is illegitimate'). Note that the questionnaire included two other open-ended questions ('what are the 4 words or expressions that come to your mind when we evoke the notion of "CONSPIRACY THEORY"?' and 'How would you define a "conspiracy theory"?'). However, given that the scope of this paper pertains to the acceptance and rejection of the label ' conspiracy theory', answers to these questions were not analyzed for the sake of brevity and readability. Indeed, a thorough analysis of this material would have been both off topic and very lengthy. The complete questionnaire is available on the OSF.
Then, just like in Study 1, participants reported their meta-conspiracy beliefs (α = 0.84) and completed the 12 items version of the Generic Conspiracist Beliefs Scale (Brotherton et al., 2013, α = 0.92). Since the factor structure of the GCBS was rehabilitated by recent research (Drinkwater et al., 2020), only the three items of the ' alien contacts' subscale were removed from the questionnaire, for the reasons developed in Study 1. An attention check was included among the items ('please tick the box "probably true" to show that you are reading the questions'). For further details, see materials and procedure section from Study 1.
Last, participants were asked to give their age, gender (male, female, or non-binary), political orientation (1 = far left, 9 = far right), current professional situation, and level of education (in France or Belgium). The questionnaire ended with the seriousness check ('Did you answer this questionnaire seriously and take the time to read the questions?', Yes/No).

Method for the Analysis of Open-ended Questions
To minimize researcher bias in the analysis of open-ended questions, we relied on statistical text analysis using IRaMuTeQ (Loubère & Ratinaud, 2014;Ratinaud, 2009), which is a software proposing analysis methods derived from the works of Reinert (1983Reinert ( , 1990. Specifically, we carried out a Descending Hierarchical Classification (DHC). Within a textual corpus (i.e., in our case, all of participants' answers to a specific open-ended question), a DHC identifies distinct classes of discourses characterized by a specific vocabulary (Reinert, 1990). For each word in a specific class, a chi-squared indicates its representativeness within the class, and therefore its potential importance in the interpretative part of the analysis. Based on the vocabulary used by participants, a DHC can also identify prototypical answers. Moreover, the DHC provides the prevalence of the different classes within the corpus (reported in percentage of the total corpus).
Crucial for the interpretative part of the analysis, a DHC also identifies variables that are significantly associated with each class. During the coding of corpuses, each answer was characterized by ideological and sociodemographic variables: 1) GCBS score (organized in deciles) 2) age, 3) gender, 4) political orientation, 5) professional situation, 6) country (Belgium or France), and 7) level of education. We expected that generic conspiracist beliefs, and particularly the extreme scores, would be associated with distinct vocabulary classes. We did not have specific hypotheses regarding the other variables; however, they were included as we know that one's representation of a social object is influenced by one's social identities, values, and beliefs (Jodelet, 2002;Marková, 2007). Besides, much research has shown that political orientation is related to conspiracist beliefs, whether linearly (e.g., Dieguez, Wagner-Egger, & Gauvrit, 2015;Featherstone, Bell, & Ruiz, 2019;Galliford & Furnham, 2017;Miller, Saunders, & Farhart, 2016) or quadratically (van Prooijen, Krouwel, & Pollet, 2015;Krouwel et al., 2017). Thus, we sought to examine whether some of these variables would be significantly associated with some vocabulary classes.
Two bodies of text were analyzed: One compiling participants' hypotheses regarding the origin of the label ' conspiracy theory' and the other compiling their opinions on the generalist versus particularist controversy. Basic spelling and grammar corrections, as well as lemmatization (Loubère & Ratinaud, 2014), were made in order to facilitate the analyses.
Note that the logic of a DHC is purely statistical. As a result, the classification of the corpus is strictly formal, and it is up to the researchers to attribute meaning (i.e., give names) to the classes and how they relate to each other. In other words, the qualitative part of the analysis is carried out on a quantitatively constituted material. The interpretation is based on the examination of the vocabulary contained in each class, the representativeness of the terms, the variables associated with the class and the content of prototypical answers. For this qualitative part of the analysis, we sought to further minimize researchers' bias by triangulating the process (Denzin, 1970). In other words, the three authors separately named the vocabulary classes returned by the DHC, then shared and discussed their interpretations until a consensus was reached.

Confirmatory Analyses
Correlations and descriptives are displayed in Table 3. Our first hypothesis was that the belief in a meta-conspiracy theory would be correlated with generic conspiracist beliefs, and that the items would fit in a single factor structure. Just like in Study 1, the two constructs were strongly correlated, r = 0.78, p < 0.001 (r = 0.78, p < 0.001 and r = 0.73, p < 0.001 for, respectively, the 'sponsored ad' and the 'Facebook pages' subsamples). Confirmatory factor analysis confirmed that the 16 items had an acceptable fit with a single factor structure, Chisq/df = 7.4, TLI = 0.92, SRMR = 0.04, CFI = 0.94, RMSEA = 0.08. This corroborates H 1 . However, a two factors structure showed a slightly better fit, Chisq/df = 5.37, TLI = 0.95, SRMR = 0.03, CFI = 0.96, RMSEA = 0.06. The difference between the fits of the single and two factors structures was significant, χ 2 difference = 186.29, p < 0.001.
Our second hypothesis was that generic conspiracist beliefs would be positively associated with a particularist view of CTs. As can be seen in Table 3, there is moderate correlation between generic conspiracist beliefs and the endorsement of a particularist view of CTs, r = 0.30, p < 0.001 (r = 0.25, p = 0.005 and r = 0.26, p < 0.001, for, respectively, the ' conspiracist sponsored ad' and the 'Facebook pages' subsamples). We carried out hierarchical regression with self-positioning on the generalist versus particularist view of CTs as dependent variable to test this hypothesis with controlled variables included in the model. At Step 1, we controlled for age and gender. At Step 2, generic conspiracist beliefs were included as independent variable.

Exploratory Analyses
For individuals' position on the 'generalist' versus 'particularist' scale, we examined the frequency table to examine the repartition of participants' position (see Table 4). As can be seen, only a minority of participants ticked an extreme value, with 16.9% ticking '1' ('The label "conspiracy theory" is legitimate', N = 182), and 6.2% ticking '9' ('The label "Conspiracy Theory" is illegitimate', N = 67). The majority of participants ticked a value below the median point (54.9%, N = 591), 22.1% ticked the median point ('undecided', N = 238), and the remaining 23% ticked a value above the median point (N = 247).

Lexicometric Analyses
Given the amount of material returned by the DHC, this section focuses only on the vocabulary classes that were relevant to our research question. Indeed, a complete restitution of the lexicometric analyses would have been both obfuscating for the reader and irrelevant given the scope of the paper. The complete material can be found on the OSF.
The DHC analysis extracted distinct vocabulary classes mobilized by participants to answer open-ended questions. For both corpuses of text, a dendrogram (Figures 1 and 2) summarizes the classification and indicates the representativeness for each of the classes within the corpuses. Note that in the following section we translated into English participants' answers. The complete answers (in French) are available on the OSF page.
How did the Label 'Conspiracy Theory' Appear?
While correcting the responses to facilitate the processing of the corpus, we realized that many participants were not making the distinction between the origin of CTs and the origin of the label ' conspiracy theory'. We took this into account and proceeded even more cautiously with the interpretative part of the analysis. The anchors were the following: 'The label "conspiracy theory" is legitimate' (1); 'undecided' (5); 'The label "conspiracy theory" is illegitimate' (9). The DHC analysis extracted six distinct vocabulary classes mobilized by participants to answer this question. For this analysis, we set the 'number of Phase 1 terminal classes' out of 15 in the classification parameters in order to obtain a more comprehensive number of classes and a more exhaustive analysis of the corpus segments. It included 85.69% of the corpus' text segments. Classes of vocabulary 1 and 4 problematized the question of the origin of CTs/the label ' conspiracy theory' in societal (Class 1) and politicized terms (Class 4). These classes are the only ones associated with higher levels of conspiracy beliefs (see Table 5), and both contain references to suspicions regarding information concealed by powerholders and governments (Class 4 being more politicized than Class 1). In both classes, some prototypical answers mention the weaponization of the label by powerholders: [The expression ' conspiracy theory'] is mainly used to define as an individual and psychological problem a social phenomenon that results from the political dispossession of a part of the population. (Participant 642, Class 4) Some criticisms of such weaponization clearly had conspiratorial overtones: To discredit this kind of 'revelation' […]. We call them conspiracy theories because it makes you smile, we don't take them seriously, it's a distraction […]. In short, we have been fooled again. (Participant 1068, Class 1) We might note that even though it was not detected by the lexicometric analyses, a participant (Number 968) copy-pasted a link toward an article explaining that the label ' conspiracy theory' was invented by the CIA. Here is an excerpt of the article that perfectly illustrates the metaconspiracy theory as defined in the introduction: Pay close attention to those using the labels and the information they are referring to. This will be your first clue that truth lies within and they are trying to deflect it.
[…] While they are busy misdirecting with their ' conspiracy theorist' labels, evidence is being dug up […] and what they claim to be theory is TRUTH. (Corey's Digs, 2018).
By contrast, others acknowledged such weaponization and at the same time distanced themselves from CTs: Because of all the lies spread by the media […], a part of the population who had lost their trust [in the media] started to believe everything they found on social media, without any reflection or fact-checking. Then [the label was] used by our leaders to silence actual callings into question. (our emphasis, Participant 936, Class 4) We might note that the only class associated with low levels of generic conspiracist beliefs is the one characterized by the expression 'no idea' (Class 6). It was also the most frequent statement of the body of text (N = 161).
Hence, congruent with our expectations, generic conspiracist beliefs were associated with the use of a specific vocabulary, which problematized the question of the origin and use of the label in societal and politicized terms. Moreover, many participants stressed the issue of the deliberate weaponization of the label by powerholders, which is a key component of the meta-conspiracy theory as we defined it. The belief in such weaponization took different forms. Some participants acknowledged the weaponization of the label while distancing themselves from CTs. These participants do not express any assumption regarding the precise origin of the label. By contrast, other participants expressed the belief that the label was purposely created as a rhetorical weapon (e.g., by the CIA) to discredit dissenting views and people about to unveil disturbing truths (e.g., who really killed JFK).
Besides generic conspiracist beliefs, the only other variable significantly related to one of the vocabulary classes was political orientation, so that the center value of the political orientation scale (5) was significantly associated  Table 5: Variables, vocabulary significantly associated with vocabulary classes for the "origin of the label" corpus. with Class 1. More polarized political orientation scores ('1' and '7', respectively radical left and moderate right) were associated with Class 5, which evokes the dissemination of CTs in the media and their temporal dimension. This class was not developed for it falls out of the scope of the paper.

Debate
For the DHC analysis 96.13% of the corpus segments were analyzed by the software. For this second analysis, we ran a 'simple classification on text segments' and the 'number of phase 1 terminal classes' was set on 14 in the classification parameters. Class 1 predominantly refers to terms that are in line with the generalist view of the CTs (see Table 6). This class is significantly associated with the lowest generic conspiracist beliefs scores in the sample and contains vocabulary emphasizing the common features of CTs, such as their structure, underlying mechanisms, or psychological underpinnings. For this class, no prototypical answer was identified by the software. However, many answers among participants who scored the lowest on the GCBS illustrate the idea that CTs share common features: Of course, every conspiracy theory has some specificities […]. However, it seems to me that for [people] who systematically question the ' official version' after every negative event, the mechanism is always the same. (Participant 20) People who endorse these theories think that they are being lied to for malevolent purposes […], cognitive biases may also be common to these theories, their content may differ but I definitely think that they all have something in common. ( Participant 27) By contrast, Classes 2 and 3 referred to criticisms of the generalist view of CTs. Class 2 appears to be articulated around the idea that real conspiracies happen, and Class 3 contains terms that refer to the vagueness of the label as well as its derogatory connotation. Hence, when asked about the legitimacy of the label ' conspiracy theory', many participants mention the importance of making a distinction between real, proven conspiracies on the one hand, and ' conspiracy theories' on the other. Interestingly, Class 2 was not associated with a specific level of generic conspiracist beliefs. Here are excerpts of prototypical answers illustrating such views: […] there are also real conspiracies […] but they are not always as sexy as a Tom Clancy novel. (Participant 816) One can be interested in conspiracies without being a conspiracy theorist […]. (Participant 145) Class 3 seems to refer to the answers criticizing the semantic vagueness of the label as well as its negative connotation. This class was significantly associated with moderate to high generic conspiracist beliefs (see Table 6). Additionally, participants stress the existence of a distinction between the pejorative use of the label in everyday language and its use by scientists -as illustrated by the following examples: […] ' conspiracy theories' is a term […] that contains a variety of phenomena […]. [It is] an issue that can be theorized, but that has a variety of forms and expressions, including in its societal dimension and in its scientific analysis […]. (Participant 775) […] I'm rather in favor of using the term in a generic way, even if it means making more precise categories […] the label took on a negative connotation in everyday language […]. (Participant 666) Note that one of the prototypical answers from Class 2 also clearly stresses the issues of the vagueness and derogatory nature of the label: I think there are too many meanings to this label… And that this label has become extremely derogatory […] any ' conspiracy theory' will first be categorized as bullshit […]. (Participant 536) Again, the hypothesis that generic conspiracist beliefs influence individuals' representations of the label ' conspiracy theory' is corroborated, for different levels of conspiracy beliefs were associated with distinct vocabulary classes. Moreover, these vocabulary classes were associated with distinct arguments reflecting different representations of the label. Participants accepting the label ' conspiracy theory' tended to mobilize vocabulary emphasizing common features (Class 1). By contrast, three broad criticisms of the generalist view of CTs emerged: historical (Class 2), conceptual and normative (Class 3). We might note that for many participants, the generalist and the particularist views are not mutually exclusive, but rather operate at different levels. This is congruent with the fact that relatively few participants ticked an extreme value on the scale measuring their position regarding the 'particularist' versus 'generalist' controversy.

General Discussion
In this research, we postulated the existence of polemical social representations of the label ' conspiracy theory', anchored in individuals' pre-existing beliefs and, specifically, their generic conspiracist beliefs. To operationalize this contention, we hypothesized a positive relation between generic conspiracist beliefs and two types of rejection of the label, namely, the belief in what we named the meta-conspiracy theory, and the endorsement of a particularist -rather than generalist -view of CTs. With open-ended questions, we moreover sought to investigate participants' hypotheses regarding the origin of the label, as well as the arguments spontaneously produced to justify their position regarding the generalist versus particularist controversy. Prototypical answers None N_536 (χ² = 13) N_816 (χ² = 9) N_145 (χ² = 9) N_245 (χ² = 9) N_361 (χ² = 9) N_747 (χ² = 9) N_294 (χ² = 9) N_638 (χ² = 9) N_171 (χ² = 12) N_775 (χ² = 12) N_849 (χ² = 10) N_666 (χ² = 9) N_729 (χ² = 9) N_266 (χ² = 9) N_922 (χ² = 9) N_737 (χ² = 9) N_468 (χ² = 8) In both studies, belief in meta-conspiracy theory and generic conspiracist beliefs were related to the point that the constructs were empirically almost indistinguishable. This resonates with the monological hypothesis (Goertzel, 1994), that is, the idea that CTs serve as mutual evidence in a belief system closed to contradictory information. Indeed, when endorsing the meta-CT, criticisms of CTs are interpreted as further evidence in favor of a conspiracy. However, analyses of open-ended questions showed that participants mentioning the deliberate weaponization of the label by powerholders did not necessarily express a positive view of CTs, nor the hypothesis that the label was specifically created for this purpose. Hence, when moderately endorsed, meta-CTs may simply refer to the (reasonable) belief that the label ' conspiracy theory' is often purposely used as a rhetorical weapon by powerholders, with no reference to the term having specifically been created for this reason. Such concerns regarding the weaponization of the label ' conspiracy theory' by powerholders are common in the academic literature (e.g., Aistrope, 2016;Basham & Dentith, 2016;Husting & Orr, 2007). By contrast, in their more radical forms, meta-CTs imply the creation of the term by a malevolent elite (e.g., the CIA) to discredit people who were about to unveil the truth hidden behind the ' official version' (e.g., about the Kennedy assassination). Hence, we could coin a distinction between 'soft' and 'hard' meta-conspiracy theories. We might note that only the latter appear as a belief protecting, and even reinforcing one's ideology: When endorsing such a belief, criticisms of CTs might be perceived as evidence of a conspiracy. While the 'monological assumption' was criticized (Franks et al., 2017;Hagen, 2017;Sutton & Douglas, 2014), this example illustrates that some conspiracy beliefs (rather than conspiracy beliefs in general) can reinforce others. As a result, future research might be particularly vigilant about the risks of a backfire effect in reaction to criticisms of CTs.
The second form of rejection of the label ' conspiracy theory' was the endorsement of a particularist view of CTs. According to this view, beliefs labeled as ' conspiracy theories' are too diverse to draw any generic conclusions. The label is therefore, to a large extent, illegitimate. In the second study, we found a robust, yet moderate, correlation between generic conspiracist beliefs and a particularist view of CTs. In other words, the more participants believed in the existence of conspiracies, the more they tended to believe that the label ' conspiracy theory' was illegitimate. We moreover found that relatively few participants fully endorsed the particularist or generalist view (i.e., ticked an extreme value on the 'generalist vs. particularist' scale). In open-ended questions, many participants leaned toward a position (e.g., generalist), while acknowledging the limitations of the advocated position (e.g., the risk of conflating 'real conspiracies' with ' conspiracy theories'). This suggests that for many participants, the label was (il)legitimate to some extent, and that the generalist and particularist views are not necessarily antagonistic, but rather work at different levels. This echoes the idea that CTs might share common underpinnings but that subcategories can be defined (e.g. Brotherton et al., 2013;Campion-Vincent, 2005;Wagner-Egger & Bangerter, 2007).
Lexicometric analyses enabled to identify arguments participants spontaneously produced when asked what they thought about the generalist versus particularist controversy. These different arguments were found to be related to different levels of generic conspiracist beliefs. Besides, both the arguments in favor and against the label echo existing literature. Arguments in favor of the generalist view of CTs were particularly prevalent among participants who reported low scores of generic conspiracist beliefs. These arguments emphasized a variety of common features of CTs, such as their systematic opposition to ' official versions', their underlying worldview (e.g., 'They' secretly plot to attain malevolent goals), or the fact that they can be explained by common psychological mechanisms (e.g., cognitive biases). As we have outlined in the introduction, the idea that CTs share social psychological features is one of the key assumptions of the social psychological research on the topic.
By contrast, the relation between generic conspiracist beliefs and arguments questioning the generalist view is more mixed. These arguments can be summarized in three main points: 1) real conspiracies happen, 2) the label has no precise meaning, and 3) the label has a negative connotation. Emphasis on the distinction between 'real' conspiracies and ' conspiracy theories' was not specific to a level of conspiracy beliefs (note that such emphasis reveals that for many participants, the label ' conspiracy theory' implies falsity: CTs are opposed to 'real' conspiracies). The issue of the prevalence of real conspiracies in history has been stressed by a number of authors willing to put in perspective the alleged irrationality of conspiracy suspicions (e.g., Aistrope, 2016;Bale, 2007;Basham & Dentith, 2016). For example, Bale (2007) has argued that among historically documented conspiracies, some of them would appear very implausible if they had not been officially acknowledged.
On the other hand, criticisms of the vagueness of the label (e.g., calling it a ' catch-all' term) and its derogatory connotation were significantly associated with moderate to relatively high levels of generic conspiracist beliefs. How the label tends to homogenize a multiplicity of beliefs and populations was emphasized in the ethnographic work by Harambam and Aupers (2016). Besides, Byford (2011) has deplored that the meaning of the label has gone from being relatively specific to being inclusive of any suspicions toward powerholders. As for the pejorative connotation of the label in its everyday use, it has been shown that one labels as ' conspiracy theories' ideas with which one strongly disagrees (Wood & Douglas, 2013). Even though the label itself does not seem to affect the credibility attributed to a piece of information (Wood, 2015), CTs are typically stigmatized beliefs (Lantian et al., 2018).

Limitations and Future Directions
A first limitation is that this research examined how participants accept or reject the label ' conspiracy theory' and the argument mobilized to justify their position, and did not dive into a more in-depth analysis of participants' representations. In other words, we have opened a line of investigation pertaining to social representations of the label ' conspiracy theory', but we did not explore the representational field associated with it. For example, we did not examine in detail the content of the hypothesized antagonistic representations, nor the organizing principles (Doise, Clemence, & Lorenzi-Cioldi, 1993) around which these representations are articulated. Neither did we examine the extent to which one's group memberships, values, or ideology influenced these representations. In Study 2, while some vocabulary classes were associated with specific political orientation scores, we lacked data to propose interpretations that would not have been excessively speculative. However, conspiracist beliefs are known to be correlated with contextual (e.g., crisis situations, van Prooijen & Douglas, 2017), ideological (e.g., holding an extreme political orientation, Krouwel et al., 2017;van Prooijen et al, 2015), and sociodemographic factors (e.g., being member of a disempowered group, Crocker et al., 1999;Uscinski & Parent, 2014). Hence, a more in-depth, empirically grounded examination of how these variables influence representations of the label ' conspiracy theory' appears like a promising lead for future research.
Second, while our questionnaire met a surprising success on Facebook, such convenience sampling has drawbacks, the main being biased sampling resulting in a lack of representativity. Despite our efforts to recruit an ideologically diverse sample, some groups were particularly interested in our research and the French skeptic movement ('Mouvement Zététique') is clearly over-represented in the sample. This over-representation may explain the prevalence of rationalist and scientific vocabulary in participants' responses. As a result, we cannot draw conclusions regarding the overall acceptance/rejection of the label in the general population based on our data. Hence, furthering the investigation in a more diverse, representative sample is an interesting lead for future research. Relatedly, in Study 2, which was introduced as a study on the label ' conspiracy theory', we recruited a large majority of male participants. While we do not have a specific hypothesis to explain this discrepancy, this gendered aspect might also be worthy of investigation.
Last, as we have already mentioned, many of the participants misunderstood the question about the potential origin of the label ' conspiracy theory'. Hence, the initial goal of this question -which was to investigate participants' hypotheses regarding how and why the label appeared in the first place -may have been obscured by this misunderstanding. Even though the analyses yielded results relevant to the scope of our research, replicating this research while making sure that the question is correctly understood by participants might lead to more robust conclusions.

Conclusion
To conclude, our results corroborate the idea that individuals' perception of the label 'conspiracy theory' varies depending on individuals' generic conspiracist beliefs.
Hence, the hypothesis that the label is associated with diverging social representations is corroborated. Specifically, we found evidence that generic conspiracist beliefs were substantially related to two distinct types of rejection of the label: the belief in meta-conspiracy theory and the endorsement of a particularist view of CTs. Moreover, we found that when answering questions about the label 'conspiracy theory', participants scoring higher on the generic conspiracist beliefs scales mobilized a distinct vocabulary, as well as distinct arguments, compared to participants scoring lower on the scale. Such diversity emphasizes the relevance of the social representations framework in the study of CTs.
On a more practical note, we found that even though rejection was stronger among participants who endorsed generic conspiracist beliefs, criticisms of the label were also found among participants with low scores of generic conspiracist beliefs. This may be an important aspect that should be acknowledged by social scientists endorsing a generalist view of CTs in their work. Specifically, when communicating their findings with the general public, such intellectual stance is likely to raise a variety of objections, some of which this contribution has highlighted (i.e., the historical, conceptual, and normative arguments). While it is -as researchers -impossible to defuse accusations of being CIA agents seeking to discredit dissenting discourses, acknowledging the limitations and strengths of both the generalist and particularist views of CTs might be a necessary step to increase the persuasiveness (and therefore the impact) of our arguments in the general public.

Ethics and Consent
Participants were informed that by participating in the study, they implicitly gave their consent for the publication of their anonymized data on an open access platform. This research and its authors are in line with the Declaration of Helsinki and the APA code of conduct. Given that the study was purely correlational, that participation was on a voluntary basis and that it did not involve any form of deception, this research was not subject to an evaluation by the ethics committee of the University.

Competing Interests
The authors have no competing interests to declare.

Authors Contribution
Kenzo Nera and Sarah Leveaux designed the research and questionnaire. Sarah Leveaux disseminated the questionnaire on Facebook and carried out the lexicometric analyses. The first draft of the manuscript was written by Kenzo Nera, with the help of [2 nd author] for the introduction and method sections. Pit Klein contributed to the qualitative analyses, revised and commented on the different versions of the manuscript.